Rights / License: Research Collection In Copyright - Non ...27500/eth... · In NMR methodological aspects, the 3D TROSY-HN(CA)HA and 4D TROSY-HACANH experiments were proposed forbackbone

Research Collection

Doctoral Thesis

Preparation and characterization of membrane proteins andchallenging oligomeric proteins for solution NMR spectroscopy

Author(s): Hu, Kaifeng

Publication Date: 2004

Permanent Link: https://doi.org/10.3929/ethz-a-004844309

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For moreinformation please consult the Terms of use.

ETH Library

https://doi.org/10.3929/ethz-a-004844309

http://rightsstatements.org/page/InC-NC/1.0/

https://www.research-collection.ethz.ch

https://www.research-collection.ethz.ch/terms-of-use

Diss.ETH Nr. 15726

Preparation and characterization of membrane

proteins and challenging oligomeric proteins for

solution NMR spectroscopy

A dissertation submitted to the

Swiss Federal Institute of Technology Zürich

for the degree of

Doctor of Sciences

Presented by

Kaifeng Hu

Master of Science

Shanghai Institute of Materia Medica, Chinese Academy of Sciences

Born on Feb. 19, 1974

Citizen of China

accepted on the recommendation of

Prof. Dr. Konstantin Pervushin, examiner

PD. Dr. Oliver Zerbe, co-examiner

2004

Acknowledgement

During these years of my study and work at the ETH, I have learnt from, cooperated and

discussed with a lot of people. Many of them helped me not only on the scientific side, but

supported me mentally with their friendship and with their generous help in everyday life. I am

very grateful to all of them.

Especially, I would like to thank Prof. Konstantin Pervushin for giving me the chance to study

NMR and to carry out my research project in his group. I appreciate his guidance and support in

my project, his explanation and his penetrating understanding of NMR theory. I thank Prof.

Andreas Plükthun for offering me the opportunity to purify FkpA in his group and his group

members for their help when I worked there.

I want to thank all Prof. Pervushin's group members for providing a nice and friendly working

environment. I should give my special thanks to Beat Vogeli, for his help and patient explanation,

which was important for me to catch up and understand physics of NMR, and for his stimulating

discussions through all these years. I should give my special thanks also to Dr. Maria lohansson

for our very pleasant cooperation on the membrane protein project and her friendly support and

help in everyday life. Simon Alioth also deserves my thanks for his explanation and practical

discussion on molecular biology. I should also thank Dr. Osvaldo Moreira, Dr. Donghan Lee,

Alexander Eletski, Veniamin Galius, Okhrimenko Oksana and Reto Walser for their help all

these years. I thank Dr. Fred Damberger for carefully reading the manuscript of my thesis.

1 owe my previous supervisor, Prof. Dr. Ronald Guiles a debt of gratitude for giving me the

opportunity to begin my Ph.D study in NMR at the Univ of Maryland, Baltimore. I also thank his

group members, Dr. Hanqiao Feng, Dr. Bindi Dangi and Dr. Nazim Shahzad for their help when

1 studied there.

Finally, I would like to thank my family for their continuous support, encouragement and

understanding all these years. I owe them a great debt for their care and love!

Table of contents

Summary 5

Zusammenfassung 7

Abbreviations 9

Chapter T Preparation and characterization of a membrane protein ß-barrel platform for

solution NMR studies 11

1.1 Introduction 13

1.1.1 Structural studies of membrane proteins 13

1.1.1.1 Membrane proteins vs. water soluble proteins 13

1.1.1.2 Micelles, bicelles and lipid bilayers 14

1.1.1.3 NMR vs. X-ray crystallography in MP structure studies 15

1.1.1.4 a- helical and ß - barrel MP studied by solution NMR 16

1.1.2 Experimental aspects: solution NMR studies of MP 20

1.1.2.1 MP sample preparation for solution NMR studies 20

1.1.2.1.1 Expression and purification 20

1.1.2.1.2 Refolding 21

1.1.2.1.3 Isotope-labeling 21

1.1.2.2 NMR spectroscopy and structure determination of MP 22

1.2 Preparation and characterization of a membrane protein ß-barrel platform of OmpA ..24

1.2.1 Introduction to the outer membrane protein A (OmpA) 24

1.2.2 Design of the ß-barrel platform (BBP) of OmpA 25

1.2.3 NMR sample preparation of BBP 26

1.2.3.1 Sequencing the reconstructed gene26

1.2.3.2 Over-expression, labeling and purification of BBP 26

1.2.3.3 Refolding of BBP into micelles 27

1.2.4 NMR spectroscopy of BBP 29

1.2.5 Conclusions and perspectives 32

References 33

Chapter II NMR studies of structure and function of FkpA 39

II. 1 Introduction 41

2

II.2 Sample preparation 43

IL3 Backbone chemical shift assignment and secondary structure determination 44

IL3.1 NMR spectroscopy and backbone chemical shift assignments 44

II.3.2 Analysis of secondary structure 49

11.4 15N Relaxation 49

11.4.1 NMR experiments 50

11.4.2 Data processing and analysis 53

11.5 Residual dipolar coupling and the dynamic molecular model 56

11.5.1 Residual dipolar coupling measurement 56

11.5.1.1 Measurements of RDCs by addition of Pfl filamentous bacteriophages 57

11.5.1.2 Measurements of RDCs in La phase 61

11.5.2 Fitting of the alignment tensor to the residual dipolar couplings 64

11.5.3 Analysis of fitting results and dynamic molecular model 67

II. 6 Chapcrone function of FkpA 71

11.6.1 Substrate protein binding 71

11.6.2 Specific polypeptide binding site of FKpA and comparison of its chaperone

function and PPIase activity 73

11.6.3 FkpA chaperone function 78

II.7 Mechanism of chaperone function of dimeric FkpA: model of "mother's arms" 81

References 82

Chapter III Backbone resonance assignment in large protonated proteins using a

combination of new 3D TROSY-HN(CA)HA, 4D TROSY-HACANH and 13C detected

HACACO experiments 85

Introduction: 13C-detection based NMR spectroscopy of proteins 87

Development of 3D TROSY-HN(CA)HA and 4D TROSY-HACANH experiments 88

Combined use of the 3D TROSY-HNCA, 3D TROSY-HNCO with 3D MQ-HACACO 97

References 99

Chapter IV Side-chain H and C resonance assignment in partially deuterated proteins

using a new 3D ,3C-detected HCC-TOCSY 101

Introduction: strategies of side chain assignment 103

Development of 3D in-phase and sensitivity enhanced HCC-TOCSY 103

3

Side-chain assignment 108

References 112

Appendix 114

A.l Table 1 BBP of OmpA: DNA sequence and amino acid sequence 114

A.2 Table 2 Backbone HN, N, Ca, C and Cp chemical shift assignments of FkpA 115

A.3 Linear least-squares fitting of RDCs 122

Back calculation of RDCs 123

Linear least-squares fitting of RDCs for homodimeric molecules 124

References 125

Curriculum vitae 127

4

Seite Leer /

Blank leaf

5

Summary

The aim of my thesis is to develop methodological aspects of NMR structure/dynamics

investigations of proteins considered to be difficult from the vista of conventional NMR. This

includes membrane proteins reconstituted into micelles of detergents or highly dynamical and

extended (non-compact) soluble proteins with low density or absence of long range structural

constraints based on NOEs. The methodological aspects discussed in the thesis include choice of

the isotope-labeling pattern, sample preparation and conditions for NMR studies, adopting both

previously developed strategies as well as developing new ones for assignment of backbone

resonances, and analysis of dynamical properties and function of the selected target proteins. The

presented results should serve as a basis for more detailed structural studies of these biologically

important systems.

Membrane proteins take part in a large number of important physiological functions. As an

example of a membrane protein suitable for NMR studies, we selected outer membrane protein A

(OmpA), which is an abundant structural protein of the outer membrane of Gram-negative

bacteria. The N-tcrminal domain of the OmpA protein from Escherichia coli, consisting of

residues 1-172, forms an antiparallel ß-barrel whose eight transmembrane ß-strands are

connected by three short periplasmic turns and four relatively large surface-exposed hydrophilic

loops. In NMR studies of OmpA, we were concerned with a structural role of the surface-

exposed loops of OmpA. An OmpA deletion variant with all four loops shortened, which we

called ß-barrel platform (BBP) of OmpA, consists of only 142 amino acid residues and

constitutes the smallest ß-struetured integral membrane protein known to date. Fractional

factorial refolding screens were used to identify refolding conditions for BBP. The shortened

protein was successfully refolded and properly labeled for NMR structure determination. An

analysis of NMR spectra shows that this artificially designed outer membrane protein refolds into

a ß-barrel structure, which is similar to the transmembrane domain of the wild type of OmpA.

This implies that the absence of the extracellular loops does not affect the structure of the

transmembrane domain. BBP therefore can serve as a good starting point for the development of

integral membrane proteins with novel engineered functions.

FkpA is a heat shock periplasmic peptidyl-prolyl cis/trans isomerase (PPIase) with chaperone

activity. The chaperone activity of FkpA is independent of its PPIase activity. The mature

dimeric FkpA protein has 245 amino acid residues per monomer. Both chain termini are not well

6

structured and were therefore removed in the reconstructed gene and the expressed protein

prepared for structure studies is denoted "shortened FkpA" (sFkpA). Both FkpA and sFkpA with

C-terminal His6-tag were overexpressed with suitable 2H/15N/l3C- labels and purified for solution

NMR studies. Backbone resonances of 94% of the residues of FKpA were assigned and

secondary structure elements were established using the CSI analysis. Primary dynamical

properties of FkpA were investigated by NMR relaxation experiments, suggesting that the

dimeric FkpA protein could be divided into three relatively rigid subunits moving relative to each

other. Residual dipolar couplings (RDC) were used to compare the global NMR structure of

FkpA in solution with the corresponding crystal structure. Experimental RDC showed significant

mobility of two FKBP domains relative to the dimerization domain. These dynamic properties of

the protein might play an important role in the chaperone activity of FkpA, where binding to

different substrates potentially requires some structural adaptations of the chaperone. Protein

substrates were then used to identify where the chaperone function resides in the dimeric

molecule. Results indicated that five residues, which are distributed between the long a helices

and the FKBP domains, could play an important role in polypeptide binding. A model, the so-

called "mother's arms" model, is then proposed to illustrate the mechanism of the chaperone

function of FkpA. In this model, FkpA "catches" the polypeptide or small protein ("baby protein")

substrates with its polypeptide binding sites (like "hands"), and then holds the substrates through

its structural adaptations, i.e. bending of its two long helical "arms", which is like a mother's

arms hugging her baby.

In NMR methodological aspects, the 3D TROSY-HN(CA)HA and 4D TROSY-HACANH

experiments were proposed for backbone resonance assignment and a new C-detected 3D HCC-

TOCSY is described which serves as an attractive experiment for simultaneous and unambiguous

assignment of the side-chain H and C resonances in partially deuterated proteins of large size. In

3D TROSY-HN(CA)HA and 4D TROSY-HACANH, the combined application of 1H-13C

multiple-quantum and the 1H-15N TROSY effect is proposed to enhance sensitivity. The C-

detected 3D HCC-TOCSY expected to be suitable for assignment of the side-chain methyl 13C

and 'H chemical shifts, containing important information for structure determination, of methyl

protonated, highly deuterated and 13C-labeled proteins with high molecular weight.

7

Zusammenfassung

Das Ziel meiner Doktorarbeit ist es, neue Methoden für die Untersuchung mittels NMR-

Spektroskopie von Struktur und Dynamik von Proteinen zu entwickeln, welche vom Standpunkt

der konventionellen NMR-Spektroskopie als schwierig angesehen werden. Dies schliesst in

Mizellen rekonstituierte Membranproteine oder auch höchst dynamische und lösliche Proteine

mit grosser Ausdehnung und geringer Dichte oder fehlenden weitreichenden strukturellen

Informationen durch NOEs mit ein. Die in dieser Arbeit behandelten Aspekte sind die Wahl einer

geeigneten Isotopenmarkierung, Herstellung einer Probe, Anpassung von schon bestehenden und

Entwicklung neuer Strategien für die Resonanzzuordnung und die Analyse von dynamischen und

funktionellen Eigenschaften des zu untersuchenden Proteins. Die präsentierten Resultate sollten

als Grundlage für detailiertere Untersuchungen über die Struktur und Dynamik von diesen

biologisch wichtigen Systemen dienen.

Membranproteine erfüllen eine grosse Anzahl physiologisch wichtiger Aufgaben. Ein Beispiel

eines Membranproteins, welches geeignet ist für die Untersuchung mittels NMR-Spektroskopie,

ist OmpA, welches in der äusseren Membran von gram-negativen Bakterien in grosser Anzahl zu

finden ist. Die N-terminale Domäne von OmpA von Escherichia coli, welche die Aminosäuren

1-172 umfasst, bildet ein antiparalleles ß-Barrel, dessen acht transmembranäre ß-Stränge durch

drei kurze periplasmatische und vier relativ lange oberflächen-exponierte hydrophile Loops

verbunden sind. Wir wollten den Einfluss der oberflächen-exponierten Loops auf die gesamte

Struktur des Proteins mittels NMR-Spektroskopie untersuchen. Eine OmpA Deletions-Variante,

in welcher alle vier Loops auf eine minimal nötige Länge gekürzt sind und von uns ß-Barrel

Plattform (BBP) benannt wurde, besteht aus nur 142 Aminosäuren und ist das kleinste bekannte

ß-strukturierte integrale Membranprotein. Um die idealen Rückfaltungsbedingungen für die BBP

zu erreichen, wurde eine grosse Anzahl verschiedener Bedingungen in einem „fractional factorial

screen" getestet. Zu diesem Zeitpunkt kann BBP erfolgreich rückgefaltet und isotopenmarkiert

werden und ist damit für NMR-spektroskopische Untersuchungen zugänglich. Ein Vergleich der

NMR-Spektren von BBP und dem Wildtyp OmpA, zeigt dass beide Proteine ähnlich gefaltet sind.

Dies führt zu der Annahme, dass die Abwesenheit der cxtrazellulären Loops keinen Einfluss auf

die Struktur der Transmembrandomäne hat.

FkpA ist eine periplasmatische Hitzeschock petidyl-prolyl cis/trans Isomerase (PPIase) mit

Chaperone-Aktivität, welche unabhängig von der PPIase Aktivität ist. FkpA liegt als ein

8

Homodimer bestehend aus zweimal 245 Aminosäuren vor. Da die beiden Termini keine feste

Struktur aufweisen, wurden sie entfernt und das daraus entstandene Protein wurde sFkpA (für

„shortened FkpA") genannt. Beide Varianten wurden mit einer C-terminalen Hisft-Extenstion und

geeigneter Isotopenmarkierung überexprimiert und für NMR-spektroskopische Untersuchungen

aufbereitet. Rückgratresonanzen von 94% der Aminosäuren von FkpA konnten zugeordnet

werden und die Sekundärstruktur wurde mittels CSI Analyse vorausgesagt. NMR-

spektroskopische Relaxationsexperimente weisen darauf hin, dass das FkpA Dimer aufgrund

dynamischer Eigenschaften in drei Domänen unterteilt werden kann. Um die Struktur von FkpA

in Lösung mit der Kristallstruktur zu vergleichen wurden dipolare Kopplungen verwendet,

welche signifikante Mobilität von zwei FKBP Domänen relativ zur Dimerisierungsdomäne

aufzeigen. Diese dynamischen Eigenschaften des Proteins könnten bei der Chaperone-Aktivität

von FkpA, wo die Bindung an verschiedene Substrate strukturelle Adaptation nötig macht, eine

wichtige Rolle spielen. Die Chaperone-Aktivität innerhalb des Dimers wurde mittels Titration

mit verschiedenen Substraten in den langen a-Helices und FKBP Domänen lokalisiert. Das

sogenannte „mother's arms" Modell, in welchem FkpA ein Polypeptid oder ein kleines Protein

(„Baby-Protein") mit den Bindungsstellen (den „Händen") hält und dieses durch strukturelle

Veränderungen in den zwei langen helikalen „Armen" umschliesst, wurde aufgrund dieser

Resultate vorgeschlagen.

In Bezug auf NMR-spektroskopische Methoden, wurden 3D TROSY-HN(CA)HA und 4D

1 ^

TROSY-HACANH Experimente zur Zuordnung von Rückgratresonanzen und ein neues C-

detektiertes 3D HCC-TOCSY Experiment als attraktive Alternative zur simultanen und

eindeutigen Resonanzzuordnung für JH und 13C in Seitenketten von partiell deuterierten,

grösseren Proteinen entwickelt. Es wird vorgeschlagen, durch kombinierte Anwendung des H-

13C „multiple-quantum" und des 'H-15N TROSY Effekts in den 3D TROSY-HN(CA)HA und 4D

TROSY-HACANH Experimenten die Sensitivität zu erhöhen. Es kann vorausgesagt werden,

dass das 13C-detektierte 3D HCC-TOCSY Experiment sehr hilfreich sein wird bei der

Resonanzzuordnung von 13C und 'H chemischen Verschiebungen in Methylgruppen von

Aminosäureseitenketten. Diese chemischen Verschiebungen enthalten für die

Strukturbestimmung wichtige Informationen in methyl-protonierten, komplett deuterierten und in

13C-markierten Proteinen mit hohem Molekulargewicht.

9

Abbreviations

ID, 2D, 3D 1-dimensional, 2-dimensional, 3-dimensionaI,

BBP ß-barrel platform

cine critical micelle concentration

CRINEPT Cross relaxation-enhanced polarization transfer

CRIPT Cross relaxation-induced polarization transfer

CSA Chemical shift anisotropy

CSI Chemical shift index

DHPC Dihexanoylphosphatidylcholine

DPC Dodecylphosphocholine

DSS 2,2-dimethyl-2-silapentane-5-sulfonate, sodium salt

FKBP FK506-binding protein

INEPT Insensitive nuclei enhanced by polarization transfer

IPTG Isopropylthiogalactoside

LpMIP Legionella pneumophila macrophage infectivity potentiator

MALDI Matrix-assisted laser desorption ionization

MAS Magic angle spin

MP Membrane protein

NMR Nuclear magnetic resonance

NOE Nuclear Ovcrhauser effect

NOESY NOE spectroscopy

OD6oo Optical density at 600 nm

Omp Outer membrane protein

PAS Principal axis system

PDB Protein data bank

PFG Pulse field gradient

PPIase peptidyl-prolyl cis/trans isomerase

ppm Parts per million

RCM-la Reduced and carboxymcthylated bovine a-lactalbumin

10

RDC Residual dipolar coupling

SDS-PAGE Sodium dodecyl sulphate-polyacrylamide gel electrophoresis

TM Transmembrane

TOCSY Total correlation spectroscopy

TROSY Transverse relaxation optimized spectroscopy

11

Chapter I

Preparation and characterization of a membrane protein ß-barrel platform

for solution NMR studies

Membrane protein ß-harrel platform for solution NMR studies 12

Membrane protein ß-barrel platform for solution NMR studies 13

1.1 Introduction

Membrane proteins take part in a large number of important physiological functions, such as

solute and macromolecular transport, chemical and electrical signaling, metabolism and

regulation, therefore, they constitute key targets for drug development. However, a deeper insight

into structure-function relationships of membrane proteins requires high-resolution structural

information, thus knowledge of their three-dimensional (3D) structures could contribute

decisively to better understanding of biological processes at the molecular level.

1.1.1 Structural studies of membrane proteins

High yield expression, purification and refolding of membrane proteins for studies either by X-

ray crystallography or by nuclear magnetic resonance (NMR) spectroscopy are still much more

demanding than the corresponding work with soluble proteins. It is generally difficulty to

crystallize membrane proteins from detergent solutions. For solution NMR experiments,

membrane proteins usually have to be indirectly solubilized in water by incorporation in model

membrane systems, such as micelles. Even for small membrane proteins, these mixed membrane

protein detergents micelles have rather large molecular masses, typically beyond 50000 Da.

Before the advent of the TROSY technique and complete deuteration of the side-chains, the size

of the resulting protein/detergent/lipid supramolecular assemblies was typically too large for

structure determination in solution by NMR.

1.1.1.1 Membrane proteins vs. water soluble proteins

Due to the above mentioned difficulties, determination of the 3D structures of membrane proteins

is still at a frontier of structural biology. Presently, less than 30 independent integral membrane

protein structures have been solved. This contrasts sharply with about 15 000 soluble proteins

solved by X-ray crystallography and NMR spectroscopy. About 30% of all proteins in eukaryotic

cells are membrane proteins [1J, however, less than 0.2% of all membrane protein structures are

known, whereas approximately 10% of soluble proteins are structurally characterized [2|.


1.1.1.2 Micelles, bicelles and lipid bilayers

For biophysical, structural and functional studies of membrane proteins, detergent micelles,

bicelles, lipid bilayers or lipid vesicles are commonly used as a replacement of the natural

membrane environment.

Micelles: Membrane proteins embedded in detergent micelles are most appropriate for studying

by solution NMR techniques as the combined demands of modest overall size of the

protein/detergent/lipid supramolecular assemblies and preservation of the functional structure of

the protein have so far most promisingly been met by reconstitution of membrane protein into

micellar structures.

Two important parameters characterize micellar solutions: the critical micelle concentration (cmc)

and the aggregation number. Detergents are monomeric below the cmc, but cooperatively

assemble into micelles above the cmc. The aggregation number describes the number of

monomers in a micelle. In the mixed membrane protein-detergent micelles, the effective

molecular weights of the membrane protein-detergent aggregates were determined more by the

structural properties of the protein than by the properties of the detergents [3], Usually there is no

general correlation between effective molecular weights for the mixed micelles and the sizes of

the corresponding protein-free micelles. Medium-chain detergents are generally preferred for use

in NMR studies of complex membrane proteins because they are no worse than short-chained

detergents in terms of increasing the effective molecular weight of the protein of interest while

they are considerably better at maintaining native-like protein conformation [4J. Cmcs and

aggregation numbers depend, sometimes quite dramatically, on environmental parameters, such

as temperature, ionic strength, pH and so on.

Bicelles are thought to be disk-shaped aggregates of phospholipid and detergent that can orient

spontaneously perpendicular to an applied magnetic field owing to their anisotropy of magnetic

susceptibility [5, 6], They were originally devised to orient membrane proteins in the magnetic

field for solid-state NMR studies. Apart from studying the structures of small membrane-bound

peptides, mostly in solid state NMR [7J, bicelles have so far not found wide application in the

solution NMR for structure determination of membrane proteins due to their overall effective


molecular weight, which is usually larger than that of corresponding mixed micelles, and their

anisotropical property, which causes additional line broadening in solution NMR. Sample

preparation of membrane protein in bicellar assemblies and their activity are also factors to be

considered. For example, diacylglycerol kinase (DAGK) exhibited a preference for

dimyristoylphosphatidylcholine or dipalmitoylphosphatidylcholine bicelles relative to those of

dilauroylphosphatidylcholine. The catalytic activity of DAGK reconstituted into several different

bicelle systems was measured and compared to the activities measured in traditional mixed

micelles and vesicles. For the most optimal bicelle systems tested, DAGK activities approached

those observed in mixed micelles or vesicles. For some other bicellar mixtures tested, activities

were much lower. [8J

Lipid bilayers are the natural environment of membrane proteins. Individual peaks can be

resolved by solid-state NMR of proteins in membranes that either are mechanically oriented in

the magnetic field or are unoriented, but spun at the magic angle (MAS) in the NMR

spectrometer [9-111- Recently, the transmembrane domain of Phospholamban (24-52) was

reported to be incorporated into phospholipid bilayers prepared from 1-palmitoyl -2-oleoyl-sn-

glycero-phosphocholine(POPC). PLB is a 52-amino acid integral membrane protein that

regulates the flow of Ca2+ ions in cardiac muscle cells [12]. Recently, the structure of GpA has

been determined in lipid bilayers by solid-state NMR [13].

1.1.1.3 NMR vs. X-ray crystallography in MP structure studies

Most structures of membrane proteins have been solved by X-ray crystallography. Despite its

relative success, X-ray crystallography of membrane proteins must still be considered a high art.

It is very difficult to crystallize membrane proteins from detergent solutions and the search for

appropriate crystallization conditions must sample a much larger space than a typical soluble

protein crystallization screen.

The use of NMR as a tool to determine structures of membrane proteins is for the most part in a

developmental stage. Compared to soluble proteins, for membrane proteins, the search for

appropriate solution conditions for the NMR samples typical must include a larger number of

variable parameters. In addition to temperature, pH and ionic strength, choice of detergents, the

Membrane protein ß-barrel platform for solution NMR studies

detergent concentration and the protein-to-detergent ratio are important parameters. Due to the

large size and long correlation time of MP in micelles or other membrane mimetic, relaxation is

efficient, resulting in reduced signal to noise and broad lines for signals detected with

conventional NMR techniques. TROSY techniques, which compensate the detrimental effects of

relaxation, open novel avenues for studies of structure, function and dynamics of large soluble

proteins, integral membrane proteins and various complexes of biomolecules [14]. Membrane

proteins can be analyzed in some detergent micelle systems using TROSY. For perdeuteratcd

proteins with high molecular weight, TROSY-based NMR experiments, such as 3D TROSY-

HNCA, TROSY-HN(CO)CA, TROSY-HN(CA)CO, TROSY-HNCACB, TROSY-

HN(CO)CACB and 4D TROSY-HNCACO and 4D TROSY-HNCOCA triple-resonance

experiments [15-17], can be recorded in order to assign backbone 1HN, 15N, 13C\ 13Ca and 13CP

resonances. Using samples uniformly 2H,15N,13C labeled with selectively protonated Val-y(l,2),

Leu-Ô(l,2) and He- 6(1) methyl groups, the NMR experiments 3D (H)C(CQ-TOCSY-(CO)-

["N.'HJ- TROSY and 3D HtCXCQ-TOCSY-fCOM^N^HJ-TROSY can be used for

assignment of the side chain methyl resonances of Val, Leu and He [18]. For uniformly

deuterated or highly deuterated and 13C- and 15N-labeled protein sample, the 13C-observe 2D and

3D spectroscopy can be employed in order to correlate the side-chain 13C chemical shifts and

assign them to particular residues along the polypeptide backbone using the'

C,lCa and C^

residue specific backbone assignment [19].

1.1.1.4 oc - helical and ß - barrel MP studied by solution NMR

With the advent of TROSY and isotope labeling techniques such as perdeuteration [20, 21J and

selective protonation of side chain[22-24J, a remarkable progress was achieved in the area of

NMR structural studies of membrane proteins.

For cc-helical membrane proteins, examples of systems studied include:

The first membrane-associated a-helical 29-residue polypeptide hormone glucagon in DPC

micelles [25], the bacteriorhodopsin fragment comprising residues 1-71 [26], native

bacteriorhodopsin [27, 28], the dimeric transmembrane domain of human glycophorin A (2 x 40


residues) in DPC micelles [29], the 81-residue human immuno-defciency virus (HIV) membrane-

associated protein Vpu [30, 31], the light-harvesting Iß subunit of Rhodobacter sphaeroides (48

residues) [32], biologically active phospholamban in lipid-mimicking dodecylphosphocholine

mice11es[33J, the 39-kDa homotrimeric protein diacylglycerol kinase (DAGK) in micellar

complexes with overall sizes larger than 100 kDa [34, 35] and most recently the atomic structure

was reported for an unusually small membrane protein Ost4p, containing only 36 residues [36].

Phospholamban is an integral membrane protein that regulates the contractility of cardiac muscle

by maintaining cardiomyocyte calcium homeostasis. The first structure of recombinant,

monomeric, biologically active phospholamban in lipid-mimicking dodecylphosphocholine

micelles was determined by multidimensional NMR experiments. The overall structure of

phospholamban is "L-shaped" with the hydrophobic domain approximately perpendicular to the

cytoplasmic portion [33],

E. coli diacylglycerol kinase (DAGK), an a-helical polytopic membrane protein is a

homotrimeric integral membrane protein comprised of 121 residue subunits, each having three

transmembrane segments. [34, 37, 38]. The aggregate detergent-protein molecular mass of

DAGK in both octyl glucoside and decyl maltoside (DM) micelles was determined to be in the

range of 100-110 kDa. Backbone NMR assignments were most recently reported for DAGK in

detergent micelles using TROSY-based pulse sequences [35].

Ost4p is a subunit of oligosaccharyltransferase (OT), which catalyzes N-glycosylation of proteins

in all eukaryotes and some prokaryotes. In Saccharomyces cerevisiae, OT is -composed of nine

nonidentical membrane proteins. Ost4p is an unusually small membrane protein containing only

36 residues, which folds into a well formed, kinked helix in the model-membrane solvent system.

The NMR structure of Ost4p helps understanding the structural basis for the function of this

protein [36].

a-helical membrane proteins can also be studied in organic solvent-water mixtures or organic

solvent, such as the subunits c and b of the Escherichia coli FoFi ATP synthase [391, subunit a of

the Escherichia coli ATP synthase [40] and the 52 residue phospholamban [41] However, there

are discrepancies between the reported structure for synthetic phospholamban in organic solvents

and its structure in lipid-mimicking dodecylphosphocholine micelles [33].

Subunit a of the Escherichia coli ATP synthase is a 30 kDa integral membrane protein, Pure

subunit a can be reconstituted with subunits b and c and phospholipids to form a functional


proton-translocating unit. NMR spectra of the pure subunit a in a mixed solvent show good

chemical shift dispersion and demonstrate the potential of the solvent mixtures for NMR studies

of the large membrane proteins that are currently intractable in aqueous detergent solutions [40].

The results must however be taken with caution considering the discrepancies observed for

phospholamban.

Bacteriorhodopsin (BR) is a 26 kDa seven-helical transmembrane protein found in the cellular

membrane of Halohacterium salinarium. Bacteriorhodopsin converts light energy into that of a

proton gradient that is subsequently used by the transmembrane protein ATP-synthase to produce

chemical energy in the form of ATP. Structures of a chymotryptic fragment C2 (residues 1- 71)

of bacterioopsin from Halohacterium halobium, solubihzed in an organic mixture of

methanol/chloroform and deuterated HCO2NH4, or in perdeuterated sodium dodecyl sulfate (SDS)

micelles in the presence of perdeuterated trifluoroethanol were determined by two-dimensional

and three-dimensional heteronuclear 15N-'H NMR techniques [26, 42]. Recently, the structures of

two forms of bacteriorhodopsin solubihzed in dodecyl maltoside with a deuterated dodecyl

moiety (dDM) present in the dark-adapted state were determined by using solution state NMR,

[28].

For ß-barrel membrane proteins, systems most widely studied include:

OmpX (148 residues) in DHPC micelles (with molecular mass of the mixed micelles of

OmpX/DHPC on the order of 60 kDa) [43, 44] and in a urea-denatured form [45], the

transmembrane domain of OmpA(177 residues) in DPC [46] and DHPC (with molecular mass of

the mixed micelles of OmpA/DHPC on the order of 80 kDa)[3], and the outer membrane enzyme

PagP (164 residues) in DPC and ra-octyl-ß-D-glucoside (OG) micelles[47].

