i
Conformational Flexibility and Amyloid Core Characterization of Human
Immunoglobulin Light Chain Domains by Multidimensional NMR Spectroscopy
DISSERTATION
Presented in Partial Fulfillment of the Requirements for
the Degree Doctor of Philosophy in the Graduate
School of the Ohio State University
By
Simon P. Pondaven,
Graduate Program in Chemistry
The Ohio State University
2012
Dissertation Committee:
Professor Christopher P. Jaroniec, Advisor
Professor Philip J. Grandinetti
Professor Ross E. Dalbey
i
Copyright by
Simon P.Pondaven
2012
ii
ABSTRACT
The development of multidimensional nuclear magnetic resonance (NMR)
spectroscopy combined with modern molecular biology methods that enable stable, NMR
active isotopes to be incorporated has increased the use of this technique for yielding
structural and dynamic information on macromolecules such as proteins and nucleic
acids.1,2
In spite of the numerous documented successes of solution-state NMR
spectroscopy in generating atomic-resolution images of large molecules, a number of
important biological systems cannot be analyzed directly using this approach because
they are too large or exist in an inherently non-crystalline solid state environment.
Examples of such systems include highly-ordered aggregates commonly referred to as
amyloid fibrils.3,4
Amyloid fibrils are characterized by an elongated thread-like morphology, and
generally believed to exhibit “cross-β” architecture with individual peptide strands
arranged perpendicular to the fibril axis. To date, some 40 human disorders, including
Alzheimer’s, Parkinson’s and Huntingon’s diseases, and type II diabetes have been
shown to be associated with the formation of intracellular or extracellular deposits
composed of amyloid fibrils.4 Light-chain amyloidosis (AL) is the most prevalent
systemic amyloidosis in the United States affecting about 100,000 persons per year in the
iii
U.S. and is characterized by the deposition of Ig VL amyloid fibrils in organs such as
kidneys or heart.5
Our initial studies, discussed in Chapter 2 of this thesis and published in 2009 in
Biomolecular NMR Assignement,6 present the
1H,
13C and
15N resonance assignments for
a recombinant 114 amino acid human immunoglobulin (Ig) κIV light-chain variable
domain (VL) LEN, which displays a high degree of sequence identity with another human
Ig κIV VL, SMA. While SMA is highly amyloidogenic in vivo and in vitro and has been
linked to the pathogenesis of light-chain amyloidosis (AL), LEN is non-amyloidogenic in
vivo and can be converted to the amyloid state only in vitro under destabilizing
conditions.
In a subsequent series of studies, described in Chapter 3 and published in 2011 in
Biochemistry,7 the conformational flexibility of LEN was investigated at physiological
and acidic pH on a residue-specific basis by multidimensional solution-state NMR
methods. Measurements of backbone chemical shifts and amide 15
N longitudinal and
transverse spin relaxation rates and steady-state nuclear Overhauser enhancements
indicate that, on the whole, LEN retains its native three-dimensional fold and dimeric
state at pH 2 and that the protein backbone exhibits limited fast motions on the
picosecond to nanosecond time scale. On the other hand, 15
N Carr-Purcell-Meiboom-Gill
(CPMG) relaxation dispersion NMR data show that LEN experiences considerable
slower, millisecond time scale dynamics, confined primarily to three contiguous
segments of about 5-20 residues and encompassing the N-terminal β-strand and
complementarity determining loop regions (CDR) 2 and 3 in the vicinity of the dimer
interface. Quantitative analysis of the CPMG relaxation dispersion data reveals that at
iv
physiological pH these slow backbone motions are associated with relatively low excited-
state protein conformer populations, in the ∼2-4% range. Upon acidification, the minor
conformer populations increase significantly, to ∼10-15%, with most residues involved
in stabilizing interactions across the dimer interface displaying increased flexibility.
In Chapter 4, we extended the 15
N-CMPG relaxation experiments to the pathogenic SMA
protein as well as eight SMA-like point mutants of LEN to assess the impact of each
mutation on the stability of the dimer interface. Overall, the residue specific μs-ms
conformational dynamics of the mutants are in good agreement with their global
thermodynamic stabilities previously determined by Raffen and co-workers8 with one
exception, being P40L, which exhibits similar conformational flexibility to LEN, despite
being able to form amyloid fibrils at physiological pH.
After investigating the amyloid aggregation pathways using soluble proteins in
their native states, structural characteristics of LEN in the amyloid state were probed
using isotopic exchange solution NMR. Specifically, we present in Chapter 5 the
characterization of the amyloid core region of the LEN fibril, using DMSO-quenched
hydrogen/deuterium (H/D) exchange combined with 2D solution-state NMR,9 and we
compare these H/D exchange data to those recorded for LEN under native conditions. For
native Len, amide protons for 33 out of 99 residues that could be assigned were fully
exchanged within the experiment dead time, while a set of ~20 residues, all located in the
most highly structured β-sheet region, did not fully exchange within a period of 7 days. In
contrast, for the amyloid state, nearly half of the assigned residues showed significant
(>80%) intensities relative to the reference fibril sample prepared in H2O for the 7 days
incubation time. The most protected residues, which presumably make up the amyloid
v
core region of LEN, were found to be located in β-strands A and B (aa ~2-15), CDR 1
and strand D (aa ~27a-40) and a region spanning aa ~70-105, encompassing strands G, H
and I as well as the CDR3 loop.
Finally, in Chapter 6 we present some ongoing work on a different amyloid
forming system, the Y145Stop mutant of the prion protein (PrP). Specifically, as part of a
larger group effort to obtain structural information on the PrP fibrils, the mass-per-length
(MPL) values, which serve as strong constraints on molecular structure,10
were calculated
by quantification of intensities in dark-field electron microscope images obtained in the
tilted-beam mode of a transmission electron microscope (TEM). The MPL mean values
for the human, mouse and Syrian Hamster PrP were found to be 50.8±1.2, 54.3±0.9 and
54.1±0.6 kDa/nm, respectively. These results indicate the presence of two molecules per
β-sheet repeat spacing in the fibril structure for each PrP variant.
vi
À la mémoire de mon père
Serge Pondaven
Docteur en Médecine
vii
ACKNOWLEDGMENTS
Je voudrais tout d’abord remercier mon directeur de thèse, Dr. Christopher
Jaroniec pour ses connaissances et conseils ainsi que les membres actuels et anciens de
mon groupe de recherche, notamment mon partenaire pendant 4 ans, Dr. Sujoy
Mukherjee. Je suis aussi reconnaissant envers Dr. Chunhua Yuan pour son aide durant les
expériences RMN.
Je remercie ma famille, tout particulièrement ma mère Jacqueline et mes deux
sœurs, Cécile et Delphine pour leur soutien sans failles ainsi que leurs nombreux conseils
tout au long de ces cinq dernières années. Je voudrais également remercier du fond du
cœur ma fiancée, Elizabeth, pour sa patience, son aide et soutien durant les bons et les
mauvais moments qui ont jalonné ce « voyage » aux Etats-Unis.
J’ai également une pensée pour les professeurs que j’ai rencontrés à Ohio State,
en particulier Dr. Stofltfus, Dr. Loza et Dr. Tatz. Merci pour votre gentillesse et votre
professionnalisme !
Enfin, je voudrais exprimer ma gratitude à CPE Lyon et en particulier à Mme
Catherine Ponthus et Mr Anthony Smith pour m’avoir permis de terminer mes études à
l’université d’Ohio State ainsi qu’à Jennifer Hambach and Judy Brown pour leurs
nombreux conseils administratifs.
viii
VITA
February 23rd
, 1984 Born – Metz, Moselle
2002 Baccalauréat, Série Scientifique (Mention Bien)
Lycée Viala-Lacoste, Salon-de-Provence
2002-2004 Classes préparatoires aux grandes écoles
Lycée Paul Cézanne, Aix en Provence
2004-2006 Ecole Supérieure de Chimie Physique
Electronique de Lyon (ESCPE)
2006-2007 Industrial Internship, Minerva Scientific Ltd.,
Derby, United-Kingdom
2009 Diplôme d’ingénieur (M.S. in Chemistry)
Ecole Supérieure de Chimie Physique Electronique
de Lyon (ESCPE)
2007 to present Graduate Teaching and Research Associate,
Department of Chemistry,
The Ohio State University, Columbus
FIELDS OF STUDY
Major Field: Chemistry
ix
PUBLICATIONS
4. S. Mukherjee, S.P. Pondaven, C.P. Jaroniec, Correlation between conformational
flexibility and thermodynamic stability of a human immunoglobulin light-chain variable
domain responsible for AL amyloidosis and its variants, in preparation 2012
3. S.P. Pondaven, S. Mukherjee, C.P. Jaroniec, Core of a human immunoglobulin
light-chain variable domain amyloid fibril probed by hydrogen/deuterium exchange
NMR spectroscopy, Phys. Chem. Chem. Phys., to be submitted, 2012
2. S. Mukherjee, S.P. Pondaven, C.P. Jaroniec, Conformational flexibility of a
human immunoglobulin light chain variable domain by relaxation dispersion nuclear
magnetic resonance spectroscopy: Implications for protein misfolding and amyloid
assembly, Biochemistry 2011, 50, 5845-5857
1. S. Mukherjee, S.P. Pondaven, N. Höfer, C.P. Jaroniec, Backbone and sidechain 1H,
13C and
15N resonance assignments of LEN, a human immunoglobulin κIV light-
chain variable domain, Biomol. NMR Assign. 2009, 3, 255-259
x
TABLE OF CONTENTS
ABSTRACT ........................................................................................................................ ii
ACKNOWLEDGMENTS ................................................................................................ vii
VITA ................................................................................................................................ viii
FIELDS OF STUDY........................................................................................................ viii
PUBLICATIONS ............................................................................................................... ix
TABLE OF CONTENTS .................................................................................................... x
LIST OF TABLES ........................................................................................................... xiii
LIST OF FIGURES ......................................................................................................... xiv
CHAPTER 1: INTRODUCTION ....................................................................................... 1
1.1 Preface .................................................................................................................. 1
1.2 Principles of NMR spectroscopy ......................................................................... 3
1.2.1 Nuclear spin and Boltzmann distribution ..................................................... 3
1.2.2 The NMR signal and spin relaxation ............................................................ 5
1.2.3 Chemical shifts and chemical shift anisotropy ............................................. 6
1.2.4 Multidimensional NMR spectroscopy .......................................................... 8
1.3 Amyloid fibrils ................................................................................................... 10
1.3.1. Role of amyloid-like structures in disease .................................................. 10
1.3.2. The structures of amyloid fibrils ................................................................. 14
1.3.3. Mechanisms of amyloid fibril formation .................................................... 16
CHAPTER 2: BACKBONE 1H,
13C AND
15N RESONANCE ASSIGNMENTS OF
LEN, HUMAN IMMUNOGLOBULIN κIV LIGHT-CHAIN VARIABLE DOMAIN .. 19
2.1 Introduction ........................................................................................................ 19
2.2 Materials and methods ....................................................................................... 22
2.2.1 Protein expression and purification ............................................................ 22
2.2.2 NMR spectroscopy...................................................................................... 26
xi
2.3 Assignments and data deposition ....................................................................... 27
CHAPTER 3: CONFORMATIONAL FLEXIBILITY OF A LEN Ig VL BY
RELAXATION DISPERSION NUCLEAR MAGNETIC RESONANCE ...................... 32
3.1 Introduction ........................................................................................................ 32
3.2 Experimental procedures .................................................................................... 34
3.2.1 NMR sample preparation ............................................................................ 34
3.2.2 NMR spectroscopy...................................................................................... 35
3.2.3 NMR data processing, analysis and modeling ............................................ 38
3.3 Results ................................................................................................................ 41
3.3.1. Effect of pH on LEN backbone chemical shifts and secondary structure .. 41
3.3.2. Fast time scale protein backbone dynamics ................................................ 45
3.3.3. Characterization of slow time scale protein motions .................................. 50
3.4 Discussion .......................................................................................................... 59
CHAPTER 4: CORRELATION BETWEEN CONFORMATIONAL FLEXIBILITY
AND THERMODYNAMIC STABILITY OF LEN, SMA AND SMA-LIKE POINT
MUTANTS OF LEN ........................................................................................................ 72
4.1 Introduction ........................................................................................................ 72
4.2 Materials and methods ....................................................................................... 73
4.2.1 NMR sample preparation ............................................................................ 73
4.2.2 Circular dichroism spectroscopy ................................................................. 74
4.2.3 NMR spectroscopy...................................................................................... 75
4.2.4 Analysis of NMR relaxation data ............................................................... 75
4.3 Results ................................................................................................................ 76
4.3.1 Resonance assignments of SMA and SMA-like point mutants of LEN ..... 76
4.3.3 Evaluation of slow timescale conformational dynamics of SMA and SMA-
like mutants of LEN .................................................................................................. 88
CHAPTER 5: CORE OF LEN AMYLOID FIBRILS PROBED BY
HYDROGEN/DEUTERIUM EXCHANGE NMR SPECTROSCOPY ........................... 96
5.1 Introduction ........................................................................................................ 96
5.2 Materials and methods ....................................................................................... 98
5.2.1 Protein expression and purification ............................................................ 98
5.2.2 Fibril formation and H/D exchange .......................................................... 100
xii
5.2.3 ThT fluorescence assay ............................................................................. 101
5.2.4 Electron microscopy experiments ............................................................. 101
5.2.5 NMR spectroscopy.................................................................................... 102
5.3 Results .............................................................................................................. 104
5.3.1 Fibril formation monitored by Thioflavin T assay ................................... 104
5.3.2 Transmission electron microscopy ........................................................... 104
5.3.3 Assignment of the LEN backbone in DMSO ........................................... 105
5.3.4 H/D exchange on the LEN native state ..................................................... 111
5.3.5 Post-trap exchange during NMR acquisition ............................................ 113
5.3.6 H/D exchange on the LEN fibril ............................................................... 113
5.4 Discussion ........................................................................................................ 116
CHAPTER 6: ON GOING WORK: CHARACTERIZATION OF HUMAN, MOUSE
AND HAMSTER PrP AMYLOID FIBRILS BY TILTED-BEAM TRANSMISSION
ELECTRON MICROSCOPY ......................................................................................... 123
6.1 Introduction ...................................................................................................... 123
6.1.1 The prion protein....................................................................................... 123
6.1.2 MPL determination by tilted beam transmission electron microscopy .... 124
6.2 Materials and methods ..................................................................................... 126
6.2.1 Protein expression and purification .......................................................... 126
6.2.2 Preparation of PrP23-144 amyloid fibrils ................................................. 126
6.2.3 Atomic force microscopy on the PrP23-144 amyloid fibrils .................... 129
6.2.4 Transmission electron microscopy on the PrP23-144 amyloid fibrils ...... 129
6.3 Results and discussion ...................................................................................... 131
6.3.1 Structural information probed by atomic force microscopy ..................... 131
6.3.2 Structural information probed by electron transmission microscopy ....... 134
LIST OF REFERENCES ................................................................................................ 143
xiii
LIST OF TABLES
Table Page
Table 1.1: Triple-resonance experiments used for sequential backbone assignments…...11
Table 1.2: Triple-resonance experiments used for side chain assignments.......................12
Table 3.1: Summary of 15
N-CPMG relaxation dispersion NMR data analysis for the three
dynamic regions of LEN at pH 6.5 and 2.0…………………………….…......…60
Table 3.2: 15
N R1, R2, {1H}-
15N NOE, and ηxy values for LEN at 600 MHz
1H frequency
and pH 6.5 and 2.0…………………………………………….…………………68
Table 3.3: Results of model-free analysis of 15
N relaxation data for LEN at 600 MHz 1H
frequency and pH 6.5 and 2.0………………..……………….……………..…...70
Table 4.1: Summary of the VL stability measurements and aggregation properties at
physiological pH under agitation……..…………………………………...….….73
Table 4.2: Summary of 15
N-CPMG relaxation dispersion NMR data analysis for the three
dynamic regions of LEN, SMA and SMA-like mutants of LEN…….…………..92
Table 5.1: Summary of the H/D exchange kinetic parameters for LEN fibrils………...120
Table 6.1: MPL measurement fit parameters for the three PrP variants………….….…141
xiv
LIST OF FIGURES
Figure Page
Figure 1.1: Schematic representation of selected conformational states that can be
adopted by polypeptide chains…………………...………………………………18
Figure 2.1: LEN crystal structure and amino acid sequence……….……………………21
Figure 2.2: Purification of 13
C, 15
N LEN analyzed by SDS-PAGE using an 18%
polyacrylamide gel.................................................................................................24
Figure 2.3: MALDI-TOF mass spectrum of pure 13
C, 15
N-LEN with an observed mass of
13,332 Da……………………………………………………………….………..25
Figure 2.4: 800 MHz 15
N-1H HSQC spectrum at pH 6.5 and 25°Cof
15N-LEN…..….…28
Figure 2.5: Representative strips from a 800 MHz 3D HNCA spectrum of 13
C, 15
N-
labeled LEN……………………………………………….……………………..29
Figure 2.6: Δδ(Cα)–Δδ(Cβ) and {1H}–
15N NOE values for LEN determined at 600 MHz
1H frequency and 25°C as a function of residue number.......................................31
Figure 3.1: Time scales for protein dynamics and NMR techniques…………………….33
Figure 3.2: Representative regions of 600 MHz 15
N-1H HSQC spectra of LEN recorded at
25°C and pH 6.5, 4.2, 3.1 and 2.1…………………………….………………….42
Figure 3.3: 800 MHz 15
N-1H HSQC spectra of LEN at pH 6.5 and 2.0, with the backbone
resonance assignments indicated…………………………………………...……42
Figure 3.4: Comparison of NMR chemical shifts for LEN at pH 6.5 and 2.0……….......44
Figure 3.5: Plots of 15
N R1, R2 , NOE and ηxy as a function of residue number for LEN at
600 MHz, 25 °C, and pH 6.5 and pH 2.0……………………………………..….46
xv
Figure 3.6: Model-free analysis of ps-ns time scale protein backbone motions and
qualitative characterization of slower time scale dynamics for LEN at pH 6.5 and
2.0.………………………………………………………………………..…...….49
Figure 3.7: Summary of 15
N-CPMG relaxation dispersion NMR data for LEN at pH 6.5
and 2.0……………………………………………………………………………52
Figure 3.8: 15
N-CPMG relaxation dispersion NMR trajectories for representative residues
from the three contiguous regions of LEN exhibiting elevated conformational
flexibility on the ms time scale at pH 6.5 and 2.0………………………………..55
Figure 3.9: 15
N CPMG relaxation dispersion trajectories at 600 MHz and 800 MHz for
residues found in regions of LEN, exhibiting ms time scale dynamics at pH
6.5…………………………………………………………………………..…….56
Figure 3.10: 15
N CPMG relaxation dispersion trajectories at 600 MHz and 800 MHz for
residues found in regions of LEN, exhibiting ms time scale dynamics at pH
2.0………………………………………………………………………...……....57
Figure 3.11: Key inter-residue contacts stabilizing the native fold of LEN and summary
of the protein regions to undergo the most considerable conformational
fluctuations of the backbone atoms on the millisecond time
scale…………………………………………………………….……..………….62
Figure 3.12: 15
N CPMG relaxation dispersion trajectories at 800 MHz at pH 6.5 for two
different LEN concentrations: 0.8 mM and 1.7 mM…………………………….66
Figure 4.1: 800 MHz 15
N-1H HSQC spectra of LEN and SMA at pH 6.5 and 25°C…....78
Figure 4.2: 800 MHz 15
N-1H HSQC spectra of Q89H at pH 6.5 and 25°C…………...…80
Figure 4.3: 800 MHz 15
N-1H HSQC spectra of T94H at pH 6.5 and 25°C…..………….81
Figure 4.4: 800 MHz 15
N-1H HSQC spectra of Y96Q at pH 6.5 and 25°C….……....….82
Figure 4.5: Plots of 13
Cα and 13
Cβ shifts for LEN vs SMA and 13
Cα shifts for LEN vs
Q89H, T94H, Y96Q at pH 6.5.…………………………………………..….…...84
Figure 4.6: Effect of protein concentration on the monomer/dimer association for LEN
and SMA-like mutants…………………………………………………………...87
Figure 4.7: Plots of Rex, for SMA, LEN and the SMA-like point mutants of LEN at pH
6.5 at 14.1 T and 18.8 T…………………………………...…………………..…89
xvi
Figure 5.1: Overview of the H/D exchange experiment……………………...……....….99
Figure 5.2: ThT fluorescence spectra of LEN amyloid fibril formation and normalized
kinetic of fibril formation at 482 nm for selected protein concentrations….…..106
Figure 5.3: Transmission Electron Microscopy (TEM) image of LEN amyloid fibrils after
an incubation of 48 hours at pH 2.0, 400 rpm and 37°C…..………………..….107
Figure 5.4: 800 MHz 15
N-1H HSQC spectrum of unfolded LEN in DMSO buffer showing
the backbone 1HN assignment of LEN dissolved in 95% d6-DMSO…….....….109
Figure 5.5: 800 MHz 15
N-1H HSQC spectra of the unfolded LEN reference and 7 days
H/D exchange ………………….…………………………...…………….....….110
Figure 5.6: 600 MHz 15
N-1H HSQC spectrum of the LEN protein after 7 days of
incubation in 20 mM Sodium Phosphate in D2O at pD* 7.0 and 25°C….….....112
Figure 5.7: Relative Intensity decay curves for residue T5 obtained from multiple 1H-
15N
HSQCs acquired over 8 hours……………………………...………………..….114
Figure 5.8: Relative peak intensities in the 7 days H/D exchanged spectrum normalized to
the reference and plotted against the residue number for the LEN Amyloid fibril
and Native State structure………………………………………….……….…..115
Figure 5.9: Relative Intensity versus incubation time for selected residues during the
amyloid fibrils H/D exchange in D2O at pD* 7.0……………….…………..…118
Figure 6.1: Amino acid sequence comparison of human, mouse, and Syrian
hamster PrP, highlighting amino acid differences in the 23-144 region…...124
Figure 6.2: MALDI-TOF mass spectrum of 20 μM natural abundance huPrP in pure
H2O.…………………………………………………………………….............127
Figure 6.3: MALDI-TOF mass spectra of 20 μM natural abundance ShaPrP and moPrP in
pure H2O.……………………………………………………………….............128
Figure 6.4: PrP23-144 Amyloid fibrils as visualized by low magnification atomic force
microscopy (AFM)………………………………………………………....…...132
Figure 6.5: Species differences in PrP23-144 morphology as revealed by high
magnification AFM………………………………………………………….….133
Figure 6.6: AFM height images reveal quantitative morphological difference in
periodicity between the PrP23-144 amyloid fibrils. ………………………..…135
xvii
Figure 6.7: High resolution bright field TEM images of negatively stained PrP
amyloid fibrils…………………………………………………………….....….137
Figure 6.8: Dark field TB-TEM of a TMV only sample and corresponding MPL
histogram extracted from TB-TEM images……………………………….....…139
Figure 6.9: Bright-field TEM images of unstained amyloid fibrils, corresponding dark
field TB-TEM images and MPL histogram for huPrP23-144, moPrP23-144 and
ShaPrP23-144………………………………………………………………..…140
Figure 6.10: Cartoon representation of a typical amyloid fibril with η = 2…….……...142
1
CHAPTER 1
1. INTRODUCTION
1.1 Preface
The characterization of the three dimensional structure and conformational of
peptides and proteins is a key to achieve an in-depth molecular level understanding of
biological interactions in living organisms, and can also play a critical role in the drug
design process. The two experimental techniques most frequently used for the atomic
level structural characterization of biological macromolecules and responsible for the vast
majority of the three-dimensional structures of proteins and nucleic acids deposited to
date in the Protein Data Bank (PDB) are X-ray crystallography and solution-state nuclear
magnetic resonance (NMR) spectroscopy.1,2
X-ray crystallography, while extremely successful and routinely employed toward
structural studies of large macromolecule complexes such as the ribosome, is ultimately
limited to systems that can be induced to form crystalline arrays displaying a high-degree
of long-range order in three dimensions.11,12
Moreover, X-ray crystallography is not able
to directly visualize the dynamics of biological macromolecules. On the other hand,
solution NMR can generally be used to study the structure and dynamics of biomolecules
and the main factor limiting this approach relates to their molecular size, which limits the
2
rate of Brownian tumbling that is essential for recording high quality NMR spectra. Over
the last decade, the developments of multidimensional NMR combined with modern
molecular biology methods using the incorporation of stable, NMR active, 13
C and 15
N
isotopes into overexpressed proteins, have increased the use of NMR for generating
structural and dynamic information on medium to large proteins (~30 to 50 kDa) and
even protein assemblies of hundreds of kDa, using specialized isotope labeling
schemes.13
A number of important biological systems cannot be analyzed directly using X-
ray crystallography or solution state NMR because they are too large or exist in an
inherently non-crystalline solid state environment. Examples of such systems include, but
are not limited to, self-assembled supramolecular peptide and protein aggregates
commonly referred to as amyloid fibrils, membrane proteins, large protein-protein and
protein-ligand complexes.