TROSY-based NMR experiments have been applied in NMR studies of the E. coli integral

membrane proteins OmpX and OmpA in mixed micelles with the detergent DHPC [3, 18, 48].

For OmpX, complete sequence-specific NMR assignments have been obtained for the

polypeptide backbone and side chain methyl resonances of Val, Leu and Ile [18J. The Creand

13CP chemical shifts and nuclear Overhauser effect data then resulted in the identification of the

regular secondary structure elements of OmpX/DHPC in solution. The structure of the integral

membrane protein OmpX from E, coli reconstituted in 60 kDa DHPC micelles calculated from


526 NOE upper limit distance constraints was recently reported [43]. The structure determination

was based on complete sequence-specific assignments for the amide protons and the Val, Leu,

and Ile(ôi) methyl groups in OmpX, which were selectively protonated on a perdeuterated

background. The solution structure of OmpX in DHPC micelles consists of a well-defined, eight-

stranded antiparallel beta-barrel, with successive pairs of beta-strands connected by mobile loops.

The topology and the protein-detergent interactions in the mixed micelles, consisting of about 90

molecules of the detergent DHPC and one molecule of the E. coli OmpX, were characterized by

intermolecular NOEs between the protein and detergent DHPC and by use of paramagnetic

probes with different physicochemical properties [44, 49]. The experimental data suggest that the

hydrophobic surface areas of OmpX are covered with a monolayer of DHPC molecules, which

appears to mimic quite faithfully the embedding of the beta-barrel in a double- layer lipid

membrane.

For the transmembrane domain of OmpA, published data indicate that very similar results were

obtained for OmpA (0-176)/DPC as for OmpA (1-176)/DHPC and OmpA (0-171)/DHPC [3, 46],

i.e. backbone assignments were obtained for 138 residues (76%), resulting in the identification of

eight ß-strands based on l3C chemical shifts and NOEs. The three-dimensional fold of the

transmembrane domain of OmpA of E. coli in DPC micelles in solution has been determined [46].

The structure consists of an eight-stranded ß-barrel connected by tight turns on the periplasmic

side and larger mobile loops on the extracellular side. The solution structure of the barrel in DPC

micelles is similar to that in rc-octyltetraoxyethylene (C8E4) micelles determined by X-ray

diffraction. NMR dynamic experiments reveal a gradient of conformational flexibility in the

structure that may contribute to the membrane channel function of this protein.

The global fold of E. coli PagP, a bacterial outer membrane enzyme which transfers a palmitate

chain from a phospholipid to lipid A, was determined in both DPC and OG detergent micelles

using TROSY. PagP consists of an eight-stranded anti-parallel ß-barrel preceded by an

amphipathic a-helix. The ß-barrel is well defined with barrel axis uniquely tilted by 30 degrees

with respect to the membrane normal, whereas NMR relaxation measurements reveal

considerable mobility in the loops connecting individual ß-strands. Three amino acid residues

critical for enzymatic activity were found to be located in extracellular loops near the membrane

interface, positioning them optimally to interact with the polar head groups of lipid A [47].


1.1.2 Experimental aspects: solution NMR studies of MP

For solution NMR studies of membrane protein, preservation of the functional structure of the

protein and restriction of the overall size of the mixed protein/detergent/lipid particles have so far

most promisingly been met by reconstitution of membrane proteins into micelles.

1.1.2.1 MP sample preparation for solution NMR studies

As the overall size of the mixed membrane protein detergent micelles is usually quite large,

typically 60-110 kDa, membrane proteins typically have to be labeled with the stable isotopes H,

13C and 15N for multidimensional heteronuclear NMR experiments. Therefore, right at the

beginning of the project, it was very important to obtain a high-yield expression system for the

desired membrane proteins due to the costs associated with isotope labeling, and deuteration

often causes a reduction in yield of the protein due to the negative influence of the deuterated

medium on the cell metabolism [50].

1.1.2.1.1 Expression and purification

Because there is so far no generally recommended refolding protocol available for a-helical

membrane proteins, the protein is usually expressed in its native form in the cell membrane.

Membrane protein then may in favorable cases be extracted from the membrane and purified in

the folded form and transferred into detergent micelles [51]. However, low yields have often

greatly increased the expense of preparing of isotope-labeled NMR samples of a-helical

membrane proteins.

So far, ß-barrel membrane proteins, such as OmpX [43, 44], OmpA [3, 46] and PagP [47], were

overexpressed with high yield in an aggregated form (inclusion bodies) in the cytoplasm of E.

coli. In some cases, the signal sequences have to be deleted to redirect the membrane protein for

cytoplasmic expression and/ or Histidine tags were engineered into the expression vectors for

easier purification. The inclusion bodies are usually first isolated, and protein is extracted by

dissolving the inclusion bodies in concentrated urea or guanidinium hydrochloride solutions for

further purification, and subsequent refolding and reconstitution in detergent micelles[43, 45].


1.1.2.1.2 Refolding

For the a-helical membrane proteins, there is so far no generally recommended refolding

protocol available, and expression and purification are usually carried out using the natively

folded form [51].

For the ß-barrel membrane proteins, purification can be performed in the presence of dénaturants

such as urea or guanidinium chloride [45]. However, the refolding step is often cumbersome.

Refolding conditions have to be carefully explored for each membrane protein. In addition to

temperature, pH, buffer and ionic strength, one has to consider the initial membrane protein

concentration in dénaturants, the choice of detergents, the detergent concentration and some other

agents which assist refolding, such as: divalent cations (MgCl2, CaCl2, orEDTA to reduce their

concentration), nonpolar additives (glycerol), polar additives (L-Arginine), chaotrope (guanidine

hydrochloride), non-detergent sulfobetaine (NDSB), ammonium sulfate, etc [52-55]. In many

cases, the lack of an appropriate refolding method is a major obstacle in the structure

determination of a membrane protein by NMR. It is hoped that in the future an efficient

expression of membrane proteins in their native form will avoid the often cumbersome refolding

step.

1.1.2.1.3 Isotope-labeling

Due to the large overall size of the membrane protein-detergent assemblies, membrane proteins

typically have to be labeled with the stable isotopes 2H, 13C and 15N for multidimensional

heteronuclear NMR experiments[35, 43],

For the nC and 15N labeling, MP can be expressed in a minimal medium or a Martek-9 medium

supplied with l3C- glucose and 15NH4C1 as the isotope sources. Deuterium labeling has been used

to improve the resolution and sensitivity of solution NMR spectra[23]. Furthermore, the use of

uniform the 2H labeling results in a relatively large improvement in the sensitivity of TROSY-

based NMR spectra of large molecules or macromolecular complexes, such as membrane

proteins in detergent micelles [56]. However, because much of the sidechain information is lost

by the complete deuteration of these moieties, methods were developed to keep methyl groups in


the predeuterated protein, such as methyl groups in Val, Leu and He, [57]. Retention of proton

labels in the amino acids of a given type in the otherwise deuterated background is an important

method to obtain sidechain NOEs and necessary for a high resolution structure determination

[22-24].

1.1.2.2 NMR spectroscopy and structure determination of MP

Solution NMR spectroscopy was recently used for structure determination of membrane proteins

in micelles [35, 40, 43, 46, 47]. Optimal spectra with good sensitivity and high resolution can be

obtained by the suppression of transverse relaxation in multidimensional hcteronuclear NMR

experiments. Suppression of transverse relaxation can be achieved by biochemical methods, such

as perdeuteration of the membrane protein samples [20, 58, 59], and NMR methods, through the

use of carefully designed TROSY-based NMR pulse sequence[14, 16, 60],

A major advance in solution NMR spectroscopy that has had a significant impact on the

determination of membrane protein structures in detergent micelles was the development of

TROSY (Transverse relaxation- optimized spectroscopy) [14], which, compared to the

conventional HSQC, enables structural studies of much larger size system such as, large protein

oligmers and membrane protein/detergent/lipid supramolecular assemblies. TROSY is an

approach for the suppression of transverse relaxation in multidimensional NMR experiments,

which is based on the constructive use of the cross-correlated interference effect between dipole-

dipole coupling (DD) and chemical shift anisotropy (CSA). The principle of TROSY combined

with cross-correlated relaxation enhanced polarization transfer (CRINEPT) [61, 62] was applied

to particles with the molecular masses up to 900 kDa [63J.

Selective and uniform deuteration of amino acid sidechains is another indispensable tool for

suppression of transverse relaxation in multidimensional heteronuclear NMR spectra of large

complexes, including membrane proteins in detergent micelles [58]. The gain in sensitivity

comes from the lower gyromagnetic ratio of 2H relative to 'H and the corresponding reduction in

the rate of transverse relaxation of the neighboring heteronuclei due to the dipole-dipole coupling

mechanism. This results in sharper resonance lines. However, a complete deuteration of the side-

chains eliminates much of the side-chain information and usually only a limited set of NOE


distance constraints are accessible among the backbone amide protons. Consequently, structural

models for ß-barrel membrane proteins which have been constructed so far are only of low

resolution, whereas for the a-helical membrane proteins usually only the secondary structure is

determined. Methyl groups are useful sources of information for structure determination.

Protonation of methyl groups in a predeuterated protein, such as methyl groups in Val, Leu and

He, results in greatly improved precision of the structure determination because more NOE

distance constraints are generated for the sidechain methyl protons. However, dipolar interactions

between the backbone amide protons and the methyl protons have a severe effect on the TROSY

efficiency in very large systems. For these systems, the sensitivity of TROSY spectra is

significantly degraded by the introduction of methyl protons into the uniformly deuterated

background. Recently, it was reported that methyl analyses with molecular weights greater than

600 kDa will complement TROSY and CRINEPT analyses of amides in NMR studies of

structure and molecular interactions of extremely large macromolecules and assemblies [64].

These results indicate that for very large systems optimal observation of amide and methyl

moieties might require separately labeled samples.

In addition to distance constraints based on the NOE (NOESY), useful additional experimental

constraints for membrane protein structure determination can presently be expected to result from

measurement of residual dipolar couplings [65] and from the use of paramagnetic spin labels

[66]. An interesting method to weakly align spherical membrane protein-detergent complexes in

the magnetic field is by the use of lanthanide metal ions to adventitious [67] or engineered [30]

sites in this class of proteins. An 'EF hand' calcium-binding site could be engineered into

membrane proteins which do not have an adventitious lanthanide-binding site.

Furthermore, for uniformly or highly deuterated membrane proteins, 13C-detected 2D and 3D

spectroscopy recently developed in our group can be employed as a method for the side-chain ' C

chemical shift assignment [19]. Another attractive feature of the 13C-observe experiments is that

multiple and redundant 'Jcc scalar couplings are resolved as the i3C multiplets in the directly

acquired 13C dimension. Measurement of multiple l3C - nC residual dipolar couplings (RDCs)

could further offer potential orientational constraints for the structure determination of membrane

proteins f681.


1.2 Preparation and characterization of a membrane protein ß-barrel

platform of OmpA

1.2.1 Introduction to the outer membrane protein A (OmpA)

Outer membrane protein A (OmpA) is an abundant structural protein of the outer membrane of

Gram-negative bacteria. It is believed to connect the outer membrane structurally to the

periplasmic peptidoglycan layer via its periplasmic domain, which consists of residues 177-325.

The N-terminal domain of the OmpA protein from Escherichia coli, consisting of residues 1-172,

is embedded in the outer membrane and forms the transmembrane domain, whose structure was

solved by X-ray crystallography [69] and NMR [46].

LI L2 L3 L4

AimAsn As»

#9

Ptie

Ihr

VMSet

(Sei

ûlySur

Glu

fîsr A$n j^ «gPm Ûty Am Ly* m «te û)yi7wa* ijW Gfyra "Ont —•&{ **P A&>î$o 7**

Ms fyr Ala to__A Thr Aap Ars»ßii a«A"^*TB^ 75«- ffiE ' —

W»^*" (Ami ] [Mal I bj».

Extracellular

Periplasmic

Figure 1.1 Two-dimensional model of the arrangement of the N-terminal ß-barrel domain of OmpA in the

outer membrane, adapted from [70]. The surface-exposed loops and periplasmic turns are labeled LI to L4

and Tl to T3, respectively. Amino acid residues are numbered according to their position in the wild-type

sequence. In the study of the ß-barrel platform of OmpA, two modifications in the periplasmic turns were

maded. At turn T2, Tle-87 was replaced by Lys-Leu-Gly. At turn T3, the peptide Arg-Arg-Arg-Ile was

introduced between 1131 and T132, and Ala-130 was converted to Val. Residues that were removed upon

loop deletion mutagenesis are shown in italics; amino acids that were introduced in their places are shown

in boldface.


In detergent micelles and lipid bilayers, the transmembrane domain of OmpA forms an

antiparallel ß-barrel whose eight transmembrane ß-strands are connected by three short

periplasmic turns and four relatively large surface-exposed hydrophilic loops (Figure 1.1) [70]. A

gradient of increasing dynamics starts from the center of the barrel towards both ends of the

barrel of OmpA, which opposes the dynamic gradient of the lipid bilayer itself [46].

Over-expression and spontaneously refolding of OmpA into detergent micelles [11] has greatly

facilitated its structure determination by NMR [46J. Refolded OmpA exhibits similar single

channel properties as native OmpA [72]. Other functions that have been attributed to OmpA are

its involvement in bacterial conjugation and its action as a receptor for various bacteriophages

and some colicins.

1.2.2 Design of the ß-barrel platform (BBP) of OmpA

The structural and functional roles of the surface-exposed loops of OmpA were extensively

studied through reconstruction of its loop-shortened mutants [701. The loops are shortened

separately and in all possible combinations. In vivo, the loop deletion mutants assembled into the

outer membrane with high efficiency and adopted the wild-type membrane topology. This

approach indicates the absence of topogenic signals (e.g., in the form of loop sizes or charge

distributions) in these loops. In our work, we prove this by direct refolding experiments followed

by NMR analysis. The shortening of surface-exposed loops did not reduce the thermal stability of

the protein. However, all loops were necessary for the OmpA protein to function in the

stabilization of mating aggregates during F conjugation[70].

An OmpA deletion variant with all four loops shortened (Figure 1.1), i.e. the ß-barrel platform

(BBP) of OmpA, consisting of only 142 residues (with the C-terminal QGEAA remaining

attached), constitutes the smallest ß-struetured integral membrane protein known to date. In our

NMR studies of this BBP, we were asked questions if the removed loops contain amino-acid

sequences critical for the membrane protein's folding and if the absence of the extracellular loops

affects the structure of the transmembrane domain. This ß-barrel platform (BBP) could serve as a

model for NMR study of membrane assembly of integral ß-struetured membrane proteins in vitro.


1.2.3 NMR sample preparation of BBP

Ï.2.3.1 Sequencing the reconstructed gene

The plasmid was extracted from the BBP mutant and transformed into E. coli XLl-Blue

competent cells. The plasmid extracted from XLl-Blue cells was sequenced with using standard

sequencing techniques offered online (www.microsynth.ch). Translation of the DNA sequence to

amino acids (Table 1 in Appendix A.l) shows that BBP is presumably four-loop shortened

mutant of OmpA, with two modifications at its periplasmic turns (T2 to T2H1 and T3 to T3S1),

Ala-130 converted to Val and QGEAA remaining attached at the C-terminaus [70].

1.2.3.2 Over-expression, labeling and purification of BBP

The sequenced plasmid was transformed into E. coli BL21-Gold (DE3)pLysS competent cells.

The freshly transformed cells were grown in the LB medium overnight. Then 4 x 2.5 ml of this

culture was inoculated into 4 x 250 ml Martek-9 medium, with 13C-glucose and 15NH4C1 supplied

as the 13C and 15N sources for the labeled samples. To prepare the deuterated samples, D20 was

used as solvent instead of H2O with a two-step adaptation of the growing cultures in 50% D2O /

H20 and 100% D20 LB medium. The culture was induced with 1 mM 1PTG at OD6(x> 0.3-0.4 in

Martek-9 medium and harvested by centrifugation when the cell growth reached the stationary

phase.

The cell pellet was frozen and then thawed on ice, resuspended in the Tris-EDTA buffer (20 mM

Tris, 5 mM EDTA), pH - 8, with a volume (ml) of the buffer corresponding to 3 times the wet

pellet weight (g). To lyse the cells, the resuspended pellet was sonicated for 20-30 minutes on

crushed ice (to avoid heating of the solution and degradation of the protein), then centrifuged for

1 hour at 6000 rpm and the white inclusion body pellet was kept. The inclusion body pellet was

resuspended in a volume (ml) of 2% Triton-XlOO Tris-EDTA buffer, pH = 8, corresponding to

about 3 times the wet pellet weight (g) shaken in the incubator for 20 minutes at 37°C to remove

cell membranes, centrifuged for 30 minutes at 4°C to collect the pellet. The supernatant was

discarded and the pellet was resuspended in 30 ml Tris-EDTA buffer, pH = 8 without Triton and

shaken again for 1 hour at 37°C to remove detergent, then centrifuged for 30 minutes at 4ÜC and

the pellet was kept. This pellet was resuspended a in minimal amount (~lml) of 6M GuHCl, Tris-


EDTA buffer, pH 6.0, shaken for a minimum of 2 hours in the incubator at 37 °C, and centrifuged

for 20 minutes at 20000 rpm, and the supernatant was kept.

SDS-PAGE is done to check the purity of the protein sample (Figure 1.2). The MALDI mass

spectrum showed a MW of 15681, which is comparable with theoretical average mass of 15673.6

(with N-terminal Met removed).

1

Figure 1.2 SDS-PAGE shows the induction and purification of BBP. Lane 1 is protein marker (14 kDa -

66 kDa). Lane 2 is the purified proteinin 6 M GuHCl solution. Lanes 3, 5, 7, 9 and 4, 6, 8, 10 are cell

extract samples of 1 ml of culture taken from four different flasks before and after induction, respectively.

1.2.3.3 Refolding of BBP into micelles

Refolding of BBP required screening of a large space of refolding conditions so that fractional

factorial [73] refolding screens were used to identify refolding conditions for BBP [53, 55]. The

protein sample obtained in 6M GuHCl, Tris-EDTA buffer, PH 6.0 was diluted to 0.4 mM and

then refolded in the presence of detergent micelles and other additives at room temperature by

slowly diluting the unfolded protein into the refolding buffer while stirring extensively followed

by dialysis and concentration by ultrafiltration.


Table 1 Fractional factorial refolding screen

Factor 1 2 3 4 5 6 7 8

Exp. 1 - - - - - - - -

Exp.2 + . - - - + + +

Exp.3 - + - - + - + +

Exp.4 + + - - + + - -

Exp.5 - - + - + + + -

Exp.6 + - + . + - - +

Exp.7 - + + - - + - +

Exp.8 + + + - - - + -

Exp.9 - - - + + + - +

Exp. 10 + - - + + - + -

Exp.ll - + - + - + + -

Exp. 12 + + - + - - - +

Exp. 13 - - + + - - + +

Exp. 14 + - + + - + - -

Exp. 15 - + + + + - - -

Exp. 16 + + + + + + + +

Factor 1 pH (MES pH 6 +/ HEPES pH 7 -)

Factor 2 detergent (Octyl-POE +/ DHPC -)

Factor 3 divalent cations (presence +/ absence, 5 mM EDTA -)

Factor 4 guanidinium hydrochloride (presence / absence)

Factor 5 Glycerol (presence / absence)

Factor 6 L-arginine (presence / absence)

Factor 7 non-detergent sulfobetaine (presence / absence)

Factor 8 ammonium sulfate (presence / absence)

With the consideration of 8 factors with 2 levels (- or + in the schematic representation shown

above) for each factor, a 16 condition fractional factorial refolding screen (Table 1) was carried

out to identify optimal refolding conditions. By examining 1/16th fraction of the full factorial (the


full factorial would require 28- 256 experiments), refolding conditions were established for BBP

in Dihexanoylphosphatidylcholine (DHPC) micelles.

Finally, the optimal refolding of BBP for solution NMR studies is carried out by diluting 0.4 mM

denatured protein in 6M GuHCl into 6 volumes of the refolding buffer: 3% DHPC in 0.1 M MES

buffer, pH 6 in the presence of 5 mM EDTA, 0.5 M guanidinium hydrochloride, 0.4 M glycerol,

0.35 M L-arginine, 0.5 M ammonium sulfate and 100 mM NaCI, at room temperature. After

concentrating and dialysis against the MES buffer, pH 6.0 containing 5% 2H20, the final protein

sample is concentrated to a volume of 300 pi by ultrafiltration and then transferred to a Shigemi

NMR tube for solution NMR studies.

1.2.4 NMR spectroscopy of BBP

NMR spectra including 2D ^H^NJ TROSY of the 2H,15N-labeled BBP and 3D TROSY-

versions of HNCA, HNCACB and 3D 15N resolved TROSY-NOESY of 2H, 15N and 13C -labeled

BBP were acquired at 30°C on a Bruker Avance 600 MHz spectrometer equipped with a TXI

cryogenic probehead and a Bruker Avance 900 MHz spectrometers. NMR data were processed

using the PROSA software [74] and analysed using XEASY [75] and CARA (unpublished,

Keller ct al., 2004). The 'H chemical shifts were referenced to the DSS signal at 0 ppm and the

l3C and 15N chemical shifts were referenced indirectly using the 13C/1H and 15N/'H gyromagnetic

ratios [76].

Figure 1.3 shows the 2D f'H,15N] TROSY of 2H, 15N and I3C -labeled BBP, which displays the

large signal dispersion typical for a folded protein. After a comparison of the peak pattern with

2D ['H,15N] TROSY of the transmembrane domain of OmpA (1-176), the primary conclusion

can be drawn that BBP with all four extracelluar loops truncated refolds in vitro in a similar way

to the transmembrane domain of wild-type OmpA. This suggests that the extracellular loops

indeed do not contain amino-acid sequences critical for the membrane protein's refolding in vitro.

The squenece specific backbone assignment is in a part achieved by the use of 3D TROSY-

versions of HNCA, HNCACB and 3D 15N resolved TROSY-NOESY of 2H, 15N and 13C-labeled

BBP together with the information from the previously reported NMR studies of the N-terminal


110

115

[ppm]

125

-130

10.0 9.0 JïFtPpm] 8.0

Figure 1.3 900 MHz 2D [*H, l5N ]-TROSY spectrum recorded with a 2 mM 2H, I5N and nC -labeled BBP

sample at 303 K. Signals corresponding to the assigned residues are annotated with number according to

their position in the wild-type OmpA sequence.

transmembrane domain of OmpA (1-176) [3, 46J. Figure 1.4 shows strips corresponding to the

residues 48- 52 of BBP in the TROSY-HNCA experiment recorded at 900MHz, 303K. The

assigned residues are indicated in figure 1.3, of which residues 10-13, 42- 45, 48- 57, 77- 85, 95-

97, 125- 128 and 164-176 are located in transmembrane ß-strand segments (except

ransmembrane ß-strand 7), ranging from the center toward the periplasmic side of the ß-barrel.

The residues residing either at the interface between the transmembrane segments and the

extracellular loops (although already shortened) or in the short periplasmic turns, have not yet

been assigned, probably due to line broadening due to the conformational exchange on ms


E52

121.79

-45

55

[ppm]

-65

7.92 7.28 8.66 9.098.67

*HJ^|»D1]

Figure 1.4 Strips corresponding to the assignment of the residues 48-52 (numbered according to their

position in the wild-type OmpA sequence) of BBP in the TROSY-HNCA spectrum recorded with a 2 mM

2H, l5N and 1JC-labeled BBP sample at 900 MHz, 303 K. Arrows indicate the sequential connectivities.

timescale [77-79]. However, the chemical shift index (CSI) analysis based on the CA chemical

shifts of the assigned residues [80] indicates the existence of the ß-strand secondary structure

elements. These results further suggest that BBP indeed refolds as a ß-barrel in DHPC in vitro, in

a similar way to the transmembrane domain of the wild type OmpA, and that the absence of the

extracellular loops does not affect the structure of the transmembrane domain. Based on the

information available so far, we constructed a 3D model of BBP using the program SWISS-

MODEL 181], as shown in figure 1.5.


Figure 1.5 A. One conformer of the solution NMR structure (PDB code: 1G90) of the N-terminal

transmembrane domain of OmpA (0-176) [46]. B. Structure of BBP modelled based on the

conformer in figure 1.5 A using SWISS-MODEL [81].

1.2.5 Conclusions and perspectives

Based on our NMR studies of BBP, this artificially designed outer membrane protein, which

constitutes the smallest ß-struetured integral membrane protein known to date, refolds into a ß-

barrel structure in vitro. The absence of the extracellular loops does not appear to affect the

structure of the transmembrane domain. Therefore, this minimal ß-barrel unit, or ß-barrel

platform, could be used as a basic model for design of an artificial membrane protein, which may

have interesting biological applications, such as, artificially controllable ion channels or signal

transduction across the cell membrane.

Successful refolding of BBP in vitro and preparation of isotope labeled samples suitable for

NMR structure determination encouraged us to engineer an 'EF hand' calcium-binding site into

this BBP, to design a new artificial membrane protein with high affinity to paramagnetic ions,

such as lanthanide. As mentioned above, binding of lanthanides to this engineered site has the

potential to weakly align the membrane protein-detergent complex in strong magnetic fields,

which in turn can provide important structural constraints based on paramagnetic relaxation rate


enhancements (distances), residual dipolar couplings (orientation), pseudocontact shifts

(distances) and Curie-dipole dipole cross-correlation (distances and orientation) [82].

Engineering of the lanthanide binding sites and aforementioned NMR data can be used as a

general method to obtain high resolution NMR structure of membrane proteins.

References

1. Walhn E, von Heijne G, Genome-wide analysis of integral membrane proteins Jrom eubacierial, archaean, and

eukaryotic organisms. Protein Sei., 1998. 7(4): p. 1029-1038.

2. Vitkup D, Melamud E, Moult J, Sander C, Completeness in structural genomics. Nat. Struct. Biol., 2001. 8(6): p.

559-566.

3. Fernandez C, Hilly C, Bonjour S, Adeishvili K, Pervushin K, Wuthrich K, Solution NMR studies ojthe integral

membrane proteins OmpX and OmpA from Escherichia coli. FEBS Lett., 2001. 504(3): p. 173-178.

4. Vinogradova, O., F. Sonnichsen, and C. R. Sanders, On choosing a detergent for solution NMR studies of

membrane proteins. J. Biomol. NMR, 1998. 11(4): p. 381-386.

5. Tjandra, N. and A. Bax, Direct measurement oj distances and angles in biomolecules by NMR in a dilute liquid

crystalline medium (vol 278, pg 1111, 1997). Science, 1997. 278(5344): p. 1697-1697.

6. Cavagnero, S., H. J. Dyson, and P. E. Wright, Improved low pH bicelle system for orienting macromolecules

over a wide temperature range. J. Biomol. NMR, 1999. 13(4): p. 387-391.

7. Fujiwara T, Todokoro Y, Yanagishita H, Tawarayama M, Kohno T, Wakamatsu K, Akutsu H, Signal

assignments and chemical-shifl structural analysis of uniformly C-13, N-15-labeled peptide, mastoparan-X, by

multidimensional solid-state NMR under magic-angle spinning. J. Biomol. NMR, 2004. 28(4): p. 311-325.

8. Czerski L, Sanders CR, Functionality of a membrane protein in bicelles. Anal. Biochem., 2000. 284(2): p. 327-

333.

9. Opella, S. J., NMR and membrane proteins. Nat. Struct. Biol., 1997. 4: p. 845-848.

10. Hu, W., N. D. Lazo, and T. A. Cross, Tryptophan Dynamics and Structural Refinement in a Lipid Bitayer

Environment - Solid-State Nmr of the Gramicidin Channel. Biochemistry, 1995. 34(43): p. 14138-14146.

11. Marassi, F. M, and S. J. Opella, NMR structural studies of membrane proteins. Curr. Opin. Struct. Biol., 1998.

8(5): p. 640-648.

12. Dave PC, Tiburu EK, Damodaran K, Lorigan GA, Investigating structural changes in the lipid bilayer upon

insertion of the transmembrane domain of the membrane-bound protein phospholamban utilizing P-31 and H-2

solid-state NMR spectroscopy. Biophys. J., 2004. 86(3): p. 1564-1573.

13. Smith, S. O., et al., Structure of the transmembrane dimer interface of glycophorin A in membrane bilayers.

Biochemistry, 2001. 40(22): p. 6553-6558.

14. Pervushin, K., et al., Attenuated T-2 relaxation by mutual cancellation ofdipole- dipole coupling and chemical

shift anisotropy indicates an avenue to NMR structures of very large biological macromolecules in solution.

Proc. Natl. Acad. Sei. U. S. A., 1997. 94(23): p. 12366-12371.


15. Salzmann, M., et al., Improved sensitivity and coherence selection for [N-15,H-1J- TROSY elements in triple

resonance experiments. J. Biomol. NMR, 1999. 15(2): p. 181-184.

16. Salzmann, M., et al., TROSY-type triple-resonance experiments for sequential NMR assignments of large

proteins. J. Am. Chcm. Soc, 1999. 121(4): p. 844-848.

17. Yang, D. W. and L. E. Kay, TROSY triple-resonance four-dimensional NMR spectroscopy of a 46 ns tumbling

protein. J. Am. Chem. Soc., 1999. 121(11): p. 2571-2575.

18. Hilty C, Fernandez C, Wider G, Wuthrich K, Side chain NMR assignments in the. membrane protein OmpX

reconstituted in DHPC micelles. J. Biomol. NMR, 2002. 23(4): p. 289-301.

19. Eletsky A, Moreira O, Kovacs H, Pervushin K, A novel strategy for the assignment of side-chain resonances in

completely deuterated large, proteins using C-13 spectroscopy. J. Biomol. NMR, 2003. 26(2): p. 167-169.

20. LeMaster, D. M., Deuteration in Protein Proton Magnetic-Resonance. Method Enzymol., 1989.177: p. 23-43.

21. LeMaster, D. M., Uniform and Selective Deuteration in 2-Dimensional Nmr of Proteins. Annual Review of

Biophysics and Biophysical Chemistry, 1990. 19: p. 243-266.

22. Rosen, M. K., et al., Selective methyl group protonation ofperdeuterated proteins. J. Mol. Biol., 1996. 263(5): p.

627-636.

23. Gardner, K. H„ et al., Solution NMR studies of a 42 KDa Escherichia coli maltose binding protein beta-

cyclodextrin complex: Chemical shift assignments and analysis. J. Am. Chem. Soc, 1998. 120(45): p. 11738-

11748.

24. Goto, N. K., et al., A robust and cost-effective method for the production of Vat, Leu, He (delta 1) methyl-

protonatedN-15-, C-13-, H-2-labeledproteins. J. Biomol. NMR, 1999. 13(4): p. 369-374.

25. Braun W, Wider G, Lee KH, Wuthrich K, Conformation Of Glucagon In A Lipid Water Interphase By II-l

Nuclear Magnetic-Resonance. J. Mol. Biol., 1983. 169(4): p. 921-948.

26. Pervushin, K. V., et al., 3-Dimensional Structure of (l-71)Bacterioopsin Solubilized in Methanol/Chloroform

and SDS Micelles Determined By N-15-H-1 Heteronuclear Nmr-Spectroscopy. Eur. J. Biochem., 1994. 219(1-2):

p. 571-583.