This dissertation investigates the aggregation mechanism and the structure of
amyloid fibril using solution-state NMR techniques and will mainly focus on the
recombinant human immunoglobulin (Ig) κIV light-chain variable domains (VL) LEN
and SMA. The SMA protein was the main component of fibrillar amyloid deposits
isolated from a patient suffering from light-chain-related amyloidosis (AL). The other
light-chain, LEN, which differs from SMA at eight amino acid positions, was excreted at
high levels in soluble form by a patient with multiple myeloma but no symptoms of AL.
Towards the end of this dissertation; we will present a work in progress on a different
amyloid system with the structural investigation of the human, mouse and hamster PrP
fibrils using dark field electron microscopy techniques.
3
This thesis is organized as follows. In the reminder of this Chapter we briefly
introduce the fundamental concepts of NMR spectroscopy as well as the biological
significance of amyloid fibrils. In Chapter 2 we describe the initial experiments to form
and purify uniformly labeled 13
C, 15
N enriched LEN as well as the protein backbone
assignment using three dimensional solution NMR techniques. In Chapter 3 we present
the results from studies aimed at providing insights in the conformational flexibility of
LEN at multiple time scales and discuss the implication of these findings for the initial
steps of LEN aggregation. Chapter 4 focuses on the analysis of conformational dynamics
of the pathogenic SMA protein as well as the eight “SMA-like” point mutants of LEN.
Chapter 5 presents an indirect method to map the core region of the LEN amyloid fibrils.
Finally, in Chapter 6 we will present results from an ongoing work with the calculation of
the molecular structure constraint Mass-Per-Length (MPL) applied to another amyloid
fibril system.
1.2 Principles of NMR spectroscopy
1.2.1 Nuclear spin and Boltzmann distribution
The nuclear magnetic resonance phenomenon was independently discovered in
1945 by the groups of Purcell and Bloch by detecting weak radiofrequency signals
generated by nuclei of hydrogen atoms in ordinary matter.14,15
With the subsequent
discovery of chemical shifts,16
which show that the molecular electronic environment at
the nucleus affects the resonance frequencies of nuclear spins, the interest for NMR
4
increased among chemists. In the 1950s, the discovery of nuclear scalar couplings,
mediated by electrons making up the chemical bonds, and their correlation with local
structure became another important step in the application of NMR spectroscopy to
molecular structure determination.
The neutrons and protons composing any atomic nucleus have the intrinsic
quantum mechanical property of spin, resulting in an overall spin of the nucleus
characterized by the spin quantum number I. If the number of protons and neutrons in a
given nucleus is even, (e.g. 12
C), then I=0, there is no overall spin and the nucleus is
NMR silent. For biological systems, the most common NMR experiments involve nuclei
such as 1H,
13C,
15N and
31P which all have I=1/2.
1H and
31P both have ~ 100% natural
abundance, while 13
C and 15
N isotopes occur at levels less than ~ 1% and samples must
typically be enriched with these last nuclei using biosynthetic methods.
Nuclear spin and nuclear magnetism are closely linked. Specifically, the spin
angular momentum, described by the vector quantity I, is proportional to the magnetic
moment, µ:2
(1.1)
The proportionality constant, γ, is the gyromagnetic ratio in units of rads-1
T-1
. In the
simplest case of a large external magnetic field, considered to be the laboratory reference
frame, the interaction energy between a nuclear spin and the magnetic field is described
by the Zeeman Hamiltonian:
(1.2)
5
where Iz in the z-component of the spin angular momentum, B0 is the magnetic field
strength, and the quantity is referred to as the nuclear Larmor frequency.
For spin -1/2 nuclei the two eigenstates |α> and |β>, with projection quantum
number m= ± 1/2, have energies where ħ is the Planck’s constant divided
by 2π. The relative population of the spins is given by the Boltzmann distribution:
(1.3)
where Nα and Nβ are the populations of spins in the |α> and |β> states, respectively, kB is
the Boltzmann constant and T the temperature. The argument of the exponential function
corresponds to the ratio of the magnetic energy to the thermal energy which defines the
nuclear spin polarization, and the NMR signal intensity is proportional to the Nα/Nβ ratio.
At room temperature, even for large magnetic fields, B0 ~ 20 T, the above ratio is very
close to 1, meaning that the population difference is exceedingly small. Physically, this
means that in at room temperature, the nuclear spin polarization is very small and thus, in
comparison to other spectroscopic methods, NMR is relatively insensitive.
1.2.2 The NMR signal and spin relaxation
When an oscillating field is applied to the sample at a frequency close to the
nuclear Larmor frequency (i.e., near resonance), the nuclear spins react to this small
magnetic field independently of the strong B0 field. For a RF pulse of a particular
duration the nuclear magnetization can be rotated to the plane perpendicular to the
direction of B0 and after the pulse is turned off. The oscillating electric current generated
by the precessing nuclear transverse magnetization is detected by a coil placed near the
6
sample. This oscillating current is called the free-induction decay (FID). The Fourier
transform of the FID converts the NMR signal from the time domain to the frequency
domain, yielding an NMR spectrum.
Throughout the NMR experiment, in addition to precessing about the state
magnetic field at their Larmor frequencies, the nuclei also interact with randomly
fluctuating magnetic fields caused by interactions with other spins and with the
vibrational and rotational motions of the lattice they are held in. These interactions result
in the relaxation of the nuclear spin magnetization and this relaxation is typically
separated into two categories. The spin relaxation along the direction parallel to B0 is
referred to as the longitudinal relaxation or spin-lattice relaxation and characterized by
the time constant T1. The spin relaxation in the plane transverse to B0 is referred to as
transverse relaxation or spin-spin relaxation and characterized by the time constant T2.
For the NMR of small molecules in liquids, T2 is typically the same order of magnitude
as T1, i.e. hundreds of milliseconds to seconds. In the case of larger molecules such as
proteins, the transverse relaxation time constant T2 may be as short as milliseconds to
tens of milliseconds.
1.2.3 Chemical shifts and chemical shift anisotropy
In a static magnetic field, each magnetically active nucleus in the molecule is
expected to give a single narrow resonance, at its characteristic Larmor frequency.
However, this is not exactly the case because the surrounding molecular environment
also has a direct impact on the NMR frequency. Since electrons are magnetic spin -1/2
particles, the nuclear Larmor frequency depends on the local electronic environment and
every nucleus does not discern the magnetic field with the same strength. This
7
phenomenon, typically referred to as chemical shielding, is anisotropic meaning that the
shielding magnitude is related to the orientation of the molecular fragment relative to B0
and can be characterized by a chemical shielding tensor, σ, generally represented by a
3x3 matrix. For isotropic solutions, rapid molecular tumbling averages parts of the
anisotropic chemical shielding interaction to zero and the effective magnetic field at the
site of the nucleus can be expressed as:
(1.4)
where σ is the shielding constant. In the NMR experiment, it is typical to report chemical
shifts instead of shielding constants. Since the nuclear Larmor frequency and the
chemical shift are both proportional to the applied magnetic field, the ratio of these two
quantities is fixed. In practice, it is convenient to specify chemical shifts in terms of this
ratio since it only depends on the sample and not on the instrument. The field-
independent expression for the chemical shift in units of parts per million, ppm, is given
by:
(1.5)
with ω and ωref corresponding to the resonance frequencies of the nuclear spin of interest
and the reference compound, respectively. For biomolecular NMR experiments, the 1H,
13C and
15N chemical shifts are typically referenced directly or indirectly using 4,4-
dimethyl-4-silapentane-1-sulfonic acid (DSS).2
8
1.2.4 Multidimensional NMR spectroscopy
All NMR spectroscopy experiments described in this thesis will use two or three
frequency dimensions. Although the fundamental concept of 2D NMR was proposed by
Jean Jeener from the Free University of Brussels, the technique was largely improved by
Richard Ernst with the development of Fourier Transform (FT) NMR in 1966 and the
principles of multi-dimensional NMR spectroscopy were eventually published in
1975.17,18
An enormous variety of 2D experimental schemes were developed by the
groups of Ernst18
and others, which expanded the applicability of NMR spectroscopy to
the characterization of complex molecules such as sugars and synthetic polymers.
In the mid-1980s, protein structure determination was added to the list of
applications by the introduction of systematic procedures for spectral analysis, primarily
developed by Wuthrich and co-workers.1 Most 2D NMR experiments involve
correlations of chemical shifts of different nuclear spins based on interactions between
the spins that allow magnetization to be transferred. In the important 2D NOESY
experiment, protons are correlated by using the dipole-dipole couplings between their
magnetic moments, giving rise to magnetization transfer via the Nuclear Overhauser
Effect or NOE. In the NOESY spectrum, correlations between a resonance of proton A
and proton B will be observed if A and B are sufficiently close in space, typically less
than 5 Å. In order to fully interpret the distance information in the NOESY spectrum, it is
necessary to assign each of the resonances in the 1H NMR spectrum to its site in the
chemical structure. To accomplish this aim, so-called J-correlation experiments, in which
magnetization is transferred between chemically bonded nuclei via the J coupling
mechanism, are required. The oldest 2D NMR pulse scheme, in which magnetization is
9
transferred from one proton to another via 1H-
1H J coupling, is known as the COSY
experiment. For large proteins, the 1H-
1H J couplings are frequently smaller than the line
width, making the COSY experiment ineffective. Another widely used 2D pulse scheme
in protein NMR is the 15
N-1H HSQC experiment where the magnetization is transferred
from the backbone amide 1HN to the attached
15N nuclei via the J-coupling. The chemical
shift is then evolved on the nitrogen and the magnetization is transferred back to the 1HN
for detection.
The next most important development was undoubtedly the addition of a third
frequency dimension to NMR spectra in 1988.19,20
The concept of 3D NMR is very
similar to 2D NMR and no new formalism for the description of such experiments was
required. For example, the 2D NOESY and 15
N-1H HSQC experiments described above
can easily be concatenated to yield a 3D 15
N-NOESY-HSQC pulse scheme. The first
problem that had to be solved was a way to record and process the enormous data
matrices associated with such experiments. The second issue was the sensitivity of 3D
experiments that are usually much lower than analogous 2D schemes unless the third
dimension corresponds to the chemical shifts of a 13
C or 15
N nucleus that was isotopically
enriched. 3D NMR schemes are especially useful in protein assignment purposes. As
indicated earlier, the 1H-
1H J correlation techniques frequently fail for large system
because of the increased 1H resonance line width and this problem is exacerbated for
13C
enriched proteins due to the additional line broadening due to 13
C-1H dipolar interactions.
As full assignments are also difficult to be obtained on the basis of the traditional NOE
approach, especially for large proteins, new experiments using through-bond J
correlations were essential. In the 1990s a novel backbone assignment procedure has
10
been developed and is applicable only to uniformly 15
N and 13
C enriched proteins.21,22
This procedure is primarily based on one-bond J couplings between adjacent atoms and
correlate the 1HN and
15N chemical shifts of the residue X with the
13CO,
13Cα,
13Cβ
shifts of residue X-1 or X. The observed correlations as well as the magnetization transfer
pathways of 3D experiments used in this dissertation can be found in Table 1.1. Once
backbone chemical shifts have been obtained, the assignments of side chain chemical
shifts can be performed using similar experiments such as (Hβ)CβCα(CO)NNH, HCCH-
TOCSY and15
N-TOCSY-HSQC. These pulse schemes provide correlations between the
15N and
1H
N shifts of residue X with all side-chains aliphatic
1H or
13C shifts in the
residue X or X-1. The magnetization transfer pathways used in this work to assign
aliphatic side chains can be found in Table 1.2. Over the last decade, additional
improvements have been made on triple resonance experiments with the development of
pulse field gradients, advanced water suppression pulse schemes as well as the increase in
field strengths and the development of commercially available cryo-probes.23
1.3 Amyloid fibrils
1.3.1. Role of amyloid-like structures in disease
The conversion of normally soluble proteins into insoluble supramolecular
aggregates underlies a number of dreaded pathological conditions, broadly referred to as
protein conformational diseases.3 To date, some 40 human disorders, including
Alzheimer’s, Parkinson’s and Huntingon’s diseases, and type II diabetes, have been
shown to be associated with the formation of intracellular or extracellular deposits
11
Table 1.1: Double and triple-resonance experiments used for the sequential backbone
assignments. Only experiments described in the present dissertation are listed.
a1JNH ~ 91 Hz,
1JNC
α ~ 7-11 Hz,
2JNC
α ~ 4-9 Hz,
1JNCO ~ 15 Hz,
1JC
αCO ~ 55 Hz,
1JC
αC
β ~ 35 Hz
12
Table 1.2: Triple-resonance experiments used for the side chain assignments. Only
experiments described in the present dissertation are listed.
13
composed of a misfolded form of one of a diverse group of unrelated polypeptides.3,4
Usually, the misfolding process results in highly-ordered aggregates, called amyloid
fibrils or amyloid plaques. These are characterized by an elongated thread-like
morphology, and generally believed to exhibit “cross-β” architecture with individual
peptide strands arranged perpendicular to the fibril axis.24-26
Although far less frequent,
granule-like deposits (i.e., non-fibrillar and apparently amorphous) have also been
observed.27-30
These conditions include pathological states in which impairment in the
folding efficiency of a given protein results in a reduction in the quantity of the protein
that is available to play its normal role. Moreover, while the precise nature of the
pathogenic aggregates responsible for certain neurodegenerative diseases has not yet been
unequivocally established,31
for many conformational disorders the extracellular
deposition in organs and tissues of insoluble amyloid fibrils has been identified as the
main cause of organ dysfunction and disease.32-34
This latter group of disorders,
commonly referred to as amyloidoses, is usually sporadic (although hereditary forms,
linked to specific protein mutations, are also known) and can be either systemic or organ-
specific. Remarkably, despite the clear relevance of amyloidogenic peptides and proteins
to human health and disease, relatively little is known, at the atomic level, about the
mechanisms underlying the assembly of amyloid fibrils, their structural features,
intermolecular interactions, and pathogenic nature. This paucity of high resolution
structural data for these systems is primarily due to difficulties in applying classical
methods of structural biology to partially-folded proteins35
and large, non-crystalline
protein aggregates.36
14
1.3.2. The structures of amyloid fibrils
For many years, the only structural information about amyloid fibrils came from
imaging techniques such as TEM, and more recently, AFM and X-ray fiber diffraction.37
These techniques enable structural insights but cannot be used to characterize amyloid
fibrils at a molecular level because they are not crystalline. Solution-state NMR
spectroscopy can be used to study amyloid fibrils architecture at a molecular level using
indirect measurement methods such as isotopic exchange followed by fibrils dissolution
in DMSO-based buffer.9,38,39
This method will be described in detail in Chapter 5 of this
thesis. Over the last decade, the situation has improved dramatically with major progress
in direct measurement techniques involving magic angle spinning (MAS) solid-state
NMR spectroscopy40-44
as well as the success in growing nano to microcrystals of peptide
fragments that have characteristics of amyloid fibrils yet are amenable to single crystal
X-ray diffraction analysis.45
Additional experimental constraints, such as the diameter of
the protofilaments observed using TEM and the mass-per-length measured by means of
scanning transmission electron microscopy (STEM) can be used to refine structure
determination. Chapter 6 of the present work will present an example of structural
constraints calculation using electron microscopy techniques.
Comparison of information about the structural properties of various fibrillar
systems tends to suggest that amyloid fibrils clearly have many structural properties in
common. Among them are the canonical cross-β structure and the frequent presence of
repetitive hydrophobic or polar interactions along the fibrillar axis. The ubiquitous
presence of a cross-β structure strongly supports the view that the physicochemical
properties of the polypeptide chain are the major determinants of the fibrillar structure in
15
each case.4 Additionally, several of the proposed structures, despite very different
sequences of their component polypeptides, suggest that the core region is composed of
two to four sheets that interact closely with each other. Nevertheless, there are also
significant differences attributable to the influence of the side chains in the structures
adopted by different systems. The lengths of the β-strands, the number and spacing of β-
sheets in the fibril and whether they are arranged in a parallel or anti-parallel arrangement
within each sheet are one of the main differences.4,26
It is also clear that the number of
residues of a polypeptide chain that are incorporated in the core structure can vary
significantly. In addition, the presence of disulfide bonds in proteins, such as insulin, may
perturb the way the sheets stack together.46
Finally, it is worth noting that polymorphism is also observed with amyloid fibrils
and that significant morphological variation exists between fibrils formed from the same
peptide or protein. A significant example of polymorphism can be found in the marked
differences in the morphology of the Aβ1-40 fibril (two or three-fold symmetric
polymorphs) that can be observed in TEM studies of sample prepared under agitation or
quiescent conditions. Solid state NMR data recorded on Aβ1-40 fibril samples provide
clear evidence that this polymorphism is linked to structural differences.47
As shown in
Chapter 6 of this dissertation, mammalian PrP23-144 fibrils from different species
(human, mouse and Syrian hamster) also vary subsequently in morphology and secondary
structure, and these differences appear to be controlled by one or two residues in a critical
region of the amino-acid sequence.48
16
1.3.3. Mechanisms of amyloid fibril formation
It is now firmly established that for many polypeptides amyloid fibril formation
proceeds via a nucleation polymerization-type pathway, with partially-structured
conformers playing a key role in the process.4,25,26,30,49
The time course of the conversion
of a protein into fibrillar form typically includes a lag phase that is followed by a rapid
exponential growth phase as shown in Chapter 5 with a ThT fluorescence assay. The lag
phase is assumed to be the time required for “nuclei” to form. Once the nucleus is
formed, fibril growth is thought to proceed rapidly by further association of monomers or
oligomers with the nucleus.4 It is also generally believed that globular proteins need to
unfold, at least partially, to aggregate into amyloid fibrils.30
Proteins have an increased
propensity to aggregate under conditions that promote their partial unfolding such as high
temperature, high pressure, low pH or moderate concentrations of denaturants. A strong
correlation between a decreased conformational stability of the native state and an
increased propensity to aggregate into amyloid-like structures has also been shown in
vitro for nondisease-associated proteins. In Chapter 3, we will study the conformational
flexibility of an immunoglobulin protein at physiological and acidic pH by solution-state
NMR relaxation experiments. The findings will provide molecular-level insights about
partial protein unfolding at low pH and point to the protein dimer dissociation, initiated
by increased conformational flexibility in several well-defined regions, as being one of
the important early events leading to amyloid assembly.
The differing features of the aggregation processes reveal that polypeptide chains
can adopt a multitude of conformational states and interconvert between them on a wide
range of timescales. Following biosynthesis on a ribosome, a polypeptide chain is
17
initially unfolded and can then populate a wide range of conformations, each of which
contains little persistent structure, as in the case of natively unfolded proteins or fold to a
unique compact structure often through one or more partly folded intermediates (Figure
1.1).4,26
Eventually, the vast majority of proteins will be degraded, usually under very
carefully controlled conditions and as part of normal biochemical processes in which
their amino acids will often be recycled.
18
Figure 1.1: A schematic representation of some of the many conformational states that
can be adopted by polypeptide chains and the means by which they can be
interconverted. Figure adapted with permission from ref.4 Copyright 2006 by Annual
Reviews.
19
CHAPTER 2
2. BACKBONE 1H,
13C AND
15N RESONANCE ASSIGNMENTS OF LEN,
HUMAN IMMUNOGLOBULIN κIV LIGHT-CHAIN VARIABLE DOMAIN
Adapted from S. Mukherjee, S.P. Pondaven, N. Höfer, C.P. Jaroniec,
NMR Assign. 2009, 3, 255-259.