27. Schubert M, Kolbe M, Kessler B, Oesterhelt D, Schmieder P, Heteronuclear multidimensional NMR

spectroscopy of solubilized membrane proteins; Resonance assignment of native bacteriorhodopsin.

Chembiochem, 2002. 3(10): p. 1019-1023.

28. Patzelt, H., et al., The structures of the active center in dark-adapted bacteriorhodopsin by solution-state NMR

spectroscopy. Proc. Natl. Acad. Sei. U. S. A„ 2002. 99(15): p. 9765-9770.

29. MacKcnzie, K. R., J. H. Prestegard, and D. M. Engelman, A transmembrane helix dimer: Structure and

implications. Science, 1997. 276(5309): p. 131-133.

30. Ma C, Opella SJ, Lanthanide ions bind specifically to an added "EF-hand" and orient a membrane protein in

micellesfor solution NMR spectroscopy. J. Magn. Reson., 2000. 146(2): p. 381-384.

31. Kochendoerfer GG, Jones DH, Lee S, Oblatt-Montal M, Opella SJ, Montai M, Functional characterization and

NMR Spectroscopy on full-length Vpu from HIV-1 prepared by total chemical synthesis. J. Am. Chem. Soc,

2004. 126(8): p. 2439-2446.


32. Sorgen PL, Cahill SM, Krueger-Koplin RD, Krueger-Koplin ST, Schenck CC, Girvin ME, Structure of the

Rhodobacter sphaeroides light-harvesting 1 beta subunit in detergent Micelles. BIOCHEMISTRY, 2002. 41(1):

p, 31-41.

33. Zamoon J, Mascioni A, Thomas DD, Veglia G, NMR solution structure and topological orientation of

monomeric phospholamban in dodecylphosphocholine micelles. Biophys. J., 2003. 85(4): p. 2589-2598.

34. Oxenoid K, Sonnichsen FD, Sanders CR, Topology and secondary structure of the N-terminal domain of

diacylglycerol kinase. Biochemistry, 2002. 41(42): p. 12876-12882.

35. Oxenoid K, Kirn HJ, Jacob J, Sonnichsen FD, Sanders CR, NMR assignments for a helical 40 kDa membrane

protein. J. Am. Chem. Soc, 2004. 126(16): p. 5048-5049.

36. Zubkov S, Lennarz WJ, Mohanty S, Structural basis for the function ofa minimembrane protein subunit ofyeast

oligo.iaccharyltransfera.se. Proc. Natl. Acad. Sei. U. S. A., 2004. 101(11): p. 3821-3826.

37. Sanders, C. R., et al., NMR structural studies of diacylglycerol kinase, a 13 kDa integral membrane protein.

Biophys. J., 1998. 74(2): p. A130-A130.

38. Vinogradova, O., et al., Escherichia coli diacylglycerol kinase: A case study in the application of solution NMR

methods to an integral membrane protein. Biophys. J., 1997. 72(6): p. 2688-2701.

39. Rastogi, V. K. and M. E. Girvin, Structural changes linked to proton translocation by subunit c of the ATP

synthase. Nature, 1999. 402(6759): p. 263-268.

40. Dmitriev OY, Altendorf K, Fillingame RH, Subunit a of the E-coli ATP synthase: reconstitution and high

resolution NMR with protein purified in a mixed polarity solvent. FEBS Lett, 2004. 556((l-3)): p. 35-38.

41. Lamberlh S, Schmid H, Muenchbach M, Vorherr T, Krebs J, Carafoli E, Griesinger C, NMR solution structure

ofphospholamban. Helv. Chim. Acta, 2000. 83(9): p. 2141-2152.

42. Orekhov, V. Y., K. V. Pervushin, and A. S. Arseniev, Backbone Dynamics of (1 -71)Bacterioopsin Studied By 2-

Dimensional H-I-N-1S Nmr-Spectroscopy. Eur. J. Biochem., 1994. 219(3): p. 887-896.

43. Fernandez C, Hilty C, Wider G, Guntert P, Wuthrich K, NMR structure ofthe integral membrane protein OmpX.

J. Mol. Biol., 2004. 336(5): p. 1211-1221.

44. Hilty C, Wider G, Fernandez C, Wuthrich K, Membrane protein-lipid interactions in mixed micelles studied by

NMR spectroscopy with the use ofparamagnetic reagents. Chembiochem, 2004. 5(4): p. 467-473.

45. Tafer H, Hiller S, Hilty C, Fernandez C, Wuthrich K, Nonrandom structure in the urea-unfolded Escherichia

coli outer membrane protein X (OmpX). Biochemistry, 2004. 43(4): p. 860-869.

46. Arora A, Abildgaard F, Bushweller JH, Tamm LK, Structure of outer membrane protein A transmembrane

domain by NMR spectroscopy. Nat. Struct. Biol., 2001. 8(4): p. 334-338.

47. Hwang PM, Choy WY, Lo EI, Chen L, Forman-Kay JD, Raetz CRH, Prive GG, Bishop RE, Kay LE, Solution

structure and dynamics of the outer membrane enzyme PagP by NMR. Proc. Natl. Acad. Sei. U. S. A., 2002.

99(21): p. 13560-13565.

48. Fernandez C, Adeishvili K, Wuthrich K, Transverse relaxation-optimized NMR spectroscopy with the outer

membrane protein OmpX in dihexanoyl phosphatidylcholine micelles. Proc. Natl. Acad. Sei. U. S. A., 2001.

98(5): p. 2358-2363.


49. Fernandez C, Hilty C, Wider G, Wuthrich K, Lipid-protein interactions in DHPC micelles containing the

integral membrane protein OmpX investigated by NMR spectroscopy. Proc Natl. Acad. Sei. U. S. A., 2002.

99(21): p. 13533-13537.

50. Marley, J., M. Lu, and C. Bracken, A methodfor efficient isotopic labeling of recombinant proteins. J. Biomol.

NMR, 2001. 20(1): p. 71-75.

51. Ren, Q. and L. Thony-Meyer, Physical interaction of CcmC with heme and the heme chaperone CcmE during

cytochrome c maturation. J. Biol. Chem., 2001. 276(35): p. 32591-32596.

52. Goldberg ME, Expert-Bezancon N, Vuillard L, Rabilloud T., Non-detergent sulphobetaines: a new class of

molecules that facilitate in vitro protein renaturation. Folding and Design, 1995. 1(1): p. 21-27.

53. Chen GQ, Gouaux E, Overexpression of a glutamate receptor (GluR.2) ligand binding domain in Escherichia

coli; Application of a novel protein folding screen. Proc. Natl. Acad. Sei. U. S. A., 1997. 94(25): p. 13431-

13436.

54. Vuillard L, Rabilloud T, Goldberg ME, Interactions of non detergent sulfobetaines with early folding

intermediates facilitate in vitro protein renaturation. Eur. J. Biochem., 1998. 256(1): p. 128-135.

55. Armstrong N, De Lencastre A, Gouaux E, A new protein folding screen: Application to the ligand binding

domains of a glutamate and kainate receptor and to lysozyme and carbonic anhydrase. Protein Sei., 1999. 8(7):

p. 1475-1483.

56. Pervushin, K. V., Impact of TROSY as a technique in structural biology. Q. Rev. Biophys., 2000. 33(2): p. 161-

197.

57. Gardner, K. H., M. K. Rosen, and L. E. Kay, Global folds ofhighly deuterated, methyl-protonated proteins by

multidimensional NMR. Biochemistry, 1997. 36(6): p. 1389-1401.

58. Gardner, K. H. and L. E. Kay, The use of H-2, C-13, N-15 multidimensional NMR to study the structure and

dynamics ofproteins. Annu. Rev. Biophys. Biomolec Struct., 1998. 27: p. 357-406.

59. Leiting, B., F. Marsilio, and J. F. O'Connell, Predictable deuteration of recombinant proteins expressed in

Escherichia coli. Anal. Biochem., 1998. 265(2): p. 351-355.

60. Salzmann, M., et al., TROSY in triple-resonance experiments: New perspectivesfor sequential NMR assignment

oflarge proteins. Proc Nail. Acad. Sei. U. S. A., 1998. 95(23): p. 13585-13590.

61. Riek, R., K. Pervushin, and K. Wuthrich, TROSY and CRINEPT: NMR with large molecular and

supramolecular structures in solution. Trends Biochem.Sci., 2000. 25(10): p. 462-468.

62. Riek, R„ et al., Polarization transfer by cross-correlated relaxation in solution NMR with very large molecules.

Proc Natl. Acad. Sei. U. S. A„ 1999. 96(9): p. 4918-4923.

63. Salzmann, M., et al., NMR assignment and secondary structure determination of an octameric 110 kDa protein

using TROSY in triple resonance experiments. J. Am. Chem. Soc, 2000.

64. Kreishman-Deitrick M, Egile C, Hoyt DW, Ford JJ, Li R, Rosen MK, NMR analysis of methyl groups at 100-

500 kDa: Model systems andArp2/3 complex. Biochemistry, 2003. 42(28): p. 8579-8586.

65. Prestegard, J. H., H. M. Al-Hashimi, and J. R. Tolman, NMR structures of biomolecules using field oriented

media and residual dipolar couplings. Q. Rev. Biophys., 2000. 33(4): p. 371-424.


66. Gaponenko V, Howarth JW, Columbus L, Gasmi-Seabrook G, Yuan J, Hubbell WL, Rosevear PR, Protein

globalfold determination using site-directed spin and isotope labeling. Protein Sei., 2000. 9(2): p. 302-309.

67. Veglia G, Opella SJ, Lanthanide ion binding to adventitious sites aligns membrane proteins in micelles for

solution NMR spectroscopy. J. Am. Chem. Soc, 2000. 122(47): p. 11733-11734.

68. Vögeli B, Kovacs H, Pervushin K, Measurements of side-chain C-13-C-13 residual dipolar couplings in

uniformly deuterated proteins. J. Am. Chem. Soc, 2004. 126(8): p. 2414-2420.

69. Pautsch, A. and G. E. Schulz, Structure of the outer membrane protein A transmembrane domain. Nat. Struct.

Biol., 1998. 5(11): p. 1013-1017.

70. Koebnik, R., Structural andfunctional roles of the surface-exposed loops of the beta-barrel membrane protein

OmpAfrom Escherichia coli. J. Bacteriol., 1999. 181(12): p. 3688-3694.

71. Kleinschmidt, J. B., M. C. Wiener, and L. K. Tamm, Refolding of outer membrane, protein a in neutral or

zwilterionic detergents requires micelles and does not depend on the detailed chemical structure of the detergent.

Biophys. J., 1999. 76(1): p. A106-A106.

72. Arora, A., et al., Refolded outer membrane protein A of Escherichia coli forms ion channels with two

conductance states in planar lipid bilayers. J. Biol. Chem., 2000. 275(3): p. 1594-1600.

73. Box GEP, Hunter WG and Hunter JS, Statisticsfor experimenters. 1978.

74. Guntert, P., et al., Processing of Multidimensional Nmr Data with the New Software Prosa. J. Biomol. NMR,

1992. 2(6): p. 619-629.

75. Bartels, C, et al., The Program Xeasy for Computer-Supported Nmr Spectral-Analysis of Biological

Macromolecules. J. Biomol. NMR, 1995. 6(1): p. 1-10.

76. Wishart, D. S., et al., U-l, C-13 and N-15 Chemical-Shift Referencing in Biomolecular Nmr. J. Biomol. NMR,

1995. 6(2): p. 135-140.

77. Pervushin, K., The use of TROSY for detection and suppression of conformational exchange NMR line

broadening in biological macromolecules. J. Biomol. NMR, 2001.

78. Orekhov, V. Y., et al., Backbone Dynamics of (1-71 jBacterioopsin and (1- 36)Bacterioopsin Studied By 2-

Dimensional H-l-N-15 Nmr- Spectroscopy. J. Biomol. NMR, 1995. 6(2): p. 113-122.

79. Li, Y. C. and G. T. Montelione, Human type-alpha transforming growth factor undergoes slow conformational

exchange between multiple backbone conformations as characterized by nitrogen-15 relaxation measurements.

Biochemistry, 1995. 34(8): p. 2408-23.

80. Wishart, D. S., Sykes, B.D., The C-13 Chemical-Shift Index - A Simple Method For The Identification Of

Protein Secondary Structure Using C-13 Chemical-Shift Data. J. Biomol. NMR, 1994. 4(2): p. 171-180.

81. Schwede T, Kopp J, Gucx N, Peitsch MC, SWISS-MODEL: an automated protein homology-modeling sei-ver.

Nucleic Acids Res., 2003. 31(13): p. 3381-3385.

82. Banci, L., el al., Partial orientation of oxidized and reduced cytochrome b(5) at high magnetic fields: Magnetic

susceptibility anisotropy contributions and consequences for protein solution structure determination. J. Am.

Chem. Soc, 1998.120(49): p. 12903-12909.

Seite Leer /

Blank leaf

Chapter II

NMR studies of structure and function of FkpA

A part of the content was published in: Kaifeng Hu, Andreas Pliickthun and Konstantin

Pervushin,./. Biomol NMR (2004), 28 (4), 405-406.

40

Seite Leer /

Blank leaf

NMR studies ofstructure and function ofFkpA 41

II.l Introduction

FkpA is a 245-residue periplasmic peptidyl-prolyl cis/trans isomerase (PPIase) from E. coli with

a chaperone activity, which is induced by heat shock [1]. Previous studies have indicated that the

chaperone activity of FkpA is independent of its PPIase activity [2] as it is also observed for

proteins devoid of cis-prolines. Overexpression of FkpA suppresses the formation of inclusion

bodies from a defective folding variant of the maltose-binding protein and promotes the

reactivation of the denaturated citrate synthase [3], Coexpression of FkpA can dramatically

improve functional periplasmic production of single-chain fragments (scFv) of antibodies, even

those not containing ds-prolines [4J. The yield of soluble and functional scFv fragment was also

increased in vitro in the presence of stoichiometric amounts of FkpA The chaperone function of

FkpA is hypothesized to be due to its interaction with early folding intermediates preventing their

aggregation, and its ability to reactivate inactive proteins, possibly by binding to partially

unfolded species |2J.

FkpA was originally discovered as a periplasmic Escherichia coli homolog of the MIP-like

FK506-binding proteins [1]. The Macrophage Infectivity Potentiator protein from Legionella

pneumophila (LpMIP) exhibits PPIase activity. The 3D structures of both LpMlP [51 and FkpA

[6J were determined by X-ray crystallography. The crystal structure of FkpA indicates that FkpA

is a dimer comprising of two domains, the N-terminal dimcrization domain and the C-terminal

FK506-binding protein (FKBP) domain (Figure 2.1). FK506 is a natural peptidomacrolide from

Streptomyces tsukubaensis [7] [8] and has been known to inhibit this family of PPIases. The N-

terminal domain of FkpA includes three helices that are interlaced with those of the other subunit

to provide all the inter-subunit contacts maintaining the dimeric species.

In solution FkpA retains apparently as a dimeric organization as it evidenced in analytical gel

filtration |21 [9] and by ultracentrifugation experiments [31,. The overall form of the dimer is V-

shaped, and a comparison of the crystal structures of FkpA and of a mutant and the mutant

complex with FK506 reveals flexibility in the relative orientation of the two C-terminal domains

located at the extremities of the V [6J. Although different reports agreed that the chaperone

activity of FkpA is independent of its PPIase activity, there is still uncertainty regarding to where

NMR studies of structure and function of FkpA 42

(a)

Figure 2.1. (a) and (b) Stereo view of the complex formed between FkpA-ACT (residue 1-224) dimer and

the immunosuppressant FK506 in two orthogonal views. The structure is shown schematically in ribbon

and loop representation. The ß strands, present in the C-lerminal domains arc shown in yellow. Helical

regions are colored green and blue for the respective monomers. The bound FK506 molecules are shown

in red (Adapted from [61).

the chaperone function resides in the dimeric molecule [3, 6, 91. In Ramm's report, experiments

with the isolated domains of FkpA imply that both the isomerase and the haperone site are

mechanistically related and confined to the the highly conserved FKBP domain (C-terminus).

The additional amino-terminal domain appears to be only utilized to mediate the dimerization,

which places the two active sites of the FKBP domains in a juxtaposition, such that they can

simultaneously interact with one protein substrate [9J. However, in Saul's report [6], the deletion

mutant FkpNL (residues 15-114), consisting principally only of the three N-terminal helices and

existing in solution as a mixture of monomelic and dimeric species, as it is established by gel

filtration and ultracentrifugation experiments [31, still exhibited chaperone activity. By contrast, a

deletion mutant comprising the C-terminal domain only is monomcric, as it is indicated by

ultracentrifugation experiments [31, and although it shows PPIase activity, it is devoid of

chaperone function. Therefore, they suggested that the chaperone and catalytic activities reside in

the N and C terminal domains, respectively. Accordingly, they proposed that the observed elative

mobility of the two C-terminal domains of the dimeric molecule could effectively adapt these two

independent folding functions of FkpA to different polypeptide substrates [6].

NMR studies of structure and function ofFkpA 43

As the nature of molecular chaperone activity of FkpA and the characteristic of the polypeptide-

binding site of this dimeric chaperone have remained poorly defined, our NMR investigations are

therefore mainly aimed at a dynamic characterization of FkpA, its interactions with polypeptide

(protein) substrates and at obtaining an explanantion of the molecular mechanism of its

chaperone activity. Due to the availability of the high resolution X-ray structrue of FkpA and the

high conservation of the sequence for the domain containing PPIase activity, our efforts are not

directed towards establishing 3D strucutre of FkpA in solution but rather concentrate on

structural dynamical aspects of its chaperone function.

II.2 Sample preparation

The nearly complete backbone chemical shift assignment was achieved by thee combined use of

triple-resonance spectra of the mature 2H/15N/13C- labeled FkpA protein (245 amino acid residues)

and an engineered variant, the so-called "shortened FkpA" (sFkpA) obtained by the removal of

the first 9 N-terminal and 18-C terminal residues of the mature FkpA, which are dispensable for

its functionality [9], 1SN relaxation, residual dipolar couplings and substrate protein titrations

were measured with the 2H/15N- labeled sFkpA.

The unlabeled protein was prepared as described [2], and MALDI mass spectra were used to

identify the proteins. For the preparation of the uniformly 2H,15N and 13C-labeled FkpA, the E.

coli strain SB536 was freshly transformed with the plasmid pHB610-His6 and the cells were

gradually adapted to D20 solution by growing cells in Luria Broth prepared in 2/1 and then 1/1

H20/D20 mixtures. Cells adapted to D20 were used to innoculate the 2H,15N and 13C-labeled

Celtone medium (Spectra Stable Isotopes) at 37°C. Due to the constitutive expression of the

FkpA encoding gene under control of its own promoter, no induction of expression was

necessary. Cells were harvested after 60 hours of growth, normally at an OD6oo of about 1-1.2.

After lysis with a French press and centrifugation, the soluble supernatant was passed over an

immobilized metal ion affinity chromatography column (IMAC, with Ni-charged Porös MC

(PerSeptive Biosystems) material on a BioCAD workstation) at pH 7.0, followed by S/Ft cation-

exchange chromatography (Poros HS/M column on a BioCAD workstation) at pH 6.0 in 20 mM

Mes buffer. FkpA elutes in one broad and one sharp peak, probably corresponding to the

monomelic and the dimeric species, respectively. After buffer exchange using 20mM Mes buffer


pH6.0 with 50mM NaCI (NMR buffer) and back exchange of amide deuterons with protons from

water, the two samples of the presumed monomer and dimer fraction showed identical NMR

spectra and thus were combined. The final NMR sample concentration was 0.6 mM of FkpA per

monomer in 20 mM Mes buffer at pH 6.0 with 50 mM NaCI.

The "shortened" sFkpA (residues 10-227) gene, also containing the C-terminal His6-tag was

subcloned into the vector pTFT74 under the control of the phage T7 promoter, and the signal

sequence was removed for cytoplasmic expression. Overproduction of sFkpA was carried out in

the E. coli strain BL21(DE3)pLysS at 25°C. After two-step adaptation of cells to D20 as

described above, expression in M9 D20 minimal media supplemented with 1SNH4C1 (99% l5N, 1

g/L) (For 2H/15N- labeled sFkpA ) or I5NH4C1 (99% 15N, 1 g/L) and l3C6-glucose (99% 13C, 4 g/L)

(Cambridge Isotope Laboratories) (For 2H/15N/13C - labeled sFkpA ) was induced by 1 mM

IPTG at OD6oo ~ 0.3 and the cells were harvested after 60 hours at an OD600 of about 1-1.2. The

purification is similar to that of FkpA as described above, except running S/H cation-exchange

chromatography at pH 4.7 in 20 mM Mes buffer.The 2H/13N/l3C-labeled NMR sample of

sFkpA protein was 0.6 mM per monomer in 20 mM Mes buffer at pH 6.0 with 50 mM NaCI, and

the 2H/15N-labeled NMR sample of sFkpA protein was 0.8 mM per monomer in 20 mM Mes

buffer at pH 6.0 with 50 mM NaCI.

II.3 Backbone chemical shift assignment and secondary structure

determination

The assignment of chemical shifts to specific sites in a macromolecule forms the basis for NMR

studies of its structure and function [10|. Among all possible resonances the backbone *H, 15N,

13Ca, 13C and 13CP chemical shift assignments of the 2H, ,5N and 13C labeled FkpA (sFkpA) in a

functional dimeric form in solution are the most informative and constitute a basis for detailed

NMR studies of its dynamic properties, its interactions with polypeptide (protein) substrates and

the molecular mechanism of its chaperone activity.

II.3.1 NMR spectroscopy and backbone chemical shift assignments

All NMR spectra of both constructs, FkpA (residues 1-245) and sFkpA (residues 10-227), were

acquired using transverse relaxation optimized spectroscopy (TROSY) [11] at 25"C and 37"C on


y -y

'HHi<t>i

15N *i

<h <l>4

-xl-xtac

*Hfr4 I éiUll

13c-a î î* MÎ I

13/

G1 Gj

V^VJ SEDUCE-1

H WALTZ16

PFG| | III 1

G2 G2 G,

Figure 2.2 Pulse sequences of 3D TROSY-HNCA with high resolution along the 15N dimension. The

radio-frequency pulses on 'H, 13Ca, 15N, l3CO, and 2H are applied at 4.7, 54.6, 119, 173.6 and 3.5 ppm,

respectively. Narrow and wide black bars indicate nonselective 90° and 180° pulses. Water suppression is

achieved by watergate. 13CO decoupling is achieved with SEDUCE-1 at a field strength ofyB2 = 1.65 kHz.

Sine bell shapes on the line marked 'H indicate water selective 90° pulses. The line marked PFG indicates

the duration and strength of pulsed magnetic field gradients applied along the z-axis: d: 0.7ms, 25 G/cm;

G2: 0.8ms, 75 G/cm; G3: 0.7ms, 45 G/cm. The delays are 7/= 12 ms, x, = 2.7ms. The phase cycle is: fa=

{y, -y, x, -x}, 4>2= {4x, 4(-x)}, <t>3 = {-y}; fa = {y}; fa = {-y}; *« = {y, -y, -x, x, -y, y, x, -x}. All other

radio-frequency pulses are applied with phase x except indicated. A phase-sensitive spectrum in the N (t{)

dimension is obtained by recording a second FID for each t\ value, with <j)i = {y, -y, -x, x}, fa = {y}, fa =

{-y} and fa = {y}. Quadrature detection in the 13C" (t2) dimension is achieved by the States-TPPI method

applied to the phase <t>2.

a Bruker Avance 600 MHz spectrometer equipped with a TXI cryogenic probehead and Bruker

Avance 800 and 900 MHz spectrometers. The experiments include 2D ['H,15N] TROSY, 3D

TROSY-versions of HNCA (Figure 2.2) ,HNCACB, HNCO and HN(CA)CO run in the high

resolution mode, utilizing both N->C and C->N polarization transfer periods to simutaneously


Figure 2.3 Scheme showing unambiguous residue specific backbone assignment achieved through

alignment of chemical shifts of 1) Cain TROSY-HNCA, 2) Ca/ Cß in TROSY-HNCACB, 3) CO in

TROSY-HNCO / TROSY-HN(CA)CO, and 4) HN in TROSY-NOESY spectra of the uniformly

^''C^N-labeled dimeric sFkpA sample.

frequency label the 15N spins to maximize spectral resolution along the 15N dimension [12, 13],

and 3D 15N resolved TROSY-NOESY (imix = 150 ms) [14]. The NMR data were processed using

the PROSA software [15J and analysed using XEASY [16] and CARA (http://www.nmr.ch/).

The 'H chemical shifts were referenced to the DSS signal and the 13C and 15N chemical shifts

were referenced indirectly using the 13C/lH and 15N/'h gyromagnetic ratios [17].

To assign the resonances unambiguously, we exploited all possible chemical shifts of the

backbone, Ca, Ö, CO and HN as fragment alignment criteria to build up sequential connetivities.

The H, N, C, i3Ca and i3C^ resonances were sequence-specifically assigned by

identification of sequentially connected [W^N] fragments of the TROSY- HNCA, TROSY-

HNCACB, TROSY-HNCO and TROSY-HN(CA)CO spectra supported by the TROSY-NOESY

connectivities (Figure 2.3), followed by global fragment mapping with the program MAPPER

[18]. With this approach ca. 60% of all resonances of the full length FkpA were sequence-

specifically assigned at 25°C. The assigned resonances were mostly located in the FKBP domain

(residue 139-224) and both N- (residues 3-14) and C- (residues 225-245) termini. The resonances

contributed by the N- and C-terminal regions 3-14 and 225-245 exhibited predominantly random

coil chemical shifts and were typically stronger by an order of magnitude than the resonances

stemming from the structured N-terminal and FKBP domains. This prevented further progress in

the assignment of resonances in the mostly a-helical N-terminal dimerization domain due to the


(a) 3D TROSY-HNCA

155 A56 G57 V58 Q59 D60

117 45 124 32

-**4D

106 29 1'2T'52

22Ä

117 19

""*|*fc

120 41 -45.0

50.0

-®

(b) 3D 15N TROSY-NOESY

155 A56 G57 V58 Q59 D60

nC

[ppm]

60.0

65 0

7.66 7.70 7.93 7.37 8.30 9.00

]HN [ppm]

7.64 7.70 7.94 7.37 8.30 9 00

'HN [ppm]

Figure 2. 4 (A) Stnps corresponding to the assignment of the residues 55-60 of sFkpA, in the TROSY-

HNCA experiment recorded at 900MHz, 37oC. Dashed lines indicate the sequential connectivities (B)

Stnps corresponding to the residues 55-60 of sFkpA, in the 15N resolved TROSY-NOESY cxpenment

recorded at 900MHz, 37oC The NOFSY pattern signifies clearly, the a-hehx secondary structure in this

region Diagonal peaks are marked with astensks

proximity of the 'H and i5N chemical shifts in a-helical and disordered structures. The other

problem hampering the assignment was a significant inhomogeneity of the line shapes of

resonances from the a-helical parts of FkpA, probably reflecting conformational exchange

processes in the milliseconds to seconds time scale for this domain.

The removal of the first 9 N-terminal and 18 C-terminal residues in the so-called "shortened

FkpA" (sFkpA), together with an increase of the temperature to 37°C and the use of the ultra high

field NMR spectrometer operating at 900 MHz permitted largerly completed (94%) the backbone

resonance assignment of sFkpA. The a-helical regions were effectively assigned by a

combination of the most sensitive experiments namely the 3D TROSY-HNCA and 3D TROSY-


.« V^fc,«»-

,„ o.o mir *

119.5

M^y »i9i mi ^y^,

<*« #" •*

J. 124.0

8.60 8.20 7.80

'HN [ppm]

10.0 9.0 HN [ppm] 70

Figure 2.5 2D [15N,'HN]-TR0SY spectrum of sFkpA (residues 10-227) measured using the Avance 900

spectrometer at 37"C. The cross-peaks are annotated according to residue numbering of the full length

FkpA. The cross-peaks stemming from the guanidinum group of Arg side-chains are labeled with asterisks.

The cross-peaks marked with open triangles are degradation products which appear with the aging of the

protein samples. An expansion of the spectrum showing resonances of the residues mostly in the a-hclical

regions is displayed in the insert.

NOESY as shown in Figure 2. 4 A and B. The spectra of sFkpA were then aligned and compared

to the corresponding spectra of FkpA showing in most cases no significant perturbations of

chemical shifts. This implies that the removal of the first 9 N-terminal and 18-C terminal residues

and the increase of the temperature does not affect the structure of FkpA. This facilitated the

assignment of the backbone resonances of FkpA (Appendix Table 2). Figure 2.5 shows the 2D

['Hn,15N] TROSY spectrum of sFkpA. The sequence-specifically assigned cross-peaks are

annotated according to the residue numbers in FkpA. The residues for which the assignment was

not unequivocally established, are predominantly located in the a-helical region of the N-

terminal domain (Figure 2.5), mostly due to weak spectral intensities and large line width,

presumably due to the conformational exchange processes on the chemical shift timescale.


11.3.2 Analysis of secondary structure

In Figure 2.6, the polypeptide sequence of FkpA from E. coli is aligned with the sequence of the

MIP from Legionella pneumophila, a PPIase for which the crystal structure was determined

(PDB code 1FD9) [5], Despite the modest sequence identity of 34.6%, the secondary structure

elements closely correspond to each other in the aligned sequences of both proteins. The

secondary structure elements in FkpA are identified by using the chemical shift index (CSI)

based on the experimental 13Ca chemical shifts [19J. A comparison of the secondary structure

perdicted from the 13Ca chemical shifts to those of LpMIP (Macrophage Infectivity Potentiator

PDB code 1FD9) [5] and the recently published X-ray structure of FkpA[6] shows a good

agreement.

These proteins are surface components of intracellularly replicating bacteria and are required for

an efficient injection into the host cell. An analysis of the NMR data indicates that the long a-

helix (Helix 3, residues 70-111) connecting the FKBP and N-terminal dimerization domains

might be dynamically disturbed in the area located between residues 84 and 91, resulting in

significant broadening of the corresponding backbone resonances. These dynamic properties of

the protein might play an important role in the chaperone activity of FkpA, where binding to

different substrates would require some structural adaptations of the chaperone. The backbone

assignment should serve as a platform to investigate details of dynamic properties and chaperone

activities of FkpA.

II.4 15N Relaxation

The structures of FkpA, a C-terminal truncation mutant FkpA-ACT (resiude 1-224) and FkpA-

ACT in complex with FK506 as determined by X-ray crystallography published recently [6]

reveal flexibility in the relative orientation of the two C-terminal domains of FkpA. While the

structure was solved for FkpA, the mechanism of sequence-specific binding in its chaperone

acitivity remains unclear. As dynamic properties of FkpA might play an important role in its

chaperone activity, which allow FkpA to bind to different substrates through structural

adaptations, our first interest is investigating details of the dynamic properties of FkpA in

solution by NMR.

NMR studies ofstructure and function of FkpA 50

Ec. FkpA X-Ray

HI

Ec. sFkpANMRFkpA 10 adskaafknddqksayalgaHgry^nslk||eklgikldkdqliagvqdafadks-klMIP 1 ATDATSLATDKDKLSYSIGADLGKNFKNQ GIDVNPEAMAKGMQDAMSGAQLAL

Lp. MIP X-Ray ^^^^^^H^H

Ec. FkpA X-Ray

Ec. sFkpANMR

H3 ßl ß2

FkpA 69 SDQEIEQTLQAFEAP§KSS0S§MEKDAADNEAKGKEYREKFAKEKGVKTSSTGLVYQVVEMIP 54 TEQQMKDVLNKFQKDLMAKRTAEFNKKADENKVKGEAFLTENKNKPGVVVLPSGLQYKVIN

Lp. MIP X-Ray ^MM^^HM^HMH^BHl^^nMHH ^~ ^^-

ß3 ß4a H4 ß4b H5

Ec. FkpA X-Ray

Ec. sFkpA NMR ,. r

FkpA 130 AGKGEAPKDSDTWVNYKGTLIDGKE&NSYTRGEPLSFRLDGVIPGWTEGLKNIKKMIP 115 SGNGVKPGKSDTVTVEYTGRLIDGTVFDSTEKTGKPATFQVSQVIPGWTEALQLMPA

Lp. MIP X-Ray"

ß6Ec. FkpA X-Ray

Ec. sFkpANMR w w ^»-

FkpA 187 GGKIKLVIPPELAYGKAGVPG-IPP§STLVFDVELLDVKPAPMIP 172 GSTWEIYVPSGLAYGPRSVGGPIGPNETLIFKIHLÏSVKKSS

Lp. MIP X-Ray

Figure 2.6 Comparison of secondary structures of FkpA with LpMIP. Sequence alignment of

FkpA(scquence in upper line) and MIP (sequence in lower line) (Macrophage Infectivity Potentiator, PDB

code 1FD9) using the SIM program (ExPASy, http://www.expasv.ch). The secondary structure elements

of FkpA shown in the top and second row are from the X-ray structure and predicted the from NMR

chemical shift index (CSI) analysis, respectively. Corresponding secondary structure elements of MIP are

shown in the bottom row. Filled boxes represent a-helices and arrows indicate ß sheets. Unassigned

amides of FkpA are enclosed in rectangular boxes.

II.4.1 NMR experiments

The pulse sequences used to record 15N Ti and T2 values are shown in Figure 2.7 a and b,

respectively, and the scheme for measurement of the steady-state 'H-15N NOE is indicated in

Figure 2.7 c. We employed the TROSY version in all three pulses sequences with the use of


pulsed field gradients in order to suppress the intense solvent resonance, and select for the

coherence transfer pathway. 15N Ti and T2 and 'H-15N NOE spectra were recorded at 37°C on a

Bruker Avance 600 MHz spectrometer equipped with a TXI cryogenic probehead, with 32 scans

for Ti and T2 experiments and 64 scans for 'H-15N NOE experiments.

(a)

EH

<h y -y y

1111

15N tiItiItiItJi*"*+HtttJBBU*iist*ft»WTWi*h<ih*JBfc*«»»»»*4w»

PFG ili I

Seduce 1

¥s

Gl &\ G4 Gb^T!-^

§4 fa

XX >"Äl**«*****wt*»

G^ Ga Gg G3

W<h y -y y

«nu

15N Til Ti I Ti|1PFG ±il

Seduce-1

WMWWWdMwniiililiiiHi i iTïrniiiili

fc

«H g, p. I^^^JCs

LUI

B MiWmwsgftWftm i uni iJ|i|im't«dWWW»«M»ih*ii»JWk

i i-i "hJ»^

Gj G2 G3 Gi

(c)

1Sain

Hal

Lock,,

PFG iLG| Gj

Mi

JUislu^nx

I'1 TiFihi|fJ1111111111 iiïiiwhhkim h

i\t(tmhJÊûtmMt*t*tJftLttttJÊir

nilGj G2 Gj G3


Figure 2.7: TROSY version pulse sequences for the measurement of (a) ISN T,, (b) 15N T2, and (c) 'H-15N

NOE. In all experiments, the radio-frequency pulses on 'H, l5N are applied at 4.7 and 119 ppm,

respectively. Narrow and wide black bars indicate nonselective 90° and 180° pulses. Delay ti = 2.7 ms.

Water suppression is achieved by a Watergate method using 1 ms water selective pulses. In (a) 13N Tls (b)

15N T2, the phase cycle is: fa = {8x, 8(-x)}, fa= {4x, 4(-x)}, fa= {y, -y, -x, x}, fa = {-y}; fa = {y};fa =

{-y}; fate = {y, -y, -x, x, -y, y, x, -x, -y, y, x, -x, y, -y, -x, x }. In (c) 'H-^N NOE, the phase cycle is: fa=

{y, -y, -x, x}, fa = {-y}; fa = {y}; fa - {-y}; fa.c = {y, -y, -x, x}. All other radio-frequency pulses are

applied with the phase x except indicated otherwise. A phase-sensitive spectrum in the 15N (t{) dimension

is obtained by recording a second FID for each t\ value, with fa = {y, -y, x, -x}, fa - (y), fa - {-y) and fa

= {y}. The line marked PFG indicates the duration and strength of pulsed magnetic field gradients applied

along the z-axis: d: 0.5ms, 19 G/cm; G2: 0.5ms, 30 G/em; G3: 0.5ms, 65 G/cm, G4: 1ms, 90 G/cm, G5:

lms, 90 G/cm. During Ti and T2, to eliminate the effects of cross-correlation between 'H-15N dipolar

coupling and chemical shift anisotropy relaxation, decoupling on 'H is applied with the SEDUCE-1

profile at the field strength of yB2 = 3.61 kHz with the offset 2583Hz (i.e. 9ppm on resonance). In (c) 'H-

N NOE, N 90" pulse and spin lock in a combination with the strong gradients are used to clean up 15N

magnetization. An overall delay between scans of 5 s was employed in both NOE and NONOE

experiments. 'H saturation is achieved by the application of 120° pulses spaced at 5-ms intervals for 3 s

prior to the 15N pulse with phase fa.

N Ti values were measured from the spectra recorded with ten different durations of the delay T:

Ti = 20, 90, 160, 200, 300, 400, 600, 800, 1000, and 1200 ms. T2 values were determined from

spectra recorded with delays T2 = 2, 8, 12, 24, 30, 35, 40, 50, 70, and 90 ms. In order to eliminate

the effects of cross-correlation between 'H-15N dipolar and 15N CSA relaxation mechanisms, 'H

were decoupled during the T relaxation times.

Relaxation delays of 1 s were employed in the measurement of 15N Ti and T2 values. The 'H-15N

steady-state NOE values are obtained by recording spectra with (NOE experiment) and without

(NONOE experiment) the use of the lU saturation applied before the start of the experiment. In

order to ensure a clean steady-state 'H-15N NOE effect and that magnetization originates on 15N

in both the NOE and NONOE experiments, the sequence begins with a 90° 13N pulse followed by

the application of a field gradient pulse to dephase any magnetization in the transverse plane, and

then magnetization is locked in the transverse plane and dephased again. The level of water


suppression in the spectra recorded without *H saturation is often significantly worse than that in

the H saturation case and can lead to difficulties in obtaining accurate values for peak intensities.

Moreover, any saturation of protons prior to the start of the NONOE experiment gives rise to a

small truncated NOE effect [20]. Therefore, the suppression of the strong H20 resonance is

achieved via the use of the Watergate. The NOE values were determined from the spectra

recorded in the presence and absence of a proton presaturation period of 3 s. 'H saturation was

achieved with the use of 120° pulses applied every 5 ms. In the case of the NONOE spectra, a net

relaxation delay of 5 s was employed, while a relaxation delay of 2 s prior to a 3-s proton

presaturation period was employed for the NOE spectra.

II.4.2 Data processing and analysis

All spectra were recorded with 60 x 512 complex matrices with spectral widths of 1500 and 9600

Hz employed in F\ and F2, respectively. The NMR data were processed using the PROSA

software [15] and analysed using XEASY [16]. All spectra were processed identically with sine

75° apodization functions applied in both dimensions. NOE and NONOE spectra are scaled by

the same factor. The intensities of the peaks in the 2D spectra were described by peak heights as

determined by the peak-picking package in XEASY[16].

The values of R| and R2 were determined by fitting the measured peak heights to a single-

exponential two-parameter decay function of the form:

7(0 = /0exp(-Äli20 (2.1)

where I(t) is the intensity after a delay of time t and I0 is the intensity at time t = 0. Figure 2.8

illustrates examples of the exponential fit to the measured longitudinal and transverse relaxation

data for sFkpA. The steady-state NOE values were determined from the ratios of the intensities of

the peaks with and without proton saturation.

The values of the relaxation parameters, Ri, R2 and NOE, are shown in Figure 2.9 for sFkpA.

Figure 2.9 displays the variations in the R| and R2 values as a function of the residue number. It

is obvious that the average value of Ri for N-terminus (calculated for residues 10-112, 0.85 S"1) is


(a) (b)

6.00E+03

0.2 04 0.6 0£

Tims (S)

T2

.26

»27

62

102

1173

£0 40

Time (mS)

Figure 2.8 Examples of (a) ^ and (b) T2 decay curves for Ala 26, Leu 27, Phe 62, Lys 102, and Val 173.

The curves indicate best fits to single-exponential decays.

100 150

Residue

ce

(C)

• *

100 150

Residue

250

Residue

Figure 2.9 Plots of (a) Rb (b) R2, (e) R^R,, and (d) NOE of sFkpA as a function of residue number.

significantly different from that for C- terminus (calculated for residues 113-226, 1.04 S"1). The

R2 values show similar sequential variations to those described for Ri above, with the average

value of R2 for the N-terminal domains, 25.8 S"1 and that for the C- terminal domains, 18.6 S"1.

Another feature that is seen from the overall pattern in R2 values is that they are more dispersive

in the N-terminal domains whereas they are more homogenous in the C-terminal domains. This

may indicate that there is different intramolecular mobility in N-terminal and C-terminal domains,


in fast (ns) and slow (ms) motional regimes reflected in Rj and R2 values. These are even more

clear when R2/Ri ratios are plotted as a function of residue number as in Figure 2.9 c. The steady-

state NOEs determined from the ratio of the peak height in NOE and NONOE spectra are plotted

in Figure 2.9 d. As low NOE values usually indicate a local high mobility, some residues within

the long a-helix (Helix 3, residues 70-111), such as Ala 90, Met 92, Glu 93 and Asp 95, with

very low NOE values (smaller that 0.5) may implicate that there is significantly high flexibility in

this long helix. This could also serve as an explanation for the interuption of our backbone

assignment in the area located between residues 84 and 91. The flexibility of the long a-helix

(Helix 3) might result in significant broadening of the corresponding backbone resonances

hampering the resonance assignment process.

As described in [20], an estimate of the overall correlation time xm may be determined from the

R2/Ri ratio for each residue. All residues with R2/Ri ratios greater than one standard deviation

from the mean were usually discarded in the evaluations of %m. Since the overall correlation time

of a molecule should be approximately proportional to its molecular weight, the molecular weight

of sFkpA can then be calculated on the basis of the estimated overall correlation time Tm using

Ubiquitin as reference upon adjustment of the difference in the temperature of measurement [21].

Table 2.1 shows approximate molecular weights of sFkpA calculated from experimental data of

N-terminal residues, C-terminal residues, and all residues, respectively.

Table 2.1 Estimates of the overall correlation time xm and approximate MW of sFkpA

(Theoretical molecular weight for sFkpA dimer is: 24.4 x 2 Kda)

Average

R2/R1

Stdev Estimated

tm (ns)

Estimated

MW (Kda)

N-term(Res. 10-112) 36.21 26.85 28.3 61.1

C-term(Res. 113-226) 18.39 7.96 17.4 37.5

Overall 26.86 21.29 18.8 40.5

The overestimation of MW of sFkpA calculated from experimental data of N-terminal residues

could be caused by the significant contribution of the conformational exchange to R2. A slight

underestimation of MW of sFkpA calculated from experimental data of all residues can be

explained by the significant intramolecular mobility of the FKBP domain relative to the


dimerization domain. The most likely explanation for a further lower estimation of MW of

sFkpA calculated from the experimental data of the C-terminal residues is that there is a relative

intramolecular mobility between the two C-terminal domains, and that they behave somewhat

like independent single globular monomeric proteins within the dimeric molecular frame. The

dispersive R2 values in the N-terminal domains may indicate that the N-terminal domain is rather

flexible in the very broad time scale.

II.5 Residual dipolar coupling and the dynamic molecular model

The introduction of residual dipolar coupling methodology has broadened the scope of structural

biological problems which can be addressed by NMR spectroscopy[22]. Residual dipolar

couplings (RDCs) represent an extremely valuable source of long-range angle information for

solution structure determinations of macromolecules by NMR spectroscopy which cannot be

obtained otherwise [23], In addition, RDCs can also be employed to determine the relative

orientation of domains and study protein dynamics [24, 25], conformational changes, and

structure of protein-protein complexes [26] [27]. The development of the residual dipolar

coupling methodology for rapid recognition of homologous protein folds and for studies of

submillisecond timescale dynamics has also seen considerable progress [28],

II.5.1 Residual dipolar coupling measurement

As dipolar couplings generally average out in an isotropic solution, an anisotropic environment is

necessary for observation of these dipolar couplings. Although high degrees of ordering can

generate strong observable dipolar interactions, their complexity makes these spectra difficult to

interpret. Thus, to make use of the residual dipolar couplings, the tumbling of the molecule must

be perturbed such that it becomes very slightly anisotropic, with the dipolar couplings (so-called

residual dipolar coupling) almost, but not completely, averaged to zero.

Methods have been proposed for generating partial and tunable degrees of alignment of

macromolecules, including dissolving macromolecules in an anisotropic solution of magnetically

aligned liquid crystalline bicelles [231 or lyotropic planar liquid crystalline bilayers [29], by


addition of magnetically aligned Pf 1 filamentous bacteriophages as a cosolute [30] or in presence

of the purple membrane of the membrane protein bacteriorhodopsin [31].

Here, we measured residual dipolar couplings (RDCs) for the 'H-15N amide groups on 0.8 mM

7 1 *î

H, N-labeled sFkpA sample either by an addition of magnetically aligned Pfl filamentous

bacteriophages or in a dilute liquid crystalline phase composed of Ci2E5 / hexanol/H20[29] and

characterized the dynamic properties of FkpA by fitting the alignment tensors [31].

II.5.1.1 Measurements of RDCs by addition of Pfl filamentous bacteriophages

The bacteriophage Pfl consists of a 7,349-nucleotide single-stranded circular DNA genome that

is packaged at a -1:1 nucleotide:coat protein ratio into a -60 Â diameter by -20,000 Â long

particle that has a negatively charged surface (pi -4.0). The coat protein forms an a-helical

structure that runs roughly parallel to the long axis of the phage. The coat proteins form a

repeating network of carbonyl groups that are believed to be the source of the phage's large

anisotropic magnetic susceptibility, with the long axis of the phage aligning parallel to the

magnetic field. It was demonstrated that the phage are already fully aligned at B0 = 7.5 Tesla [30],

The solution of oriented phage particles forms a liquid crystalline medium that can impart

alignment of biomolecular cosolutes. For some RNA, DNA and protein systems, the Pfl phage

appears to align the macromolecule by a steric mechanism with no evidence of any binding of the

macromolecules to the phage. In this case, the macromolecule does not bind to the phage but

instead alignment is induced by collisions. Alignment has also been reported through binding

interaction with phages, however, there are no significant structural changes in the

macromolecule in the presence of phages. The proteins are in fast exchange between the free and

phage-bound forms. The phages do not affect the microscopic rotational correlation times of

individual macromolecules dissolved in the phage solution in spite of the fact that they lead to a

high macroscopic viscosity [30].

The magnetic alignment of the Pfl phages was monitored by observation of the 2H quadrupolar

splitting in a ID H NMR spectrum of the water as a function of the phage concentration (Figure

2.10 B). The splitting of the HOD signal arises from the large deuterium quadrupolar moment


Figure 2.10 (A) TROSY (red) and H-anti-TROSY (black) spectra (recorded at 37"C at 900 MHz) of 0.8

mM 2H, 15N-labeled sFkpA aligned with the Pfl filamentous bacteriohages by monitoring (B) 2H

quadrupolar splitting in ID 2H NMR spectrum of the solvent. (C) An expanded portion shows the effect of

phage on the splitting between the TROSY (red) and H-anti-TROSY(black) components. (D) ID slice cut

from (C) at the dotted line shows fitting of Lorentzian lineshapes on the top of the peaks to obtain more

precise measurements of the splitting between the TROSY (red) and H-anti-TROSY (black) components.

that is not isotropically averaged for water bound to the aligned phage particles. The observed

quadrupolar splitting varies approximately linearly with the phage concentration (up to 60 mg/ml

of phage). The TROSY and H-anti-TROSY spectra (Figure 2.10 A) were recorded using the 0.8

mM 2H, ^N-labeled sFkpA in 20 mM Mes buffer at pH 6.0 with 50 mM NaCI at 37°C using a

Bruker Avance 900 MHz spectrometer, equipped with a cryogenic Z-gradient TXI probe. The

sample with the lightly aligned protein was obtained by titration of the Pfl phage from a 50

mg/mL stock solution until the observed quadrupolar deuterium splitting of the solvent signal

was about 4.4 Hz. RDCs were obtained by calculating the changes (which could be either


positive or negative) of the frequency difference (splitting) between the TROSY and H-anti-

TROSY components upon the addition of the Pfl phages.

As sFkpA is in homodimeric form, in an isotropic solution (before addition of the Pfl phages),

each single peak, both TROSY and H-anti-TROSY, can be assumed as an equally weighted

superposition of chemical shifts from both subunits. When the protein is weakly aligned in an

anisotropic solution (after the addition of Pfl phage), it could concomitantly induce a slightly

non-equivalent chemical environment between these two subunits. However, this small

difference in chemical shift is usually averaged by the fast tumbling of the macromolecule. In

most cases, an almost complete overlap of the chemical shifts from the two subunits appears as a

single broadened peak [32, 33]. In our experiments, we measured the chemical shifts with high

precision at the center of the broadened peaks by fitting of Lorentzian lineshapes to the sum of

the overlapped peaks, even for weak peaks with very low signal-to-noise ratio as shown in

(Figure 2.10 D). For a homodimeric protein, such as sFkpA, it is reasonable to assume that two

subunits are equally weighted in the overlapped peak and thus the center of the apparent single

broaden peak occurs exactly at the average of the chemical shifts of the two subunits, which is

given by equation 2.2:

J + D,-(J + D')+2(y + D')(2-2)

Dl +Drand Deiip=__ (2.3)

Measured RDCs on sFkpA aligned with the Pfl filamentous bacteriophages are shown in Table

2.2. Figure 2.11 shows the statistic distribution of the measured RDCs.

Figure 2.11 Histogram of RDCs measured with sFkpA aligned with the Pfl filamentous bacteriophage


Table 2.2 Measured RDCs on sFkpA aligned with Pfl filamentous bacteriophages at 37°C

Residue Pfl_RDC Residue Pfl RDC Residue Pfl RDC

18 -9.9 107 -6.3 169 -2.7

20 -1.8 109 0.9 170 26.1

21 0.0 110 0.0 172 -15.3

22 -12.6 111 1.8 173 22.5

24 0.9 112 26.1 174 2.7

25 -9.0 116 0.9 176 20.7

26 -14.4 117 -11.7 180 17.1

27 -9.9 118 0.0 182 7.2

28 -0.9 119 15.3 183 27.0

29 -11.7 121 -0.9 184 -25.2

34 0.9 123 27.0 185 -10.8

43 2.7 124 8.1 186 19.8

44 -10.8 125 0.9 187 -5.4

45 -7.2 126 -5.4 188 2.7

46 -11.7 127 -8.1 189 -0.9

47 6.3 137 -10.8 190 -12.6

50 -16.2 138 -3.6 191 -2.7

53 2.7 139 24.3 192 0.9

54 -23.4 141 34.2 193 10.8

56 -20.7 142 1.8 197 -17.1

58 -18.0 144 -6.3 198 9.9

67 -12.6 145 -4.5 199 0.0

68 -15.3 146 4.5 201 -4.5

69 -15.3 147 2.7 202 15.3

70 -16.2 148 -1.8 204 20.7

71 -8.1 149 20.7 205 -5.4

74 1.8 150 10.8 207 19.8

75 -4.5 151 -22.5 208 -19.8

81 -19.8 152 -15.3 212 -14.4

82 -12.6 153 13.5 213 27.0

95 -18.0 154 -10.8 214 26.1

96 -1.8 155 12.6 215 24.3

98 -9.0 157 -18.9 216 1.8

100 -2.7 158 -17.1 218 -10.8

101 5.4 159 -8.1 219 -7.2

102 -7.2 160 -7.2 220 -17.1

103 -15.3 161 -12.6 221 3.6

104 2.7 166 -20.7 222 8.1

105 4.5 168 0.9 224 17.1

Data are divided into N-terminus (in bold, residues 18 - 112) and C-terminus (FKBP domains, residues

116 - 224). RDCs from flexible parts are already excluded.


II.5.1.2 Measurements of RDCs in La phase

ra-Alkyl-poly(ethylene glycol)/n-alkyl alcohol and glucopone/rc-hexanol mixtures are shown to

form dilute liquid crystalline phases in an aqueous solution and were introduced as a suitable

medium for partial alignment of biological macromolecules in a magnetic field [29]. The

poly(ethylene glycol)-based systems are uncharged and thus insensitive with respect to pH

variation, only slightly sensitive to salt, and tolerant to high protein concentrations. Different

alkyl-poly(ethylene glycol) molecules are denoted CmEn, where m is the number of carbons in

the rc-alkyl group and n is the number of glycol units in the poly(ethylene glycol) moiety. Under

certain conditions, the mixture of w-Alkyl-poly(ethylene glycol)/«-alkyl alcohol systems form a

lyotropic liquid crystalline phase referred to as La, where the hydrophobic n-alkyl chains

aggregate into planar bilayers with the hydrophilic poly(ethyleneglycol) headgroups directed into

the water phase. The La phase is optically clear with slight opalescence, forming a lamellar-like

superstructure where the spacing of the stacked bilayers and thus the degree of alignment of guest

molecules in the water-rich layers can be tuned by the surfactant/water ratio. In a magnet, the

bilayer surfaces are oriented parallel to the direction of the magnetic field, presumably with a

superstructure in which the bilayers bend into a set of concentric tubes of different radii with the

axis aligned along the magnetic field. The La phase is relatively viscous, immobilizing any small

air bubbles present in solution. However, they can be readily removed by centrifugation.

In our experiment, a C12E5/n-hexanol/ H20 system was chosen to form the La phase for partial

alignment of sFkpA in a magnetic field. To obtain a sample with the aligned protein, 15%(w/v)

concentrated liquid crystal composing of C|2E5/n-hexanol in 90% H2O/10%D2O was first

prepared under vigorous shaking and then the protein sample was carefully added in during

further vigorous shaking to form a suitable final concentration (around 4%) of the liquid crystal.

In the case that addition of the protein leads to a phase separation, more n-hexanol should be

added to the mixture to re-establish the liquid crystal. Just as with the measurement of RDCs on

sFkpA aligned with Pfl phage, TROSY and H-anti-TROSY spectra (Figrue 2.12 A) were

recorded on 0.8 mM 2H, 15N-labeled sFkpA in 20 mM Mes buffer at pH 6.0 with 50 mM NaCI at

25°C at 900 MHz in the presence and absence of the liquid crystal of Ci2E5/n-hexano1/ H20.

Deuterium quadrupolar splitting of the solvent signal in ID 2H NMR spectrum, finally adjusted


to 18.6 Hz (Figrue 2.12 B), was observed to tune the degree of alignment of sFkpA. RDCs were

calculated in the same way as above.

A) B)

1 18.5

15.0 0 '5-0 2II[Hz]

1 19.5

[ppmj

120.5

7.3 7.2 7.1

'HNfppml

7.07.3 7.2 7.1

'HN[ppmJ

7.0

Figure 2.12 (A) An expanded portion of the TROSY (red) and H-anti-TROSY(black) spectra (recorded

at 25°C at 900 MHz) of 0.8 mM 2H, 15N-labeled sFkpA and (B) corresponding 2H quadrupolar splitting

observed in the ID 2H NMR spectrum of the solvent in liquid crystalline phase composed of C12E5 /

hexanol/H20. (C) 1D slices cut from (A) at the dotted line show that fitting of Lorentzian lineshapes on

the top of the peaks gives reliable measurements of the splitting between the TROSY (red) and H-anti-

TROSY (black) components even for the peaks with very low signal-to-noise ratios.

25

20

i153

I 10

40 -35 -25 5 15 25 35

Figure 2.13 Histogram of RDCs measured in C!2ES / hexanol/H20 liquid crystalline phase

Measured RDCs on sFkpA aligned in C|2ES / hexanol/H20 liquid crystalline phase are shown in

Table 2.3. Figure 2.13 shows the statistic distribution of the measured RDCs.


Table 2.3 Measured RDCs on sFkpA aligned in C!2ES / hexanol/H20 liquid crystalline at 25 °C

Residue OP RDC Residue OP RDC Residue OP_RDC

15 2.7 118 -15.3 167 -11.7

16 16.2 119 -6.3 169 -36.0

17 30.6 120 0.9 170 -17.1

18 5.4 122 -21.6 171 -6.3

20 -5.4 123 9.0 172 6.3

25 34.2 124 -18.0 173 -13.5

28 -20.7 125 1.8 174 27.9

29 1.8 126 -16.2 176 -25.2

38 -6.3 127 1.8 179 -23.4

45 -16.2 130 -27.9 184 -24.3

48 38.7 131 -0.9 185 -7.2

59 0.0 133 -18.0 187 -36.0

66 -13.5 134 11.7 188 3.6

67 -2.7 137 21.6 189 -23.4

69 18.9 138 18.9 190 -18.9

70 -37.8 139 -8.1 191 -13.5

75 -26.1 142 -18.9 192 8.1

80 0.0 143 -24.3 197 -13.5

86 27.9 144 -37.8 198 -0.9

87 6.3 146 -9.9 199 20.7

94 3.6 147 -8.1 202 11.7

103 -21.6 148 -1.8 204 -1.8

104 -25.2 149 -9.0 205 -7.2

105 -30.6 151 17.1 207 -9.9

106 -2.7 152 -27.0 208 -22.5

107 -23.4 153 15.3 213 -2.7

108 -18.9 154 27.0 214 8.1

109 -28.8 155 26.1 215 -8.1

110 0.0 157 -25.2 216 6.3

111 -13.5 158 -25.2 218 -8.1

112 -21.6 159 3.6 219 -28.8

114 7.2 160 3.6 220 -36.0

115 11.7 162 19.8 221 -24.3

116 -18.0 163 -0.9 222 -7.2

117 5.4 164 5.4 224 -13.5

Data in normal (non-bold), which arc from the flexible part of the protein, are excluded in calculating the

final tensor alignment. The data are divided into N-terminus (dimerization part) and C-terminus (FKBP

domains) subsets. Here, the N-terminal domain is defined as residues 18 - 112, which are shown in italics.

The C-terminal domain includes residues 116 - 224.


II.5.2 Fitting of the alignment tensor to the residual dipolar couplings

The method used to fit RDCs to alignment tensors is given in Appendix 3. Although it has been

reported that the effect of nonuniform S2 values, even those of few residues that exhibit large

amplitude motions (S2 ranging from 0.3 to 0.8) on the calculated alignment tensor is minimal [34]

[35], care is taken in the analysis of the RDC data and only residues with low internal mobility

(as judged from the secondary structure of FkpA and using the data from the residues with

steady-state !H- 15N NOE larger than 0.6) were considered.

As shown in equation 2.3, for a homodimeric molecule, such as sFkpA, measured RDCs are in

fact an averaged value of RDCs from two different subunits / and r. With this in mind, three

types of molecular models with different dynamic properties are considered for fitting of

alignment tensors to the measured RDCs:

1) With the assumption that the whole dimeric molecule is rigid, so that the whole molecule

should be aligned with a single unique tensor. The alignment tensor can be calculated

using equation 5 in Appendix 3 with an average matrix X containing the structural

information from both subunits / and r, as shown in equation 15 in Appendix 3. The

calculated results for fitting the alignment tensor are shown in the tables 2.4 and 2.5 with

the notation of "Dimer" in the column of "Molecule Model".

2) In the second extreme situation, two monomers are assumed to be able to move

completely independently relative to each other. When the dimeric molecule is aligned,

RDCs from two subunits, Dl!theo and DTIJheo are then similar to RDCs from the same two

unassociated monomers, i.e. D\theo is equal to D\fhf„ We can then easily derive:

Di,theo - DUheo - T>ifhro (2.4)

In this case, alignment tensors can be calculated by fitting to the experimental data using

as equation 5 or 13 in Appendix 3 where the coordinate matrices Y and X contain the

structural information from only a single subunit, monomer / or monomer r. The

calculated results of alignment tensor fitting are shown in the tables 2.4 and 2.5 with the

notation of "Mon /" or "Mon r" in the column of "Molecule Model".


Talbe 2.4 Results of alignment tensor fitting to RDCs measured on sFkpA aligned with the Pfl

filamentous bacteriophage at 37°C

#

Data Data

Size

Molecule

Model

Alignment Tensors Correlation

CoefficientAxx Ayy Azz

1 Exjoop 117 Dimer -10.53 -14.97 25.50 0.747

2 Ex_loop 117 Mon / -12.73 -19.74 32.47 0.872

3 Exjoop 117 Dimer

2 Tensors

11.89 55.02 -66.91 0.909

-0.57 -43.07 43.64

5 Ex_loop_Nterm 44 Dimer -11.69 -39.20 50.88 0.684

6 Ex_loop_Nterm 44 Mon / -10.03 -15.36 25.39 0.714

7 Ex_loop_Nterm 44 Mon r -5.12 -16.27 21.39 0.606

8 Ex_loop_Nterm 44 Dimer

2 Tensors

23.97 75.70 -99.67 0.816

-29.24 -44.46 73.69

9 Ex_loop_Cterm 73 Dimer -8.04 -89.12 97.16 0.766

10 Ex_loop_Cterm 73 Mon/ -15.32 -21.93 37.24 0.949

11 Ex_loop_Cterm 73 Mon r -15.20 -21.20 36.41 0.950

12 Ex_loop_Cterm 73 Dimer

2 Tensors

-5.45 -45.73 51.18 0.954

-7.65 -43.12 50.77

Ex_loop indicates that fits were performed with RDCs from the flexible part of the protein excluded.

Nterm indicates that data are from the N-terminal domain (residues 18 - 112) were used whereas Ctcrm

indicates that data from the C-terminal residues 116 - 224 were used. AM, Ayy, and A^ are the main axis

components of the Cartesian alignment tensor in the principle axis system (PAS), where |AXX| < \A.yy\ <

\A7l | and Axx+ Ayy + A„ =0.