2.1 Introduction
Light-chain amyloidosis (AL) is one of the most prevalent systemic diseases of
this kind, and is characterized by the extracellular deposition of amyloid fibrils in organs
and tissues, most commonly in the kidneys and heart, which leads to rapidly progressive
organ failure.5,50
The amyloidogenic precursor proteins in AL are usually
immunoglobulin (Ig) light-chain variable domains (VL) of the κ or λ type, produced by
clonal plasma cells in the bone marrow; AL patients typically have elevated plasma cell
burdens, secrete in urine homodimers composed of free monoclonal Ig light-chains,
commonly referred to as Bence Jones proteins, and in ~10 – 15% of the cases also suffer
from multiple myeloma.5,50,51
Interestingly, only a relatively small fraction of Ig VLs are
amyloidogenic in vivo,5,50
and multiple studies have established the links between
mutations in Ig VL amino acid sequences, reduced protein stability and the capacity of
20
these proteins to misfold and assemble into amyloid fibrils under physiological
conditions.8,52,53
High resolution X-ray structures of several amyloidogenic and non-
amyloidogenic Ig VL domains have also been reported,54-58
revealing the existence of
both large and more subtle structural perturbations - typically involving residues near the
dimer interface - that appear to be associated with the reduced thermodynamic stability
and increased amyloidogenicity of certain Ig VLs relative to others. However, relatively
little remains known at atomic resolution about the partially folded intermediates
involved in Ig VL aggregation and protein conformation in amyloid fibrils.
In order to provide such insights, we have recently commenced solution and
solid-state NMR studies focused on two highly homologous (~93% sequence identity)
114-residue human Ig κIV VL domains, LEN and SMA, which can be readily prepared in
recombinant form and exhibit drastically different amyloidogenic propensities in vivo.59
Specifically, SMA was the main component of amyloid deposits isolated from an AL
patient, whereas LEN, which differs from the germline sequence κIV by a single somatic
mutation (S29N) and from SMA at eight amino acid positions (S29N, K30R, P40L,
Q89H, T94H, Y96Q, S97T, I106L) was extracted as a Bence Jones protein from a
multiple myeloma patient with no symptoms of AL.59
The crystal structure of LEN and
the locations of the SMA-type mutations are shown in Figure 2.1. These amyloidogenic
propensities have been correlated with different thermodynamic stabilities of the two
proteins8 and are also mirrored in vitro; at physiological pH, SMA assembles into
amyloid fibrils with relative ease, while LEN is a stable dimer under these conditions and
can be converted to the amyloid state only under destabilizing conditions (e.g., in the
presence of denaturant, ~1 – 2 M guanidinium hydrochloride).8,59
21
Figure 2.1: LEN crystal structure, A),Left) Ribbon representation of one LEN subunit.
Right) Dimer structure with individual subunits shown in yellow and cyan. Residues
where LEN and SMA differ are labeled by residue number and indicated by red spheres
on the Cα atoms. B) Amino acid sequences of LEN and SMA VLs. Residue numbering is
according to Kabat et al.60
22
Extensive subsequent studies of SMA and LEN aggregation carried out using a
variety of biophysical techniques by Fink and coworkers61-63
revealed that the dimer
dissociation is a critical step in initiating protein aggregation and that multiple partially
folded intermediates appear to play a role in the conversion of these Ig VL domains to
amyloid fibrils.
As an initial step in our NMR studies aimed at elucidating the detailed atomic
level mechanisms of SMA and LEN aggregation, we present the solution-state backbone
assignments of LEN at pH 6.5. We note here that 1H,
13C and
15N resonance assignments
of a related fibrillogenic Ig κVI VL were also recently reported.64
2.2 Materials and methods
2.2.1 Protein expression and purification
The plasmid encoding for LEN59
was kindly provided by Dr. Fred J. Stevens
(Argonne National Laboratory). Uniformly 15
N- and 13
C, 15
N-labeled samples were
prepared according to published procedures59
with minor modifications. Briefly, for
13C,
15N-LEN, electro-competent Escherichia coli C41(DE3) cells transformed with the
corresponding plasmids were grown at 30°C and 110 rpm in a modified M9 minimal
medium65
containing 100 lg/mL carbenicillin, 1 g/L 15
NH4Cl, 3 g/L 13
C-D-glucose
(Cambridge Isotope Laboratories) and 0.5 g 13
C, 15
N-enriched Isogro growth medium
(Isotec/Sigma-Aldrich); for 15
N LEN the minimal medium contained 3 g/L natural
abundance D-glucose and 0.5 g 15
N-Isogro. Proteins expression were induced at OD600 ~
0.8 by the addition of isopropyl-β-D-thiogalactoside to a final concentration of 1 mM.
Cells were grown for an additional 16 h and harvested by centrifugation at 4,000 g and
23
4°C for 10 min. The periplasmic extract which contained the proteins was isolated as
described previously59
by incubating the cells for 1 h at 4°C in Tris EDTA–sucrose pH
8.0 buffer containing lysozyme. Following centrifugation at 27,000 g and 4°C for 15 min,
the supernatant that included the periplasmic fraction was dialyzed against 10 mM Tris,
pH 8.0 and applied at a flow rate of 0.5 mL/min to two 1 mL HiTrap Q XL cartridges
(GE Healthcare) connected in series and equilibrated with 10 mM Tris, pH 8.0. The flow,
through fraction containing LEN was subsequently dialyzed against 10 mM sodium
acetate, pH 4.6 and applied at a flow rate of 0.5 mL/min to two 1 mL HiTrap SP XL
cartridges (GE Healthcare) connected in series and equilibrated with 10 mM sodium
acetate, pH 4.6. The proteins were then exchanged into 20 mM Na-Phosphate, 100 mM
NaCl, and pH 6.5 buffer by using an Amicon Ultra-15 3,000 molecular-weight-cut-off
centrifugal device (Millipore) and applied to a HiLoad 16/60 Superdex 75 gel filtration
column (GE Healthcare) equilibrated with the same Phosphate buffer. Fractions
containing the proteins were collected and pooled together, and the protein identity and
purity confirmed by SDS–PAGE (Figure 2.2) and MALDI–TOF mass spectrometry
(Figure 2.3). Typical yields of isotopically enriched protein (determined from absorbance
at 280 nm with an extinction coefficient of 24,535 M-1
cm-1
) were 12–14 mg for LEN per
liter of cell culture.
24
Figure 2.2: Purification of 13
C, 15
N LEN analyzed by SDS-PAGE using an 18%
polyacrylamide gel. Protein samples were mixed in a 1:1 (v/v) ratio with a standard gel
loading buffer and heated at 96°C for 15 min prior to loading onto the gel. Lanes 1 and
2: periplasmic fractions; Lane 3: Q load impure LEN protein; Lanes 4 and 5: molecular
weight marker (GE healthcare); Lanes 6 and 7: pure LEN protein after Q and S
purification; Lane 8: pure LEN protein with small amount of high molecular weight
impurities.
25
Figure 2.3: MALDI-TOF mass spectrum of pure 13
C, 15
N-LEN with an observed mass of
13,332 Da. The minor peaks observed at 6,663 Da and 26,691 Da correspond
respectively to the di-charged ions (M2+
) and LEN dimer (2M+).
26
2.2.2 NMR spectroscopy
Uniformly 15
N- and 13
C, 15
N-labeled samples used for NMR studies consisted of a
concentration of respectively 1.7 mM and 1.5 mM for LEN. All samples were in aqueous
solution containing 20 mM sodium phosphate, 100 mM NaCl, 7% (v/v) D2O and 0.02%
(w/v) NaN3 at pH 6.5 in a total volume of 280 μL in Shigemi microcells. NMR spectra
were recorded at 25°C on Bruker DMX-600 and DRX-800 MHz spectrometers equipped,
respectively, with a TXI room-temperature probe with triple-axis gradients and a QXI
cryogenic probe with z-axis gradients. The sequential backbone 1HN,
15N,
13C’,
13Cα and
13Cβ assignments were obtained using a suite of triple-resonance 3D HNCO, HNCA,
HN(CO)CA and HN(CA)CB experiments based on the pulse schemes of Kay and co-
workers.66
For LEN, the 1Hα and side-chain
1H,
13C and
15N resonances were assigned
using 3D H(C)CH-TOCSY, (H)CCH-TOCSY, 15
N-TOCSY-HSQC and 15
N-NOESY-
HSQC spectra.2 A schematic representation of these 3D experiments can be found in
Table 1.1 for the backbone assignments and Table 1.2 for the side chains assignments.
NMR spectra were processed using NMRPipe and analyzed using NMRDraw.67
1H
chemical shifts were referenced to external TSP, and 13
C and 15
N shifts were referenced
indirectly.2 Residue-specific longitudinal (R1) and transverse (R2) relaxation rates and
steady-state heteronuclear {1H}–
15N NOEs for backbone amide
15N nuclei were obtained
at 600 MHz 1H frequency using the pulse schemes of Farrow et al.
68 Spectra were
processed and analyzed using the NMRPipe/NMRDraw software package.67
27
2.3 Assignments and data deposition
Figure 2.4 shows a 2D 15
N-1H HSQC spectrum of LEN recorded at 800 MHz
1H
frequency. Representative strips from the 3D HNCA spectrum illustrating sequential
backbone assignments is shown in Figure 2.5. The backbone amide 1H and
15N chemical
shifts were obtained for 106 of 108 non-proline residues, with the exceptions being D1
and Y96. Several residues in the HSQC spectrum, including S27e, S27f, Y32, G41, W50,
S77, Q90, Y91, S97, F98 and G99, exhibit notably attenuated intensities most likely due
to conformational exchange. According to the X-ray structure of LEN55
all of these
residues are found in the loop regions with the majority present in the vicinity of the
dimer interface. Nearly complete assignments (97.2%) of the 1Hα,
13Cα,
13Cβ and
13C’
chemical shifts were obtained, with P43 and P95 being the only residues that could not be
assigned. In addition, chemical shifts were obtained for 94.1% of the remaining aliphatic
1H and
13C side-chain resonances, as well as for
15Nδ–
1Hδ of N22 and N28 and
15Nε–
1Hε
of Q27 and Q79. The chemical shifts have been deposited in the BioMagResBank
(http://www.bmrb.wisc.edu) under the accession number 16463.
The 13
Cα and 13
Cβ secondary shifts, Δδ(Cα) and Δδ(Cβ), respectively, were used
to evaluate the secondary structure of LEN. The plot of Δδ(Cα)-Δδ(Cβ), shown in Figure
2.6A, indicates that the majority of residues adopt a β-strand conformation and that the
locations of different secondary structure elements predicted on the basis of chemical
shifts are in good agreement with the X-ray structure of LEN.
28
Figure 2.4: 800 MHz 15
N-1H HSQC spectrum at pH 6.5 and 25°C of
15N-labeled LEN,
the Asn and Gln amide side-chain resonances are indicated by horizontal lines and
aliased Arg 15
Nε signals are marked by asterisks.
29
Figure 2.5: Representative strips from a 800 MHz HNCA spectrum of 13
C, 15
N-labeled
LEN.
30
Finally, in order to probe the oligomeric state of LEN, the 15
N R2/R1 ratios for 96
residues with {1H}–
15N NOE values greater than 0.75 (Figure 2.6B) were used to
estimate the overall protein rotational correlation time within the ModelFree software
package.69
These measurements provided average amide 15
N relaxation time constants of
T1 = R1-1
= 1,056 ± 8 ms and T2 = R2-1
= 54.0 ± 0.4 ms, yielding an estimated rotational
correlation time of τc = 13.9 ± 0.1 ns for LEN. This τc value indicates that at
physiological pH LEN (M ~ 13 kDa) is a dimer at concentrations required for NMR
studies. This observation is consistent with the dimerization constant of approximately
4x105M
-1determined for LEN,
59 both published
59 and our own gel filtration
chromatography data (not shown) which indicate that LEN elutes as a dimer at
concentrations > 0.5 mM, as well as the LEN crystal structure.
31
Figure 2.6: A) Plot of Δδ(Cα)–Δδ(Cβ) as a function of residue number. Above the plot,
the schematic secondary structure diagram of LEN indicates the disulfide bond in yellow
and the location of β-strands and α-helix found in the X-ray structure (PDB entry
1LVE)55
. Residue numbering is according to Kabat et al.60
B) Heteronuclear {1H}–
15N
NOE values for LEN determined at 600 MHz 1H frequency and 25°C as a function of
residue number. For residues where the NOE could not be determined, the NOE value
was set to zero.
32
CHAPTER 3
3. CONFORMATIONAL FLEXIBILITY OF A LEN Ig VL BY RELAXATION
DISPERSION NUCLEAR MAGNETIC RESONANCE
Adapted from S. Mukherjee, S.P. Pondaven, C.P. Jaroniec,
Biochemistry 2011, 50, 5845-5857.
3.1 Introduction
In this chapter, the conformational flexibility of LEN, which can undergo
conversion to amyloid fibrils under destabilizing conditions, will be investigated at
physiological and acidic pH on a residue specific basis by multidimensional solution-
state nuclear magnetic resonance methods. The biological relevance of the LEN protein
was explained in the previous chapter and will not be discussed further. This introduction
will instead focus on the importance of studying protein dynamics and why NMR plays a
major role.
Numerous biological processes rely on transduction of information through
conformational changes in proteins and nucleic acids associated with folding and
assembly, ligand binding, molecular recognition and catalysis.70
One of the main
challenges in understanding biological processes at a molecular level is the elucidation of
33
how the active conformation of macromolecules is achieved on time scale necessary for
function. As a result, over the past 20 years, there has been an explosion of research on
the dynamical properties of proteins, driven by the emergence of a handful of techniques
that are sensitive to protein motion.71
Solution-state nuclear magnetic resonance
spectroscopy offers a unique set of tools for investigating the structure and motions of
protein molecules over a wide range of time scales, from picoseconds to seconds,2,70,71
with recent advances enabling studies of “invisible” protein excited states at populations
as low as 0.5% of the major, ground-state conformer.72-74
Protein motions and NMR spin
relaxation techniques for studying them span more than 12 orders of magnitude in time
scale as shown in Figure 3.1. This methodology facilitates comprehensive analysis of
protein unfolding phenomena frequently associated with aggregation and is highly
complementary to other techniques traditionally used to probe amyloid formation and
architecture, including hydrogen-deuterium exchange,9,38,39,75-77
electron paramagnetic
resonance78,79
and solid-state NMR.80,81
In this chapter, we apply a suite of
multidimensional solution-state NMR techniques toward the quantitative, site-resolved
analysis of the conformational flexibility of LEN on two different time scales.
Figure 3.1: Time scales for protein dynamics and NMR techniques.70
34
Measurements of backbone chemical shifts and amide 15
N longitudinal and transverse
spin relaxation rates (R1 and R2) and steady-state nuclear Overhauser enhancements
(NOE) are used to probe the fast motions, from picoseconds to nanoseconds time scale.
All of these parameters are typically measured using two-dimentional, HSQC-type,
experiments in which the intensities of peaks are modulated as a function of a time delay
placed at a point in the sequence when the relevant relaxation process is active. For the
slower dynamics, on the milliseconds time scale, 15
N Carr-Purcell-Meiboon-Gill (CPMG)
relaxation dispersion NMR data are acquired. The findings of these experiments will
provide insights into the initial stages of low pH-induced protein misfolding and
assembly into amyloid fibrils.
3.2 Experimental procedures
3.2.1 NMR sample preparation
The LEN plasmid59
was kindly provided by Dr. Fred J. Stevens (Argonne
National Laboratory). 15
N- and 13
C,15
N-labeled proteins were overexpressed in
Escherichia coli and purified as described previously.6 Two samples were used for the
NMR relaxation measurements, each consisting of 15
N-labeled LEN at a concentration of
1.7 mM in aqueous solution with a total volume of ∼280 μL loaded into a Shigemi
microcell. The first sample was prepared in pH 6.5 sodium phosphate buffer (20 mM
sodium phosphate, 100 mM NaCl, 7% (v/v) D2O and 0.02% (w/v) NaN3). The second
sample contained ∼15 mM HCl, 100 mM NaCl, 7%D2O, and 0.02% NaN3 at pH 2.0.
Two additional samples were utilized to establish the backbone resonance assignments of
35
LEN at pH 2. One sample consisted of 1.5 mM 13
C,15
N-LEN in the pH 2.0 HCl/NaCl
solution in a Shigemi microcell, and the second sample contained 0.8 mM 15
N-labeled
LEN in the sodium phosphate buffer described above in a conventional 5 mm sample
tube (Wilmad-Labglass, Buena, NJ); the latter sample was used to record a successive
series of two-dimensional (2D) 15
N-1H heteronuclear single quantum coherence (HSQC)
spectra at pH values of 7.4, 7.2, 6.9, 6.7, 6.5, 6.1, 5.2, 4.2, 3.1, 2.1, 2.9, 4.0, 5.0, 6.0, and
7.0 ± 0.1, with the pH adjusted by adding small amounts of concentrated HCl or NaOH
and measured directly inside the NMR tube by using a specialized electrode (Hamilton,
Reno, NV).
3.2.2 NMR spectroscopy
NMR measurements were performed at 25°C on Bruker DMX-600, DRX-600,
and DRX-800 spectrometers equipped with either a room-temperature TXI probe with
triple-axis gradients (DMX-600) or cryogenic probes with z-axis gradients (DRX-600
and DRX-800). All pulse schemes used were gradient and sensitivity enhanced82,83
and
employed water flip-back and 3-9-19 binomial pulses for optimum solvent suppression.2
In addition, for all experiments that employed periods during which a variable amount of
high-power radiofrequency (RF) irradiation was applied to the sample, a constant duty
cycle was maintained by incorporating additional RF pulses following the free-induction
decay (FID) acquisition to compensate for potential sample heating effects.84
Residue-specific backbone amide 15
N longitudinal (R1) and transverse (R2)
relaxation rates and steady-state heteronuclear {1H}-
15N nuclear Overhauser
enhancements (NOEs), which report on protein motions on the picosecond (ps) to
36
nanosecond (ns) time scale,2 were obtained at 14.1 T (600 MHz
1H frequency) using the
pulse schemes of Kay and co-workers.68
The R1 and R2 measurements employed recycle
delays of 1.2 s. For the R1 measurements, 15
N-1H correlation spectra were collected with
longitudinal relaxation delays of 5, 65, 145, 245, 365, 525, 750, and 1150 ms, and for the
R2 measurements a Carr-Purcell-Meiboom-Gill (CPMG) π-pulse train was applied on the
15N channel during transverse relaxation delays of 7.2, 21.6, 43.2, 57.6, 72, 108, and 144
ms. For the {1H}-
15N NOE measurements, the reference and NOE spectra were acquired
by using a 5 s relaxation delay and a 1 s relaxation delay followed by a 4 s proton
presaturation period, respectively.
Measurements of ηxy, the 1H-
15Ndipole-dipole/
15Nchemical shift anisotropy
(CSA) interference rate constant for transverse 15
N magnetization,85,86
were carried out at
14.1 T using the pulse scheme described by Palmer and co-workers86
with recycle delays
of 2.5 s and used to assess the extent of chemical exchange at different protein sites.71,87,88
The measurements employed relaxation periods T of 32.0, 53.4, 74.8, and 106.8 ms. The
ratios of spectral intensities obtained from separate experiments that detect either the in-
phase (Ny) or antiphase (2HzNy) 15
N magnetization were used to calculate the residue-
specific ηxy values according to: 85
(3.1)
Quantitative studies of protein conformational dynamics on the millisecond (ms)
time scale were undertaken using CPMG relaxation dispersion NMR techniques,70,73,74
with data collected at magnetic field strengths of 14.1 and 18.8 T, corresponding to 1H
frequencies of 600 and 800 MHz, respectively. A constant-time, relaxation-compensated
37
pulse sequence developed by Kay and co-workers,89
with a 1H continuous wave spin-lock
field applied during the 15
N CPMG π-pulse train, was used. Spectra were recorded in an
interleaved manner with a constant relaxation delay Trelax = 30 ms and a 15
N CPMG π-
pulse width t180,N = 90 μs. The number of 15
N CPMG π-pulses applied during the Trelax
period determined the CPMG frequency, νCPMG, according to:89
(3.2)
where 2τCP is the spacing between successive 15
N CPMG π-pulses. In our experiments
νCPMG was varied between 33.33 and 1000 Hz. The 1H spin-lock fields used were in the
8.5-11 kHz range, with the exact RF amplitude fine-tuned for each νCPMG value to ensure
that an integral number of 1H π-pulses are accommodated within Trelax, and an
equilibration delay τeq = 5 ms was used before and after the water alignment as
recommended.89
For each data set, a total of 22-24 15
N-1H correlation spectra,
corresponding to different νCPMG values, were recorded. This included two sets of
triplicate measurements—one for a relatively low CPMG frequency (νCPMG = 100 Hz)
and the other for a high CPMG frequency (νCPMG = 800 - 1000 Hz)—as well as one
control experiment with no CPMG relaxation delay (i.e., Trelax = 0). The 15
N-1H spectra
were collected with a recycle delay of 3 s and a total experiment time of ∼ 48 - 60 h per
data set. For each backbone amide 15
N the effective transverse relaxation rate, R2,eff, as a
function of νCPMG was calculated as: 90,91
(3.3)
38
where Trelax is the fixed 30 ms relaxation delay, I(νCPMG) is the peak intensity for a
particular CPMG frequency, and I0 is the peak intensity in the control spectrum recorded
with Trelax = 0.
3.2.3 NMR data processing, analysis and modeling
All NMR spectra used to extract the protein motional parameters were processed
in NMRPipe67
with Lorentzian-to-Gaussian apodization. Peak intensities were obtained
by fitting the spectra to Gaussian lineshapes using the nlinLS routine in NMRPipe.