3) In the intermediate situation between 1) and 2), there is some intramolecular motion

between subunits / and r, with none-the-less some correlation due to the connection

between the monomers. In this situation, subunits / and r are assumed to be independent

in the alignment medium, and Dliäheo and Drutheo should be calculated from two different

alignment tensors, that is: Dilhm = (D[lheo + D\thfg)l2, and


A) Dimer / one unique tensor B) Monomer /

-30 -20 -10

y=1.1474x-4.2329

R2 = 0 5872

0 10 20 30 40

RDCexp

C) Monomer r

40

30

20

8r 10

r>Q 0DC

-10

-20

-30

y = 0.9024x-1.8318

R2 = 0.902

30 -20 -10 0 10 20 30 40

RDCexp

40

30

20

10

0

-10

-20

-30

y = 08985x-19869

^ ^ R2 = 0 8998

-30 -20 -10 0 10 20 30 40

RDCexp

D) Dimer / two tensors

y = 1 2511X + 0 74S8

y = 0 9118k- 1.6867

Rf = 0.91

40 -30»

Figure 2.14 Correlation between measured dipolar couplings (RDCexp) for the C-terminal domain

(residues 116 - 224) of sFkpA in the presence of the Pfl filamentous bacteriophages and calculated

dipolar couplings (RDCtheo ) based on different alignment tensor fitting models, (A) fitting experimental

data to a single tensor using rigid dimeric molecular model (No. 9 in table 2.4), (B) and (C) fitting

experimental data using monomer / or monomer rasa molecular model (No. 10 and 11 in table 2.4) (D)

fitting experimental data to two independent tensors using a dimeric molecular model (No. 12 in table 2.4).

RDC1 and RDC2 are back-calculated dipolar couplings using tensors T' and Y for subunit / and subunit r,

respectively.

D',*o - Kh<o <7" ) ,and /),;„„„ ~ D[Mu (T ) (2.5)

Equation 2.5 states that Dlitheo and D\Mo depend on two different alignment tensors T'

and T'. Alignment tensors were fit to the experimental data using equation 16 in

Appendix 3. The results are shown in the tables 2.4 and 2.5 with the notation of "Dimer 2

Tensors" in the column of "Molecule Model".


To further investigate the detailed dynamic properties of the N- and C- terminal domains of the

dimeric sFkpA. The experimental data of RDCs are divided into a subset from the N-terminal

residues 18-112 and a subset from the C-terminal residues 116 - 224. Fitting of these data using

different molecular models and corresponding alignment tensors are shown in Table 2.4 and 2.5.

II.5.3 Analysis of fitting results and dynamic molecular model

Fitting of RDCs measured at 37°C on sFkpA aligned with Pfl filamentous bacteriophage (Table

2.4) and RDCs measured at 25°C on sFkpA in the presence of C|2E5 / hexanol/H20 liquid

crystalline phase (Table 2.5) gave very good agreement. The results are shown in Table 2.4 and

2.5 and are plotted in Figure 2.14 in 2.15, respectively.

In Table 2.5, a comparison of fitting No. 3 and 4 using the data, in which RDCs from the flexible

part of the protein are excluded, to fitting No.l and 2 using all measured RDCs (Table 2.3) shows

an improvement in the correlation coefficients, which it is probably due to the removal of the

effect of nonuniform S2 values on the fitting. Thus, in the later stages of fitting of the alignment

tensors, only residues with low internal mobility were considered [31].

However, the low correlation coefficients for fittings No. 3 and No. 4 in Table 2.5, which are

0.756 and 0.843, respectively, suggested that there is intramolecular mobility within the dimeric

molecular frame and that there may even be flexibility in each monomeric subunit, as well. No. 1,

2 and No. 3 in Table 2.4 show the same results. Experimental RDCs are then divided into subsets

from the N-terminal and from the C-tcrminal residues for alignment tensor fitting to illustrate the

detailed dynamic properties of N- and C- terminal domains of the dimeric sFkpA.

In No. 9, 10, 11 and No. 12 in Table 2.4 and 2.5, fitting of the measured RDCs from the C-

terminal residues are carried out using different molecular models as described above. A

significant improvement in the correlation coefficients from No. 9 to No. 10 and/or No. 11 is

observed, where the correlation coefficient improves from 0.766 (0.857 in Table 2.5), obtained


Table 2.5 Alignment tensor fitting results from RDCs of sFkpA measured in Q2E5 / hexanol/H20

liquid crystalline phase

#

Data Data

Size

Molecule

Model

Alignment Tensors Correlation

CoefficientAxx Ayy A„

1 All 105 Dimer 0.59 13.32 -13.91 0.727

2 All 105 Mono / -5.42 -25.51 30.94 0.778

3 Exjoop 88 Dimer 2.73 11.59 -14.33 0.756

4 Exjoop 88 Mono / -5.14 -26.82 31.96 0.843

5 Ex_loop_Nterm 23 Dimer 6.04 43.93 -49.97 0.708

6 Ex_loop__Nterm 23 Mono / 7.77 16.71 -24.48 0.707

7 Ex_loop_Nterm 23 Mono r 7.96 14.90 -22.85 0.643

8 Ex_loop_Nterm 23 Dimer

2 Tensors

-6.55 -48.50 55.05 0.757

13.51 45.54 -59.04

9 Ex„loop_Cterm 65 Dimer -10.24 -26.87 37.11 0.857

10 Ex_loop_Cterm 65 Mono / -5.63 -31.58 37.21 0.952

11 Ex_loop_Cterm 65 Mono r -5.10 -30.66 35.75 0.947

12 Ex_loop_Cterm 65 Dimer

2 Tensors

3.21 45.91 -49.13 0.954

-14.20 -19.61 33.81

Exjoop is data (Table 2.3) in which RDCs from the flexible part of the protein are excluded. Nterm

indicates data from the N-terminal domain (residues 18 - 112) and Cterm stands for data from the C-

terminal residues 116 - 224. Axx, Ayy, and Azz are the main axis components of Cartesian alignment tensor

in the principle axis system (PAS), where |AXX| < [Ayy| < |AZZ | and Axx+ Ayy + Azz =0.

using a unique tensor for the dimeric molecule, to 0.949 (for monomer I) / 0.950 (for monomer r)

(0.952 /0.947 in Table 2.5), calculated using a single tensor with monomeric molecular model,

monomer / or monomer r, indicated that there is relative motion between the two C-terminal

domains within the molecular frame of the dimer. However, the high correlation coefficients

resulting from fitting using a single tensor with a monomeric molecular model indicates that

these two C-tcrminal domains are quite rigid within each subunit, thus fitting No. 10 and No. 11

in both table 2,4 and 2.5 gave a satisfactorily high correlation coefficient as rigid bodies. Fitting


A) Dimer / one unique tensor B) Monomer /

y = 0 7514x -1 205

Ff = 0 7349

-20 0

RDCsxp

C) Monomer r

20

: o

i

-20

-40

y = 09142x + 00696

# = 0.8964

40 -20 0

RDCexp

20 40

ys09177x-0 1621

tf = 0 9058

O

RDCaxp

D) Dimer / two tensors

-50 -40 -30

y=1 3113X + 228

y-0 9244x 0 0443

(^ = 09109

y=0 5376x 2 3686

* Avgrago of RDC1 and RDC2

• RDC1

BUC2

Figure 2.15 Correlation between measured dipolar couplings, RDCexp for the C-terminus (residues 116-

224) of sFkpA in liquid crystalline phase composed of Ci2E5 / hexanol/H20 and calculated dipolar

couplings (RDQheo ) based on different alignment tensor fitting models, (A) fitting experimental data to a

single tensor using rigid dimeric molecular model (No. 9 in table 2.5), (B) and (C) fitting experimental

data using monomer / or monomer r as a molecular model (No. 10 and 11 in table 2.5) (D) fitting

experimental data to two independent tensors using a dimeric molecular model (No. 12 in table 2.5).

RDC1 and RDC2 are back-calculated dipolar couplings using tensors T' and Tr for subunit / and subunit r,

respectively.

No. 12 using simultaneously two independent alignment tensors with a dimeric molecular model

gave a correlation coefficient of 0.954. This suggested that the two C-terminal domains moved

independently within the dimeric molecular frame rather than being rigidly fixed at the end of the

long helix (helix 3). One thing to be noted is that the different results calculated using monomer /

and monomer r as molecular model are probably due to differences in the structures of the / and r


monomers in the homodimeric molecular model. In our RDC fitting, we use, as a structural

model, the published X-ray crystal structural coordinates (PDB code: 1Q6H), which can be

considered as a "snapshot" of a dynamically tumbling molecule. At a certain time point when the

"snapshot" is taken, the homodimcr most probably exhibits a somewhat asymmetric relative

orientation. This may explain why the two tensors of No. 12 in Table 2.4 and 2.5 are not exactly

same. Correspondingly, the theoretical RDCs predicted on the basis of these two distorted tensors

T' and T' are also a little bit different from each other. The fitting results are plotted in Figure

2.14 and Figure 2.15.

The relative motion between the two C-terminal domains could be caused by somewhat higher

flexibility of the long helix (helix 3), presumably between residues 84 and 91. This result agrees

with the previous data obtained during backbone assignment and relaxation studies of sFkpA.

Similarly, in No. 5, 6, 7 and No. 8 in Table 2.4 (and 2.5), fitting of measured RDCs from N-

terminal residues are carried out using different molecular models. Low correlation coefficients

of fittings No. 5 and No. 6 / No. 7 in Table 2.4 (and 2.5), which are 0.684 (0.708) and 0.714

/0.606( 0.707/0.643), respectively, suggested that there is intramolecular mobility within the two

N-terminal domains of the dimeric molecule, which could be intra-subunit or inter-subunit

motion. However they are 1) neither relatively fixed in orientation and rigid within each subunit,

as fittings No. 5 gave low correlation coefficient, which means that the molecular model 1 ) is not

satisfied, i.e. there is relative motion between the two N-terminal domains of the dimeric

molecule or the individual monomers are not rigid. 2) nor highly free and independent to each

other and rigid within each subunit, as fittings No. 6 and 7 gave also low correlation coefficients,

which means that the molecular model 2) above and equation 2.4 are not satisfied. 3) Thus,

fitting is carried out using molecular model 3) and equation 2.5 described above. However fitting

No. 7 in Table 2.4 and 2.5 still gave rather low correlation coefficients of 0.816 and 0.757. These

results can only be explained by the existence of backbone flexibility within each monomeric

subunit for the N-terminal domains, rather than due to relative motion between these two

subunits. That is, the backbone of N-terminal residues is highly flexible. This could also imply

that the dimerization is generally due to the interaction between sidechains of the interlaced N-

terminal domains of the two subunits rather than two monomers are hold together through

backbone interactions, such as H-bonds between the backbone groups.


Conclusion: The dynamic model for this dimeric molecule can be described as follows:

1. The backbone of the N-terminal domain is highly flexible and the dimerization is

suspected mainly due to an interaction between sidechains of the interlaced N-terminal

domains (mainly helices 1 and 2) of the two subunits rather than through backbone

interactions between two subunits.

2. There is high relative mobility between the two C-terminal domains (i.e. the FKBP

domains) due to the existence of flexibility in the long a-helix (helix 3), presumably

between residues 84 and 91. However the skeleton of each FKBP domain (excluding the

disordered loops) is quite rigid.

II.6 Chaperone function of FkpA

Our NMR studies on the dimeric sFkpA in solution showed that the long a-helix (helix 3) is

flexible, presumably between residues 84 and 91. It has been reported that the X-ray structures of

LpMIP the bending of the long a-helix around residue 81 changes before and after the protein

binds FK506[5J. The published X-ray structure of the dimeric FkpA-ACT, which can be looked

upon as a snapshot of the dynamically tumbling dimeric FkpA in solution, also showed a visible

bending in the long a-helix around residue 91 (Figure 2.20 B, "elbow") [6]. Considerable

plasticity of the long a-helix might be a key to the chaperone activity of FkpA, which allows it to

bind a wide range of targets.

However, the location of the chaperone-funtion-related polypeptide-binding site resides within

the dimeric molecule is subject to some dispute [6, 9]. Here, using solution NMR techniques, we

investigate the interactions between FkpA and substrate polypeptides (proteins) and the

molecular mechanism of its chaperone activity.

11.6.1 Substrate protein binding

To locate the chaperone-funtion-related polypeptide-binding site in its 3D structure and

investigate the interactions between FkpA and substrate polypeptides (proteins), the candidate

substrate proteins should ideally be: 1) partially folded or misfolded proteins or relatively stable


B)

115.0 ,..-''

10.0 9-0 W[ppm] 7-°

7.50 7.30

'HN[ppm]

7.10

* '

r a » Iff à 7.50

10.0 90 <HN£ppm] 7.0

7.30

'HNfrpm]

7.10


Figure 2.16 Overlay of the 2D 'H-'^N TROSY spectra (900MHz at 37°C) of sFkpA alone (black) and

titrated with RNase AS-Protein (A, and B) and RCM-la (C and D) at the molar ratios of 1:0.25 (magenta),

1:0.5 (blue), 1:0.8(green), and 1:1 (red). The crosspeaks showing shifts are marked with the corresponding

residue numbering for FkpA. B and D show an expansion of the 'H/15N chemical shift correlation spectra.

intermediates on a refolding pathway, but 2) still water soluble and 3) preferably with low

molecular weight, making possible study of the complex with solution NMR.

According to these criteria, we choose RNase AS-Protein and reduced and carboxymethylated

bovine a-lactalbumin (RCM-la) as protein substrates to investigate the chaperone function of

FkpA. RNase AS-Protein is a 104 amino acid protein, derived from RNase A, which retains some

residual structure stabilized by four disulfide-bonds [36] [371. By limited proteolysis, the peptide

bond between A20-S21 of RNase A can be selectively cleaved, resulting in only a slightly

altered structure [38], but with a fully cnzymatically active complex. By removing the N-terminal

20 amino acid ploypeptide, the enzymatically inactive, but folding competent so-called RNase

AS-Protein is generated. RNase AS-Protein has a native-like residual structure with good water

solubility under slightly acidic conditions [39]. Reduced and carboxymethylated bovine a-

lactalbumin (RCM-la) is a denatured protein with the molecular weight of 14 kDa and is

unfolded under physiological conditions. However, it remains water soluble under the condition

chosen for our solution NMR study [9]. In our experiments, RNase AS-Protein and

carboxymethylated bovine a-lactalbumin were obtained from Sigma. The latter was then reduced

with dithiothreitol (DTT, also from Sigma) to generate RCM-la. 2D 1H-I5N TROSY spectra are

recorded on 2H/15N-1abeled sFkpA (0.8 mM per monomer in 20 mM Mes buffer at pH 6.0 with

50 mM NaCI) at 37°C at 900MHz at the following molar ratios of sFkpA to substrate: 1:0, 1:0.25,

1:0.5, 1:0.8, and 1:1, with either RNase AS-Protein or RCM-la as substrate (Figure 2.16).

Through a careful comparison, the residues showing 'H/15N chemical shift perturbation during

the titration are mapped to the amino acid sequence of FkpA, as shown in Figure 2.17 A.

II.6.2 Specific polypeptide binding site of FKpA and comparison of its chaperone function

and PPIase activity

As shown in Figure 2.17 A, there is an extensive scope of residues showing 'h/'^N chemical shift

perturbation when titrated with either RNase AS-Protein or RCM-la. This suggested that there


A)

sFkpA+Sprotein 10

sFkpA+RCM-la 10

sFkpA+Fk506 10

sFkpA+Sprotein 69

sFkpA+RCM-la 69

sFkpA+Fk506 69

HI H2

adskaafknddqksayalgaeiIgryESnslkEBeklgikldkdqliagvqdafadkskl

adskaafknddqksayalga|3ljgry£[i nslk îçeklgikldkdqliagvqdafadkskladskaafknddqksayalgaslgrymenslkeqeklgikldkdqliagvqdafadkskl

H3 ßl ß2

sdqeieqtlqafeafvJkssESsISmekdaadneakgkeyrekfakekgvktsstglvtqvve

sdqeieqtlqafeab\jkss^qfifmekdaadneakgkeyrekfakekgvktsstglvyqvvesdqeieqtlqafearvkssaqakmekdaadneakgkeyrekfakekgvkt sstglvyqwe

ß3 ß4a H4 ß4b H5

sFkpA+Sprotein 130 AGKGEAPKDSDTWVNYKGTLIDGKEFDNSYTRGEPLSFRLDGVIPGWTEGLKNIKK

sFkpA+RCM-la 130 AGKGEAPKDSDTWVNYKGTLIDGKEFDNSYTRGEPLSFRLDGVIPGWTEGLKNIKK

sFkpA+Fk506 130 AGKGEAPKDSDTVVVNYKGTLIDGKEFDNSYTRGEPLSFRLDGVIPGWTEGLKNIKK

ß5 ß6

sFkpA+Sprotein 187 GGKIKLVIPPELAÏGKAGVPGIPPM3TLVFDVELLDVKPAP

SFkpA+RCM-la 187 GGKIKLVIPPELAYGKAGVPGIPPN3TLVFDVELLDVKFAP

sFkpA+Fk506 187 GGKIKLVIPPELAYGKAGVPGIPPNSTLVFDVELLDVKPAP

B)

ANW


Figure 2.17 A) Residues showing 'H/15N chemical shift perturbation upon titration with RNase AS-

Protein and RCM-la are highlighted in yellow. Residues N99, K102, F168, Y200 and V205 showing

chemical shift changes, AS > 0.1 (see Equation 2.6) with RNase AS-Protein are in red. For comparison,

residues located in FK506 binding pocket are in green. B) Comparison of structure function relation of

FkpA as chaperone with its PPIase function. Residues related to chaperone function are in red (left) and

residues related to its PPIase function are in green (right). The molecule is shown perpendicular to and

along the crystallographic 2-fold axis. The molecular model is based on the X-ray structure (PDB code:

1Q6H).

might be a background of chemical shift perturbations due to non-specific binding of RNase AS-

Protein and RCM-la to sFkpA. In order to locate potential specific polypeptide binding sites and

to illustrate the thermodynamics of its chaperone activity, detailed quantitative analysis of the

chemical shift changes upon addition of RNase AS-Protein and RCM-la is necessary. The

following expression defining changes between the spectrum with the sFkpA only and a

spectrum with a substrate protein and sFkpA was proposed [40]:

A£ = ^/(A#/)2+0.04(A<W)2 (2.6)

Here, A5H and AON represent the difference in chemical shifts of the amide hydrogen, and the

nitrogen, respectively, of the free sFkpA vs. a mixture of substrate proteins and sFkpA, where all

chemical shifts are in ppm. In Tables 2.6 A and B, residues are listed in the descending order

according to the calculated chemical shift changes using equation 2.6 upon addition of equimolar

RNase AS-Protein and RCM-la, respectively.

Six residues showing the most significant peak shifts upon addition of equilmolar RNase AS-

Protein (bold in the Table 2.6 A) are chosen for quantitative investigation on sFkpA and substrate

binding. Residue 200 is excluded due to the overlapping with the residue 33, which makes it not

suitable for the quantitative characterization. Similarly, five residues showing the most

significant peak shifts upon addition of equimolar RCM-la (bold in the Table 2.6 B) are chosen.

Residue 82 and 24 are excluded due to overlap and weak peak intensity.

For all chosen residues, chemical shift changes of sFkpA are measured during the process of

titration at molar ratios of 1:0.25, 1:0.5, 1:0.8, and 1:1, with RNase AS-Protein and RCM-la. The


measured chemical shift changes (AÔ) are plotted against the concentrationof the substrates,

RNase AS-Protein and RCM-la, respectively, as shown in Figure 2.18 A and B.

tO

012 ~*~-Residue49—,

01Residue 143

-*- Residue 157- y" 3

008~*~ Residue 1S9

-«— Residue 207

~ - ~yyxy

~

ys—

^

006

004- - éé

yy^d&-

002~

<

01 02 03 04

IS] («M)

OS 06

£tesidus_t57

0.3 04

[Sj (mM)

oe

Figure 2.18 (A) Plots of peak shifts, AÖ, in 2D 'H-15N TROSY spectra (900MHz at 37°C) of sFkpA, as a

function of the concentration of RCM-la, |5*| .(B) Peak shifts, A5, in 2D 'H-15N TROSY spectra

(900MHz at 37"C) of sFkpA, as a function of the concentration of the substrate RNase AS-Protein, [5].

The concentration of 0.55 mM of substrates corresponds to the molar ratio of 1:1 of sFkpA (concentration

based on monomer) to substrates. The lines through the experimental points represent fits to the data with

Equation 2.13.

As we can see from Figure 2.18 A, RCM-la shows weak binding to sFkpA at the concentrations

on the order of mM. Probably there is a positive cooperative effect when the concentration of

RCM-la increases because the chemical shift changes vs. the concentration of RCM-la shows an

"S" (sigmoidal) curve. However, overall the chemical shift changes AÔ are small with maximum

of 0.101 of residue 207. Using AÔ of 0.1 ppm as cutoff, as previously reported criteria [41], we

only consider perturbations greater than 0.1 ppm significant, Thus, we assume that RCM-la

weakly binds to sFkpA and note that this type of weak binding could induce the chemical shift

perturbation AÔ of up to 0.1 ppm. However, RNase AS-Protein shows strong binding to sFkpA

at the concentration on the order of mM as there is an obvious trend of saturation of the binding

as the concentration of RNase AS-Protein increases (Figure 2.18 B).


Table 2.6 A Chemical shift changes of sFkpA upon addition of RNase AS-Protcin at molar

ratio of 1:1

Residue ÔH _ref ON _ref SH ON AôH A5N AÔ

205 7.639 115.149 7.598 114.583 -0.041 -0.566 0.120396

99 8.567 119.593 8.679 119.446 0.112 -0.147 0.115794

102 8.35 118.909 8.251 118.628 -0.099 -0.281 0.11384

168 8.376 120.785 8.273 120.59 -0.103 -0.195 0.110136

200 8.565 122.35 8.464 122.158 -0.101 -0.192 0.108054

207 8.393 113.699 8.34 113.295 -0.053 -0.404 0.096631

157 7.055 120.789 7.04 120.339 -0.015 -0.45 0.091241

Table 2.6 B Chemical shift changes of sFkpA upon addition of RCM-la at molar ratio of 1:1

Residue ÔH _ref ON _ref 5H ON A5H AON AÖ

207 8.393 113.699 8.335 113.232 -0.058 -0.467 0.109943

157 7.055 120.789 7.044 120.288 -0.011 -0.501 0.100802

82 7.465 120.961 7.469 120.539 0.004 -0.422 0.084495

49 8.353 124.702 8.32 124.406 -0.033 -0.296 0.067776

24 7.086 122.545 7.019 122.497 -0.067 -0.048 0.067684

169 8.524 123.007 8.46 122.902 -0.064 -0.105 0.067357

143 7.64 122.027 7.595 121.785 -0.045 -0.242 0.066088

28 8.09 108.165 8.137 107.999 0.047 -0.166 0.057543

ÔH and 8N are the chemical shifts of the amide hydrogen and nitrogen of sFkpA measured for a mixture

of sFkpA and substrate proteins, RNase AS-Protein or RCM-la. 8H_ref and ÖN_rcf are the corresponding

chemical shifts measured for the free sFkpA. All chemical shifts are in ppm.

Considering only perturbations greater than 0.1 ppm significant, therefore, only residues N99,

K102, F168, Y200 and V205 show significant chemical shift changes and they are believe to be

important for specific binding with the substrate of RNase AS-Protein and reflect the chaperone


function of FkpA. These residues are marked in red in Figure 2.17 A and mapped to the 3D X-ray

structure (Figure 2.17 B). These residues involved in the polypeptide binding and potentially

involved in the chaperone activity are compared with the FK506 binding pocket, which is related

to the PPIase activity of FkpA, as shown in Figure 2.17 B. Inspection of Figure 2,17 A and B,

reveals that residues Y200 and V205 are important for both chaperone function and PPIase

activity; and that residue F168, which is located adjacent to the FK506 binding pocket, is

important for specific polypeptide (protein) binding, although it is not involved in the PPIase

activity; N99 and K102, which are located in the long a-helix, are also important for the

chaperone function but have no relation to the PPIase activity.

II.6.3 FkpA chaperone function

We have shown that the chaperone FkpA can specifically bind to non-natively structured RNase

AS-Protein. The residues N99, K102, Fl 68, Y200 and V205 are believe to be important for this

specific binding and the chaperone function of FkpA. Our next interest is trying to characterize

the equilibrium constant of the substrate binding and the number of substrate molecules per dimer

of FkpA in the binding. The binding reaction can be written as:

[sFkpA] + n[S] — [sFkpA^SJ (2.7)

A B

Where free and bound forms of FkpA are noted as spieces A and B and the integral constant n

gives the number of substrate molecules that bind to a single chaperone molecule sFkpA. For a

homodimeric molecule, sFkpA, the constant n is likely to be equal to either 1 or 2, which means

that two or only one substrate molecule binds to a single dimeric molecule of FkpA. As only a

single set of NMR signals is observed for the mixture of RNase AS-Protein and sFkpA, one can

assume that the exchange between the free and bound forms of sFkpA is fast on the NMR

timescale. Thus, without further detailed consideration of chemical shift perturbation due to non¬

specific binding, our measured chemical shift is a weighted average value between the free and

bound forms of FkpA expressed by:

ST=-^-SA+-^—SB (2.8)[A\ + [B] [A] + [B]


where or is the measured chemical shift and SA and SB are chemical shifts for the free and bound

forms of sFkpA. Thus, the chemical shift changes can be expressed as:

AS= ÖT -SA =

[A]3A +

{B]5B-ÖA

T A

[A] + \B] [AJ + [B1

=_^1_(^ -SA) = -W—ASm[A] + [B\

B A

[A] + [B]

where Aô = S„- SÀ is the chemical shift difference between the bound and free form sFkpA.TTL1X 15 r\.

It can easily be derived that,

(2.10)[B] AS _ôT- SA

[A] ASm-AS 50-ôT

The observable equilibrium constant Kapp can be written as:

Kaa,=â = Kx*[S] + K2»lS]2 (2.11)

K; and K2 are the binding constant when integral constant n is equal to 1 and 2.

For non-competitive binding the average number of substrates bound per sFkpA protein molecule,

N,can be calculated as:

( AS \

N = d\nKapp/dln(lS\) = d\n /d\n(\S]) (2.12)V^^max A<? J

If ln(AÔ/AÔmax - A5) is plotted as a linear function of ln(rS'l), then the slope is equal to the

constant N,i.e., the average number of substrates bound per sFkpA protein molecule. Chemical

shift changes for residues N99, K102, F168. V205 at different concentration of RNase AS-

Protein were used to calculate lnCAS/AS* - AS). Figure 2.19 shows the plot of ln(AÔ7A<w -

A8) for residues N99, K102, F168, V205 and their average vs. ln([5j). Linear regression gives

an average value for constant N of about 0.7, which indicates that n is equal to 1 (n must be an

integer).

The intrinsic binding constant for the titratable residues, Klt was evaluated by fitting the data in

Figure 2.18 to a simple two-state binding equation:


AS = (2.13)

The approximate value for K1 is 1.57 mM'1 and the corresponding inverse value of the binding

constant is around 640 u.M. This matches previously reported results of 730±100 u,M in steady-

state kinetic analysis using Suc-Ala-Phe-Pro-Phe-pNA as substrate [3].

y = 0.692ÔX + 0.438 R8Sidua-_S9

y = 0.6702X + 0.4554 * Resïdue_102

y = 0.7535X + Q.47Ô6 ResidueJSS

y = 0.6842* + 0.438S x ReSldU6„205

S*

Mi

J -2<3

» Average

y = 07ÛÛ2x+Û.4S3

0.5

Figure 2.19 Linear regression analysis of ln(AÔ/A8111£lx - AS) for residues N99, 1C102, F168, V205 and

their average as a function of ln([^]). [S] is the concentration of substrate RNase AS-Protein.

Conclusion: Our NMR studies on the complex of dimeric sFkpA with substrate proteins

characterize the chaperone function of FkpA as follows:

1. Five residues, N99, K102, Fl 68, Y200 and V205 are probably important for specific

polypeptide (or substrate protein) binding and may play an important role in the FkpA

function, of which N99 and K102 locate in the long a-helix are not involved in the PPIase

activity, whereas F168, Y200 and V205 are located in the FKBP domain, which are in or

nearby the PPIase activity related FK506 binding pocket.

2. Thermodynamic property of the complex of dimeric sFkpA with substrate protein RNase

AS-Protein shows that sFkpA dimer binds RNase AS-Protein at a 1:1 molar ratio and is in


the fast exchange regime on the NMR time scale with an equilibrium constant on the order

of 100 p.M .

II.l Mechanism of chaperone function of dimeric FkpA: model of "mother's

arms"

Based on our NMR studies of the structure and dynamics of FkpA and its complex with substrate

proteins, we propose a molecular model, the so-called "mother's arms" model, to explain the

mechanism of chaperone function of FkpA, as shown in Figure 20.

Figure 2.20 Model of "mother's arms" to illustrate the mechanism of chaperone function of FkpA. The

curved double arrows indicate the flexibility in the long a helices.

In this model, FkpA has two long flexible "arms" (the two long a helices) which can bend at the

"elbows" (presumably between residues 84 and 91 in the long a helices). The two "hands" are

the specific polypeptide binding sites on the two subunits and each of them has five

"fingers"(residues N99, K102, F168, Y200 and V205). These two hands using their five fingers

can "catch" (bind) "one" (constant « is 1) polypeptide or small protein (called "baby protein")

substrate. The "mother", i.e., dimeric FkpA, through bending of her two long a-helical "arms",

can then hold the "baby" (polypeptide or small protein substrate), which is like a mother's arms

hugging her baby. The considerable plasticity in the long a helices renders FkpA able to adapt its

structure and bind a wide range of targets, which might be very important for the chaperone

activity of FkpA.


Perspectives:

To further confirm the refolding assistant function of FkpA on partially folded or misfolded

proteins or relatively stable intermediates on a refolding pathway, the binding experiments can be

carried out, however using fully folded protein, such as RNase A, as substrate. This type of

negative control experiment will exclude the possibility of false-positive results in our substrate

binding experiment in II 6.1. RDCs and 15N relaxation can be measured with the sFkpA complex

with RNase AS protein. The results will help further understand the structural and dynamical

properties of the chaperone-substrate complex, which in turn can provide information to illustrate

the mechanism of the chaperone function of the dimeric FkpA.

References

1. Home Sm, Young KD, Escherichia-Coli And Other Species Of The Enterobacteriaceae Encode A Protein

Similar To The Family OfMip-Like Fk506-Rinding Proteins. Archives Of Microbiology, 1995. 163(5): p. 357-

365.

2. Kathrin Ramm, Andreas Pliickthun, The Periplasmic Escherichia coli Peptidylprolyl cis,lrans-Isomerase FkpA.

II. 1SOMERASE-INDEPENDENT CHAPERONE ACTIVITY IN VITRO. J. Biol. Chem., 2000. 275: p. 17106-

17113.