For the R1 and R2 experiments, the relaxation trajectories were modeled as single-
exponential decays to extract the residue-specific relaxation rate constants. The site-
resolved {1H}-
15N NOEs were obtained as ratios of cross-peak intensities determined in
the presence and absence of the presaturation pulses. Errors in R1, R2, and NOE
parameters were estimated using the root-mean-square (rms) spectral noise obtained in
NMRPipe. For modeling the ps-ns time scale protein backbone dynamics, an initial
rotational correlation time estimate for LEN was obtained from the 15
N R2/R1 ratios92
by
using the r2r1_tm program kindly provided by Dr. Arthur G. Palmer (Columbia
University).93
Subsequently, the rotational diffusion tensor was estimated using the
method of Tjandra et al.,94
implemented in the r2r1_diffusion program.93
This calculation
utilized the rotational correlation time estimate obtained above and the X-ray structure of
LEN (PDB entry 1LVE),55
which was used to construct the LEN dimer and appropriately
rotated and translated with the pdbinertia program.93
For this initial estimation of the
diffusion tensor only residues located in the β-strands of LEN were considered;
furthermore, from this subset of residues we excluded those displaying elevated fast
39
internal motions (characterized by {1H}-
15N NOE values of <0.7) or significant chemical
exchange contributions to R2 according to the CPMG relaxation dispersion measurements
(vide infra).94
An F-value analysis was used to select the appropriate model for the
diffusion tensor. This diffusion tensor estimate was then used as the input for an initial
round of Lipari-Szabo model-free analysis,95,96
followed by the simultaneous
optimization of both the diffusion tensor and model-free parameters. This process was
iterated within the FASTModelFree97
implementation of the ModelFree program98,99
until
a self-consistent set of parameters was obtained. All model-free calculations assumed an
N-H bond length of 1.02 Å and a 15
N CSA of -160 ppm.
For the CPMG relaxation dispersion experiments, uncertainties in the measured
R2,eff values for each residue were calculated according to:
[
] (3.4)
where ΔI is the average standard deviation in the peak intensity estimated from the two
sets of triplicate measurements described above. In cases where ΔR2,eff was found to be
less than 2% of the corresponding R2,eff, a minimum value of 2% was imposed for the
subsequent calculations.91
The residue-specific CPMG relaxation dispersion trajectories
were obtained by plotting R2,eff vs νCPMG, and data from both magnetic field strengths
were fit simultaneously to the Carver-Richards model for two-site exchange100
using the
CPMGfit program kindly provided by Dr. Arthur G. Palmer.93
Results of this calculation
were used to perform an initial identification of backbone amide 15
N spins subject to
chemical exchange as follows. For each residue, the simulated CPMG relaxation
dispersion trajectories at 600 and 800 MHz 1H frequency were used to calculate the
40
corresponding exchange contributions to 15
N R2 according to Rex = R2,eff(νCPMG = 0) -
R2,eff(νCPMG = 1000); residues with Rex > 3 s-1
at 600 MHz were classified as exhibiting
significant chemical exchange, while those with Rex < 3 s-1
were assumed to display
negligible motions on the ms time scale. This analysis identified several contiguous
regions of ~5-20 residues in LEN, displaying elevated chemical exchange. For each of
these regions, the 600 and 800 MHz CPMG relaxation dispersion trajectories were used
simultaneously to perform independent global two-state fits within the Catia program
kindly provided by Dr. Lewis E. Kay (University of Toronto).101
These calculations
yielded the exchange parameters for each region, including the exchange rate chemical
shift differences between the two exchanging states (|Δω|). Calculations were performed
with an array of starting values for kex and pB. For each region the final kex and pB
parameters and their uncertainties were obtained as the averages of the corresponding
values in the 20 lowest χ2 fits. The |Δω| value for each residue was subsequently
recalculated using these mean kex and pB parameters. The robustness of the fits and
extracted exchange parameters was verified by removing trajectories for individual
residues from the data set, re-estimating kex, pB, and |Δω| and comparing the resulting
values to the exchange parameters obtained using the complete data sets. The signs of
chemical shift differences between the major and minor conformers were subsequently
determined by comparing the 15
N chemical shifts in a pair of HSQC and heteronuclear
multiple quantum coherence (HMQC) spectra recorded at 800 MHz 1H frequency as
described by Skrynnikov et al.102
41
3.3 Results
3.3.1. Effect of pH on LEN backbone chemical shifts and secondary structure
A series of 2D 15
N-1H HSQC spectra of LEN were recorded at pH values ranging
from 2.0 to 7.4 as described in the Experimental Procedures section. LEN is a soluble
dimer at around physiological pH and is known to form amyloid fibrils at low pH under
agitation.59,63,103,104
The complete backbone and side-chain resonance assignments of
LEN at pH 6.5 have been established previously,6 and an assigned HSQC spectrum under
these conditions is shown in Figure 3.3 (blue contours). The spectra of LEN at pH values
between 7.4 and 5.2 were found to be virtually superimposable. Upon further reduction of
pH, to as low as ∼2, small changes in many backbone amide chemical shifts were
observed, with the most pronounced changes occurring at pH ∼3-4 as shown in Figure
3.2. These low pH-induced chemical shift changes were completely reversible upon
returning the pH value back to ∼7.
The 2D 15
N-1H HSQC spectrum of LEN at pH 2.0 (Figure 3.3, red contours)
could be largely assigned from the pH titration experiments. However, in order to
confirm these resonance assignments and unambiguously assign the remaining amide
signals (as well as establish the 13
Cα chemical shifts), a 3D HNCA spectrum of 13
C,15
N-
LEN at pH 2.0 was also recorded (data not shown). Altogether, backbone amide 15
N-1H
correlations for 89 of 108 non-proline residues could be unambiguously assigned for
LEN at pH 2.0. The 19 non-proline residues that could not be assigned include D1, I2,
K30, K45-L47, Y49-T53, S56, Q89-Y91, T94, and Y96-F98.
42
Figure 3.2: Representative regions of 600 MHz 15
N-1H HSQC spectra of LEN recorded
at 25°C and pH 6.5 (blue), 4.2 (cyan), 3.1 (green) and 2.1 (red).
Figure 3.3: 800 MHz 15
N-1H HSQC spectra of LEN at pH 6.5 (blue contours) and 2.0
(red contours), with the backbone resonance assignments indicated.
43
These residues, the majority of which are found in complementarity determining (CDR)
loop regions 2 and 3 near the LEN dimer interface, were broadened beyond detection at
low pH most likely due to significant conformational exchange dynamics as discussed
below. Interestingly, a number of the same residues were also found to exhibit
pronounced exchange broadening at pH 6.5.6
For each residue in LEN that could be assigned in the pH 6.5 and 2.0 15
N-1H
HSQC spectra, the overall change in amide chemical shift was obtained from the
expression:
√
⁄ (3.5)
where ΔδHN and ΔδN are the observed differences in the 1H
N and
15N chemical shifts,
respectively, between the pH 6.5 and 2.0 spectra. By and large, the perturbations in
backbone amide chemical shifts of LEN upon acidification were found to be relatively
minor, with the average Δδamide of 0.17 ± 0.17 ppm and only six residues (S14, G41, Q42,
E81, Q100, and R108) with Δδamide > 0.4 ppm. Figure 3.4A shows the plot of Δδamide vs
residue number, and in Figure 3.4B,C the amide chemical shift changes are mapped onto
the X-ray structure of LEN. These results suggest that LEN does not undergo major
structural changes between physiological and acidic pH.
To probe more directly for the presence of pH-related structural differences in
LEN, we also examined the pH dependence of 13
Cα shifts, which are highly sensitive
reporters of the protein backbone conformation.105
Figure 3.4E shows that 13
Cα shifts at
pH 6.5 and 2.0 exhibit a nearly linear correlation (R2 = 0.991).
44
Figure 3.4: (A) Plot of residue-specific amide chemical shift changes observed for LEN
upon changing the pH from 6.5 to 2.0. (B, C) Amide chemical shift changes from panel
(A) mapped onto the X-ray structure of (B) the LEN dimer and (C) one of the monomer
subunits (PDB entry 1LVE). The observed Δδamide values range from ∼0 ppm (blue) to
∼1 ppm (green), as indicated by the color bar. The N-terminal residue (D1) and six
prolines (aa 8, 40, 43, 44, 59, 95), which are undetectable in 15
N-1H HSQC spectra, are
colored in gray, while residues for which a Δδ amide value could not be determined due
to excessive chemical exchange broadening in pH 6.5 and/or 2.0 spectra are colored in
red. (D) Plot of residue-specific 13
Cα chemical shift differences for LEN at pH 6.5 and
2.0 (E) Plot of 13
Cα shifts for LEN at pH 2.0 vs pH 6.5
45
Calculation of residue-specific 13
Cα shift differences, ΔδCα = δCα(pH 2.0) - δCα(pH 6.5),
yields an average ΔδCα of -0.1 ± 0.4 ppm and shows that nearly all of the few outliers
characterized by |ΔδCα| in the ~1-2 ppm range (Figure 3.4D) correspond to loop regions
and amino acid residues having carboxylate groups (Asp, Glu, C-terminus) that become
protonated at low pH. Taken together, the comparison of backbone 1H
N,
15N, and
13Cα
chemical shifts at pH 6.5 and 2.0 strongly indicates that the three-dimensional fold of
LEN remains effectively unperturbed upon acidification. This is consistent with previous
suggestions by Fink and co-workers,63,103,104
based on results of low-resolution
biophysical measurements, that structures of LEN at physiological and low pH are
similar. Concurrently, the facts that LEN is capable of assembling into amyloid fibrils at
pH 263,103,104
and that partial unfolding of proteins is generally believed to be a critical
step in amyloid formation4,106
imply that thorough knowledge of LEN dynamics is likely
to be important for understanding the initial steps along the protein aggregation pathway.
With this premise we have commenced a comparative study of the backbone dynamics of
LEN at physiological and acidic pH.
3.3.2. Fast time scale protein backbone dynamics
The ps-ns time scale dynamics of LEN were probed at pH 6.5 and 2.0, by
measuring the R1 and R2 rates and {1H}-
15N NOEs for backbone amide
15N nuclei (Table
3.2 and Figure 3.5). At pH 6.5, relaxation data could be obtained for 103 of 108
nonproline residues. (Data for D1, S27f, Q37, L47, and Y96 were not accessible due to
excessive exchange broadening or peak overlap in 2D 15
N-1H HSQC spectra.)
46
Figure 3.5: Plots of 15
N R1 (A), R2 (B), NOE (C) and ηxy (D) as a function of residue
number for LEN at 600 MHz, 25 °C, and pH 6.5 (filled circles) and pH 2.0 (hollow
circles).
47
In order to extract quantitative motional parameters from these data, the extended Lipari-
Szabo model-free formalism was employed,95,96,107
and the analysis was performed using
isotropic and anisotropic motional models. While a reasonable fit to the experimental
data could be obtained with an isotropic tumbling model, the statistical F-value analysis
clearly indicated an axially symmetric anisotropic rotational diffusion model to be more
appropriate; the use of a fully asymmetric diffusion tensor did not provide a further
statistically significant improvement to the fit. As noted in the Experimental Procedures
section, the initial rotational diffusion tensor parameters were estimated according to an
established approach,94
with 15
N R2/R1 ratios for a conservative set of 40 residues located
in the most highly structured β-strand regions of the protein and the X-ray structure of
LEN used to construct the dimer. Model-free calculations were then initiated for the 103
residues with available relaxation data, with concurrent optimization of the diffusion
tensor parameters. In summary, the model-free analysis yielded generalized order
parameters, S2, for 84 residues (Table 3.3) with an average S
2 of 0.90 ± 0.06, as well as
the diffusion tensor parameters including the rotational correlation time, τc, of 13.48 ±
0.01 ns and Dǁ/D┴ = 0.825 ±0.006. The remaining 19 residues could not be satisfactorily
assigned to any one of the five possible motional models.98
A similar analysis of fast time scale motions was performed using 15
N R1, R2, and
{1H}-
15N NOE data for LEN at pH 2.0 (Table 3.2 and Figure 3.5). In this case an axially
symmetric diffusion tensor was also found to provide the best fit to the relaxation
measurements, with the final optimized parameters τc = 13.42 ± 0.01 ns and Dǁ/D┴ =
0.880 ± 0.005. Note that the diffusion tensor estimation assumed that the fold of LEN at
pH 2.0 is accurately represented by the X-ray structure under native conditions; this
48
assumption appears to be reasonable based on the backbone chemical shift data discussed
above. In summary, order parameters were obtained for a total of 78 residues (Table 3.3)
with an average S2 of 0.90 ± 0.05; ten residues could not be assigned to any of the five
motional models, and relaxation data could not be accurately extracted for S14 due to
partial peak overlap in the HSQC spectra.
The comparative analysis of ps-ns time scale dynamics of LEN at pH 6.5 and 2.0
allows several conclusions to be made. First, we note that 15
N R1, R2, and {1H}-
15N NOE
measurements at both pH values are optimally fit with effectively identical diffusion
tensor parameters and that the calculated τc values are characteristic of the presence of a
predominantly dimeric protein species. In other words, the native three-dimensional fold
and oligomeric state of LEN are not significantly altered at low pH. Second, the residue-
specific order parameters obtained using the model-free approach (Figure 3.6A) reveal
that, on the whole, the LEN backbone does not undergo significant motions on the ps-ns
time scale irrespective of pH, with the average S2 of ∼0.9 at both pH 6.5 and 2.0. We
note, however, that a small number of residues in LEN at pH 6.5 (A12, S14, A19, S29,
E81, V83, K107, and R108) and pH 2.0 (A12, L15, G41, S77, I106, and K107) do appear
to experience elevated internal motions on the ∼500-2000 ps time scale (Table 3.3).
Most of these residues (except S29 and G41) are located at the periphery of the LEN
dimer near the C-termini and are thus unlikely to be of major functional relevance. Third,
and most significantly, the model-free calculations for LEN at both pH values required
the inclusion of phenomenological chemical exchange correction factors (Rex) in order to
adequately model the 15
N R2 relaxation data for a number of residues.
49
Figure 3.6: (A) Generalized backbone amide order parameters, S2, for LEN at pH 6.5
(filled circles) and pH 2.0 (hollow circles) obtained by fitting 15
N spin relaxation (R1, R2,
and NOE) data at 600 MHz 1H frequency to one of the five possible motional models
according to Mandel et al.98
(see text and Supporting Information Tables S1 and S2 and
Figure S2 for additional details). (B, C) Plots of phenomenological chemical exchange
correction factors, Rex, required to properly account for the experimental 15
N R2 values
within model-free analysis for LEN at pH 6.5 (B) and pH 2.0 (C). (D, E) Plots of R2/ηxy,
the ratio of the 15
N transverse relaxation rate constant to the 1H-
15N dipole-dipole/
15N
CSA interference rate constant, determined at 600 MHz 1H frequency for LEN at pH 6.5
(D) and 2.0 (E). The average R2/ηxy values were found to be 1.38 ±0.21 at pH 6.5 and 1.6
± 0.4 at pH 2.0; for residues that are not subject to slow conformational exchange R2/ηxy
is expected to be approximately constant.
50
As shown in Figure 3.6B, at physiological pH, residues that display increased likelihood
of undergoing chemical exchange dynamics are located primarily in the three CDR loops.
At pH 2.0 the largest exchange contributions to R2 were found for residues in the N-
terminal β-strand and the CDR1 region (Figure 3.6C), while a number of residues in the
CDR2 and CDR3 regions exhibited severe exchange broadening and were not detectable
in HSQC spectra. These observations point to the possible presence in LEN of slower
time scale conformational dynamics, which become more pronounced at acidic pH.
Motions of this type are frequently associated with partial unfolding of proteins74
and are
therefore likely to be relevant for understanding protein aggregation phenomena. The fact
that no doubling of backbone amide resonances is readily detected in HSQC spectra of
LEN at either pH suggests that populations of the minor protein conformers are relatively
low and/or that exchange dynamics are in the fast or intermediate regime on the chemical
shift time scale. Consequently, we have focused our attention on the characterization of
the slower, ms time scale backbone dynamics of LEN as a function of pH.
3.3.3. Characterization of slow time scale protein motions
Prior to quantitative analysis of ms time scale backbone dynamics by CPMG
relaxation dispersion NMR techniques, measurements of 1H-
15N dipole_dipole/
15N CSA
interference rate constants, ηxy, were performed for LEN at pH 6.5 and 2.0 (Table 3.2
and Figure 3.5). In the absence of chemical exchange, 15
N R2 rates are given by κηxy,86
where κ is a constant characteristic of the protein. Therefore, determination of residue-
specific R2/ηxy ratios provides a convenient approach to rapidly and directly probe for the
presence of slow conformational exchange along the protein backbone—namely,
assuming that most sites are relatively unaffected by chemical exchange phenomena, the
51
aa residues with considerably elevated R2/ηxy ratios with respect to the average are likely
to be subject to significant dynamics on the ms time scale.71
At pH 6.5, most residues in
LEN displayed very similar R2/ηxy ratios (Figure 3.6D), with an average value of 1.38 ±
0.21. A total of 12 residues, including Y32, L46, I48, A51, R54, S56, V58, L78, Q90,
G99, Q100, and T102, have R2/ηxy ratios of at least one standard deviation larger than the
average. For LEN at pH 2.0, the R2/ηxy average and dispersion (1.6 ± 0.4) both increased
compared to pH 6.5, with Q6, K24, S25, V27b, S27e, Q42, L78, G99, and L104 having
R2/ηxy ratios of at least one standard deviation larger than the mean (Figure 3.6E).
Although qualitative, the R2/ηxy analysis confirms that multiple residues in LEN, many
located in or around the CDR loop regions, indeed experience exchange dynamics at pH
6.5 and pH 2.0 and that these effects appear to become more pronounced as the pH is
decreased.
Millisecond time scale motions of LEN were subsequently examined
quantitatively by using 15
N-CPMG relaxation dispersion NMR methods, which are able
to probe chemical exchange phenomena in proteins that involve minor conformer
populations on the order of a few percent or less.73,74
Residue-specific relaxation
dispersion trajectories at magnetic field strengths of 14.1 and 18.8 T were recorded for
LEN at pH 6.5 and 2.0. Initially, the relaxation dispersion data for individual residues
were fit to the Carver-Richards model for two-site exchange100
to obtain estimates for the
chemical exchange contribution, Rex, to 15
N R2. Figure 3.7A and Figure 3.7C show plots
of Rex at 600 and 800 MHz 1H frequency as a function of residue number for LEN at pH
6.5 and 2.0, respectively. In Figure 3.7B and Figure 3.7D, the Rex values at 600 MHz
and pH 6.5 and 2.0, respectively, are mapped onto the X-ray structure of LEN.
52
Figure 3.7: Summary of 15
N-CPMG relaxation dispersion NMR data for LEN at pH 6.5
and 2.0. (A) Plot of Rex, the estimated chemical exchange contribution to 15
N R2, for LEN
at pH 6.5 and 14.1 T (hollow circles) and 18.8 T (filled circles). (B) The Rex values at
14.1 T mapped onto the X-ray structure for one of the monomer subunits of LEN, with
magnitudes indicated by the color bar. Residues that were undetectable or displayed
severe peak overlap in the 15
N-1H HSQC spectrum are colored in gray, while residues for
which Rex could not be determined due to excessive chemical exchange broadening are
colored in red. (C, D) Same as panels (A, B) for LEN at pH 2.0.
53
At pH 6.5, the vast majority of residues characterized by Rex > 3 s-1
at 600 MHz and
interpreted as undergoing significant exchange dynamics (Figure 3.7A and Figure 3.7B)
were found to cluster roughly in three contiguous stretches: (1) aa V3-D9 near the N-
terminus (denoted “region 1” and containing β-strand A), (2) aa Q42-G57 (denoted
“region 2” and encompassing β-strand E and the entire CDR2 loop), and (3) aa Q89-
G101 (denoted “region 3” and containing the CDR3 loop). Note that several isolated
residues within these dynamic stretches (e.g., S52) displayed relatively low Rex values—
this is likely due to similar chemical shifts for the exchanging conformers (vide infra). On
the other hand, large segments of the protein backbone, spanning residues ~10-35 and
~60-85 as well as ~5-6 C-terminal amino acids, exhibit generally negligible ms time scale
motions. Analogous studies at acidic pH reveal that all LEN residues in regions 1-3
subject to chemical exchange at pH 6.5 are also dynamic at pH 2.0, with the majority
characterized by much larger Rex values, or cannot be detected at all in HSQC spectra due
to excessive exchange broadening (Figure 3.7C and Figure 3.7D). Several additional
residues, most found in the vicinity of region 2 and a few isolated aa in the CDR1 loop,
also showed elevated Rex values relative to pH 6.5. In summary, the quantitative CPMG
relaxation dispersion measurements reveal that residues located near the N-terminus of
LEN as well as in and around the CDR2 and CDR3 loops in proximity of the dimer
interface are subject to considerable ms time scale dynamics, which become more
significant at low pH. Meanwhile, for the most part, the structured β-sheet core region of
the protein and the CDR1 loop experience minimal conformational flexibility. These
findings are generally in agreement with the more qualitative evaluation of chemical
exchange phenomena by the model-free and R2/ηxy approaches discussed above.
54
In order to extract the rates of exchange between the major and minor protein
conformers, kex, within the two-site exchange model, and the populations of minor
conformers, pB, from the CPMG relaxation dispersion trajectories, we performed global
two-state fits of data at both magnetic fields for residues located in regions 1-3 identified
above. Initial data fitting was done by assuming the same kex and pB parameters for all
three dynamic regions of LEN at each pH—this analysis yielded kex = 491 ± 19 s-1
and pB
= 2.40 ± 0.10% at pH 6.5 and kex = 500 ± 13 s-1
and pB = 6.84 ± 0.11% at pH 2.0.
Although, on the whole, reasonable quality fits to the relaxation dispersion data were
obtained in this manner, it was also apparent that trajectories showing the largest
relaxation dispersion were fit quite poorly, especially for small νCPMG frequencies (data
not shown). Experimental data were therefore reanalyzed by considering dynamic regions
1-3 independently from one another, which resulted in significantly improved fits for all
relaxation dispersion trajectories (see representative trajectories and fits in Figure 3.8 and
complete data sets in Figure 3.9 and Figure 3.10) and yielded independent exchange
parameters, kex and pB, for each region as follows: pH 6.5, region 1: kex = 200 ± 40 s-1
, pB
= 4.0 ± 0.40%; region 2: kex = 630 ± 30 s-1
, pB = 2.30 ± 0.06%; region 3: kex = 250 ± 30 s-
1, pB = 2.30 ± 0.21%; pH 2.0, region 1: kex = 690 ± 30 s
-1, pB = 15.6 ± 1.8%; region 2: kex
= 540 ± 20 s-1
, pB = 11.6 ± 0.8%; region 3: kex = 390 ± 30 s-1
, pB = 6.6 ± 0.3%. Since the
three flexible regions of LEN are well separated from each other in the primary amino
acid sequence, even if their motions are correlated it is reasonable to expect that each
region may experience conformational dynamics characterized by somewhat different
exchange parameters.
55
Figure 3.8: 15
N-CPMG relaxation dispersion NMR trajectories for representative
residues from the three contiguous regions of LEN exhibiting elevated conformational
flexibility on the ms time scale at pH 6.5 and 2.0. Experimental (circles) and simulated
(lines) CPMG trajectories at pH 6.5 and 2.0 are shown in blue (left column) and red
(right column), respectively, with experimental data corresponding to magnetic field
strengths of 14.1 T (600 MHz 1H frequency; hollow (circles) and 18.8 T (800 MHz
1H
frequency; filled circles).
56
Figure 3.9: 15
N-CPMG relaxation dispersion trajectories at 600 MHz (hollow circles)
and 800 MHz (filled circles) for residues found in regions of LEN, exhibiting ms time
scale dynamics at pH 6.5. As discussed in the main text (c.f., Table 3.1), for each of the
three dynamic regions the relaxation dispersion data at both magnetic fields were fit
simultaneously using a two-site exchange model with a single exchange rate (kex) and
population of minor conformer (pB). The best fits are shown as solid lines.
57
Figure 3.10: 15
N-CPMG relaxation dispersion trajectories at 600 MHz (hollow circles)
and 800 MHz (filled circles) for residues found in regions of LEN, exhibiting ms time
scale dynamics at pH 2.0.
58
Analysis of the CPMG relaxation dispersion trajectories also yielded the differences in
15N chemical shifts between the two exchanging protein states, Δω, for each residue, and
these data are summarized in Table 3.1. At both pH 6.5 and 2.0, most residues for which
relaxation data are available exhibit |Δω| on the order of ∼1 ppm. Given that the
exchange rates for dynamic regions 1-3 of LEN are in the ∼200-700 s-1
range, these |Δω|
values indicate that the vast majority of residues are in the intermediate (kex ∼ |Δω|) to
fast (kex > |Δω|) exchange regime with respect to 15
N chemical shifts. Thus, the severe
line broadening observed at pH 2.0 for multiple residues in flexible regions 2 and 3 is
likely caused by increased populations of the minor protein conformers combined with
exchange dynamics being in the intermediate regime on the 15
N and/or 1H chemical shift
time scales.
Taken together, the 15
N relaxation measurements indicate that LEN exhibits
limited conformational flexibility with the exception of three well-defined regions of ∼5-
20 aa positioned near the dimer interface. Around physiological pH, the slow protein
backbone dynamics in these regions are associated with relatively low excited-state
protein conformer populations, in the ∼2-4% range. Upon acidification, the CPMG
relaxation dispersion data are compatible with a significant, ∼3-4 fold, increase in the
higher-energy minor protein conformer populations, accompanied by similar or slightly
accelerated rates of exchange between the major and minor conformers for most residues.
These findings provide valuable molecular-level insights into why LEN does not form
amyloid under physiological conditions and does so only with considerable difficulty at
low pH. In addition, our data clearly point to the LEN dimer dissociation process,
correlated with increased ms time scale motions for a small set of specific residues, as
59
being one of the important initial steps involved in protein misfolding and amyloid
assembly. This conclusion is consistent with the fact that the amide 15
N chemical shifts
for most of the dynamic residues in the excited state of LEN, readily obtained from the
corresponding major conformer shifts and Δω values (cf. Table 3.1), tend toward random
coil values.108
3.4 Discussion
LEN is structurally homologous to other Ig VL domains of the κ subgroup.55
The
protein fold consists of two antiparallel β-sheets—made up of strands A,C, G, F and D,E,
H (Figure 3.4) — comprising the framework region and three hypervariable CDR loops
located between strands C and D (CDR1), E and F (CDR2), and H and I (CDR3). The
tertiary structure is stabilized by hydrogen bonds between the antiparallel strands and the
shielding of hydrophobic β-sheet residues from solvent. Formation of dimeric quaternary
structure is facilitated primarily by two sets of intermolecular hydrogen bonds involving
side chains of residues Q38 and Q38, and Y36 and Q89,55
while intramolecular contacts
between the Q38 and Y87 side chains further reinforce the structural integrity of the
interface region as shown in Figure 3.11. This architecture balances the electrostatic
interactions involving surface residues and minimizes unfavorable interactions of
nonpolar side chains with water molecules, with contacts between the Q38 side chains
being especially important for stabilizing the dimer conformation. In fact, a single
mutation, Q38E, has been found to alter the charge balance so severely that a distinct
dimer interface is obtained, with one of the monomer subunits rotated by ∼180 relative to
the native protein.109
60
Table 3.1: Summary of 15
N-CPMG relaxation dispersion NMR data analysis for the three
dynamic regions of LEN at pH 6.5 and 2.0. The signs of the chemical shift differences
were obtained according to the method of Kay and co-workers.102
Region 1 pH 6.5
kex = 200 ± 40 s-1, pB = 4.0 ± 0.4 %
pH 2.0
kex = 690 ± 30 s-1, pB = 15.6 ± 1.8 %
Residue Δω (ppm) Δω (ppm)
V3 -0.99 ± 0.04 -1.22 ± 0.01
M4 -0.80 ± 0.03 -1.13 ± 0.01 T5 -0.56 ± 0.04c -0.34 ± 0.01
Q6 -1.26 ± 0.04 -0.95 ± 0.01
S7 1.03 ± 0.04 -0.74 ± 0.01 D9 -0.18 ± 0.09c ~0
Region 2 pH 6.5
kex = 630 ± 30 s-1, pB = 2.30 ± 0.06 %
pH 2.0
kex = 540 ± 20 s-1, pB = 11.6 ± 0.8 %
Residue Δω (ppm) Δω (ppm)
Q37 –a -0.58 ± 0.01c
Q38 – 0.50 ± 0.01
K39 – -0.40 ± 0.02 P40 n/ab n/a
G41 – -0.70 ± 0.01
Q42 1.88 ± 0.04 -1.08 ± 0.01 P43 n/a n/a
P44 n/a n/a
K45 1.99 ± 0.06 n/a L46 2.57 ± 0.06 n/a
L47 n/a n/a
I48 -0.32 ± 0.07 0.20 ± 0.03c Y49 -2.01 ± 0.05 n/a
W50 2.82 ± 0.07 n/a
A51 1.19 ± 0.04 n/a
S52 ~0 n/a
T53 -0.48 ± 0.06 n/a
R54 1.78 ± 0.04c -0.74 ± 0.01 E55 1.21 ± 0.04 -1.47 ± 0.02
S56 -2.18 ± 0.06 n/a
G57 -1.39 ± 0.04 -0.20 ± 0.02 V58 – -0.40 ± 0.01
Region 3 pH 6.5
kex = 250 ± 30 s-1, pB = 2.30 ± 0.21 %
pH 2.0
kex = 390 ± 30 s-1, pB = 6.6 ± 0.3 %
Residue Δω (ppm) Δω (ppm)
Q89 -1.13 ± 0.06c n/a
Q90 1.29 ± 0.07c n/a
Y91 -0.70 ± 0.06c n/a Y92 -0.84 ± 0.06 n/a
S93 1.10 ± 0.06c n/a
T94 -0.74 ± 0.06 n/a P95 n/a n/a
Y96 n/a n/a
S97 2.88 ± 0.12 n/a
F98 -2.02 ± 0.09 n/a
G99 1.67 ± 0.07 -1.10 ± 0.02
Q100 -2.92 ± 0.15 3.30 ± 0.05 G101 -0.44 ± 0.07c -0.33 ± 0.03
aResonance present in spectrum but not included in data fitting due to low Rex value (see Figure 3).
bResonance not detectable due to excessive exchange broadening, spectral overlap or a proline residue.
cDifference in
15N resonance frequencies between HSQC and HMQC spectra < 0.3 Hz.
61
At physiological pH LEN does not readily assemble into amyloid fibrils,8
presumably due to the high degree of stability of the native tertiary and quaternary
structure. Indeed, the 15
N relaxation data, which show that overall the LEN dimer
displays rather limited backbone motions, are consistent with the notion that the most
critical intra- and intermolecular hydrogen-bonding interactions remain effectively intact
under these conditions. In spite of this apparent structural integrity, however, the 15
N-
CPMG relaxation dispersion measurements reveal the existence of three extended
dynamic “hot spots” along the protein backbone; consisting of ∼5-20 amino acids each,
subject to slow motions on the ms time scale (Figure 3.7). These dynamic regions
correspond roughly to (1) N-terminal β-strand A and loop between strands A and B, (2)
loop between strands D and E, β-strand E, and CDR2 loop, and (3) the CDR3 loop.
Interestingly, with the exception of residue S27f which is subject to pronounced
conformational exchange broadening in 15
N-1H HSQC spectra
6 and residues K24, Y27d,
Y32, and L33 which show some degree of relaxation dispersion, the large CDR1 loop
located at the periphery of LEN is relatively inflexible on the ms time scale. This is likely
due to the presence of an extensive network of intramolecular hydrogen bonds involving
the backbone HN and O atoms of residues V27b, Y27d, S29, K30, and Y32 and the N28
and Y32 side chains.55
62
Figure 3.11: (A) Cartoon representation of the LEN dimer X-ray structure, highlighting
important hydrogen-bonding interactions between side chains of residues Y36, Q38, and
Q89 across the dimer interface. Residues located in the three contiguous dynamic
regions of LEN at pH 6.5 and/or 2.0, as identified by CPMG relaxation dispersion NMR
methods are colored in red. Surface representations of (B) the LEN dimer corresponding
to the cartoon diagram in panel (A), and (C) one of the monomer subunits, with the
conformationally flexible residues listed above colored in red. Selected residues are
labeled for reference.
63
Although quantitative analysis of the CPMG relaxation dispersion data indicates
that populations of higher-energy protein states associated with elevated ms regime
backbone motions are very low at physiological pH (∼2-4% of the major, ground-state
conformer), the location of the three flexible regions within the LEN structure furnishes
valuable molecular-level insights into the likely initial events leading to protein
misfolding and amyloid assembly under destabilizing conditions. Specifically, dynamic
regions 2 and 3 encompass the vast majority of residues lining the dimer interface, and aa
V3-D9, making up dynamic region 1, pack directly against region 3 residues ∼S97-G101
(Figure 3.11A); all three flexible regions are terminated primarily by interactions
involving the framework β-strands. Altogether, it is evident that amino acids located in
regions 1-3 of LEN discussed above cluster to form a large area predisposed to undergo
considerable conformational fluctuations, covering effectively the entire dimer interface
(Figure 3.11B,C). These conformational fluctuations appear to play an important role in
destabilizing the native fold of LEN, triggering dimer dissociation and, ultimately,
amyloid formation once a sufficiently high concentration of appropriately misfolded
protein states is achieved.4,106
In order to further probe this issue, an analogous set of 15
N relaxation
measurements was performed at pH 2.0, where in the presence of agitation LEN is known
to undergo conversion to the amyloid state.63,103,104
Measurements of ps-ns time scale
dynamics and backbone chemical shifts (Figure 3.4 and Figure 3.6) show that, even at
low pH where it is capable of forming amyloid, LEN by and large retains its native three-
dimensional fold, in agreement with results of previous low-resolution biophysical
measurements63,103,104
which suggest that the structure and dimeric state of LEN are
64
relatively unperturbed at pH ∼2. Most remarkable, however, are the findings that all
residues along the LEN dimer interface found to exhibit slow conformational dynamics at
pH 6.5 (as well as several additional amino acids in the CDR1 loop and in the loop
between β-strands D and E, bordering region 2) are subject to significant ms time scale
motions at pH 2.0 (Figure 3.7) and secondly, partially unfolded, excited-state protein
conformer populations increase significantly, from ∼2-4% at physiological pH to ∼10-
15% at low pH. The latter finding, which is qualitatively consistent with the observed
decrease in the dimerization constant for LEN from ∼4 ∙ 105 M
-1 under physiological
conditions59
to ∼2 ∙ 104 M
-1 at pH 2,
103 suggests that at acidic pH the key hydrogen-
bonding interactions involving residues Q38 and Q89 as well as neighboring water
molecules become progressively more disrupted, allowing the protein to occupy higher-
energy, partially unfolded states with greater frequency.
It is important to reiterate here that the main assumption underlying the
interpretation of 15
N CPMG relaxation dispersion data—namely, that the observed
relaxation dispersions report on internal protein backbone dynamics within the LEN
dimer—is reasonable and that these dispersions and the significant differences between
them at physiological and acidic pH are not simply the result of the monomer-dimer
equilibrium process and any associated chemical shift changes. First, as already discussed
in detail above, the strong correlations in 1H,
15N, and
13Cα chemical shifts, which
include shifts for a number of residues found near the dimer interface, and effectively
identical rotational correlation times obtained from the ps-ns time scale dynamics
analysis indicate no significant differences in the structure and oligomeric state of LEN at
pH 6.5 and 2.0. In addition, to probe this issue more directly we have recorded a set of
65
15N CPMG relaxation dispersion trajectories at 800 MHz
1H frequency for a sample of
15N-LEN at pH 6.5 diluted to 0.8 mM—i.e., having a protein concentration that is lower
by more than a factor of 2, and thus an increased fraction of LEN in the monomeric state,
relative to samples used to record the relaxation data in Figure 3.8. The fact that no
significant differences are observed for any of the residue-specific relaxation dispersion
trajectories between the 0.8 and 1.7 mM LEN samples (Figure 3.12) confirms that at
protein concentrations employed in this study these dispersions can be interpreted
exclusively in terms of internal dynamics of the LEN dimer and that any contributions
due to the monomer-dimer equilibrium process are negligible.
Under agitation, protein molecules including antibody fragments have been found
to adsorb at the air-water interface, resulting in hydrophobic stress, destabilization of
native structure to form partially unfolded states, and, ultimately, aggregation.110,111
At
acidic pH, though still largely a properly folded dimer, LEN exists in an aggregation-
competent state, where a number of amino acid residues along the dimer interface display
significant conformational flexibility that interferes with the maintenance of proper
hydrogen-bonding interactions between the individual monomer subunits. Nonetheless,
since LEN is incapable of forming amyloid in the absence of agitation,8,62,63,103,104
a ~10-
15% population of the excited-state protein conformers present at pH 2 is alone clearly
insufficient to tip the system toward aggregation. The agitation process thus appears to be
required to overcome the additional energy barrier for aggregation by generating a higher
concentration of misfolded amyloidogenic protein states, possibly with further residues in
the vicinity of the three most flexible regions experiencing elevated dynamics on the ms
time scale.
66
Figure 3.12: 15
N CPMG relaxation dispersion trajectories at 800 MHz for residues found
in regions of LEN, exhibiting ms time scale dynamics at pH 6.5. Data are shown for two
different LEN concentrations: 0.8 mM (hollow circles) and 1.7 mM (filled circles). The
similarity of the residue specific dispersion trajectories for the two protein
concentrations indicates that they report on internal backbone dynamics within the LEN
dimer, and that the contribution of the LEN monomer-dimer equilibrium process to the
observed relaxation dispersions is negligible in this protein concentration regime.
67
Moreover, in retrospect, these findings imply that mere agitation of LEN under
physiological conditions does not lead to aggregation because the partially unfolded,
higher-energy states are accessed for only a small fraction of time, owing to the
sufficiently high degree of dimer interface stabilization by the network of hydrogen
bonds.
While additional NMR studies, including site-resolved measurements of 1H
N,
13CO, and
13Cα relaxation dispersion,
74 are necessary to generate more precise models of
the partially unfolded states of LEN through which conversion from the native state to
amyloid can occur, the current study clearly identifies the key amino acid residues
involved in the earliest stages of protein misfolding and lays the foundation for future
investigations along these lines. Interestingly, the light-chain variable domain SMA,
which is highly homologous to LEN and associated with development of AL
amyloidosis,59,112
has been found to exhibit, under physiological conditions, many
characteristics of LEN at pH ∼2.61,63,103,104
This suggests that the main findings of this
work, linking the conformational dynamics of a specific set of residues colocalized at the
dimer interface to populating higher energy partially misfolded protein conformers, are
relevant for understanding, at the molecular level, the primary driving forces behind
amyloid formation for disease-related Ig VL variants and the effects of site-specific
mutations on the aggregation propensities of these proteins.8,52
Finally, the identification
of specific residues and regions involved in the earliest stages of protein unfolding may
aid in the rational design of therapeutics that stabilize the tertiary and quaternary structure
and prevent aggregation of these immunoglobulin light-chain domains, in analogy to
transthyretin,113
superoxide dismutase,114
and other amyloidogenic proteins.106
68
Table 3.2: 15
N R1, R2, {1H}-
15N NOE, and ηxy values for LEN at 600 MHz
1H frequency
and pH 6.5 and 2.0. Complementarity determining regions (CDR1, 2, and 3) are
highlighted in gray.