3. Jean-Philippe Arié, Nathalie Sassoon, Jean-Michel Betton, Chaperone function of FkpA, a heat shock prolyl

isomerase, in the periplasm ofEscherichia coli. Mol. Microbiol, 2001. 39(1): p. 199 -210.

4. Hendrick Bothmann, Andreas Pliickthun, The Periplasmic Escherichia coli Peptidylprolyl eis, trans-Isomerase

FkpA. I. INCREASED FUNCTIONAL EXPRESSION OF ANTIBODY FRAGMENTS WITH AND WITHOUT cis-

PROLINES. J. Biol. Chem., 2000. 275: p. 17100-17105.

5. Riboldi-Tunnicliffe A, Konig B, Jessen S, Weiss MS, Rahfeld J, Hacker J, Fischer G, Hilgenfeld R, Crystal

structure ofMip, a prolylisomerase from Legionella pneumophila. Nat. Struct. Biol., 2001. 8(9): p. 779-783.

6. Saul FA, Arie JP, Normand BVL, Kahn R, Betton JM, Bentley GA, Structural and functional studies of FkpA

from Escherichia coli, a cis/trans peptidyl-prolyl isomerase with chaperone activity. J. Mol. Biol., 2004. 335(2):

p. 595-608.

7. Schreiber SL, Crabtree GR, The Mechanism OfAction OfCyclosporine-A AndFk506. Immunology Today, 1992.

13(4): p. 136-142.

8. Tanaka H, Nakahara K, Halanaka H, Inamura N, Kuroda A, Discovery and development of a novel

immunosuppressant, tacrolimus hydrate. Yakugaku Zasshi-Journal Of The Pharmaceutical Society Of Japan,

1997.117(8): p. 542-554.


9. Kathrin Ramm, Andreas Pliickthun, High Enzymatic Activity and Chaperone Function are Mechanistically

Related Features of the Dimeric E. coli Peptidyl-prolyl-isomerase FkpA. J. Mol, Biol., 2001. 310(2): p. 485-498.

10. Wuthrich, K, NMR ofProteins and Nucleic Acids. 1986, New York: Wiley.

11. Pervushin, K., et al., Attenuated T-2 relaxation by mutual cancellation of dipole- dipole coupling and chemical

shift anisotropy indicates an avenue to NMR structures of very large biological macromolecules in solution.

Proc. Natl. Acad. Sei. U. S. A., 1997. 94(23): p. 12366-12371.

12. Salzmann, M., et al., TROSY in triple-resonance experiments: New perspectivesfor sequential NMR assignment

of large proteins. Proc. Natl. Acad. Sei. U. S. A., 1998. 95(23): p. 13585-13590.

13. Salzmann, M., et al., TROSY-type triple-resonance experiments for sequential NMR assignments of large

proteins. J. Am. Chem. Soc., 1999. 121(4): p. 844-848.

14. Zhu, G., et al., 2D and 3D TROSY-enhanced NOESYofN-15 labeled proteins. J. Biomol. NMR, 1999. 14(4): p.

377-381.

15. Guntert, P., et al., Processing of Multidimensional Nmr Data with the. New Software Prosa. J. Biomol. NMR,

1992. 2(6): p. 619-629.

16. Bartels, C, et al., The Program Xeasy for Computer-Supported Nmr Spectral-Analysis of Biological

Macromolecules. J. Biomol. NMR, 1995. 6(1): p. 1-10.

17. Wishart, D. S., et al., H-l, C-13 and N-15 Chemical-Shift Referencing in Biomolecular Nmr. J. Biomol. NMR,

1995. 6(2): p. 135-140.

18. Guntert P, Salzmann M, Braun D, Wuthrich K, Sequence-specific NMR assignment of proteins by global

fragment mapping with the program MAPPER. J. Biomol. NMR, 2000. 18(2): p. 129-137.

19. Wishart, D. S., Sykes, B.D., The C-13 Chemical-Shift Index - A Simple Method For The Identification Of

Protein Secondary Structure Using C-13 Chemical-Shift Data. J. Biomol. NMR, 1994. 4(2): p. 171-180.

20. Farrow, N. A., et al., Backbone Dynamics of a Free and a Phosphopeptide-Complexed Src Homology-2 Domain

Studied By N-15 Nmr Relaxation. Biochemistry, 1994. 33(19): p. 5984-6003.

21. Cavanagh, J., et al., Protein NMR Spectroscopy: Principles and Practice. 1996, New York: Academic Press.

22. Bax, A., G. Kontaxis, and N. Tjandra, Dipolar couplings in macromolecular structure determination, in Nuclear

Magnetic Resonance ofBiological Macromolecules, Pt B. 2001. p. 127-174.

23. Tjandra, N. and A. Bax, Direct measurement ofdistances and angles in biomolecules by NMR in a dilute liquid

crystalline medium [see comments] [published erratum appears in Science 1997 Dec 5;278(5344):1697],

Science, 1997. 278(5340): p. 1111-4.

24. Tolman JR, Al-Hashimi HM, Kay LE, Prestegard JH, Structural and dynamic analysis of residual dipolar

coupling data for proteins. J. Am. Chem. Soc, 2001. 123(7): p. 1416-1424.

25. Bernado P, Blackledge M, Anisotropic small amplitude peptide plane dynamics in proteins from residual

dipolar couplings. J. Am. Chem. Soc, 2004. 126(15): p. 4907-4920.

26. Matsuda T, Ikegami T, Nakajima N, Yamazaki T, Nakamura H, Model building of a protein-protein complexed

structure using saturation transfer and residual dipolar coupling without paired intermodular NOE. J. Biomol.

NMR, 2004. 29(3): p. 325-338.


27. Clore GM, Schwieters CD, Docking of protein-protein complexes on the basis of highly ambiguous

inlermolecular distance restraints derived from H-1(N)/N-15 chemical shift mapping and backbone N-15-H-1

residual dipolar couplings using conjoined rigid body/torsion angle dynamics. J. Am. Chem. Soc, 2003.

125(10): p. 2902-2912.

28. Tolman, JR, Dipolar couplings as a probe ofmolecular dynamics and structure in solution. Curr. Opin. Struct.

Biol., 2001. 11(5): p. 532-539.

29. Ruckert M, Otting G, Alignment of biological macromolecules in novel nonionic liquid crystalline media for

NMR experiments. J. Am. Chem. Soc, 2000.122(32): p. 7793-7797.

30. Hansen, M. R., L. Mueller, and A. Pardi, Tunable alignment of macromolecules by filamentous phage yields

dipolar coupling interactions. Nat. Struct. Biol., 1998. 5(12): p. 1065-1074.

31. Sass, J., et al., Purple membrane induced alignment of biological macromolecules in the magnetic field. J. Am.

Chem. Soc, 1999.121(10): p. 2047-2055.

32. Byeon, I. J. L., J. M. Louis, and A. M. Gronenbom, A captured folding intermediate involved in dimerization

and domain-swapping ofGBl. J. Mol. Biol., 2004. 340(3): p. 615-625.

33. Yuzawa, S., et al., Solution structure of the tandem src homology 3 domains ofp47(phox) in an autoinhibited

form. J. Biol. Chem., 2004. 279(28): p. 29752-29760.

34. Vogeli B, Kovacs H, Pervushin K, Measurements of side-chain C-13-C-13 residual dipolar couplings in

uniformly deuterated proteins. J. Am. Chem. Soc, 2004. 126(8): p. 2414-2420.

35. Clore GM, Schwieters CD, How much backbone motion in ubiquitin is required to accountfor dipolar coupling

data measured in multiple, alignment media as assessed by independent cross-validation? J. Am. Chem. Soc,

2004. 126(9): p. 2923-2938.

36. Richards, F.M. and H.W. Wyckow, Bovine pancreatic ribonucleases, in The Enzymes, Boyer, Editor. 1971,

Academic Press, New York. p. 647-806.

37. Ritter, C. and A. Helenius, Recognition of local glycoprotein misfolding by the. ERfolding sensor UDP-glucose ;

glycoprotein glucosyltransferase. Nat. Struct. Biol., 2000. 7(4): p. 278-280.

38. Kim, E. E., et al., Refinement of the Crystal-Structure of Ribonuclease-S - Comparison with and between the

Various Rihonuclease-a Structures. Biochemistry, 1992. 31(49): p. 12304-12314.

39. Pervushin K, Gallius V, Ritter C, Improved TROSY-HNCA experiment with suppression of conformational

exchange induced relaxation. J. Biomol. NMR, 2001. 21(2): p. 161-166.

40. Damberg CS, Orekhov VY, Billeter M, Automated analysis of large sets ofheteronuclear correlation spectra in

NMR-based drug discovery. J. Med. Chem., 2002. 45(26): p. 5649-5654.

41. Hajduk PJ, Boyd S, Nettesheim D, Nienaber V, Severin J, Smith R, Davidson D, Rockway T, Fesik SW,

Identification of novel inhibitors of urokinase via NMR-based screening. J. Med. Chem., 2000. 43(21); p. 3862-

3866.

85

Chapter III

Backbone resonance assignment in large protonated proteins using a

combination of new 3D TROSY-HN(CA)HA, 4D TROSY-HACANH and 13C

detected HACACO experiments

Kaifeng Hu, Alexander Eletsky and Konstantin Pervushin, J. Biomol NMR, (2003) 26 (1),

66-77.

86

Seite Leer /

Blank leaf

87

Introduction: X-detection based NMR spectroscopy of proteins

In the recent years a method for the residue-specific assignment of the backbone resonances in

large proteins has been developed, which is based on the implementation of transverse relaxation-

optimized spectroscopy (TROSY) (Pervushin et al. 1997) in 3D and 4D sets of the triple-

resonance experiments (Salzmann et al. 1999a; Salzmann et al. 1999b; Yang and Kay 1999;

Mulder et al. 2000). The sensitivity of the basic TROSY-type HNCA, HNCOCA, HNCACB,

HNCO and HNCACO experiments constituting this method is critically important for the

successful backbone assignment. The uniform or partial replacement of non-labile protons with

deuterons reducing transverse relaxation of the *HN and 13Ca'^ spins has been an indispensable

tool for resolving and assigning NMR spectra of large complexes, including membrane proteins

in detergent micelles (LeMaster 1994; Yamazaki et al. 1994; Grzesiek et al. 1995b; Shan et al.

1996; Gardner and Kay 1998; Salzmann et al. 1998). In this case much of the valuable side chain

information is lost by the complete deuteration of these moieties. With the use of only partial

deuteration the assignment process can be hampered by 13C line broadening due to ^^H isotope

effects (Venters et al. 1996). Other problems associated with the *HN "out- and-back" method

underlying this line of experiments are the use of smaller scalar couplings Vnc« (relative to

I

i n

Jnco) to establish sequential correlations (Cavanagh et al. 1996), strong "C transverse

relaxation due to the large 13C CSA tensor, limiting the use of the HNCOCA-type experiments

to spectrometers operating at lower polarizing magnetic field strength, and the need to back-

exchange 2HN to protons.

Recently we proposed a new 3D multiple-quantum HACACO experiment (Eletsky and Pervushin

2002) designed to complement this method. The use of the 3D MQ-HACACO experiment with

proteins of larger molecular weight relies on the favorable relaxation properties of the 1Ha-

13Ca multiple quantum coherences (Grzesiek et al. 1995a; Swapna et al. 1997; Xia et al. 2000) to

record 1Ha and 13Ca chemical shifts and transfer the 'Ha magnetization to the 13C spins using

one double- constant-time evolution period (Swapna et al. 1997) and the direct acquisition of the

13C resonances (Serber et al. 2000; Serber et al. 2001). High sensitivity and a simple anti-phase

doublet structure of the 13C resonances enable effective signal acquisition even for large

88

protonated proteins on the order of 50-60 kDa at room temperature. Sequential connectivities

derived from weak sequential cross peaks in the 3D TROSY-HNCA experiments are

complemented or completely replaced by strong one bond 13Ca-13C correlation cross- peaks

obtained from the MQ-HACACO experiment, thereby "bridging" intra-residual cross- peaks of

the HNCA experiment with the strong inter-residual cross peaks derived from the HNCO

experiment. Although 3D MQ-HACACO effectively resolves the problem of overlapping inter-

and intra-residual 13CU resonances (Salzmann et al. 1999b), bridging of the HNCA and HNCO

experiments still relies on the consecutive matching of single pairs of the C

(HNCO/HACACO) and 13Ca (HACACO/HNCA) chemical shifts. While good spectral resolution

and large dispersion of the 13C chemical shifts facilitate the correlation between HNCO and

HACACO experiments, broader lines along the 13Ca dimension in the 3D HACACO and 3D

HNCA experiments might result in significant ambiguities in the assignment.

Development of 3D TROSY-HN(CA)HA and 4D TROSY-HACANH

experiments

A new method for backbone resonance assignment suitable for large proteins with the natural 'H

isotope content is proposed based on a combination of the most sensitive TROSY-type triple-

resonance experiments. These techniques include TROSY- HNCO, 13C detected 3D multiple-

quantum HACACO and the newly developed 3D TROSY multiple-quantum-HN(CA)HA and 4D

TROSY multiple-quantum-HACANH experiments. The favourable relaxation properties of the

multiple-quantum coherences, signal detection using the l3C anti phase coherences, and the use

of TROSY optimize the performance of the proposed set of experiments for application to large

protonated proteins. The method is demonstrated with the 44 kDa uniformly 15N,13C-labe1ed and

fractionally (35%) deuterated trimeric B. Subtilis Chorismate Mutase and is suitable for proteins

with large correlation times but a relatively small number of residues, such as membrane proteins

embedded in micelles or oligomeric proteins.

The remarkably good sensitivity obtained in the MQ-HACACO experiment (Eletsky and

Pervushin 2002) applied to large proteins with natural lW isotope content encouraged us to

89

(a)y -y

lu * >l

h

H

ÏH t2 t2il i

*k

15,

y b

'N x\%\ T2 | t2 I

13C<* III â

Î3CO

PFG j_iiG1 Gl t^3

SEDIiCË

AJ.U.Ci Ci Gj Gj

(b) t> ,i ,

'H | j j |2 lT5,

4>5<t>6

LUIL<f>r

ISN HI 44W

13>-ia *3

13.

Hr^li 1

CO

PFG

SEKUCE:

ic3

j_t4-4

90

Figure 1. Experimental scheme for (a) 3D TROSY MQ-HN(CA)HA and (b) 4D TROSY MQ-HACANH

experiment. The radio-frequency pulses on 'H, 13Cu, isN and l3C are applied at 4.7, 54.6, 119 and 173.6

ppm, respectively. Narrow and wide black bars indicate high power 90° and 180° pulses. Filled sine bell

shapes on the line marked *H indicate water-selective 1 ms Gaussian 90° pulses applied to align the water

magnetization in +z direction before data acquisition (Grzesiek and Bax 1993). The residual transverse

water magnetization is dephased using 3-9-19 composite pulses (Liu et al. 1998). Selective 13C

decoupling is achieved by SEDUCE-1 (Shaka ct al. 1983) at a field strength of yfi2 = 1.65 kHz. The line

marked PFG indicates the duration and strength of pulsed magnetic field gradients applied along the z-

axis: d: 0.9 ms, 40 G/cm; G2: 1 ms, 80 G/cm; G3: 0.5 ms,50 G/cm and G4: 0.9 ms, 40 G/cm. In (a) open

shapes on the line marked 'H indicate the 'HN band-selective 1.5 ms excitation E-Burp2 pulse (Geen and

Freeman 1991) with y#y = 2.73 kHz with the phase (J)2(time point b) and the time-reversed excitation E-

Burp2 pulse with the phase x (time point c). The centre of excitation for the *HN band-selective pulses is

placed at 9 ppm, so that all amide protons resonating between 11.5 and 6.5 ppm are returned to the +z

axis. A filled sine bell shape on the line marked 13Ca indicate 1.6 ms refocussing 180° Re-Burp pulse

(Geen and Freeman 1991) with yBt = 3.91 kHz. The delays are T = 14 ms, ô] = 2.7 ms, ô2 = 14 ms, ô3 =

3.5 ms. The phase cycle is: (J), = {x}, fa= {-x}, fa= {4(-x), 4x}, fa= {y,-y,-x, x}; fa = {-y};fa= {y}; fa

= l-yl; <|W = (y. -y. -x, x, -y, y, x, -x}. A phase-sensitive spectrum in the 15N(f0 dimension is obtained by

recording a second FID for each ty value, with fa = {y, -y, x, -x}, fa = {y}, fa - {-y} and fa = {y}.

Quadrature detection in the 'H (f2) dimension is achieved by the States-TPPI method (Marion et al. 1989)

applied simultaneously to the phases fa and fa. In (b) the delays arc 7-14 ms, ô\ = 2.7 ms, 62 = 12 ms

and Ô3 = 3.5 ms. The phase cycle is: fa = {x}, fa= {x}, fa= {4x, 4(-x)}, fa- {y, -y, -x, x}; fa= {-y}; fa -

ly}; $7 = {-y}; <t>rec - (y> -y> _x* x> -y» y> x» -*} A phase- sensitive spectrum in the 15N(fi) dimension is

obtained by recording a second FID for each ^ value, with <J>4 = {y, -y, x, -x}, fa = {y}, fa = {-y} and fa =

{y}. Quadrature detection in the 'H (f2) and nC(t}) dimensions is achieved by the States-TPPI method

(Marion et al. 1989) applied to the phases fa and <j>2, respectively.

explore the possibility of using 'Ha as well as l3Cn chemical shifts to reduce ambiguities in

correlating the backbone resonances. Here, we describe two experiments, 3D multiple-quantum

TROSY HN(CA)HA, correlating the chemical shifts of the 'H^, 15N, and intra-residual 'H

spins, and 4D multiple-quantum TROSY HACANH, correlating 'H,N, 15N;, 13C" and intra-

residual LH,a spins. Previously, enhancement factors of 1.5 to 1.7 were observed for 24 kDa

91

(a) 3D TROSY HN(CA)HA (b) 3D TROSY HA(CA)NH a-helix

EI7125.23

T16in.65

t~?

0

&

fci

HIS116.74

fr g-Ô,y-r-

(7

o

E19119.66

C&-

116113 65

£17125.23

<m

EIS116,74

va

E19119,66

0»3,6

-4.0

0

-4.4

O-4,8

7.57 9.16 8 85 '52 7.57 9.16 8,85

tt^H^lppm]7.52

(c) 3D TROSY HN(CA)HA (d) 3D TROSY OA(CA)MH ß-sheet

12474

K

' Y?!11*73

O

^

M92127 25

*O

M93109,90

/1X

<D

00

,R90124 74

O

&

O

&

&

V9111973

©

* i

M92127,25

„0

*J! (

O

C

M93109,90

Co>

-4,4

-4,8

[Pp]

-5,2

5.6

8,71 8,88 8,92

%{'H**)fppm]1019 8 71 8.92

»4{lH«)[ppm]10,19

92

Figure 2. Comparison of corresponding 2D fœ2(1Hct), (o3,4('HN)] strips from 3D TROSY MQ-

HN(CA)HA (a and c) and 3D TROSY MQ-HA(CA)NH experiments (b and d) recorded with a 1 mM

solution of the uniformly lsN,13C-labeled and fractionally 35% deuterated trimeric 44 kDa B. Subtilis

Chorismate Mutase (BsCM). The strips were taken at the 15N chemical shifts that assigned to residues 16-

19 and 90-93, located in the a-helical and (3-sheet regions of BsCM, respectively. They are centered about

the corresponding amide proton chemical shifts and have a width of 131 Hz along 'H dimension. The 4D

TROSY MQ-HACANH experimental scheme of Fig. lb was acquired with fj = 0 resulting in the 3D

TROSY MQ-HA(CA)NH experiment. For both experiments 40(fi) x 55(r2) x 512(f3/f4) complex points

were accumulated, with fi^^N) = 26.64 ms, t2aisa(]Ha) = 22.89 ms and r3/4nu*(1HN) = 53.25 ms and 8

scans per increment resulting in a total measuring time of - 22 hours per spectrum. The sequential

connectivities are indicated by dashed lines.

13C,15N-labeled protein (FimC) when the standard triple-resonance schemes were upgraded to

TROSY (Salzmann et al. 1998). We expect similar factors in our case.

Figures 1 a and lb show the experimental schemes of 3D TROSY MQ-HN(CA)HA and 4D

TROSY MQ-HACANH, utilizing the "out-and-back" and "out-and-stay" methods, respectively

(Cavanagh et al. 1996). The product operator description of the TROSY coherence transfer

pathway in 3D TROSY MQ-HN(CA)HA is given by Eq. (1 ).

^-4C?H^NA--H?)^-4C^HfNz(-~H?)[t2]^-2C^Nz(-~H?) (1)

^_2N±C?(|-tff)[rJ^//^ + A^)[r1]

where H, N and C represent product operators of !H, 15N and 13Ca spins, respectively. Single

E E

transition operators H"(~ + N7) and N±(— + Hz) represent slowly relaxing TROSY

components of the 15N-'HN multiplet (Pervushin et al. 1997). The experimental scheme of Fig. la

deviates from the corresponding TROSY HNCA experiment (Salzmann et al. 1998) at time point

a, where the polarization transfer period x^ is inserted, resulting in the development of the C

coherence antiphase relative to the directly attached [Ha spin. This coherence is subsequently

converted by the pulse fa to the 13ca-'Ha multiple quantum operator with favourable relaxation

93

properties, used to record the !H chemical shift. The high power 90° ]H excitation pulses applied

at time points b and c are bracketed by lHN band-selective 90° E- Burp2 pulses (Geen and

Freeman 1991), in order to preserve the spin state of the *HN spins and to prevent exchange

between TROSY and anti-TROSY l5N multiplet components. In the middle of t2 13Ca chemical

shift is refocused by a 180" 13Ca band selective Re-Burp pulse, (Geen and Freeman 1991)

reducing signal attenuation due to the evolution of 13C homonuclear passive /-couplings between

13Ca and 13CP. The standard ST2-PT element (time points d and e) (Pervushin et al. 1998) is used

to correlate the frequency of the TROSY component of the 15N multiplet with the corresponding

component of the 'H^ multiplet.

Figure lb shows the 4D TROSY MQ-HACANH experiment, based on the "cVH" HMQC

double-constant-time magnetization transfer scheme and TROSY detection of the 15N and 1UN

chemical shifts. The product operator description of the relevant coherence transfer pathways is

given by Eq. (2).

H" -> 2HaxC* -» -2H*Cay[t2,t^ -> 2CayNz -> 2CfN7

F F(2)

->-2CfN±(--Hf)[r1]^H_A'(- + iV()[f4J

The experimental scheme of Fig. lb is best suited for the methine Ca carbons, since full use of

the double-constant-time period can be made for the 'Ha and 13Ca chemical shift evolution. A

modification of the experimental scheme of Fig. lb is required in order to detect with optimal

sensitivity the methylene Ca carbons which occur in glycine residues. This version of the

HACANH corresponds to the first part of the (HA)CA(CO)NH experiment introduced by

(Swapna et al. 1997) and will not be described in detail in this paper. After the double-constant-

time period 2x2, the resulting signal is determined by magnetization transfer from the lHa to l3Ca

and subsequently to 15N spins, as well as attenuation due to passive homonuclear J- couplings of

1Ha to !Hß, 'HN spins and 13Ca to 13Cpspins. Neglecting effects of relaxation and small

heteronuclear J-couplings between 'H and 13Cß and13Ca and 'Hp spins, the HMQC part of the

transfer function is given by Eq. (3):

/(T2, r, ) = sin(8/nx:, t3

)2 sin(2^/cf^

r2 ) cos(2tc/h?h, t2 )

x cos(2tiJh,,h!I t2 )cos(27tJc«n ^

r2 ) cos(2nJc;,cj1t2 )

94

4D TROSY HACANH

(a) T16

«>2dH«H,5£2«>3( Ca)=57,'513

T16

(c) E17

W3(13ca^5gg26J

(e) El8

<92f%aH.*9ö)3(13faHÖ201 |

' EIS i

l I

12,00 9,00

118.00

0,(«N)

112.00

112 00

118,00

124.00

6,00

pr

E17

(b)

El?

f»l(,3N>=125.210'»4<lHN)=9J63

« T16

EIS

»I(15M)r» 16,710

ca4(1HIV 8,835 ,

*

*E17

(f)*

B.E18

El9

- E19

8>1(15N>ÎI9.96Q

«4J,1 ^7.537

.?

4.00

5.00

5.50

-1

4,00

(jppml

5.00

5,50

4,00

©at'H11)

iPpm]

5.0Û

^f 5,50

ontWXppml

60.00 56.00 52,00

©rf^ppm]

95

Figure 3. 2D [œ,(15N), co4 ('HN)] (a, c and e) and Lcù/H"), g>3 (13Ca)l (b, d and f) planes taken at the

positions of the 'Ha, l3Ca chemical shifts and the 15N, 'HN chemical shifts, respectively, of residues 16-19

showing the procedure of sequential assignment of the backbone resonances using the 4D TROSY MQ-

HACANH experiment of Fig. lb. 16 (fi) x 30 (r2) x 40 (f3) x 512 (f4) complex points were accumulated,

with rim,lx(,5N) = 10.65 ms, WLHU) - 22.89 ms, r3mjx(13CH) = 22.21 ms, fw('HN) = 53.25 ms and 4

scans per increment resulting in the total measuring time of 5 days and 20 hours. Sequential connectivities

are indicated by arrows. In each 2D plane two chemical shifts of each residue associated with an

individual cross-peak are indicated.

Transfer efficiencies of 0.28 and 0.21 are obtained for the a-helical and ß-sheet secondary

structure elements of a protein at the optimal transfer delays (x3-3.4 ms and 2t2 = 27 ms),

determined using V(Ha-HN) = 4 Hz and 9 Hz for a-helix and ß-sheet, respectively, (Wuthrich

1986) and V(Ha-Hß2) - 12 Hz, '/(Ca-Cp) - 37 Hz, J(Ca-N,) ~ 9 Hz and /(Ca-N,_y) - 7 Hz for both

secondary structure elements (Cavanagh ct al. 1996). In the experiment, a somewhat shorter

delay (2i2 = 24 ms) was used to compensate the shift of the maximum in Eq. (3) due to transverse

relaxation. Since the full delay can be used to record chemical shifts of 'H" and l3Ca resonances,

excellent resolution along the corresponding 'Ha and 13Ca dimensions can be obtained.

Sufficiently sensitive 3D TROSY MQ-HN(CA)HA, 3D TROSY MQ-HA(CA)NH and 4D

TROSY MQ-HACANH spectra are obtained for the 44 kDa uniformly 15N,13C-labeled and

fractionally 35% deuterated chorismate mutase from Bacillus subtilis (BsCM) (Eletsky et al.

2001) using the experimental schemes of Figure 1. Figure 2 compares the intensities of cross-

peaks stemming from residues in the a-helical and ß-sheet secondary structure elements of

BsCM, obtained with the 3D TROSY MQ-HN(CA)HA and 3D TROSY MQ-HA(CA)NH

experiments. Although sufficiently strong intra-residual 1HN-1Ha cross-peaks are identified in

both experiments for most residues (see Fig. 2), significantly different performance of these two

experiments is observed for residues located in a-helices and ß-sheets. Overall, higher sensitivity

is detected for the a-helical regions of BsCM compared to the ß-sheet or turn regions in the

spectra measured with the 3D TROSY MQ-HA(CA)NH experiment (Fig. 2 b and d). Rather

similar relative cross-peak amplitudes for a-helical and ß-sheet are observed in spectra measured

with the 3D TROSY MQ-HN(CA)HA experiment (Fig. 2 a and c). Although the average S/N

96

over all residues of BsCM is slightly higher for MQ-HA(CA)NH than MQ- HN(CA)HA, the long

double-constant-time period x2 renders the former experiment more sensitive to the deleterious

effects of homonuclear passive J-couplings and transverse relaxation induced by dipole-dipole

interactions of the 'Ha spins with remote proton spins. These factors result in a significant

decrease of S/N in the ß-sheet regions. Thus, for practical purposes the choice between the

experimental schemes of Fig. 1 a and b can be determined by the predominance of a certain type

of the secondary structure in the studied protein or both experiments can be tried for the best

results.

The use of TROSY in the proposed experiments is justified by a direct comparison of signal-to-

noise ratio in TROSY and conventional 'HN-dccoupled MQ-HA(CA)NH experiments. For all

residues of 13N-, 13C-, 2H(35%)-labeled 44 kDa BsCM the TROSY version exhibited S/N

enhancement factors over the conventional implementation in the range of 1.1 to 1.5. These

enhancement factors agree with to those previously observed for a 24 kDa 13C,15N- labeled

protein (FimC) when the standard triple-resonance schemes were replaced with TROSY

(Salzmann et al. 1998). It should be noted that for fully protonated large proteins smaller and

none-uniform S/N enhancement factors in TROSY versions can be expected. This is attributed to

the dominant contribution of the proton-proton spin-flip interactions to the TROSY 1SN

relaxation rate (Kontaxis et al. 2000).

The presence of the ^c^H* HMQC double-constant-time period in the experimental scheme of

Fig. lb opens an attractive possibility of running TROSY MQ-HACANH in a 4D mode without

compromising sensitivity of the experiment due to additional time incrementations, besides the

v2 loss in S/N due to the quadrature detection in the 13Ca dimension. Figure 3 illustrates the

quality of the 4D spectrum measured using the 4D TROSY MQ-HACANH experiment with the

44 kDa uniformly 15N,13C-labeled and fractionally (35%) deuterated BsCM. Provided that

sensitive sequential correlations are observed, the 4D TROSY MQ-HACANH experiment alone

is sufficient to assign backbone resonances. In this experiment both 'Ha and 13Ca chemical shifts

are used as reference for sequential connectivity assignment, which is less ambiguous and more

reliable than only 13Cn chemical shift, as used in the conventional HNCA experiment. Figure 3

97

illustrates the process of sequential assignment obtained for a difficult case of Thrl6, Glul7,

Glu18 and Glu19 repeat.

Overall, the sole use of the 4D TROSY MQ-HACANH spectrum allowed establishing the

sequential connectivities between 51 residues. The longest found continuous stretches of the

connected residues are Thrl6 to Glu34, Ala59 to Ser66, Trp68 to Val71 and ArglOS to Argll6.

For the other residues either only intraresidual cross-peaks were found due to the lower intensity

of the sequential cross-peaks (39 residues) or the corresponding strips were void of any cross-

peaks (15 residues). For 20 residues the found connectivities were ambiguous. The use of the 3D

HN(CA)HA in addition to the 4D TROSY MQ-HACANH spectrum helped to establish the

sequential connectivities for Arg4, Ile6, Lys38 and His54.