pH 6.5 pH 2.0
Residue R1 (s-1
) R2 (s-1
) NOE ηxy (s-1
) R1 (s-1
) R2 (s-1
) NOE ηxy (s-1
)
D1 I2 0.891 ± 0.024 16.6 ± 0.3 0.612 ± 0.024 V3 0.971 ± 0.015 17.6 ± 0.3 0.851 ± 0.022 11.5 ± 1.0 0.95 ± 0.06 23.2 ± 1.5 0.80 ± 0.12 M4 0.935 ± 0.009 17.22 ± 0.16 0.821 ± 0.016 13.0 ± 0.5 1.00 ± 0.04 26.8 ± 1.6 0.93 ± 0.11 T5 0.930 ± 0.009 17.95 ± 0.14 0.803 ± 0.014 12.9 ± 0.8 0.899 ± 0.009 18.47 ± 0.16 0.86 ± 0.03 13.1 ± 1.6 Q6 1.003 ± 0.016 17.38 ± 0.24 0.827 ± 0.022 14.7 ± 3.6 0.89 ± 0.03 22.2 ± 0.6 0.86 ± 0.08 10 ± 3 S7 0.988 ± 0.010 18.51 ± 0.16 0.847 ± 0.016 14.9 ± 0.4 0.987 ± 0.018 21.7 ± 0.4 0.88 ± 0.05 P8 D9 0.965 ± 0.007 18.2 ± 0.1 0.832 ± 0.011 15.0 ± 0.6 0.912 ± 0.004 16.28 ± 0.06 0.831 ± 0.012 10.75 ± 0.08 S10 0.968 ± 0.004 17.24 ± 0.06 0.795 ± 0.007 12.82 ± 0.11 0.974 ± 0.003 16.76 ± 0.04 0.808 ± 0.010 10.72 ± 0.16 L11 0.927 ± 0.006 17.5 ± 0.1 0.778 ± 0.012 14.5 ± 0.8 0.938 ± 0.005 16.74 ± 0.08 0.801 ± 0.017 12.5 ± 0.3 A12 0.965 ± 0.004 15.60 ± 0.06 0.716 ± 0.009 12.7 ± 0.6 0.966 ± 0.003 14.61 ± 0.04 0.733 ± 0.011 10.74 ± 0.23 V13 0.936 ± 0.004 17.54 ± 0.07 0.650 ± 0.010 11.9 ± 0.7 0.954 ± 0.004 16.77 ± 0.06 0.692 ± 0.016 11.5 ± 0.4 S14 1.011 ± 0.004 16.33 ± 0.06 0.619 ± 0.008 11.22 ± 0.4 L15 0.927 ± 0.004 18.29 ± 0.07 0.792 ± 0.010 13.73 ± 0.24 0.960 ± 0.003 17.41 ± 0.05 0.775 ± 0.012 11.81 ± 0.19 G16 0.847 ± 0.007 18.51 ± 0.14 0.819 ± 0.016 12.8 ± 0.6 0.877 ± 0.006 17.92 ± 0.10 0.829 ± 0.022 11.08 ± 0.3 E17 1.003 ± 0.004 18.65 ± 0.07 0.814 ± 0.009 13.5 ± 0.3 0.970 ± 0.003 18.20 ± 0.05 0.781 ± 0.011 12.68 ± 0.19 R18 0.837 ± 0.004 19.99 ± 0.10 0.767 ± 0.010 13.9 ± 0.5 0.880 ± 0.004 18.80 ± 0.07 0.741 ± 0.014 12.17 ± 0.21 A19 0.955 ± 0.006 16.65 ± 0.10 0.739 ± 0.011 14.8 ± 0.5 0.977 ± 0.005 15.68 ± 0.07 0.804 ± 0.016 10.8 ± 1.05 T20 0.866 ± 0.003 17.40 ± 0.06 0.820 ± 0.008 12.14 ± 0.3 0.910 ± 0.003 16.16 ± 0.05 0.798 ± 0.012 10.20 ± 0.08 I21 0.986 ± 0.011 17.72 ± 0.19 0.838 ± 0.017 0.963 ± 0.008 17.42 ± 0.14 0.82 ± 0.03 11.90 ± 0.19 N22 0.945 ± 0.009 18.88 ± 0.17 0.877 ± 0.017 13.4 ± 0.6 0.992 ± 0.011 18.69 ± 0.17 0.84 ± 0.03 12.7 ± 0.9 C23 0.928 ± 0.006 18.71 ± 0.11 0.840 ± 0.010 13.5 ± 0.5 0.909 ± 0.007 19.61 ± 0.14 0.862 ± 0.020 10.8 ± 0.5 K24 0.972 ± 0.009 19.13 ± 0.18 0.789 ± 0.015 14.02 ± 0.18 0.939 ± 0.013 22.8 ± 0.3 0.87 ± 0.04 10.34 ± 1.17 S25 0.969 ± 0.009 18.88 ± 0.16 0.824 ± 0.015 14.24 ± 0.4 0.924 ± 0.013 20.4 ± 0.3 0.85 ± 0.04 7.1 ± 1.8 S26 0.983 ± 0.009 18.15 ± 0.15 0.848 ± 0.015 12.5 ± 0.3 0.943 ± 0.015 18.4 ± 0.3 0.86 ± 0.04 12.8 ± 1.6 Q27 1.012 ± 0.005 17.99 ± 0.08 0.810 ± 0.009 15.2 ± 0.6 1.018 ± 0.005 17.66 ± 0.07 0.821 ± 0.015 13.0 ± 0.3 S27a 1.02 ± 0.03 16.8 ± 0.4 0.81 ± 0.03 0.946 ± 0.012 16.16 ± 0.17 0.80 ± 0.03 11.5 ± 0.9 V27b 0.838 ± 0.013 20.8 ± 0.3 0.823 ± 0.020 16.19 ± 0.20 0.943 ± 0.016 19.0 ± 0.3 0.87 ± 0.04 7.2 ± 1.5 L27c 0.893 ± 0.009 18.67 ± 0.18 0.912 ± 0.016 14.2 ± 0.4 1.006 ± 0.024 16.8 ± 0.4 0.73 ± 0.06 Y27d 0.960 ± 0.017 19.8 ± 0.4 0.851 ± 0.022 16.2 ± 0.5 0.99 ± 0.03 22.4 ± 0.8 0.94 ± 0.08 S27e 1.00 ± 0.03 23.9 ± 0.8 0.84 ± 0.03 0.91 ± 0.03 24.2 ± 1.0 0.79 ± 0.06 6.3 ± 1.5 S27f 0.86 ± 0.03 14.9 ± 0.5 0.750 ± 0.06 11.38 ± 2.18 N28 1.053 ± 0.012 17.06 ± 0.17 0.826 ± 0.014 12.7 ± 0.8 0.872 ± 0.007 17.89 ± 0.12 0.840 ± 0.022 16.6 ± 0.8 S29 0.950 ± 0.010 16.77 ± 0.16 0.777 ± 0.014 11.8 ± 1.14 0.915 ± 0.010 16.01 ± 0.15 0.76 ± 0.03 10.92 ± 0.21 K30 0.872 ± 0.007 19.58 ± 0.15 0.758 ± 0.012 17.59 ± 4.14 N31 0.940 ± 0.011 17.97 ± 0.18 0.833 ± 0.016 13.4 ± 2.5 0.912 ± 0.011 19.71 ± 0.22 0.91 ± 0.03 10.3 ± 0.7 Y32 1.00 ± 0.03 17.7 ± 0.5 0.85 ± 0.03 11 ± 4 1.02 ± 0.04 17.3 ± 0.7 0.86 ± 0.08 L33 0.984 ± 0.012 19.24 ± 0.22 0.860 ± 0.016 15.22 ± 1.01 0.94 ± 0.06 27.1 ± 2.3 0.99 ± 0.15 A34 1.032 ± 0.013 17.0 ± 0.2 0.849 ± 0.018 14.4 ± 0.3 0.98 ± 0.03 18.3 ± 0.5 0.90 ± 0.06 11.0 ± 0.3 W35 0.938 ± 0.015 17.57 ± 0.24 0.833 ± 0.022 15.7 ± 1.4 0.94 ± 0.03 16.5 ± 0.4 0.79 ± 0.06 10.10 ± 2.04 Y36 0.953 ± 0.015 17.74 ± 0.24 0.820 ± 0.020 13.3 ± 0.9 0.931 ± 0.017 19.6 ± 0.3 0.85 ± 0.04 14.6 ± 2.5 Q37 0.938 ± 0.014 18.57 ± 0.24 0.88 ± 0.04 13.97 ± 5.15 Q38 0.939 ± 0.011 18.91 ± 0.21 0.777 ± 0.018 17.7 ± 0.9 0.967 ± 0.014 17.51 ± 0.22 0.88 ± 0.04 13.1 ± 0.4 K39 0.853 ± 0.014 18.8 ± 0.3 0.859 ± 0.024 14.9 ± 2.4 0.943 ± 0.021 16.7 ± 0.4 0.8 ± 0.05 P40 G41 0.94 ± 0.08 14.7 ± 1.0 0.76 ± 0.08 0.918 ± 0.012 13.28 ± 0.16 0.71 ± 0.03 9.0 ± 0.4 Q42 0.962 ± 0.009 19.12 ± 0.16 0.775 ± 0.014 13.3 ± 0.7 1.022 ± 0.018 17.6 ± 0.3 0.76 ± 0.03 7.0 ± 1.5 P43 P44 K45 0.949 ± 0.018 22.3 ± 0.4 0.760 ± 0.023 19 ± 4 L46 0.899 ± 0.018 20.6 ± 0.4 0.87 ± 0.03 9.31 ± 2.23 L47 I48 0.962 ± 0.010 17.60 ± 0.16 0.812 ± 0.017 10.02 ± 1.17 0.958 ± 0.012 18.62 ± 0.22 0.88 ± 0.03 11.2 ± 1.4 Y49 1.034 ± 0.013 23.5 ± 0.3 0.855 ± 0.021 14.9 ± 0.9 W50 1.04 ± 0.04 25.3 ± 1.2 0.85 ± 0.04 A51 1.018 ± 0.019 19.3 ± 0.4 0.92 ± 0.03 10.0 ± 2.5 S52 0.842 ± 0.008 17.85 ± 0.14 0.821 ± 0.014 14.30 ± 0.18 T53 0.982 ± 0.010 18.58 ± 0.18 0.840 ± 0.016 13.94 ± 0.02 R54 0.938 ± 0.010 18.48 ± 0.18 0.785 ± 0.016 8.7 ± 1.6 0.922 ± 0.024 18.3 ± 0.4 0.99 ± 0.07 E55 0.971 ± 0.013 18.77 ± 0.25 0.858 ± 0.021 14.2 ± 1.16 1.02 ± 0.04 20.3 ± 0.7 0.80 ± 0.09 S56 0.932 ± 0.013 21.9 ± 0.3 0.816 ± 0.021 13.5 ± 1.3 G57 0.98 ± 0.03 16.4 ± 0.4 0.83 ± 0.03 0.841 ± 0.006 15.83 ± 0.09 0.735 ± 0.020 10.15 ± 0.12 V58 0.927 ± 0.004 17.97 ± 0.07 0.757 ± 0.009 11.0 ± 0.5 0.901 ± 0.009 17.54 ± 0.14 0.82 ± 0.03 10.4 ± 0.9
Continued
69
D60 0.979 ± 0.006 17.16 ± 0.09 0.829 ± 0.011 12.4 ± 0.4 0.952 ± 0.006 16.63 ± 0.09 0.816 ± 0.020 10.2 ± 0.4 R61 1.007 ± 0.011 19.42 ± 0.22 0.864 ± 0.016 12.52 ± 1.22 0.994 ± 0.008 16.75 ± 0.12 0.843 ± 0.022 10.3 ± 1.0 F62 0.990 ± 0.007 20.33 ± 0.14 0.850 ± 0.012 15.7 ± 0.8 1.020 ± 0.005 19.57 ± 0.09 0.851 ± 0.015 11.74 ± 1.22 S63 0.935 ± 0.007 18.13 ± 0.12 0.808 ± 0.013 12.6 ± 0.8 0.963 ± 0.008 17.27 ± 0.11 0.843 ± 0.023 8.8 ± 0.8 G64 0.953 ± 0.008 18.41 ± 0.14 0.840 ± 0.015 11.8 ± 0.9 0.994 ± 0.009 17.92 ± 0.13 0.87 ± 0.03 11.0 ± 0.8 S65 0.987 ± 0.008 18.03 ± 0.13 0.876 ± 0.012 13.6 ± 0.4 0.973 ± 0.007 18.03 ± 0.11 0.848 ± 0.017 12.2 ± 0.3 G66 1.018 ± 0.007 15.96 ± 0.09 0.845 ± 0.010 12.61 ± 0.20 0.899 ± 0.006 15.61 ± 0.10 0.809 ± 0.018 10.85 ± 0.11 S67 0.832 ± 0.004 16.70 ± 0.08 0.804 ± 0.008 12.0 ± 0.3 0.877 ± 0.006 15.80 ± 0.09 0.817 ± 0.018 8.74 ± 0.23 G68 0.943 ± 0.013 17.84 ± 0.24 0.899 ± 0.020 0.991 ± 0.013 15.81 ± 0.18 0.87 ± 0.03 11.6 ± 1.3 T69 0.901 ± 0.009 17.96 ± 0.14 0.866 ± 0.014 11.5 ± 0.4 0.891 ± 0.011 16.72 ± 0.19 0.84 ± 0.03 9.1 ± 0.9 D70 0.946 ± 0.005 18.24 ± 0.09 0.814 ± 0.010 14.2 ± 0.4 0.940 ± 0.009 18.46 ± 0.16 0.82 ± 0.03 12.8 ± 0.4 F71 0.952 ± 0.009 17.44 ± 0.14 0.885 ± 0.014 12.7 ± 1.6 0.936 ± 0.009 17.95 ± 0.13 0.860 ± 0.024 11.0 ± 0.8 T72 0.950 ± 0.010 18.02 ± 0.16 0.829 ± 0.014 13.3 ± 0.6 0.936 ± 0.008 17.80 ± 0.12 0.858 ± 0.019 12.0 ± 0.7 L73 0.965 ± 0.009 16.96 ± 0.15 0.861 ± 0.014 14.8 ± 0.9 0.972 ± 0.009 16.84 ± 0.15 0.858 ± 0.024 13.1 ± 0.8 T74 0.923 ± 0.008 19.32 ± 0.17 0.844 ± 0.014 16.4 ± 0.7 0.912 ± 0.007 17.56 ± 0.12 0.793 ± 0.022 13.9 ± 0.5 I75 0.915 ± 0.006 16.77 ± 0.11 0.852 ± 0.013 13.80 ± 0.3 0.970 ± 0.005 16.25 ± 0.08 0.852 ± 0.017 11.02 ± 0.12 S76 0.864 ± 0.008 19.86 ± 0.19 0.816 ± 0.015 16.2 ± 0.6 0.923 ± 0.007 18.12 ± 0.12 0.807 ± 0.021 12.96 ± 1.08 S77 0.893 ± 0.017 17.6 ± 0.3 0.82 ± 0.03 14.0 ± 1.7 1.128 ± 0.016 14.4 ± 0.4 0.71 ± 0.04 10.6 ± 0.6 L78 0.952 ± 0.008 22.86 ± 0.22 0.843 ± 0.017 11.9 ± 0.8 1.007 ± 0.009 16.90 ± 0.16 0.72 ± 0.03 5.5 ± 1.6 Q79 0.872 ± 0.009 18.49 ± 0.17 0.804 ± 0.018 15.0 ± 1.7 0.902 ± 0.007 17.62 ± 0.12 0.814 ± 0.022 10.5 ± 0.9 A80 0.967 ± 0.005 17.54 ± 0.08 0.834 ± 0.010 12.5 ± 0.3 0.938 ± 0.004 17.6 ± 0.06 0.837 ± 0.015 10.7 ± 0.3 E81 1.003 ± 0.006 17.00 ± 0.09 0.777 ± 0.011 12.1 ± 0.3 0.939 ± 0.003 16.70 ± 0.05 0.811 ± 0.012 11.38 ± 0.1 D82 0.895 ± 0.005 20.44 ± 0.12 0.804 ± 0.010 16.3 ± 0.4 0.953 ± 0.005 19.36 ± 0.09 0.834 ± 0.016 13.39 ± 0.18 V83 1.002 ± 0.004 18.15 ± 0.07 0.804 ± 0.009 14.8 ± 0.5 0.996 ± 0.004 17.67 ± 0.06 0.815 ± 0.015 11.53 ± 0.16 A84 0.919 ± 0.007 17.22 ± 0.12 0.765 ± 0.014 12.4 ± 0.7 0.942 ± 0.005 16.45 ± 0.08 0.777 ± 0.015 10.73 ± 0.19 V85 0.903 ± 0.007 18.77 ± 0.12 0.824 ± 0.013 14.04 ± 0.12 0.937 ± 0.007 17.8 ± 0.1 0.806 ± 0.021 11.8 ± 0.4 Y86 0.910 ± 0.012 18.74 ± 0.23 0.874 ± 0.021 14.8 ± 1.0 0.932 ± 0.019 17.3 ± 0.3 0.81 ± 0.04 Y87 0.960 ± 0.010 18.24 ± 0.16 0.848 ± 0.016 14.1 ± 1.6 0.970 ± 0.013 18.00 ± 0.18 0.89 ± 0.04 11.6 ± 0.4 C88 0.967 ± 0.009 18.77 ± 0.16 0.845 ± 0.014 13.0 ± 0.8 0.972 ± 0.010 17.65 ± 0.15 0.856 ± 0.023 13.8 ± 0.6 Q89 0.953 ± 0.014 19.0 ± 0.3 0.807 ± 0.018 Q90 1.036 ± 0.023 19.0 ± 0.4 0.84 ± 0.03 10.5 ± 2.1 Y91 0.973 ± 0.020 21.5 ± 0.4 0.850 ± 0.023 18 ± 8 Y92 0.955 ± 0.019 21.6 ± 0.5 0.855 ± 0.022 16.8 ± 0.5 0.85 ± 0.06 18.4 ± 1.2 0.76 ± 0.11 S93 0.900 ± 0.011 20.14 ± 0.24 0.838 ± 0.017 16.6 ± 1.0 0.89 ± 0.08 23.1 ± 1.8 0.89 ± 0.18 T94 0.928 ± 0.006 18.87 ± 0.12 0.786 ± 0.011 13.3 ± 0.6 P95 Y96 S97 1.085 ± 0.018 9.45 ± 0.18 0.431 ± 0.020 F98 0.902 ± 0.023 19.5 ± 0.5 0.93 ± 0.03 G99 0.976 ± 0.019 17.9 ± 0.3 0.90 ± 0.03 11.0 ± 1.2 1.021 ± 0.023 18.2 ± 0.3 0.86 ± 0.04 8.84 ± 1.02
Q100 1.020 ± 0.013 21.1 ± 0.3 0.818 ± 0.017 13.0 ± 1.7 1.00 ± 0.03 19.1 ± 0.5 0.82 ± 0.06 10.2 ± 1.9 G101 0.979 ± 0.009 19.40 ± 0.17 0.858 ± 0.014 16.7 ± 1.4 0.989 ± 0.013 19.06 ± 0.22 0.81 ± 0.03 12.4 ± 0.5 T102 0.894 ± 0.014 18.8 ± 0.3 0.866 ± 0.024 11.6 ± 0.9 0.940 ± 0.020 18.1 ± 0.4 0.81 ± 0.05 15 ± 4 K103 0.991 ± 0.007 16.83 ± 0.12 0.835 ± 0.015 14.5 ± 0.6 0.991 ± 0.006 15.76 ± 0.08 0.812 ± 0.018 11.0 ± 0.6 L104 0.902 ± 0.008 19.32 ± 0.17 0.757 ± 0.015 17.5 ± 0.6 0.939 ± 0.007 17.74 ± 0.12 0.79 ± 0.03 8 ± 2 E105 0.954 ± 0.006 17.6 ± 0.1 0.753 ± 0.014 13.6 ± 0.3 0.966 ± 0.005 17.16 ± 0.08 0.737 ± 0.018 12.1 ± 0.4 I106 0.930 ± 0.004 17.78 ± 0.07 0.752 ± 0.010 13.7 ± 0.4 0.948 ± 0.002 16.42 ± 0.04 0.678 ± 0.008 11.22 ± 0.14 K107 1.015 ± 0.005 16.07 ± 0.08 0.664 ± 0.010 13.56 ± 0.4 1.037 ± 0.004 14.85 ± 0.06 0.646 ± 0.014 10.65 ± 0.11 R108 1.000 ± 0.002 11.327 ± 0.021 0.114 ± 0.004 8.3 ± 0.3 0.997 ± 0.001 9.841 ± 0.013 0.077 ± 0.005 6.64 ± 0.20
Table 3.2: continued
70
Table 3.3: Results of model-free analysis of 15
N relaxation data for LEN at 600 MHz 1H
frequency and pH 6.5 and 2.0. Residues that could not be fit to any one of the five
motional models are indicated by ‘*’, and residues for which no relaxation data were
available are indicated by blank spaces. Complementarity determining regions (CDR1, 2,
and 3) are highlighted in gray.
pH 6.5 pH 2.0
Residue S2 S2f e (ps) Rex (s
-1) S2 S2f e (ps) Rex (s
-1)
D1 I2 0.851 ± 0.014 54 ± 9 V3 0.933 ± 0.010 0.91 ± 0.05 6.2 ± 1.8 M4 0.888 ± 0.009 0.73 ± 0.22 0.98 ± 0.04 8.1 ± 1.8 T5 0.895 ± 0.009 1.08 ± 0.22 0.887 ± 0.009 1.34 ± 0.23 Q6 0.941 ± 0.010 0.882 ± 0.025 5.2 ± 0.8 S7 0.964 ± 0.006 0.945 ± 0.017 4.0 ± 0.5 P8 D9 0.945 ± 0.004 0.871 ± 0.002 S10 0.919 ± 0.003 17 ± 3 * L11 0.891 ± 0.006 18 ± 4 0.51 ± 0.15 0.897 ± 0.003 A12 0.796 ± 0.004 0.867 ± 0.003 1260 ± 70 0.786 ± 0.003 0.845 ± 0.003 1200 ± 100 V13 0.865 ± 0.004 52 ± 3 1.18 ± 0.11 0.879 ± 0.005 45 ± 5 0.38 ± 0.11 S14 0.862 ± 0.003 0.934 ± 0.003 620 ± 30 L15 0.930 ± 0.003 0.898 ± 0.004 0.924 ± 0.003 960 ± 160 G16 0.789 ± 0.007 4.14 ± 0.18 0.861 ± 0.006 1.36 ± 0.15 E17 0.978 ± 0.003 0.943 ± 0.002 38 ± 6 R18 * 0.839 ± 0.005 19 ± 3 2.81 ± 0.11 A19 0.844 ± 0.006 0.895 ± 0.005 1160 ± 110 * T20 0.866 ± 0.003 0.41 ± 0.08 0.863 ± 0.002 9 ± 3 I21 * 0.925 ± 0.005 N22 0.955 ± 0.006 0.976 ± 0.007 C23 0.892 ± 0.006 1.94 ± 0.15 0.897 ± 0.007 2.30 ± 0.19 K24 0.940 ± 0.009 26 ± 10 1.15 ± 0.25 0.924 ± 0.013 5.0 ± 0.4 S25 0.929 ± 0.009 1.48 ± 0.23 0.908 ± 0.013 2.9 ± 0.4 S26 0.928 ± 0.008 1.04 ± 0.22 0.915 ± 0.015 1.1 ± 0.4 Q27 * 0.959 ± 0.003 S27a * 0.881 ± 0.007 V27b 0.778 ± 0.012 6.7 ± 0.4 0.930 ± 0.016 1.0 ± 0.4 L27c * 0.932 ± 0.017 Y27d 0.956 ± 0.017 1.2 ± 0.5 0.94 ± 0.03 5.1 ± 1.0 S27e 0.95 ± 0.03 6.2 ± 1.0 0.881 ± 0.027 7.6 ± 1.1 S27f 0.816 ± 0.019 N28 * 0.832 ± 0.007 2.39 ± 0.18 S29 0.866 ± 0.009 0.897 ± 0.007 1200 ± 300 0.860 ± 0.006 16 ± 6 K30 0.803 ± 0.007 13 ± 2 4.87 ± 0.19 N31 0.908 ± 0.011 0.8 ± 0.3 * Y32 0.943 ± 0.019 0.95 ± 0.03 L33 0.984 ± 0.008 0.90 ± 0.06 10 ± 3 A34 * 0.960 ± 0.018 W35 0.914 ± 0.009 0.892 ± 0.015 Y36 0.925 ± 0.010 0.871 ± 0.016 3.7 ± 0.4 Q37 0.890 ± 0.013 2.1 ± 0.4 Q38 0.901 ± 0.011 20 ± 7 1.8 ± 0.3 0.931 ± 0.009 K39 0.842 ± 0.014 2.6 ± 0.4 0.897 ± 0.014 P40 G41 0.80 ± 0.04 0.718 ± 0.011 0.786 ± 0.009 1230 ± 230 Q42 0.910 ± 0.009 23 ± 6 2.06 ± 0.23 0.949 ± 0.011 70 ± 30 P43 P44 K45 0.902 ± 0.018 27 ± 10 5.2 ± 0.5 L46 0.863 ± 0.017 4.5 ± 0.5 L47
Continued
71
I48 0.929 ± 0.006 0.920 ± 0.012 1.4 ± 0.3 Y49 1.000 ± 0.001 4.51 ± 0.01 W50 0.97 ± 0.04 7.6 ± 1.4 A51 * S52 0.831 ± 0.008 1.79 ± 0.21 T53 0.939 ± 0.010 1.01 ± 0.25 R54 0.867 ± 0.010 12 ± 4 2.65 ± 0.25 0.930 ± 0.016 E55 0.964 ± 0.009 1.000 ± 0.025 S56 0.889 ± 0.012 5.3 ± 0.4 G57 0.905 ± 0.017 0.816 ± 0.004 16 ± 3 V58 0.914 ± 0.003 31 ± 4 0.900 ± 0.006 P59 D60 * 0.901 ± 0.004 R61 0.999 ± 0.008 * F62 0.984 ± 0.007 1.18 ± 0.20 0.959 ± 0.005 1.94 ± 0.13 S63 0.930 ± 0.005 0.923 ± 0.005 G64 0.945 ± 0.005 0.953 ± 0.005 S65 0.953 ± 0.005 0.948 ± 0.004 G66 * * S67 * 0.832 ± 0.006 0.37 ± 0.14 G68 0.926 ± 0.009 * T69 * 0.873 ± 0.007 D70 0.938 ± 0.003 0.923 ± 0.009 0.72 ± 0.23 F71 0.919 ± 0.006 0.930 ± 0.005 T72 0.932 ± 0.006 0.904 ± 0.008 0.72 ± 0.19 L73 * 0.913 ± 0.006 T74 0.923 ± 0.008 1.22 ± 0.23 0.880 ± 0.007 0.94 ± 0.18 I75 * * S76 0.851 ± 0.008 3.42 ± 0.24 0.882 ± 0.007 1.65 ± 0.17 S77 0.831 ± 0.016 2.5 ± 0.4 0.723 ± 0.020 0.872 ± 0.019 2000 ± 400 L78 0.929 ± 0.008 5.1 ± 0.3 0.919 ± 0.007 65 ± 13 Q79 0.844 ± 0.009 2.48 ± 0.24 0.890 ± 0.007 0.44 ± 0.18 A80 0.931 ± 0.003 0.903 ± 0.004 0.64 ± 0.10 E81 0.915 ± 0.005 0.936 ± 0.004 790 ± 180 * D82 * 0.940 ± 0.005 1.22 ± 0.13 V83 0.928 ± 0.004 0.954 ± 0.004 1900 ± 400 0.948 ± 0.002 A84 0.877 ± 0.007 19 ± 4 0.56 ± 0.19 0.885 ± 0.003 V85 0.872 ± 0.007 2.29 ± 0.18 0.896 ± 0.007 1.03 ± 0.16 Y86 * 0.908 ± 0.013 Y87 0.943 ± 0.006 0.910 ± 0.012 1.3 ± 0.3 C88 0.962 ± 0.006 0.918 ± 0.009 0.70 ± 0.23 Q89 0.919 ± 0.014 1.7 ± 0.4 Q90 1.000 ± 0.015 Y91 0.920 ± 0.019 4.5 ± 0.6 Y92 0.950 ± 0.019 3.1 ± 0.6 0.89 ± 0.04 S93 0.893 ± 0.011 2.8 ± 0.3 0.88 ± 0.08 6.1 ± 2.4 T94 0.899 ± 0.006 16 ± 4 1.61 ± 0.17 P95 Y96 S97 * F98 * G99 0.942 ± 0.012 0.978 ± 0.013
Q100 0.972 ± 0.012 3.0 ± 0.3 0.994 ± 0.018 G101 0.985 ± 0.006 0.982 ± 0.008 T102 0.894 ± 0.014 1.2 ± 0.4 0.886 ± 0.019 1.8 ± 0.5 K103 * * L104 0.853 ± 0.008 18 ± 4 3.27 ± 0.23 0.896 ± 0.007 1.02 ± 0.17 E105 0.919 ± 0.004 34 ± 6 0.894 ± 0.006 34 ± 6 0.62 ± 0.14 I106 0.874 ± 0.004 23 ± 3 1.39 ± 0.10 0.860 ± 0.002 0.904 ± 0.002 510 ± 40 K107 0.828 ± 0.004 0.913 ± 0.004 1000 ± 50 0.751 ± 0.004 0.874 ± 0.004 1270 ± 60 R108 0.578 ± 0.001 0.828 ± 0.001 641 ± 4 *
Table 3.3: continued
72
CHAPTER 4
4. CORRELATION BETWEEN CONFORMATIONAL FLEXIBILITY AND
THERMODYNAMIC STABILITY OF LEN, SMA AND SMA-LIKE POINT
MUTANTS OF LEN
4.1 Introduction
Light-chain amyloidosis (AL) is the most prevalent systemic amyloidosis in the
United States and results in the deposition of Ig VL amyloid fibrils in certain organs such
as the heart and kidneys. The immunoglobulin light chain VL SMA was the main
component of amyloid deposits isolated from a patient suffering from AL. This protein
differs from LEN at eight aa positions (S29N, K30R, P40L, Q89H, T94H, Y96Q, S97T
and I106L) and was found to aggregate readily in vitro at both physiological and low pH.
In a series of thermodynamic studies, Stevens and co-workers8 (Table 4.1) have probed
the thermodynamic stabilities of each SMA-like mutant and established that only two of
them, P40L and Y96Q were able to aggregate readily at physiological pH. The present
chapter aims to present the sequential backbone assignments of SMA and SMA-like
mutants and probe their slow-motion dynamic properties using constant time CPMG
relaxation dispersion experiment. In particularly, one objective is to obtain a residue-
73
specific map of the dynamic properties of each mutant and establish which mutations
have the most impact on the dimer stability.
Table 4.1: Summary of the VL stability measurements and aggregation properties at
physiological pH and under agitation.8
ΔΔGunif (kcal/mol) Fibril formation
LEN - 0
S29N -1.0 0
T94H -0.7 0
S97T -0.6 0
I106L -0.2 0
K30R 0.1 0
P40L 0.7 +
Q89H 1.0 0
Y96Q 3.2 +
SMA 2.6 +
4.2 Materials and methods
4.2.1 NMR sample preparation
The SMA and LEN plasmids were kind gifts from Prof. Fred J. Stevens (Argonne
National Laboratory). Eight distinct plasmids, each corresponding to an SMA-like
mutation (S29N, K30R, P40L, Q89H, T94H, Y96Q, S97T and I106L) were prepared by
inserting the relevant mutation into the LEN plasmid using the Quikchange II site-
directed mutagenesis kit (Stratagene). All proteins were over-expressed in E. coli C41
(DE3) cells as described previously.6 All
15N as well as
13C,
15N labeled samples for
NMR were prepared by exchanging the buffer of the relevant protein to a commonly
prepared phosphate buffer at pH 6.5 containing 20 mM sodium phosphate and 100 mM
sodium chloride. The final samples were also added with 7% (v/v) D2O and 0.02% (w/v)
74
sodium azide prior to loading them in NMR tubes. Approximately 260 to 300 µL of
protein samples were loaded into regular or thin-walled Shigemi microcells. The
concentration of 15
N labeled SMA was 1.2 mM while that of the 15
N labeled single
mutants were between 1.2 to 1.8 mM. In addition, 13
C, 15
N labeled samples were also
prepared for SMA (1.2 mM) and three mutants of LEN, Q89H, T94H and Y96Q with
final concentrations of 2.0 mM. Finally, two additional 15
N labeled diluted samples were
prepared for S29N and Y96Q mutants at approximately half of their original
concentrations. Unless otherwise mentioned, all NMR samples were in pH 6.5.
4.2.2 Circular dichroism spectroscopy
Near-UV CD spectroscopy was performed at 37°C on a JASCO J-815 CD
spectrometer using rectangular cuvettes of 1 mm and 10 mm path length. For SMA, LEN
at pH 6.5 and pH 2.0 as well as the LEN variants, concentrations of 0.05, 0.1, 0.25, 0.5, 1,
2 and 3 mg/ml were used. The pH of all samples, except LEN at pH 2, was maintained at
6.5. In some cases, where higher concentrations did not result in precipitation of proteins,
spectra were also recorded at > 3 mg/ml; for example, in LEN at pH 6.5 (5.0 and 8.0
mg/ml), S29N (5.0 and 8.6 mg/ml), K30R (5.0 and 10 mg/ml), Q89H (5.0 mg/ml), T94H
(3.8 mg/ml), Y96Q (6.2 mg/ml) and S97T (4.2 mg/ml). Each CD measurement was
performed with the continuous scanning mode with a speed of 100 nm/min and 3 scans
accumulation. For each sample, the data represents an average of a set of three distinct
scans acquired consecutively. The association constant (Ka) for each protein was
calculated by fitting the observed ellipticities at 282.5 nm (Θobs) as a function of the
concentration of the protein using a non-linear regression fitting routine, as has been
performed previously for LEN protein by Souillac et. al.103
75
4.2.3 NMR spectroscopy
NMR experiments were performed at 25°C on either a Bruker DRX-600 or a
DRX-800 spectrometer, each equipped with a cryo-probe having z-axis gradient. The
assignment of resonances corresponding to the backbone 1HN,
15N,
13Cα and
13Cβ spins
of SMA as well as some of the single mutants of LEN was accomplished using one or
more of HNCA, HN(CO)CA, HN(CA)CB pulse schemes.66
Experiments probing the
slow (ms) time scale motion were performed using techniques as described in earlier
work.7 Backbone
15N CPMG relaxation dispersion experiments were performed at 14.1 T
and 18.8 T field strengths, corresponding to 1H frequencies of 600 MHz and 800 MHz,
respectively, employing a 30 ms relaxation delay (Trelax) in all cases. Data was acquired in
an interleaved manner along with the application of compensatory 15
N (and, if necessary,
1H spinlock) pulses in the middle of recycle delay to maintain a constant duty cycle
throughout the experiment. The frequency of 15
N CPMG pulses (νCPMG) was varied
between 33.3 Hz to 1000 Hz while an accurately calibrated 1H spin-lock frequency of
8.5–12 kHz was applied during the relaxation delay as recommended.89
Each relaxation
dataset comprised of 23 interleaved HSQC spectra, corresponding to 18 different CPMG
frequencies, one control spectrum and two pairs of triplicate spectra, one recorded at a
low frequency (100 Hz) and another at the highest frequency (typically, 1 kHz).
4.2.4 Analysis of NMR relaxation data
NMR data was processed using NMRPipe67
while Sparky (Goddard TD, Kneller
DG) was used for visualization and assignment of resonances. Processing of all
76
relaxation data, including error propagation, data fitting and modeling was performed as
described previously,7 with the following exceptions. The dispersion profile for each
residue, obtained by plotting R2,eff against CPMG frequency, νCPMG, at both the field
strengths was first fitted with the program Catia, kindly provided by Prof. Lewis Kay
(University of Toronto), to obtain simulated trajectories. For each simulated trajectory,
the difference in the value of R2,eff at the lowest and the highest frequency was used to
calculate the effective line-broadening due to chemical exchange, i.e. Rex = {R2,eff (νCPMG
= 0 Hz) – R2,eff (νCPMG = 1000 Hz)}. A Rex value of 3 Hz or more for the simulated
trajectory corresponding to the dispersion profile at 600 MHz (1H frequency) was taken
as a cut-off to identify residues with significant chemical exchange. As in the case of
LEN reported previously7, elevated conformational dynamics were limited to three
distinct regions. For each of these three regions, dispersion profiles for SMA and all
mutants of LEN were scanned to obtain the largest set of contiguous residues with
significant chemical exchange. For consistency, global fits was performed on the same
set of residues for all proteins, unless, of course, it was found that conformational
dynamics have quenched altogether, in which case further calculation was not performed
on the particular protein.
4.3 Results
4.3.1 Resonance assignments of SMA and SMA-like point mutants of LEN
Two dimensional 15
N-1H HSQC spectra were recorded for SMA as well as the
eight SMA-like mutants of LEN. As can be observed from Figure 4.1, the spectrum of
77
SMA (magenta) was found to vary significantly from that of LEN (black). While some of
the resonances in the HSQC spectrum of SMA could be assigned based on its similarity
to LEN, a set of HNCA, HN(CO)CA and HN(CA)CB spectra were recorded
unambiguously assign the resonances. Of the 109 non-proline residues in SMA, 80 (~73
%) of the 1H-
15N backbone amide resonances could be unambiguously assigned based on
the triple resonance experiments; the resonances that could not be assigned includes D1,
S27e, S27f, Y32 – Q38, K45 – L47, Y49 – A51, R54, E55 and Y87 – F98. While a few
of the unassigned residues could be putatively assigned to some of the remaining peaks,
an overall analysis of the 15
N-1H HSQC suggest that the unassigned residues undergo
significant line-broadening, many of them were broadened beyond detection, pointing to
the possibility of chemical exchange as was observed for LEN under acidic conditions.7
For the HSQC spectra recorded for eight single mutants of LEN, S29N, K30R, P40L,
S97T and I106L yielded virtually super-imposable spectra (data not shown) with minor
variation in the chemical shift of the mutant residue as well as residues adjoining the
point of mutation. Additionally in case of S97T, residues I2 – Q6 as well as in I106L,
residues L11 – E17 and S76 – Q79 was found to exhibit small to moderate variation in
their resonance frequencies; however, as evident from the x-ray crystal structure of LEN
(1LVE), all these residues are located in spatial proximity to the point of mutation. For
these five mutants, over 95 % of the non-proline backbone amide resonances could be
unambiguously assigned based on the isolated position of the resonances. In contrast to
the mutants mentioned above, the 15
N-1H HSQC spectra for Q89H, T94H and Y96Q
(Figure 4.2, Figure 4.3, Figure 4.4, respectively) was found to be different from either
LEN or SMA in many of the resonances, prohibiting straightforward assignment.
78
Figure 4.1: 800 Mhz 15
N-1H HSQC spectra of LEN (black) and SMA (magenta) at pH 6.5
and 25°C.
79
For these three mutants, a combination of a triple resonance HNCA spectrum along with
the LEN spectrum was used to assign resonances. Residues D1, S27e, S2f, K30, L33 –
W35, K45 – L47, Y49 – S52, S56 and H89 – F98 for Q89H; residues D1, S27f, Y32,
K45, L46, Y49, W50 and H94 – F98 for T94H; and residues D1 – V3, I21, K24, Y27d –
N28, Y32, A34, Y36 – W50, R54, S56, A84, Y86 – F98, Q100, T102 for Y96Q mutant
of LEN could not be unambiguously assigned.
Interestingly and in contrast to SMA, LEN and its other mutants, some of the
residues in two mutants of LEN (i.e. Q89H and T94H) were found to exhibit two
different peaks in their respective 1H-
15N spectrum, indicating to the existence of two
different conformers of these proteins. In Q89H, residues V3 – Q6, L11 – V13, N22 –
K24, I48, D60, F62, G68, T72, L73, L78, E81, A84, V85 – Y87 and L104 exhibit a pair
of resonances, where either the 1H or
15N, or in some case, both
1H and
15N resonance
frequencies were different. In case of T94H, residues M4 – Q6, K24, L27c, S29, L33 –
Q37, Q42, L47, E55, S56, V58, D60, F62, G64, S65 – G66, T72, E81, Y86, F98 – T102
and L104 exhibit a pair of resonances in the 15
N-1H HSQC spectrum. Whether these
conformers are in chemical exchange with each other or with another minor, invisible
species is difficult to conclude from the above studies, but they suggest at least minor
modification of the environment around these residues. In our subsequent studies, the
bigger and predominant resonance was considered. Overall then, the results of the 15
N-1H
HSQC experiments indicated nearly identical spectra for LEN and five of its mutants
(S29N, K30R, P40L, S97T and I106L) whereas SMA and Q89H, T94H and Y96Q
variants of LEN exhibit non-overlapping spectra, indicating to the possibility that there
may be change in the structure.
80
Figure 4.2: 800 MHz 15
N-1H HSQC spectra of Q89H at pH 6.5 and 25°C.
81
Figure 4.3: 800 MHz 15
N-1H HSQC spectra of T94H at pH 6.5 and 25°C.
82
Figure 4.4: 800 MHz 15
N-1H HSQC spectra of Y96Q at pH 6.5 and 25°C.
83
4.3.2 Analysis of secondary and tertiary structure of SMA and LEN mutants
The change in the secondary and tertiary structure for SMA and three
aforementioned LEN mutants were further investigated using the secondary chemical
shifts of the backbone carbon, which is known to be a very sensitive reporter of the
secondary structure. As shown in Figure 4.5, there is an excellent correlation for both the
Cα (Figure 4.5A, R2 = 0.943) and the Cβ (Figure 4.5B, R
2 = 0.955) shifts of SMA with
that of LEN, with almost all the outliers being that of the mutated residues that are
reasonably expected to have significant deviation due to change in the side-chains.
Removing the mutated residue from the correlation result in R2 values of > 0.99 for both
the Cα and Cβ spins which suggests that the secondary structural elements in SMA are
similar to that of LEN. As for the LEN mutants, analysis of secondary shifts of the
backbone Cα spin was at first carried out for the three mutants that exhibited different 1H-
15N spectra, i.e. Q89H (Figure 4.5C), T94H (Figure 4.5D) and Y96Q (Figure 4.5E).
Here too, a very high correlation (R2 > 0.99) was obtained for all of them. Since SMA,
having all the aforementioned three individual mutations of LEN, was found to exhibit a
linear correlation of Cβ shifts, further analysis with Cβ chemical shift was not necessary,
especially since the Cα shifts were already similar to that of LEN.
84
Figure 4.5: A) Plots of 13
Cα shifts for LEN vs SMA at pH 6.5 (R2=0.943). B) Plots of
13Cβ
shifts for LEN vs SMA at pH 6.5 (R2=0.955). Plots of
13Cα shifts for LEN vs C) Q89H, D)
T94H, E) Y96Q at pH 6.5 (R2>0.99 for all three). F) CD spectra for LEN at pH 6.5 and
2.0 and SMA-like mutants corresponding to a concentration of 3 mg/mL.
85
Moreover, since mutations (SMA, Q89H, T94H, Y96Q) that gave rise to non-
identical 15
N-1H HSQC spectra yield similar Cα shifts, it is very unlikely that the other
mutants would have significantly different Cα chemical shift; hence, no further analysis
was necessary to conclude that the secondary structure of LEN, SMA and the SMA-like
mutants of LEN were essentially similar. This however does not necessarily indicate that
the tertiary and quaternary structure of SMA or other variants of LEN would be similar.
To probe the tertiary structure, we performed near-UV CD experiments for SMA
and all the SMA-like mutants of LEN and the spectra corresponding to a concentration of
3 mg/ml are show in Figure 4.5F; additionally, the spectra for LEN at pH 6.5 and pH 2.0
(3 mg/ml concentration for both) were also plotted for comparison. The results of the CD
experiments show that of the 11 spectra recorded, LEN (pH 6.5) and five of its mutants
(S29N, K30R, P40L, S97T and I106L) have essentially the same features with minima at
274 nm and maxima at 278, 288 and 296 nm, as have been observed by Souillac et. Al.103
In comparison, the spectra for SMA, LEN (pH 2.0) and three variants of LEN (Q89H,
T94H and Y96Q) appear different. While the minima at 274 nm in Q89H spectrum
(Figure 4.5F, cyan) significantly deepens, practically eliminating the peak at 278 nm, for
T94H (Figure 4.5F, magenta), the bands at 288 and 296 broaden and lower in intensity.
In comparison, the spectra for Y96Q and LEN at pH 2.0, shown in Figure 4.5F as yellow
and black (dashed) lines, respectively, are similar to each other albeit with reduced
intensity with respect to LEN at pH 6.5. Finally for SMA (Figure 4.5F, black dotted
line), the spectrum matches closely to that of the Y96Q, with reduced intensity of the
spectrum and the peak at 288 nm practically eliminated. While SMA and three mutants
have distinct CD spectra, they probably indicate minor rearrangement of the environment
86
of the aromatic residues and not an unfolded state of these proteins, since unfolding of
LEN would lead to a flat, featureless spectrum devoid of peaks. Overall, the results of 1H,
15N,
13C chemical shift analysis performed on a residue specific basis combined with near
UV circular dichroism spectroscopy clearly indicate that for SMA and all SMA-like
mutants of LEN, the secondary structural elements remain identical and that there may be
minor changes in the tertiary fold in SMA and three LEN-mutants. These results are
consistent with observations by Fink and co-workers where they found that the relative
tertiary structure of SMA remains relatively native-like at pH ranging from 4 to 6, below
which it starts losing most of its tertiary structure and becomes unfolded at pH 2.
The oligomeric state of SMA and LEN mutants under the conditions used in
experimental NMR was additionally verified using circular dichroism (CD) experiments
where the CD trajectory against a varying protein concentration was fitted to extract the
rate of association. Initially, the trajectory for LEN at pH 6.5 and pH 2.0 were fitted, for
which detailed information on the oligomeric state was previously obtained from accurate
measurement of rotational correlation time and showed LEN at both pH to be dimers. For
LEN at pH 6.5 and pH 2.0, the association constant, (Ka) was found to be 2.3 ∙ 105 M
-1
and 1.4 ∙ 105 M
-1, respectively, which is consistent with dimer being the predominant
species of LEN under experimental conditions. For SMA, results of data fitting was
performed with data from five distinct values of protein concentration (data point
corresponding to concentration of 0.05 mg/ml was erroneous and was not included in
regression) which revealed a Ka value of 12.8 ∙ 105 M
-1 suggesting that SMA too remains
a stable dimer under concentrations used for NMR experiments.
87
Figure 4.6: Effect of protein concentration on the monomer/dimer association for LEN
and the SMA-like mutants
88
With exception to Q89H, where reliable fits could not be obtained, association constants
could be extracted for the following mutants, S29N (133 ∙ 105 M
-1), K30R (6.5 ∙ 10
5 M
-1),
P40L (18.5 ∙ 105 M
-1), T94H (2.7 ∙ 10
5 M
-1), Y96Q (3.4 ∙ 10
5 M
-1), S97T (1.0 ∙ 10
5 M
-1)
and I106L (1.8 ∙ 105 M
-1) all of which clearly indicate dimer to be the dominant species at
~1 mM or more concentrations used in this study.
4.3.3 Evaluation of slow timescale conformational dynamics of SMA and SMA-
like mutants of LEN
To quantitatively probe the existence of slow (ms) time scale conformational
dynamics in the backbone of SMA we performed 15
N CPMG relaxation dispersion
experiments on backbone amide nitrogen spin at 14.1 and 18.8 T magnetic field
strengths, corresponding to 1
H frequencies of 600 and 800 MHz, respectively. After
fitting the trajectories of effective transverse relaxation rate, R2,eff against varying CPMG
frequencies for each residue, the contribution of chemical exchange to transverse
relaxation (Rex) was obtained for each observed residue in SMA from the difference in
R2,eff at the lowest and highest CPMG frequencies (Figure 4.7A). For comparison, the
residue specific Rex values of LEN is also shown in Figure 4.7B. Similar relaxation
dispersion measurements were performed for all the 8 SMA-like mutants of LEN and are
shown in Figure 4.7C (S29N) to Figure 4.7J (I106L) in the order of the position of
mutation in the LEN sequence. With the exception of T94H where exchange broadening
seems to have been practically quenched, most mutants as well as SMA exhibit
significant Rex values (> 3 Hz at 600 MHz) at 3 distinct locations.
89
Figure 4.7: Plots of Rex, the estimated chemical exchange contribution to 15
N R2 for A)
SMA, B) LEN and from C) to J) the SMA-like mutants. All plots show the Rex values at pH
6.5 at 14.1 T (hollow circles) and 18.8 T (filled circles). For each protein, the Rex values
at 14.1 T are mapped onto the X-ray structure for one of the monomer subunits of LEN,
with the magnitude indicated by the color bar.