Combined use of the 3D TROSY-HNCA, 3D TROSY-HNCO with 3D MQ-

HACACO

The combined use of the 3D TROSY-HNCA, 3D TROSY-HNCO, 3D MQ-HACACO (Eletsky

and Pervushin 2002) experiments for residue specific assignment of the 'Hu, 'Hn, 15N,

13Caand13C backbone resonances is now effectively complemented by the 3D TROSY MQ-

HN(CA)HA and 3D TROSY MQ-HA(CA)NH experiments. The 3D HA(CA)NH experiment is a

reduced version of 4D HACANH of Fig. 1. In this method sequential connectivities derived from

weak sequential cross peaks in the 3D TROSY-HNCA experiments are replaced by strong one

bond 13ça.1:,c' correlation cross-peaks obtained from the MQ-HACACO experiment. As shown

in Fig. 4, sequential correlations between protein backbone resonances can be obtained by

1 ^

matching the C chemical shifts in the HNCO/MQ-HACACO pair of experiments, obtained

with sufficiently high resolution along the 13C dimension, with subsequent matching of the 13Ca

and 'Ha chemical shifts in the MQ-HACACO/MQ-HACANH or MQ-HACACO/ MQ-

HN(CA)HA experiments.

98

L21 L2I Q22 Q22 K.23 K23

i^Tl

4.'il

TV,1$

"HEdl)r*v

«-*-5=iLi?

Z<;

oïU'H«) < CIppml 3:

>- <!rt u

<pi ~

5i—

o

-r s

55-o

4

-

ü

£0^ *

•* A.--!

ü>2( 1\'il) ..»^

'B2P'

[ppml £-

<

*

65

.—\ 178.9 /•£ 179 3 181 2

ÎJ** .ffl!

S**ï^,.

1745

1 "'-y

c^t'-^cO)ûi

©C-

|ppm| s s

178

180

«ôs?..**

8 69 8.06

W3(>HN>[ppm]8 62

Figure 4. Method of assignment of the backbone 'H'\ l3C\ uCa and 15N resonance in uniformly l3C, l5N-

labeled and fractionally deuterated large proteins using a combination of 3D TROSY-HNCA (Salzmann et

al. 1998; Eletsky et al. 2001), 3D TROSY-HNCO (Salzmann et al. 1999b; Salzmann et al. 1999c), 3D

MQ-HACACO (Eletsky and Pervushin 2002) and 4D HACANH experiments (shown in red) of Fig. lb.

Arrows indicate a sequential walk through 2D l3Ca-'HN, 13C'-'HN, 13ca-13C and 'Ha-'HN strips taken at

the positions of the corresponding ,5N, 13Ca and 'Haresonances in 3D TROSY-HNCA, 3D TROSY-

HNCO, 3D MQ-HACACO and 4D HACANH spectra, respectively. The assignment relies mostly on the

connectivities shown by solid blue, green and magenta arrows. Occasionally the sequential connectivities

of TROSY-HNCA, shown by dashed green arrows, are used.

99

The assignment process begins with the identification of the intraresidual and sequential 13Ca and

l3C chemical shifts in HN strips taken from the TROSY-HNCA and TROSY-HNCO

experiments, respectively. The intraresidiual 'Ha chemical shifts are identified using the 3D

HA(CA)HN spectrum. If the 4D HACANH spectrum is available, the intraresidual 'Hând

l3Ca chemical shifts are found simultaneously and then 13Ca chemical shifts are verified using the

TROSY-HNCA spectrum. At the next stage, connectivities between HN strips are established

using 3D MQ-HACACO spectrum and verified for the cases where the sequential cross-peaks are

resolved in TROSY-HNCA (see Fig. 4). Thus connected HN strips are then mapped to the

primary sequence of protein using the program MAPPER (Guntert et al. 2000). At this point it is

very helpful to obtain the TROSY-HNCACB spectrum (Salzmann et al. 1999b) to reduce

ambiguities in mapping. Thus, the minimal set of the triple-resonance experiments includes 3D

TROSY-HNCA, 3D TROSY-HNCO performed with standard transfer delays (Cavanagh et al.

1996), 3D MQ-HACACO experiment and 3D HA(CA)NH experiment of Fig. 1. If spectrometer

time and sample stability permit, the 3D TROSY-HNCA and 3D HA(CA)NH experiments can be

replaced by the single 4D HACANH experiment.

An attractive feature of this method is the requirement of only one fully protonated or partially

deuterated protein sample to assign backbone resonances, which can be subsequently used for 3D

structure determination. For proteins with large correlation times but a relatively small number of

residues, such as membrane proteins embedded in micelles or oligomeric proteins this method

alleviates the need to acquire the low sensitivity 13Cot-constant-time TROSY-HNCA experiment

(Salzmann et al. 1999a) and two or more of time-consuming 4D TROSY-type triple-resonance

experiments (Konrat et al. 1999).

References

Cavanagh, J., Fairbrolher, W. J., Palmer, A. G. and Skelton, N. J. (1996) Protein NMR Spectroscopy: Principles and

Practice. New York, Academic Press.

Chook, Y. M., Gray, J. V., Ke, H. M. and Lipscomb, W. N. (1994) J. Mol. Biol., 240, 476-500.

Eletsky, A., Heinz, T„ Moreira, O., Kienhöfer, A., Hilvert, D. and Pervushin, K. (2002) J, Biomol NMR.

Eletsky, A„ Kienhöfer, A. and Pervushin, K. (2001) J. Biomol. NMR, 20, 177-180.

Eletsky, A. and Pervushin, K. (2002) J. Biomol. NMR.

100

Gardner, K. H. and Kay, L. E. (1998) Annu. Rev. Biophys. Biomolec. Struct., 27, 357-406.

Geen, H. and Freeman, R. (1991)./. Magn. Reson., 93, 93-141.

Grzesiek, S. and Bax, A. (1993) /. Am. Chem. Soc, 115, 12593-12594.

Grzesiek, S., Kuboniwa, H., Hinck, A. P. and Bax, A. (1995a) J. Am. Chem. Soc, 117, 5312- 5315.

Grzesiek, S., Wingfield, P., Stahl, S., Kaufman, J. D. and Bax, A. (1995b) J. Am. Chem. Soc, 117, 9594-9595.

Guntert, P., Dötsch, V„ Wider, G. and Wuthrich, K. (1992),/. Biomol. NMR, 2, 619-629.

Guntert, P., Salzmann, M., Braun, D. and Wuthrich, K. (2000) J. Biomol. NMR, 18, 129-137.

Konrat, R„ Yang, D. W. and Kay, L. E. (1999) J. Biomol. NMR. 15, 309-313.

Kontaxis, G., Clore, G. M. and Bax, A. (2000) J. Magn. Reson., 143, 184-196.

LeMaster, D. M. (1994) Prou. Nucl. Magn. Reson. Spectrosc, 26, 371-419.

Liu, M. L., Mao, X. A., Ye, C. H., Huang, H., Nicholson, J. K. and Lindon, J. C. (1998) /. Magn. Reson., 132, 125-

129.

Marion, D., Tkura, M., Tschudin, R. and Bax, A. (1989) J. Magn. Reson., 85, 393-399.

Mulder, F. A. A., Ayed, A., Yang, D. W., Arrowsmith, C. H. and Kay, L. E. (2000) /. Biomol. NMR, 18, 173-176.

Pervushin, K., Riek, R„ Wider, G. and Wuthrich, K. (1997) Proc Natl. Acad. Sei. U. S. A., 94, 12366-12371.

Pervushin, K., Wider, G. and Wuthrich, K. (1998) J. Biomol. NMR, 12, 345-348.

Salzmann, M., Pervushin, K., Wider, G., Senn, H. and Wuthrich, K. (1998) Proc. Natl. Acad. Sei. U. S. A., 95,

13585-13590.

Salzmann, M., Pervushin, K., Wider, G., Senn, H. and Wuthrich, K. (1999a) J. Biomol. NMR, 14, 85-88.

Salzmann, M., Wider, G., Pervushin, K., Senn, H. and Wuthrich, K. (1999b) J. Am. Chem. Soc, 121, 844-848.

Salzmann, M., Wider, G., Pervushin, K. and Wuthrich, K. (1999c) J. Biomol. NMR, 15, 181- 184.

Serber, Z., Richter, C. and Dotsch, V. (2001) Chembiochem, 2, 247-251.

Serber, Z., Richter, C, Moskau, D., Bohlen, J. M., Gerfin, T., Marek, D„ Haberli, M., Baselgia, L., Laukien, F.,

Stern, A. S., Hoch, J. C. and Dotsch, V. (2000) J. Am. Chem. Soc, 122, 3554-3555.

Shaka, A. J., Keeler, J., Frenkiel, T. and Freeman, R. (1983) .1. Magn. Reson., 52, 335-338.

Sban, X., Gardner, K. H., Muhandiram, D. R., Rao, N. S., Arrowsmith, C. H. and Kay, L. E. (1996) J. Am. Chem.

Soc, 118, 6570-6579.

Swapna, G. V. T., Rios, C. B., Shang, Z. G. and Montclione, G. T. (1997) J. Biomol. NMR, 9, 105-111.

Venters, R. A., Farmer, B. T., Fierke, C. A. and Spicer, L. D. (1996) J. Mol. Biol, 264, 1101- 1116.

Wuthrich, K. (1986) NMR ofProteins and Nucleic Acids. New York, Wiley.

Xia, Y. L., Kong, X. M., Smith, D. K., Liu, Y., Man, D. and Zhu, G. (2000) J. Magn. Reson., 143, 407-410.

Yamazaki, T., Lee, W., Arrowsmith, C. H., Muhandiram, D. R. and Kay, L. E. (1994) J. Am. Chem. Soc, 116,

11655-11666.

Yang, D. W. and Kay, L. E. (1999) J. Am. Chem. Soc, 121, 2571-2575.

101

Chapter IV

Side-chain H and C resonance assignment in partially deuterated proteins

using a new 3D 13C-detected HCC-TOCSY

Kaifeng Hu, Beat Vogeli and Konstantin Pervushin

Laboratorium für Physikalische Chemie, Swiss Federal Institute ofTechnology, ETH- Honggerberg, CH-8093

Zürich, Switzerland.

102

Seite Leer /

Blank leaf

103

Introduction: strategies of side chain assignment

Recently," C-detected NMR spectroscopy [2-5] has been proposed as an attractive alternative for

studying large macromolecules [1, 6], Thus, two-dimensional (2D) 13C-start and 13C-observe

TOCSY NMR experiments were successfully used for the assignment of side-chain aliphatic 13C-

resonances in a completely deuterated protein with a molecular weight of 44 kDa [1] and

11 1 ^

measurement of a number of C- C residual dipolar couplings [7], As the protein size increases,

the peak overlap becomes a significant problem, especially for monomeric proteins, which have a

large number of residues. Simultaneous multiple-band-selective homonuclear 13C- and 15N-

decoupling during both chemical shift evolution periods and signal acquisition can alleviate this

problem to some extent by simplifying the multiplet patterns (Vögeli et al., submitted). The

introduction of a third chemical shift dimension in a combination with extensive spin decoupling

is expected to offer further reduction in peak overlap.

Development of 3D in-phase and sensitivity enhanced HCC-TOCSY

Here, we propose a new 'H-start and 13C-observe 3D HCC-TOCSY for the unambiguous

resonance assignment of the side-chain *H and 13C resonances. This experiment extends 2D C-

13-start and C-13-observe TOCSY type experiments [1]. Introduction of the third lH dimension

to 2D TOCSY i) reduces the peak overlap and ii) increases the sensitivity per unit time, even for

highly deuterated ( > 85%) protein samples, which makes this new method also an attractive tool

for the side-chain H and C assignment of large partially deuterated proteins. This method is

demonstrated with a 16 kDa 15N,13C-labeled non-deuterated apo-CcmE and a 48 kDa uniformly

15N,13C-labeled and fractionally (-90%) deuterated dimeric sFkpA. It is predicted that this

method is suitable for the assignment of methyl 13C and *H chemical shifts of methyl protonated,

highly deuterated and ~C-labeled proteins with even higher molecular weight. 3D HCCH-

TOCSY experiments were designed for protonated proteins in order to correlate the chemical

shifts of H1, C1 and H', which is bonded to the adjacent nonfrequency-labeled C1 [8, 9].

Comparing to the HMCM[CG]CBCA and amino-acid specific HMCM(CGCBCA)CO "out-and-

back" experiments proposed by Kay and coworkers for assignment of methyl groups in a 723

residue protein [10], we expect that the 13C-detected "out-and-stay" HCC is better-suited and can

104

be regarded as a general route for the assignment of methyl 13C and lU chemical shifts for methyl

protonated, highly deuterated and 13C-labeled proteins with high molecular weight [11] [12]. A

comparison to a primary single-quantum 13C-detected 3D HCC-TOCSY reported on a protonated

14 kDa protein alluded to potential application of 13C-detected spectra [13] [1]. Here, we

correlate H1, C and Cj in large partially deuterated or protonated proteins. The use of the *H-13C

multiple-quantum coherence during the chemical shift labeling period for the indirect *H and 13C

dimensions offers more favorable relaxation properties compared to the 13C single-quantum

coherence, as demonstrated for the 13C'-detected 3D multiple-quantum-HACACO, 3D TROSY

multiple-quantum-HN(CA)HA and 4D TROSY multiple-quantum-HACANH experiments [14].

The main advantage of the proposed approach is a possibility to use the resolved 13C multiplet

pattern along the directly acquired dimension in order to match resonances from different strips.

Besides reducing the overlap by introducing the *H dimension, an additional advantage of this

H-start 3D experiment is a significant reduction of the interscan delay from 2.5 ms to less than 1

ms in comparison to its 13C-start 2D counterpart [1] due to faster equilibrium magnetization

recovery of *H compared to 13C in the 2H-13C moieties [15]. Neglecting transverse relaxation,

theoretically, this new 'H-start and 13C-observe 3D HCC-TOCSY can be expected to have

comparable sensitivity per unit time to its 13C-start and 13C-observe 2D version as follows:

T = (S/N)hcc/(S/N)cc=\^.Zl.P (1)

where (S/N)HCC and (S/N)cc are signal-to-noise ratio in 3D HCC-TOCSY and 2D CC-

TOCSY spectra, respectively. Tf and T" are longitudinal relaxation times for 13C and 'H, with

typical values for Tf of about 3- 4s and for T" of 0.5-Is. Yc and Yh are gyromagnetic ratios for

13C and ]H, P indicates the proton level in the partially deuterated protein sample. For 90%

deuterated protein samples, P is equal to 0.1, resulting T range from 0.7 to 1.1. Optimal

sensitivity for the given protein size could be obtained by varying the deuteration level [16]. In

the construction of HCC-TOCSY, we pursued two alternative goals resulting in two

complementary experiments. In the first experiment, referred further on as IP-HCC-TOCSY, a

clean, in-phase multiplet pattern in the directly acquired dimension is produced, facilitating

sequential matching of 2D strips in the process of resonance assignment. In the second

experiment, referred as SE-HCC-TOCSY (SE stands for "sensitivity-enhanced"), the spectral

105

sensitivity is maximized, which is achieved by relaxing the requirement of "in-phase" appearance

of the resulting 3D spectra in the directly acquired dimension.

4>l <j>3

| vl| l|!H iwr 2

+, +<

,3cîhJJLdi

SE A,

WALTZ-16

FLOPSY-16

H WAIT/.-161/2H WALTZ-16

PFG (IP only)

Gi G!

Figurel. Pulse sequence of 13C-detected 3D-HCC TOCSY, the radio-frequency pulses on ]H, 13C and 2H

are applied at 2.5, 35 and 3.0 ppm, respectively. Narrow and wide bars indicate non-selective 90° and 180°

pulses (black pulses are applied in both experiments; white pulses are only applied in the indicated

experiment). 'H- and 2H- decoupling is achieved with WALTZ-16 [25] at a field strength around yBi = 2.5

kHz. Unless indicated otherwise, all radio-frequency pulses are applied with phase x. The phase cycle is:

fa = {X}; fa. = {x, -x, x, -x, x, -x, x, -x}; fa = {y, y, y, y, -y, -y, -y, -y }; <j>4 = {x, x, -x, -x }; fatc = {x, -x, x,

-x, -x, x, -x, x}. In the IP experiment: The delays are Ti = l/(2/Cc) = 14 ms, r2 = 1/(27Ch) = 4 ms. The

FLOPSY-16 mixing time is 16.96 ms at yB2 = 0.83 kHz. Pulsed field gradients indicated on the line

marked PFG are applied along the z-axis with duration of 0.9 ms and strength of 40 G/cm. Quadrature

detection in the indirect 13C (fj) dimension and 'H (t2) dimension is achieved by the States-TPPI method

[26] applied to the phases fa and fa, respectively. In the SE experiment: The delays are: t\ is set to

smaller than l/(27Cc) = 14 ms (depending on the relaxation properties of the multiple-quantum coherence),

T2 = 1/(1/Ch) = 4 ms. The FLOPSY-16 mixing time is shortened to 16.96/2 ms - 8.48 ms at yB2 = 0.83

kHz. Pulsed field gradients are not applied. A phase-sensitive spectrum in the indirect ]'C(ti) dimension is

obtained using the ECHO/ANTIECHO method by recording two FIDs for each h value with fa = {x} and

fa = {-x}, respectively. Quadrature detection in the !H (t2) dimension is achieved by the States-TPPI

method applied to the phase fa.

106

Figure 1 shows the experimental scheme of the 3D 13C-detected HCC-TOCSY. The initial *H

polarization is transferred to 13C by the INEPT step generating the 'H-13C multiple-quantum

coherence during the chemical shift labeling period for the indirect 13C(r;) and lH(t2)

dimensions. In the IP experiment, t\ is set to 14 ms = l/(2/Cc) in order to maximize the pure 13C

in-phase operator at the beginning of the FLOPSY-16 13C-13C mixing period [20]. The pulse field

gradients (PFG) are applied to select the in-phase operators before and after the TOCSY mixing.

It should be noted that even with PFG based operator selection homonuclear 13C zero-quantum

terms insensitive to PFG can survive[21, 22], resulting distortion of the pure in-phase appearance

of the crosspeaks [7, 23]. Currently no efforts are made to prevent this. After a 13C-read pulse,

the NMR signal is detected while decoupling of lH or 2H (for deuterated samples) spins. In the

SE-HCC-TOCSY experiment, we aim to minimize relaxation and to preserve all magnetization

transfer pathways. Therefore, zx is set to much less than 14 ms = l/(2/cc) (e.g. 5 ms was used in

the spectrum of Figure 4). No gradients are applied before and after the mixing period to retain a

superposition of in-phase and antiphase coherence significantly improving the signal-to-noise

ratio. The product operator description of the coherence transfer pathway is given by diagrams 2

and 3 for the IP and the SE experiment, respectively:

FLOPSY-16 >CJ^CJ[t]

H[ ->-H'v -*2H\C\ ->-2//;c;n(amE/2 + bmC'")[t[,t2]m

~^-Ci±ll(am\2Tl]E/2 + bm[2Tl]C'zn)tWPSY~* >CiU(c'E/2+ d'Clz)[t3]

(2)

(3)

IF and C stand for the spins of the hydrogen and carbon atom of the bond i, from which the

magnetization pathway starts. C stands for the directly detected carbon atom. Cm and C1 are spins

involved in the ./-coupled network of the spins C and Ö. Due to the coherence order selective

mixing achieved by isotropic FLOPSY sequence [24], the "minus" operators evolving during ti

are transferred exclusively to "minus" operators, which is used for signal acquisition. Quadruture

detection is achieved by flipping "plus" and "minus" operators with a 13C 180° pulse before

TOCSY mixing. The mixture of in-phase or antiphase operators of C and C with respect to Cm

107

(a) Val 80 3D IP-HCC-TOCSY (c)

3D TROSYHNCACB

Wn:126.1

»H»1 iH» W2CH) JBJ 2.28

30.0

40.0

w,(13Q[ppm]

50.0

60.0

9.52 59.5 34.4 22.1 21.2

Wlppm] 13C* "C? »C1 "O3

wi(nC)[ppni]

**

r-

iHai iH9ï1.16 1.22

59.5 34.4 22.1 21.2

13(2« ^C* "C51 "C9*

Wl(13C)[ppm]

(b) lie 84

3DTROSYHNCACB

Wfc:124.2

^j

3D IP-HCC-TOCSY

»H» W" »H«2 ^ wzOH)5.72 1.47 0.88 0.63 -«—fo^

*—

-20.0

30.0

w3(13Q[ppm]

40.0

-50.0

9.40 58.8 40.5 17.7 14.5

Wfopm] 13Ctt "C* W »CV

wi(13Ç)[ppm]

(d)

iH«5.72

58.8

Iff» Ijjgî1.47 0.88

40.5

"C"17.7

"C2

wi(13C)[ppm]

»H*0.63 -

—

'

*dL :'

-

__

:

H —.

'

.

—

"*

;

w2(»H)[ppm]

20.0

30.0

w3(13C)[ppm]

-40.0

50.0

14.5

108

Figure 2. H-C strips taken from the 3D 13C-detected IP-HCC-TOCSY spectrum and HN strips taken from

the 3D HNCACB spectrum for (a) Val80 and (b) Ile84 of apo-CcmE-His6. The corresponding indirect nC

chemical shifts and their bonded 'H chemical shifts are assigned and labeled for each strip. Diagonal

peaks are marked with asterisks. 13C" and 13Cß chemical shifts are aligned to corresponding assignmed

signals in the 3D HNCACB. Slices taken along the direct 13C dimension are shown in c and d. The region

of 20.5 - 22.5 ppm is magnified to present the resolved multiplets due to 7CC couplings. Alignment of the

splitting pattern resolved at the high resolution obtained in the directly detected 13C dimension, helps to

confirm the assignment of the spin systems. The side chain 'H and i3C chemical shifts of Val80 are

completely assignable. The spin system of Ile84 can be identified without difficulty although the strip

corresponding to H7l-CYl is not visible. The experiment was performed at 600 MHz. 75(fi) x 24(t2) x

2048(r3) complex points were accumulated, with flmax(indirect 13C) = 9.93 ms, t2taJ1}i) = 9.99 ms and

?3max(direct 3C) = 168.7 ms, the interscan delay of 1 s and 56 scans per increment resulted in a total

measurent time of 117 hours. The time domain data was multiplied with a cosine function in all

dimensions and zero-filled to 256 x 128 x 2048 complex points.

and C1 is reflected by the time-dependent coefficients am, bm and c\ d] with

^(am)2 +(b'")2 =land yj(c')2 + (d')2 =1, respectively. In general, the splitting pattern becomes

very complex due to the superposition of the in-phase and antiphase terms, as well as the

different spin networks.

Side-chain assignment

As an example for the application to a protonated protein, Figure 2 shows data from a spectrum

of the 3D 13C-detected IP-HCC-TOCSY measured on 16 kDa uniformly 15N,13C-labeled apo-

CcmE-His6. CcmE (Cytochrome c maturation heme chaperone protein E) is a heme chaperone

active in the cytochrome c maturation pathway of Escherichia coli, protecting the cell from

premature activities of the highly reactive metalloorganic cofactor, which could cause oxidative

damage.Uniformly 15N-, 13C-labeled apo-CcmE- His6 (residues 30-159) was expressed and

purified as described in [17J. The NMR sample contained 350 pi of 1 mM protein solution in 20

mM sodium phosphate buffer at pH = 6.0 containing in addition 300 mM NaCI. Figures 2a and b

show H-C strips from the 3D 13C-detected IP-HCC-TOCSY spectrum and HN strips from the 3D

HNCACB spectrum for Val80 and Ile84, respectively. After the sequence-specific backbone

109

(a) Leu 123

3HNCACBY 3D IP-HCC-TOCSY

WH 'ff1 iff ^ W-

122.14

SC»

«ente

3.83

»C« 13(J> 13(^g | 13çdl | 13Q02

^[ppm] WlC3C)[ppm]

iHaa

7.40 56.1 41.1 25.4 23.2 22.0

.W2('H)[ppm]

25.0

35.0

w3C3C)[Ppm]

45.0

55.0

(C)

'Ha iff» ^ »H31 »ff

«C* | °&

m

l3çg|j 13çdl [j 13ÇJ2

1.02

56.1 41.1 25.4 23.2

Wl(»C)[ppm]

W2CH)[ppm]

22.0

25.0

35.0

|w3(13Ç)[ppm]

-45.0

55.0

(b) Lys 154

3D TROSYHNCACB

,

wn iff»

122.66

V/.v

3D IP-HCC-TOCSY

iffj_jH^ iHd_ 1H"

7.82 57.0

Wfcpm]

w3(13C)[ppm]

31.4 22.5 27.7

Wi(13C)[ppm]

40.9

(d)

Jffi

r*pc*

iff» iHg 'Hd1J96

»W»

1.52

13C«] »CI

57.0 31.4 22-5 27-7

Wi(13C)tppm]

lg.

PC

.w2CH)[ppm]

25.0

35.0

w3(13C)IPPm]

145.0

[55.0

40.9

110

Figure 3. H-C strips taken from the 3D 13C-detected IP-HCC-TOCSY spectrum and HN strips taken from

the 3D HNCACB spectrum for (a) Leul23 and (b) Lys 154 of sFkpA- His6. The corresponding indirect 13C

chemical shifts and their bonded 'H chemical shifts are assigned and labeled for each strip. Diagonal

peaks are marked with asterisks. ,3Cn and 13Cß chemical shift are aligned to their assignments in the 3D

HNCACB. Slices taken along the direct 13C dimension are shown in c and d. The systems of both Leul23

and Lysl54 are completely assigned. Broken lines indicate negative peaks. The peaks from the strips of

I3Cß of Leul23 and the 13Cß, 13CY and 13C5 of Lys 154 are opposite in sign to the other strips, which further

confirms the assignment of the spin systems. The experiment was performed at 500 MHz. 75(f,) x 24(t2) x

2048(?3) complex points were accumulated, with flmiiX(indirect 13C) = 11.93 ms, t2waxCU) = 11.99 ms and

^max(direct KC) = 203 ms, the interscan delay of 1 s and 56 scans per increment resulted in a total

measurement time of 117 hours. The time domain data was multiplied with a cosine function in all

dimensions and zero-filled to 256 x 128 x 2048 complex points.

assignment, 13Ca and 13Cß chemical shifts can be aligned for identification of the side-chain spin

systems. The complete spin system of Val80 can be clearly recognized. In the case of Ile84, all

matching 2D H-C strips are found with the sole exception of the strip corresponding to y1 13C

resonance. As an application to a large highly deuterated protein, a 3D IP-HCC-TOCSY was

recorded on 0.8 mM uniformly 15N-,13C-,2H(~90%)-labeled dimeric 48 kDa sFkpA- His6. Figures

3a and b show the 2DH-C strips from the 3D IP-HCC-TOCSY spectrum and HN strips from the

3D HNCACB spectrum of the complete spin systems of Leu 123 and Lys 154. As shown in Figure

2, the'C" and 13CP chemical shifts can be aligned for identification of bonded H and C groups

belonging to the same residue. The sign of the peaks in the H-C strips reports on the number of

carbon neighbors helping the assignment. For example, the H-C strips for the ß of Leul23 and

the ß, y and 5 of Lysl54 are opposite in sign to those in the other strips[l].

The high sensitivity of the SE version of the 3D 13C-detected HCC-TOCSY was demonstrated on

the 16 kDa uniformly 15N,13C-labeled apo-CcmE-His6 sample. Figure 4a shows H-C strips from

the 3D 13C-detected SE-HCC-TOCSY spectrum for Ile84. Although the maximum of the

components may shift by up to several Hertz due to the mixture of the in-phase and antiphase

coherences and concomitant lineshape distortion, identification of spin systems is obtained with

little difficulty and the 13C and 13CP chemical shifts can still be aligned with the corresponding

signals from the HNCACB. Slices taken along the direct dimension are shown in Figure 4b. An

Ill

(a) ne 84 3D SE-HCC-TOCSY

iHdi

lHa5.72 1.47

(b)

lH»m*2 'H" „vim0.88 a63^"tPPl5

5'72

iff» iff* ^ m

1.47 0.88 0.63 -«-W "'

[ppm]

20.0 L

30.0

W3(13C)[ppm]

r 40.0

50.0

r -:* i *

*-

>* lr ir

-20.0

(c)

58.5 40.5 17.5 144 58.5 40.5 17.5 14.4

«c 130» 13çg2wi(13C)[ppm]

l3Cdi 13c* 130> 13ça2

w^Qfppm]13031

Figure 4. (a) H-C strips taken from the 3D ,3C-detected SE-HCC-TOCSY spectrum for Ile84 of apo-

CcmE-Hiss. The corresponding indirect 13C chemical shifts and their bonded 'H chemical shifts are

assigned and labeled for each strip. Diagonal peaks are marked with asterisks. Compared with Figure 2

(b), the much cleaner spectrum demonstrates the significant gain in signal-to-noise ratio compared to the

IP experiment, which facilitate the recognition of the spin systems, (b) Corresponding slices taken along

the direct 13C dimension (c) Expanded regions showing the distorted lineshape of the peaks due to the

mixture of in-phase and antiphase coherence. The experiment was performed at 600 MHz. 75(r0 x 26(t2) x

2048(f3) complex points were accumulated, with flmM(indirect I3C) - 9.93 ms, ^max^H) = 10.82 ms and

?3max(direct I3C) = 168.8 ms, the interscan delay of 1 s and 56 scans per increment resulted in a total

measurement time of 126 hours. The time domain data was multiplied with a cosine function in all

dimensions and zero-filled to 256 x 128 x 2048 complex points and Fourier transformation was applied in

power mode.

expansion shows the distorted multiplet patterns. Using approximately same measurement time

for the SE version as for IP-HCC-TOCSY, as a comparison of Figures 2d and 4b reveals, affords

approximate 8 times gain of signal-to-noise ratio.

112

In conclusion, the new method presented here can serve as an attractive alternative to the

standard H-detected side-chain H and C assignment strategies. High resolution and good

sensitivity can overcome problems associated with 13C-spectroscopy such as peak overlap.

Especially for proteins with high molecular weight that requires partial deuteration, this method

proved to be very useful as exemplified with the 48 kDa sFkpA. Because the recovery delay can

be considerably shortened, and due to the start on 'H instead of 13C, the experiment can yield

comparable or higher sensitivity per unit experimental time. In addition to chemical shifts, the

clear splitting pattern of the peaks along the directly detected 13C dimension in IP type 3D 13C-

detected HCC-TOCSY can also be used to identify and confirm peak alignment. If the splitting

pattern of the peaks is not of interest, the SE experiment can greatly increase the signal-to-noise

ratio. More sophisticated homonuclear decoupling can further increase the signal-to-noise ratio

(Vögeli et al., submitted). The very high sensitivity indicates that the measurement time can be

further shortened.

References

1. Eletsky A, Moreira O, Kovacs H, Pervushin K, A novel strategyfor the assignment of side-chain resonances in

completely deuterated large proteins using C-13 spectroscopy. J. Biomol. NMR, 2003. 26(2): p. 167-169.

2. Maudsley, A. A., L. Müller, and R. R. Ernst, Cross-Correlation of Spin-Decoupled Nmr-Spectra by

Heteronuclear 2-Dimensional Spectroscopy. J. Magn. Reson., 1977. 28(3): p. 463-469.

3. Chan, T. M., et al., C-13 Nmr Sub-Spectra of a Protein Based on the Number ofAttached Protons - Ferredoxin

from Anabaena-Variabilis. J. Am. Chem. Soc, 1982. 104(14): p. 4008-4010.

4. Westler, W. M., et al., Two-Dimensional Nmr Strategies for Carbon Carbon Correlations and Sequence-

Specific Assignments in C-13 Labeled Proteins. J. Am. Chem. Soc, 1988. 110(12): p. 4093-4095.

5. Oh, B. H., et al., Protein C-13 Spin Systems by a Single Two-Dimensional Nuclear Magnetic-Resonance

Experiment. Science, 1988. 240(4854): p. 908-911.

6. Bermel W, Bertini I, Felli IC, Kummerle R, Pierattelli R, C-13 direct detection experiments on the paramagnetic

oxidized monomeric copper, zinc superoxide dismutase. J. Am. Chem. Soc, 2003. 125(52): p. 16423-16429.

7. Vögeli B, Kovacs H, Pervushin K, Measurements of side-chain C-I3-C-13 residual dipolar couplings in

uniformly deuteratedproteins. J. Am. Chem. Soc, 2004. 126(8): p. 2414-2420.

8. Bax, A., G. M. Clore, and A. M. Gronenborn, H-l-H-1 Correlation Via Isotropic Mixing ofC-13 Magnetization,

a New 3-Dimensional Approach for Assigning H-l and C-13 Spectra of C-13-Enriched Proteins. J. Magn.

Reson., 1990. 88(2): p. 425-431.

9. Olejniczak, E. T., R. X. Xu, and S. W. Fesik, A 4d-Hcch-Tocsy Experiment for Assigning the Side-Chain H-l-

Resonance and C-13-Resonance ofProteins. J. Biomol. NMR, 1992. 2(6): p. 655-659.

113

10. Tugarinov, V. and L. E. Kay, lie, Leu, and Val methyl assignments ofthe 723-residue malate synthase G using a

new labeling strategy andnovelNMR methods. J. Am. Chem. Soc, 2003. 125(45): p. 13868-13878.

11. Gardner, K. H. and L. E. Kay, Production and incorporation ofN-15, C-13, H-2 (H-l-delta I methyl) isoleucine

into proteinsfor multidimensional NMR studies. J. Am. Chem. Soc, 1997. 119(32): p. 7599-7600.

12. Goto, N. K., et al., A robust and cost-effective method for the production of Val, Leu, lie (delta 1) methyl-

protonatedN-15-, C-13-, H-2-labeledproteins. J. Biomol. NMR, 1999. 13(4): p. 369-374.

13. Serber, Z., C. Richter, and V. Dotsch, Carbon-detected NMR experiments to investigate structure and dynamics

of biological macromolecules. Chembiochem, 2001. 2(4): p. 247-251.

14. Hu KF, Eletsky A, Pervushin K, Backbone resonance assignment in large protonated proteins using a

combination ofnew 3D TROSY-HN(CA)HA, 4D TROSY-HACANH and C-13-detected HACACO experiments. J.

Biomol. NMR, 2003. 26(1): p. 69-77.

15. Pervushin, K., B. Vogeli, and A. Eletsky, Longitudinal H-l relaxation optimization in TROSY NMR

spectroscopy. J. Am. Chem. Soc, 2002. 124(43): p. 12898-12902.

16. Richter, G., et al., NMR studies on the 46-kDa dimeric protein, 3,4-dihydroxy-2- butanone 4-phosphate synthase,

using H-2, C-13, and N-15- labelling. Eur. J. Biochem., 1999. 261(1): p. 57-65.

17. Enggist E, Thony-Meyer L, Guntert P, Pervushin K, NMR structure of the heme chaperone CcmE reveals a

novelfunctional motif. STRUCTURE, 2002. 10(11): p. 1551-1557.

18. Hu KF, Pluckthun A, Pervushin K, Letter to the Editor: Backbone H-N, N, C-alpha, C'

and C-beta chemical

shift assignments and secondary structure of FkpA, a 245-residue peptidyl-prolyl cis/trans isomerase with

chaperone activity. J. Biomol. NMR, 2004. 28(4): p. 405-406.

19. Delaglio, F., et al., Nmrpipe - a Multidimensional Spectral Processing System Based On Unix Pipes. J, Biomol.

NMR, 1995. 6(3): p. 277-293.

20. Kadkhodaie, M., et al., Broad-Band Homonuclear Cross Polarization Using Flip-Flop Spectroscopy. J. Magn.

Reson., 1991. 91(2): p. 437-443.

21. Sorensen, O. W., M. Rance, and R. R. Ernst, Z-Filters for Purging Phase-Distorted or Multiplet-Distorted

Spectra. J. Magn. Reson., 1984. 56(3): p. 527-534.

22. Braunschweiler, L. and R. R. Ernst, Coherence Transfer by Isotropic Mixing - Application to Proton

Correlation Spectroscopy. J. Magn. Reson., 1983. 53(3): p. 521-528.

23. Kramer, F., B. Luy, and S. J. Glaser, Offset dependence of homonuclear Hartmann-Hahn transfer based on

residual dipolar couplings in solution state NMR. Appl. Magn. Reson., 1999. 17(2-3): p. 173-187.

24. Parella, T., A complete set of novel 2D correlation NMR experiments based on heteronuclear J-cross

polarization. J. Biomol. NMR, 2004. 29(1): p. 37-55.

25. Shaka, A. J„ J. Keeler, and R. Freeman, Evaluation ofa New Broad-Band Decoupling Sequence - Waltz-16. J.

Magn. Reson., 1983. 53(2): p. 313-340.

26. Marion, D. and A. Bax, Baseline Correction of 2d Ft Nmr-Spectra Using a Simple Linear Prediction

Extrapolation of the Time-Domain Data. J. Magn. Reson., 1989. 83(1): p. 205-211.

Appendix

A.l Table 1 BBP of OmpA: DNA sequence and amino acid sequence

DNA sequence (Complimentary strand) Result of translation

ATGCCGAAAGATAACACC Mot P K D N T

TGGTACACTGGTGCTAAACTGGGCTGGTCCCAGTAC WYTGAK.LGWSQY TM1

TCTAGAGAA SRE Ll

AACCAACTGGGCGCTGGTGCrrTTGGTGGTTACCAGGTT NQLGAGAFGGYQV TM2

AACCCGTAT NPY Tl

GTTGGCTTTGAAATGGGTTACGACTGGTTAGGTCGTATG V G FE MM G Y DW LG R Met TM3

CCTAGGAAA PRK L2

GCTCAGGGCGTTCAACTGACCGCTAAACTGGGTTAC AQGVQLTAKLGY TM4

CCCAAGCTTGGGACTGACGAC PKLGTDD T2H1

CTGGACATCTACACTCGTCTGGGTGGCATGGTATGGCGTGCA LDIYTRLGGMctVWRA TM5

GACACTAGT DTS L3

GTTTCTCCGGTCTTCGCTGGCGGTGTTGAGTACGTGATC VSPVFAGGVEYV1 TM6

CGTCGACGGATCACTCCTGAA RRRITPE T3S1

ATCGCTACCCGTCTGGAATACCAGTGGACCAACAAC IATRLEYQWTNN TM7

GCTAGCGAC ASD LA

AACGGTATGCTGAGCCTGGGTGTTTCCTACCGTTTC N G Met L S L G V S Y R F TM8

GGTCAGGGCGAGGCAGCTTGA G Q G E A A stop

TMl-8: transmembrane ß-strands, Ll-4: extracellular loops, Tl, T2H1 and T3S1 : periplasmic turns

115

A.2 Table 2 Backbone HN, N, Ca, C and Cß chemical shift assignments of

FkpA

Number Residue H N CA CB CO

1 ALA

2 GLU- 55.50 29.20 176.132

3 ALA 8.415 126.499 51.62 18.52 177.357

4 ALA 8.239 124.187 51.62 18.52 177.478

5 LYS+ 8.255 122.645 53.50 31.47 174.725

6 PRO 62.32 30.92 176.835

7 ALA 8.412 124.958 52.06 18.41 178.101

8 THR 8.042 113.589 60.59 69.56 174.565

9 ALA 8.280 126.692 52.29 18.41 177.799

10 ALA 8.132 122.969 52.29 18.41 177.853

11 ASP- 8.075 119.376 53.76 40.47 176.427

12 SER 8.129 117.190 58.40 62.72 174.920

13 LYS+ 8.257 123.279 55.96 31.56 176.487

14 ALA 7.957 124.520 51.73 18.32 177.150

15 ALA 7.845 123.539 52.00 18.95 176.929

16 PHE 7.721 116.683 56.46 40.42 176.246

17 LYS+ 8.705 121.974 57.24 33.01 176.266

18 ASN 7.439 110.636 51.87 39.49 175.663

19 ASP- 8.764 120.076 56.71 40.15 178.456

20 ASP- 8.327 123.946 57.47 39.10 179.862

21 GLN 8.205 121.498 59.19 29.42 177.110

22 LYS+ 7.445 119.533 59.19 32.25 178.134

23 SER 8.252 114.546 62.40 174.599

24 ALA 7.086 122.514 54.68 19.41 178.174

25 TYR 7.854 118.596 61.28 38.50 177.973

26 ALA 8.502 120.034 55.03 17.35 178.978

27 LEU 8.073 120.009 57.99 40.87 175.081

28 GLY 8.104 108.026 47.46 175.041

29 ALA 9.907 125.526 53.98 16.85 180.364

30 SER

31 LEU 57.60 179.480

32 GLY 8.982 107.901 47.34 175.302

33 ARG+ 8.531 122.429 58.02 28.57 178.717

116


34 TYR 7.933 121.439 61.15 37.95 177.110

35 MET

36 GLU- 55.78 32.38 176.667*

37 ASN 8.265 121.695 54.04 40.87 176.507

38 SER 8.806 115.562 58.06 63.69 174.478

39 LEU 8.263 122.609 54.44 40.88 176.366

40 LYS+ 8.134 121.908 56.55 29.42 180.284*

41 GLU-

42 GLN 58.08 28.56 177.15*

43 GLU- 8.339 121.905 59.51 32.58 181.288

44 LYS+ 7.446 120.324 58.60 31.59 177.893

45 LEU 7.390 117.073 53.82 41.66 177.190

46 GLY 7.839 107.025 44.91 173.835

47 ILL 7.507 123.157 60.37 38.24 174.177

48 LYS+ 8.572 130.391 54.17 32.18 175.422

49 LEU 8.359 124.658 53.47 42.52 176.333

50 ASP- 7.721 121.278 53.62 41.66 176.875

51 LYS+ 8.606 128.494 59.44 31.85 177.893

52 ASP- 8.157 118.621 56.94 39.62 180.163

53 GLN 7.780 120.259 56.94 28.29 177.672

54 LEU 7.356 121.773 58.45 41.73 179.153

55 ILE 7.624 117.478 62.74 36.46 177.391

56 ALA 7.694 124.378 54.88 18.70 179.460

57 GLY 7.924 106.179 46.48 175.683

58 VAL 7.368 121.445 66.15 31.07 177.833

59 GLN 8.299 117.300 58.92 28.23 179.922

60 ASP- 8.997 120.502 56.45 38.61 179.239

61 ALA 8.079 122.915 55.10 18.98 181.449

62 PHE 8.144 118.938 59.24 38.24 176.145

63 ALA 7.368 119.906 50.94 19.01 176.145

64 ASP- 7.866 117.809 54.92 38.93 175.422

65 LYS+ 8.409 117.759 53.99 32.52 176.306

66 SER 8.806 114.176 58.93 63.87 177.933

67 LYS+ 9.236 127.530 57.55 32.76 175.603

68 LEU 7.670 116.018 51.95 45.31 176.909

117


69 SER 9.526 121.647 56.60 64.24 174.297

70 ASP- 9.036 122.332 57.93 38.67 179.159

71 GLN 8.422 119.872 59.18 27.83 178.456

72 GLU- 7.763 121.054 59.02 30.17 181.228

73 ILE 8.864 124.615 66.95 37.45 177.029

74 GLU- 7.793 119.579 59.44 28.94 179.641

75 GLN 8.275 118.235 58.71 28.30 179.641

76 THR 8.249 118.525 66.95 67.82 177.270

77 LEU 8.665 123.367 57.99 39.82 179.601

78 GLN 8.230 120.946 59.18 27.58 179.440

79 ALA 7.632 123.004 54.36 16.71 180.525

80 PHE 8.441 122.389 60.94 38.50 176.893

81 GLU- 8.620 119.880 59.24 28.56 178.964

82 ALA 7.470 120.846 54.44 17.10 180.344

83 ARG+ 7.866 120.469 58.60 31.92 175.081*

84 VAL 62.73 31.32 177.082*

85 LYS+ 8.324 122.125 55.82 32.45 176.808

86 SER 7.720 122.088 59.18 64.58 175.904*

87 SER 7.836 123.600 59.44 64.68

88 ALA

89 GLN 55.43 30.08 176.045

90 ALA 8.229 127.066 50.29 17.89 175.543

91 LYS+ 57.27 32.05 177.872*

92 MET 8.042 122.297 53.99 29.68 178.636*

93 GLU- 8.312 125.429 56.02 32.25 176.748

94 LYS+ 8.317 121.370 56.68 29.55 177.411

95 ASP- 8.323 121.026 55.23 40.48 177.307

96 ALA 7.927 123.648 52.92 18.29 178.556

97 ALA 7.992 122.757 52.92 18.29 181.148

98 ASP- 8.564 122.188 56.90 40.35 178.889

99 ASN 8.567 119.478 55.76 41.53 179.139

100 GLU- 7.953 122.777 57.74 28.33 178.436

101 ALA 7.995 124.733 55.06 18.80 181.409

102 LYS+ 8.365 118.805 58.62 32.12 180.806

103 GLY 8.654 110.723 47.57 175.583

118


104 LYS+ 8.495 124.346 60.08 32.15 178.476

105 GLU- 7.443 118.201 59.18 29.06 178.837

106 TYR 8.006 120.298 61.67 38.69 178.837

107 ARG+ 8.837 118.979 60.41 29.43 178.054

108 GLU- 8.249 118.131 59.05 28.82 179.139

109 LYS+ 7.689 119.611 58.98 32.39 180.083

110 PHE 8.201 122.907 59.98 38.81 176.527

111 ALA 7.850 115.790 53.25 17.97 177.913

112 LYS+ 6.931 114.894 55.62 32.15 177.719

113 GLU- 7.513 121.289 56.05 28.26 176.855

114 LYS+ 8.305 122.934 57.38 31.03 178.315

115 GLY 8.664 112.887 44.54 173.795

116 VAL 7.199 122.641 62.36 30.03 175.784

117 LYS+ 8.525 128.197 53.62 34.13 174.117

118 THR 7.972 114.295 60.45 69.79 175.523

119 SER 9.447 123.376 55.72 64.02 177.190

120 SER 8.874 120.618 60.94 62.88 175.801

121 THR 8.245 112.968 62.03 68.67 175.764

122 GLY 7.697 108.236 43.43 J73.815

123 LEU 7.397 122.138 56.16 41.10 174.297

124 VAL 8.161 129.166 60.37 33.58 175.282

125 TYR 9.271 121.853 55.12 41.93 172.088

126 GLN 9.245 121.940 53.17 33.10 175.081

127 VAL 9.228 130.022 64.76 30.79 175.322

128 VAL 8.350 133.431 64.30 31.92 176.427

129 GLU- 7.997 119.621 55.10 32.58 175.623

130 ALA 8.778 127.876 55.10 19.12 179.038

131 GLY 8.156 102.901 43.65 172.891

132 LYS+ 8.455 118.403 55.27 34.72 175.061

133 GLY 8.298 110.750 44.16 172.831

134 GLU- 8.439 123.723 54.94 30.14 176.145

135 ALA 8.370 125.737 49.22 17.42 176.085

136 PRO 62.11 31.90 174.799

137 LYS+ 9.005 121.001 53.43 33.73 177.270

138 ASP- 8.097 118.305 58.22 41.52 174.558

139 SER 7.212 105.375 57.30 63.47 174.799

119


140 ASP- 7.857 123.327 54.72 40.55 176.246

141 THR 8.367 117.126 62.25 69.00 173.654

142 VAL 8.815 120.985 57.82 32.47 172.992

143 VAL 7.597 121.669 60.42 32.58 176.929

144 VAL 9.233 118.600 58.01 34.90 175.362

145 ASN 8.308 117.545 51.16 41.75 174.446

146 TYR 8.824 115.912 56.20 41.88 171.826

147 LYS+ 8.595 118.287 55.05 35.46 175.121

148 GLY 8.783 113.896 44.54 172.067

149 THR 9.404 ] 17.969 58.71 71.66 174.217

150 LEU 8.494 120.194 53.06 39.89 181.529

151 ILE 9.383 116.639 64.35 36.45 176.668

152 ASP- 7.528 119.117 52.73 39.44 177.391

153 GLY 7.961 108.475 44.09 174.337

154 LYS+ 7.818 122.658 56.60 31.35 176.266

155 GLU- 8.621 128.245 56.31 31.07 176.607

156 PHE 56.03 174.860

157 ASP- 7.056 120.588 54.39 42.88 174.799

158 ASN 8.159 122.313 52.95 38.56 175.462

159 SER 8.824 122.019 60.48 61.81 176.427

160 TYR 8.298 126.039 60.55 35.01 180.163

161 THR 7.615 112.801 64.02 67.63 176.306

162 ARG+ 7.098 119.417 57.27 30.14 177.652

163 GLY 7.609 106.252 45.20 173.333

164 GLU- 7.091 118.622 52.51 31.25 172.690

165 PRO 62.22 31.94 175.864

166 LEU 8.635 123.531 53.72 44.77 175.241

167 SER 8.196 118.810 55.71 64.12 174.317

168 PHE 8.323 120.701 55.71 40.28 173.675

169 ARG+ 8.502 122.823 55.59 30.40 179.380

170 LEU 8.582 128.739 58.49 40.00 176.587

171 ASP- 8.089 115.723 53.17 38.67 177.210

172 GLY 8.355 107.452 44.98 174.498

173 VAL 6.563 112.230 59.24 33.76 174.840

174 ILE 6.899 111.716 59.55 35.85 175.683

120


175 PRO 65.08 34.49 179.862

176 GLY 9.005 102.748 47.43 175.724

177 TRP 7.837 120.861 60.00 30.40 176.909

178 THR 7.740 117.664 66.70 68.68 174.458

179 GLU- 7.954 115.739 57.86 29.43 179.741

180 GLY 7.616 105.544 48.10 175.503

181 LEU 8.407 118.451 56.31 39.67 175.663

182 LYS+ 6.592 112.379 58.15 31.47 177.049

183 ASN 7.201 113.727 53.94 39.00 174.016

184 ILE 7.304 112.996 59.48 42.65 171.585

185 LYS+ 7.550 114.742 53.06 34.13 175.985

186 LYS+ 7.935 120.881 60.04 32.12 176.306

187 GLY 8.703 116.318 44.54 175.583

188 GLY 8.876 109.924 44.20 171.907

189 LYS+ 9.260 118.737 54.83 36.56 174.980

190 1LE 9.310 125.344 58.60 42.77 170.641

191 LYS+ 8.853 129.418 54.27 33.46 175.703

192 LEU 9.613 125.557 52.88 44.86 174.599

193 VAL 9.517 125.988 61.89 31.40 174.599

194 ILE 9.529 126.179 59.05 39.92 173.675

195 PRO

196 PRO 64.24 29.70 180.103

197 GLU- 9.848 119.823 58.72 27.27 177.773

198 LEU 7.631 118.891 52.73 41.55 174.498

199 ALA 7.708 125.771 50.96 18.07 175.422

200 TYR 8.557 122.314 58.49 37.34 176.849

201 GLY 8.388 107.879 45.86 175.041

202 LYS+ 8.641 125.050 58.04 31.42 176.808

203 ALA 8.222 118.599 52.84 18.52 179.862

204 GLY 7.290 103.205 44.29 172.148

205 VAL 7.650 115.072 58.73 31.80 173.353

206 PRO 65.44 29.82 177.491

207 GLY 8.387 113.556 44.77 173.474

208 ILE 8.400 121.497 57.42 40.48 173.293

209 PRO

210 PRO

121


211 ASN 52.83 174.518

212 SER 7.857 114.349 60.22 63.12 174.197

213 THR 8.663 124.841 62.62 68.99 173.233

214 LEU 8.416 125.203 52.34 44.50 175.041

215 VAL 8.619 122.926 60.52 30.58 176.045

216 PHE 9.806 123.407 55.32 43.10 174.880

217 ASP- 8.740 123.511 54.08 42.87 176.286

218 VAL 9.041 123.842 60.93 33.69 174.036

219 GLU- 9.165 127.974 54.33 32.11 174.558

220 LEU 8.420 127.616 53.62 42.14 174.634

221 LEU 8.993 129.296 56.34 41.90 177.933

222 ASP- 7.797 114.032 52.84 43.10 173.634

223 VAL 8.301 121.229 61.14 34.46 174.337

224 LYS+ 9.345 128.400 52.06 32.47 173.233

225 PRO 61.68 31.72 177.029

226 ALA 8.522 126.294 50.16 17.30 176.326

227 PRO 62.25 31.77 176.931*

228 LYS+ 8.264 122.067 55.50 31.77 176.594

229 ALA 8.320 126.307 51.95 18.52 177.478

230 ASP- 8.201 119.948 53.50 40.33 175.871

231 ALA 7.989 124.572 51.73 18.06 177.297

232 LYS+ 8.238 122.645 53.44 31.49 175.489*

233 PRO 62.47 30.87 177.277

234 GLU- 8.458 121.489 56.24 29.25 176.634

235 ALA 8.240 125.150 52.33 18.41 177.739

236 ASP- 8.175 119.755 53.88 40.38 176.574

237 ALA 8.143 125.343 52.77 18.11 178.442

238 LYS+ 8.142 119.562 56.31 31.46 177.237

239 ALA 8.002 124.765 52.48 18.33 178.422

240 ALA 8.130 123.031 52.67 18.06 178.322

241 ASP- 8.124 119.562 54.60 40.47 177.056*

242 SER 8.079 116.479 58.60 62.58 175.086

243 ALA 8.066 124.958 52.62 18.18 178.442

244 LYS+ 7.813 118.984 56.38 31.58 177.056

245 LYS+ 7.878 120.911 56.05 31.69

Data are from FkpA (in bold) and sFkpA (in plain) samples, respectively. Assignments marked with asterisk were not unambigously determined.

122

A.3 Linear least-squares fitting of RDCs

The theoretical dipolar coupling D^heo for a certain dipolar vector NHi in the absence of internal

motions of the molecule, i.e with order parameter S = 1 (or with a certain generalized order

parameter S2 for internal motion of the vector N-H) can be expressed using irreducible tensor

notation by a set of linear equations 11 ] :

A3L«W) = (4W5)1/2£ £< D\(Ci) >Y2m{dl,<t>i)m=-2,2

= (Axi5r2kJjsj2m(elA) (D

m=-2,2

m=-2,2

where D2n0{Q.) are elements of the Wigner rotation matrix for Euler angles Q, - (a, ß, y) of the

molecule with respect to the laboratory frame, which have properties Dfn0 (Q.) = (~l)m D2m0 (Q).

< > represents an ensemble average over the orientations of the molecule, which incorporates the

information about strength and shape of the alignment tensor, k = S f, ,

with the

8x-rNH

assumption that the order parameter S2 and rNH are constant, k is a constant which subsumes the

dipolar interaction energy and an overall order parameter, the spherical harmonics Y2m, and ($, 8,)

are the polar angles of the dipolar vector NHj in the molecular frame, and Sm is an irreducible

representation of the Saupe order matrix with S*m - (-l)mS_m. The five independent parameters

Sm relate to the five independent elements of the traceless, symmetric Saupe order matrix by a

linear transformation:

Sxx=éU2(S2+S_2)-S0/28

5,>--^),/2(52+S-2)-V2sa = So

1 (2)

Sxy=Syx=-i(-)U2(S2-S_2)o

'

^=^=-/(|)1,2(5,+S_I)Equation (1) can be written explicitly in matrix form:

123

D2

vAy

X-2(î) r2,-i(0iM) Y2fi(0ith) Yîo^) Yê^S

Y2,„2(.e2,4>2) Y2^(e2,<p2) Y2iO(02,02) Y2M(S2,<f>2) Y2t+2(02,</>2)

*-2

*+l

(3)

\A+2jy2.-2(0, A) Y^(0,^) YM(e„h) Y2M(0t,^) YÂ)

In brief form,

D = Y*A (4)

Linear least-squares fitting can be applied by solving:

(X'»D) = (Y'»Y)*Ä (5)

where Y - (Y*)T .Once the constants Am = (4n/5)mkSm are calculated from the measured

couplings and the molecular coordinates by a linear least-squares fit, Aap = kSap can be calculated

with equation (2). The principal axis system (PAS) and eigenvalues of Aap can then be derived by

standard methods of linear algebra.

Back calculation of RDCs

The individual dipolar coupling Dlitneo (in Hz) for a given NHi dipolar vector (tp, 9) can be

calculated from Azz and rhombicity r\ by:

DM) = Azz(-3cos26>-l A

+—sin 0cos20)2

(6)

where Axx, Ayy, and Azz are the main axis components of Cartesian alignment tensor in the

principle axis system (PAS), Axx| < |Ayy| < |AZZ |, Axx4- Ayy + Azz =0 and 0 < Ar = (Axx - Ayy)/Az

<1, or by [2] [3]:

LmLlBAa (3cos2#-1 {

3

%7t*rL °

2 4

flX^W^fÂar"V 1+47sin2gcos2^) (7)'NH

where S is the generalized order parameter for internal motion of the vector N-H, |Xo is the

magnetic permeability of vacuum, Yn and Yh are the magnetogyric ratios of nuclei N and H, h is

Planck's constant, rNH is the distance between nuclei N and H, and 6 and § are spherical

coordinates describing the orientation of the N-H vector in the principal axis system(PAS) of A.

Values for Aa and r\ describe the amplitude and the shape of the alignment tensor. The degree of

alignment is adjusted to yield an value of Aa of about 103.

Comparison equation of (6) and (7) gives:

124

rNrHßoh

Sx'r,

2A_

3_3Aa, r, = ^~

= 2{Axx-Ayy)l3Az (8)

NH

In weakly aligned medium, Az/ is on the order of about 10 Hz as the amplitude of the alignment

tensor Aa is adjusted to about 10"3.

Linear least-squares fitting of RDCs for homodimeric molecules

Linear least-squares fitting of RDCs as in equation (5) is carried outin complex space. Equation

(3) can be rewritten in five dimensional real space as:

D:

2A, 2Ä 2C, - 20, \

2An -2B7 2C2 -2D, £,

2A,. 2B, 2C; -2D,

Ct\

T5

(9)

.VV'-V

by linear transformation of the molecular coordinates:

Y2,+2 =A + iB, Y2,+1 = C+iD, Y2,o = E, Y2rl = -C + i D, Y2,.2 =A-iB

and of the tensor components:

A+2 = T, + i T2,A+I = T3 + i T4,A0 = T5,A ., =- T3 + i T4, A.2 = T} - i T2,

Similarly, in brief form,

D = X»T

(10)

(11)

(12)

where X is a matrix containing the structural information and f is the alignment tensor in real

space.

Similarly, linear least-squares fitting is obtained by solving:

(X •D) = (i'»X)*f (13)

For homodimeric molecules, such as FkpA, measured RDCs are assumed to correspond to the

average value of RDCs from two different subunits / and r.

DUheo=(Dltheo+D[lhJ/2 (14)

If the molecule is rigid, the whole molecule can be aligned with a single tensor, with an average

matrix X containing the structural information from both subunits / and r.

X =(X' +Xr)/2 (15)

125

If there is intramolecular relative mobility between subunits / and r, subunits / and r are assumed

to be independent in the alignment medium. In this case, D'j[heo and Drithei> should be calculated

from two independent alignment tensors T' and Tr, respectively. Equation (9) can be expanded as:

Dn

yDtj

f2A\

2Â2

2B\ 2C'i -2D'i E\ 2Ar} -2B\ 2C\ -2D'i £r>

2D'2 E'22Ar2 -2Br2 2C2 -2Dr2 E\

1 i

2B' 2 2Cl2

2A'i -2B't 2C't -2D'i É, 2A\ -2Bri 2d -2Drt Ert

r,

r-pl10

(16)

where superscripts / and r stands for two different subunits / and r. Linear least-squares fitting is

then similar to equations (12) and (13) but in a 10 dimensional real space. Tensor components for

subunits / and r in irreducible tensor notation can be recovered using equation (11). The principal

axis system (PAS) and eigenvalues of T' and Tr can then be derived by standard methods of

linear algebra for each domain, respectively. It should be noted that in the case of a homodimer,

linear least-squares fitting can not be done in complex space by expansion of equation (4) to 10

dimensions because the solution to equation (5) in the 10 dimensional complex space may no

longer have the properties required by the traceless, symmetric Saupe order matrix in the

irreducible representation, i.e. S*m = (-l)mS.m, which is mathematically correct but is not a

physically meaningful tensor.

References

1. Sass, J., et al., Purple membrane induced alignment of biological macromolecules in the magnetic field. J. Am.

Chem. Soc, 1999.121(10): p. 2047-2055.

2. James H. Prestegard, J.R.T., Hashim M, Al-Hashimi, and Michael Andrée, Structure Computation and

Dynamics In Protein NMR. Biological Magnetic Resonance, edited by Krishna and Berliner. Kluwer Academic /

Plenum Publishers, New York, 1999., 17: p. 314-320.

3. Vögeli B, K.H., Pervushin K, Measurements of side-chain C-13-C-13 residual dipolar couplings in uniformly

deuterated proteins. J. Am. Chem. Soc, 2004. 126(8): p. 2414-2420.

126

Seite Leer /

Blank leaf

127

Curriculum vitae

Personal Data:

Name: Kaifeng Hu

Date of birth: Feb. 19, 1974

Place of birth: Hubei, People's Republic of China

Citizenship: China

Education:

1991 - 1995 School of Pharmaceutical Science, Beijing Medical University.

Beijing, People's Republic of China.

Degree: Bachelor of Science

1995- 1998 Shanghai Institute of Materia Medica, Chinese Academy of Sciences.

Shanghai, People's Republic of China.

Degree: Master of Science

1998 - 2000 School of Pharmacy, University of Maryland at Baltimore.

Baltimore, Maryland, U. S. A.

Ph.D study, PhD Candidate.

2000 - 2004 Institute of Physical Chemistry, Swiss Federal Institute of Technology.

Zurich, Switzerland.

Degree (to be awarded): Doctor of Science.

Publications:

Kaifeng Hu, Alexander Eletsky and Konstantin Pervushin, (2003) J. Biomol NMR, 26 (1), 66-77.

Kaifeng Hu, Andreas Plückthun and Konstantin Pervushin, (2004) J. Biomol NMR, 28 (4), 405-

406.

Thereza Soares, Markus Christen, Kaifeng Hu and Wilfred F. van Gunsteren, Alpha- and beta-

polypeptides show a different stability of helical secondary structure, (2004) Tetrahedron, 60,

7775-7780

Documents

Rights / License: Research Collection In Copyright - Non ...27500/eth... · In NMR methodological aspects, the 3D TROSY-HN(CA)HA and 4D TROSY-HACANH experiments were proposed forbackbone