90
The first region (denoted region-1) consists of the N-terminal β-strand A and up to the
beginning of β-strand B including residues from I2 – D9 and is essentially the same for
what was observed for LEN at pH 6.5. In comparison, the second region (region-2) has a
variable span centered on residue I48. This region encompasses the β-strand E and the
complementarity determining region (CDR2) while ends with residue D60 (except in
Y96Q). The beginning of region-2, however, varies with the mutant type. For LEN and
its 5 mutants having similar 15
N-1H HSQC spectra, residues from Q42 exhibit significant
increase in Rex value whereas for Q89H and Y96Q this extends up to Y32 encompassing
the β-strand E. For SMA, though, many of the residues in this region are broadened out
making it difficult to ascertain the extent of the sequence undergoing chemical exchange.
The last region exhibiting elevated ms time scale motion (denoted region-3) extends from
residue C88 up to T102 and includes CDR 3. The choice of regions is dictated by the
consensus obtained from observing the residue specific Rex profiles of SMA, LEN and its
8 mutations, in addition to scanning for contiguous regions of elevated relaxation
dispersion (> 3 Hz at 600 MHz). In S29N (Figure 4.7C), the exchange broadening
appears to be essentially similar to that of LEN with a marginal increase in Rex in region-
2 suggesting that S29N mutation leaves the conformational flexibility of LEN backbone
unperturbed. While the pattern of dynamics in the three regions of K30R (Figure 4.7D)
are similar to LEN, the extent of ms time scale motion is clearly subdued with none of the
residues showing Rex > 15 Hz at 600 MHz. Similar results are also observed in the P40L
mutation (Figure 4.7E) where the extent of residues exhibiting elevated ms time scale
motions and the magnitude of Rex resemble that of LEN very closely. In contrast, the
results for Q89H, T94H and Y96Q shown in Figure 4.7F, G and H, are very different.
91
While Q89H and Y96Q have elevated dynamics compared to LEN, in T94H these
motions are practically quenched (except residues Y27d and L33 having Rex values > 3
Hz). Unlike LEN and most other mutants, where majority of residues outside the three
regions have negligible Rex values, in Q89H and Y96Q, many of the residues in S10 –
Y31 and R61 – Y87 have Rex values > 3 Hz indicating elevated motions in these parts of
the sequence. For both mutants, these residues are non-contiguous and exhibit Rex
magnitudes smaller in comparison to residues in the three dynamic hotspots. In case of
S97T and I106L (Figure 4.7I and J, respectively) the residue specific Rex profile are
similar to that of LEN although the magnitude is significantly stymied for S97T.
Qualitatively speaking, the pattern of the residue specific values of Rex indicate that
amongst all the LEN mutants, T94H exhibit the least ms time scale dynamics followed by
(in increasing order of dynamics) S97T, K30R, I106L, P40L, S29N, Q89H and Y96Q.
The fact that the sequence specific Rex profile of P40L appears to be similar to LEN is
surprising since P40L is able to form amyloid fibrils at physiological pH.8 However, the
Rex value is obtained from the difference in the relaxation dispersion trajectory at the
maximum and minimum CPMG frequencies and is qualitative indictor of chemical
exchange. More quantitative dynamic parameters were be extracted by fitting the
dispersion trajectories for each of the three regions to obtain the rate of exchange (kex)
between the major and minor conformers, the population of the minor conformer (pB) and
residue specific difference in chemical shifts (|Δω|) and show in Table 4.2.
92
Table 4.2 : Summary of 15
N-CPMG relaxation dispersion NMR data analysis for the
three dynamic regions of LEN, SMA and SMA-like mutants of LEN. The signs of the
chemical shift differences were obtained according to the method of Kay and co-
workers.102
LEN¶ S29N K30R P40L Q89H T94H Y96Q S97T I106l SMA
Region 1
kex pB 200±40 s
-1,
4.0±0.4 %
60±15 s-1
,
9.0±1.9 %
60±22 s-1
,
8.0±2.3 %
340±53 s-1
,
3.0±0.5 %
289±18 s-1
,
51±13 % n/a
420±15 s-
1,
17.0±0.4
%
35±19 s-1
,
9.0±4.6 %
64±17 s-1
,
10.0±2.1 %
1100±80 s-
1, n/a
Δω (ppm) Δω (ppm) Δω (ppm) Δω (ppm) Δω (ppm) Δω
(ppm) Δω (ppm) Δω (ppm) Δω (ppm) Δω (ppm)
I2 –a 0.37±0.05
c -0.37±0.06 -0.66±0.03
c -0.26±0.01 – n/a 1.0±0.08
c 0.39±0.04 n/a
V3 -0.99±0.04 -1.00±0.04 -0.88±0.05 -0.93±0.03 -0.41±0.01 – n/a -0.57±0.07 -0.77±0.04 n/a
M4 -0.80±0.03 -0.78±0.04 -0.66±0.05 -0.79±0.03 -0.40±0.01 – 1.09±0.01 -0.41±0.07 -0.71±0.04 n/a
T5 -0.56±0.04c -0.59±0.04 -0.47±0.05 -0.53±0.04
c -0.24±0.01
c –
-0.40±0.01
-0.56±0.06 -0.46±0.04 n/a
Q6 -1.26± 0.04 -1.43±0.06 -1.18±0.06 -1.12±0.04 -0.56±0.01 – -
2.54±0.02 -1.32±0.09 -1.27±0.04 n/a
S7 1.03±0.04 1.09±0.05c -0.94±0.05
c 0.96±0.03 -0.36±0.01 n/a n/a -1.1±0.09 1.04±0.04 n/a
P8 n/ab n/a n/a n/a n/a n/a n/a n/a n/a n/a
D9 -0.18±0.09c ~0 ~0 -0.29±0.06
c ~0 – 0.39±0.01 ~0 ~0 n/a
Region 2
kex pB 630±30 s
-1,
2.3±0.06 %
500±21 s-1
,
3.2±0.1 %
290±24 s-1
,
2.6±0.2 %
620±28 s-1
,
2.2±0.1 %
92±9 s-1
,
10±1 % n/a
260±17 s-
1, 6.8±0.3
%
550±38 s-1
,
1.6±0.1 %
410±22 s-1
,
2.7±0.1 %
850±70 s-1
,
1.8±0.8 %
Δω (ppm) Δω (ppm) Δω (ppm) Δω (ppm) Δω (ppm) Δω
(ppm)
Δω
(ppm) Δω (ppm) Δω (ppm) Δω (ppm)
Y32 – -0.88±0.03 -1.12±0.05 -0.88±0.04 -4.36±0.11 n/a n/a -0.31±0.02 -0.98±0.04 n/a
L33 – -0.86±0.03 -0.55±0.05 -0.84±0.04c n/a – 1.55±0.03
c -0.3±0.02 -0.75±0.04 n/a
A34 – -0.30±0.06 -0.14±0.15 -0.19±0.12c n/a – n/a ~0 -0.17±0.11 n/a
W35 – -0.53±0.04 -0.48±0.05 -0.67±0.04 n/a n/a n/a -0.18±0.03 -0.60±0.04 n/a
Y36 – 0.40±0.06 -0.16±0.12 -0.65±0.04c -2.12±0.07 – n/a ~0 ~0 n/a
Q37 – n/a n/a n/a -1.37±0.04 – n/a n/a n/a n/a
Q38 – -0.75±0.03 -0.64±0.05 -0.77±0.04c -1.28±0.05 – n/a -0.15±0.05 -0.69±0.04 n/a
K39 – -0.43±0.05c -0.42±0.06 -0.67±0.04 -0.77±0.03 – n/a -0.10±0.06 -0.41±0.05 1.15±0.05
c
P40d n/a n/a n/a n/a n/a n/a n/a n/a n/a 0.69±0.06
c
G41 – -0.67±0.05 -1.0±0.05 -0.77±0.04 -0.79±0.05 – n/a -0.44±0.01 0.94±0.06 -1.03±0.04
Continued
93
Q42 1.88±0.04 1.71±0.03c -1.69±0.06 2.05±0.05 -1.94±0.06 – n/a -0.52±0.02 -1.91±0.05
c -1.25±0.05
P43 n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a
P44 n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a
K45 1.99±0.06 1.8±0.04c -1.86±0.08 1.88±0.05
c n/a n/a n/a -0.53±0.02 1.72±0.05
c n/a
L46 2.57±0.06 2.42±0.06c -2.75±0.12 2.31±0.06 n/a n/a n/a -0.63±0.01 2.96±0.09 n/a
L47 n/a n/a 0.58±0.05c 0.51±0.06 n/a – n/a 0.18±0.04
c -0.58±0.04 n/a
I48 -0.32±0.07 -0.32±0.05 -0.25±0.09 -0.4±0.06 -0.47±0.03c – n/a ~0 -0.26±0.07 0.93±0.11
c
LEN S29N K30R P40L Q89H T94H Y96Q S97T I106l SMA
Region 2, continued
Y49 -2.01±0.05 -1.68±0.04 -2.93±0.12 -2.02±0.05 n/a n/a n/a -0.49±0.02 -2.02±0.06 n/a
W50 2.82±0.07 3.25±0.08 -2.06±0.08 3.43±0.10 n/a n/a n/a 0.76±0.01 3.24±0.14 n/a
A51 1.19±0.04 1.00±0.03 1.11±0.05 1.10±0.04 n/a – 0.57±0.03 0.34±0.02 1.12±0.04 n/a
S52 ~0 ~0 ~0 ~0 n/a – 0.46±0.02 ~0 ~0 ~0
T53 -0.48±0.06 -0.44±0.04 -0.34±0.07 -0.61±0.05 -0.63±0.03 – 0.92±0.02 ~0 -0.39±0.05 1.84±0.06c
R54 1.78±0.04c -1.58±0.03
c -2.16±0.08 1.85±0.05 2.85±0.10
c – n/a 0.52±0.01
c 1.85±0.05 n/a
E55 1.21±0.04 -1.10±0.03c -1.1±0.05
c -1.02±0.04
c 2.84±0.10
c – 2.81±0.06 -0.36±0.02
c -1.11±0.04 n/a
S56 -2.18±0.06 -2.03±0.05 -2.46±0.10 -2.20±0.06 n/a – n/a -0.54±0.01 -2.48±0.07 4.21±0.16c
G57 -1.39±0.04 -0.95±0.03 -1.02±0.04 -1.04±0.04 -0.86±0.03 – 1.17±0.02 -0.34±0.02 -0.95±0.03 0.82±0.08c
V58 – -0.40±0.04 -0.31±0.07 -0.40±0.07 -0.38±0.03 – 0.54±0.02 ~0 -0.40±0.05 -0.62±0.06
P59 n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a
D60 n/a -0.56±0.03 -0.43±0.05 -0.62±0.04c -0.61±0.03 – 0.88±0.02 -0.24±0.03 -0.52±0.04 0.78±0.05
c
Region 3
kex pB 250±30 s
-1,
2.3±0.21 %
170±24 s-1
,
3.5±0.4 %
110±27 s-1
,
3.9±0.8 %
530±54 s-1
,
4.4±1.4 %
540±33 s-1
,
4.7±0.2 % n/a n/a
330±60 s-1
,
1.8±0.2 %
310±35 s-1
,
3.5±0.2 %
1390±90 s-
1, n/a
Δω (ppm) Δω (ppm) Δω (ppm) Δω (ppm) Δω (ppm) Δω
(ppm)
Δω
(ppm) Δω (ppm) Δω (ppm) Δω (ppm)
C88 ± -0.42±0.05 -0.36±0.08 -0.14±0.08 -0.86±0.03 – n/a ~0 -0.31±0.05 n/a
Q89d -1.13±0.06
c -0.90±0.05 -0.75±0.06 0.71±0.02 n/a – n/a ~0 -0.72±0.03
c n/a
Q90 1.29±0.07c 1.20±0.05
c 0.95±0.06
c 0.73±0.02 n/a – n/a -0.34±0.02
c -1.05±0.03 n/a
Y91 -0.70±0.06c 0.76±0.05
c -0.51±0.07 -0.56±0.03 n/a n/a n/a -0.22±0.03 -0.68±0.04 n/a
Table 4.2: continued
Continued
94
Y92 -0.84±0.06 -0.78±0.05 -0.53±0.07 -0.52±0.03 n/a – n/a ~0 -0.54±0.04 n/a
S93 1.10±0.06c 1.08±0.05 -0.85±0.06 0.61±0.03 n/a – n/a -0.25±0.03 -0.75±0.03
c n/a
T94d -0.74±0.06 0.53±0.05
c -0.51±0.06 -0.60±0.03 n/a n/a n/a -0.29±0.02 -0.43±0.04 n/a
P95 n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a
Y96d n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a
S97d 2.88±0.12 2.09±0.07
c 1.82±0.07
c n/a n/a n/a n/a n/a n/a n/a
F98 -2.02±0.09 0.94±0.05c 1.66±0.09
c n/a n/a n/a n/a 0.47±0.01
c -1.26±0.04 n/a
G99 1.67±0.07 1.60±0.06 1.74±0.09 -1.0±0.02c -1.51±0.05 – – -0.50±0.01 1.47±0.04 n/a
Q100 -2.92±0.15 -2.4±0.1 -2.37±0.13 -0.71±0.02c -2.11±0.08 – n/a -0.35±0.02 -1.01±0.03 n/a
G101 -0.44±0.07c -0.46±0.05
c -0.52±0.06 -0.27±0.04 -0.67±0.03 – – -0.16±0.04 -0.36±0.05 n/a
T102 – 0.53±0.05c 0.33±0.08 0.39±0.03 0.72±0.03 – n/a ~0 n/a n/a
¶Values taken from previous work
7
aResonance present in spectrum but not included in data fitting due to low Rex value
bResonance has been ignored either due to excessive exchange broadening, spectral overlap, a proline
residue or not applicable for other reason (e.g. Δω cannot be reliably extracted, simulation did not
converge)
cDifference in
15N resonance frequencies between HSQC and HMQC spectra < 0.3 Hz
dResidue will change in case of SMA and appropriate mutant of LEN
For region-1, the mutants exhibit two sets of motions; LEN, P40L, Q89H, Y96Q have
moderate rates of exchange (~200 – 420 Hz) whereas in case of S29N, K30R, S97T and
I106L, the value of kex is reduced several fold between to 35 – 64 Hz range and have 8 –
10% minor species. In contrast, dynamic parameters for region-1 of SMA appears very
distinct with an exchange rate of ~1100 Hz which brings it closer to fast exchange limit
(kex > |ω|). For region-2, the kinetic parameters do not appear to be clustered with the kex
Table 4.2: continued
95
varying from 92 Hz in Q89H up to 850 Hz for SMA while the population of minor
conformer remains small ~ 1 – 2% for most domains, except Q89H (10 %) and Y96Q (7
%). And in region-3 too, the value of kex seem to be well distributed amongst LEN
mutants spreading between 110 Hz for K30R to 530 Hz for Q89H while that for SMA is
~1390 Hz clearly putting it in fast exchange regime.
96
CHAPTER 5
5. CORE OF LEN AMYLOID FIBRILS PROBED BY
HYDROGEN/DEUTERIUM EXCHANGE NMR SPECTROSCOPY
Adapted from S.P. Pondaven, S. Mukherjee, C.P. Jaroniec,
Phys. Chem. Chem. Phys. 2012, to be submitted
5.1 Introduction
Amyloidoses or protein conformational diseases result from the conversion of
normally soluble proteins into non-crystalline supramolecular aggregates called amyloid
fibrils or amyloid plaques.4 These fibrils are characterized by an elongated thread-like
morphology and believed to exhibit ‘cross-β’ architecture with individual peptide strands
arranged perpendicular to the fibril axis.25,26
Light-chain amyloidosis (AL) is one of the
most prevalent systemic diseases of this kind and is characterized by the extracellular
deposition of amyloid fibrils in organs and tissues, most commonly in the kidneys and
heart.5 The LEN protein, originally isolated from the urine of a patient suffering from
multiple myeloma, is the 114-residue Ig kIV VL variable domain of an immunoglobulin
light chain. Only a relatively small fraction of human Ig VL fragments are amyloidogenic
in vivo and many studies have been conducted on the specific features of the Ig VL
domains which lead to certain sequences being amyloidogenic and others benign.115,116
97
LEN fibrils were only observed in vitro under destabilizing conditions, low pH or 1-2 M
guanidinium hydrochloride.8 Previous studies of LEN aggregation carried out using
biophysical techniques by Fink and co-workers showed that the dimer dissociation is a
critical step in initiating protein aggregation.61,62,103
Ultimately, an improved
understanding of amyloid diseases at a molecular level will require high resolution
methods like NMR spectroscopy to characterize fibrils and their intermediates.
Hydrogen-Deuterium exchange combined with high resolution NMR
spectroscopy can provide information on protein stability, dynamics, folding and
association at a residue level.38,117,118
This method has been widely applied not only to the
native structures but also to various conformational states of proteins, such as chemically
or physically denatured states and kinetic folding intermediates.119,120
Quenched
hydrogen experiments are used when direct measurements of hydrogen exchange are
impossible due to the insoluble form of proteins or fibrils that is undetectable by liquid-
state NMR spectroscopy. To be applicable to amyloid fibrils, quench hydrogen
experiments need to satisfy two requirements. First, the amyloid fibrils need to be
converted to a soluble form of the protein amenable to NMR. Second, hydrogen
exchange must be sufficiently slow under the conditions of NMR detection to prevent
loss of hydrogen trapped into the fibril core. DMSO/D2O mixtures containing more than
90% of DMSO have been shown to efficiently solubilize amyloid fibrils and
subsequently reduce hydrogen exchange with the solvent during NMR detection.9,121
Indeed, the intrinsic exchange rates in unfolded model peptides are shifted to a minimum
near pD* 5.0 in 90% DMSO / 10% D2O and are reduced 100-fold as compared to those
in pure D2O.122
98
This chapter presents the quenched hydrogen exchange measurements for the
LEN amyloid fibril. The main experimental steps are summarized in Figure 5.1. After
synthesizing 15
N-labeled fibrils under vigorous agitation and low pH, the H/D exchange
experiment is performed by incubating the fibrils in D2O buffer at pD* 7.0 and 25°C.
After six different times of incubation ranging from 10 minutes to 7 days, the fibrils are
placed in liquid nitrogen to quench the exchange and subsequently dissolved in DMSO
during the NMR acquisition. The comparison of the peak intensities for each residue to
the non-deuterated reference can give us information about the LEN amyloid fibril
stability and structure.
5.2 Materials and methods
5.2.1 Protein expression and purification
The plasmid encoding for LEN116
was kindly provided by Dr. Fred J. Stevens
(Argonne National Laboratory). Uniformly 15
N- and 13
C, 15
N-labeled Len samples were
prepared according to published procedures.6 Briefly, electro-competent Escherichia coli
C41 (DE3) cells transformed with LEN plasmid were grown at 30°C and 110 rpm in a
modified M9 minimal medium65
containing 100 μg/mL carbenicillin, 1 g/L 15
NH4Cl, 3
g/L 13
C-D-glucose (Cambridge Isotope Laboratories) and 0.5 g/L 13
C, 15
N-renriched
Isogro growth medium (Isotec/Sigma-Aldrich). Protein expression was induced at OD600
~ 0.7 by the addition of isopropyl-β-D-thiogalactoside to a final concentration of 1 mM.
The periplasmic fraction which contained LEN was isolated as described
previously.116
The purification was then carried out with two ion-exchange columns
Hitrap Q XL and SP XL (GE Healthcare) and a final gel filtration with a HiLoad 16/60
99
Figure 5.1: Overview of the H/D exchange experiment.
100
Superdex 75 column (GE Healthcare). Protein identity and purity were confirmed by
SDS-PAGE and MALDI-TOF mass spectrometry. Typical yields of isotopically
enriched LEN (determined from absorbance at 280 nm with an extinction coefficient of
24,535 M-1
cm-1
) were ~ 10-12 mg per 1L of cell culture.
5.2.2 Fibril formation and H/D exchange
Amyloid fibrils were formed in 50mL Nalgene tubes containing 4mL aliquots of a
purified 1 mg/mL 15
N-labeled Len stock solution (10 mM HCl, 100 mM NaCl, pH 2.0).
Fibril samples were incubated in a thermo-shaker for 48 hours in acidic pH with 2 μL of
seed fibrils at 37°C and 400 rpm. The resulting solutions were then centrifuged at 15,000
g and 4°C for 5 min and the OD280 of the supernatant was tested for presence of soluble
LEN and discarded.
30 mL of deuterated buffer (10 mM DCl, 100 mM NaCl) at pD* 2.0 and 4°C was
added to each tube and the fibril pellets were resuspended to remove any residual water.
Resulting solutions were centrifuged down at 15,000 g and 4°C for 5 min and the
supernatant discarded. The fibrils were resuspended in 30 mL of deuterated H/D
exchange buffer (20 mM sodium phosphate, 100 mM NaCl) at pD* 7.0 and 25°C for 10
min, 1 hour, 9 hours, 1 day, 3 days or 7 days. After each period of incubation, the isotopic
exchange was quenched by centrifuging the sample at 15,000 g and 4°C for 5 min.
Supernatants were subsequently discarded and fibril pellets frozen in liquid nitrogen and
lyophilized for 24 hours. Samples were stored at -80°C until NMR experiments.
For the reference sample, the pellet was directly frozen in liquid nitrogen and
lyophilized for 24 hours after the fibril formation. To ensure a good reproducibility,
101
amyloid fibrils were left in the same Nalgene tube throughout the experiment until
DMSO dissolution and NMR acquisition.
5.2.3 ThT fluorescence assay
The formation of LEN amyloid fibrils was monitored using a fluorescence assay
based on the enhanced fluorescence properties of the dye Thioflavin T (ThT) when bind
to amyloid fibrils.62
The kinetic of fibril formation was determined using four samples
with different protein concentrations of 3 mg/mL, 1 mg/mL, 0.5 mg/mL and 0.25 mg/mL.
All samples were kept in 20 mM HCl, 100 mM NaCl and pH 2.0 and incubated in a
thermo-mixer at 37°C and 1400 rpm. After a time t of incubation, aliquots of 1.67 μL, 5
μL, 10 μL and 20 μL, were respectively taken and diluted in eppendorfs containing 500
μL of sodium phosphate buffer at pH 2.0 to a final protein concentration of 0.01 mg/mL.
5 μL of a freshly made 1 mM ThT was then added to each solution to a final
concentration of 10 μM. ThT fluorescence measurements were acquired with a Perkin-
Elmer Spectrometer LS 50 B at 37°C using an excitation wavelength of 450 nm and
recording the emission spectra from 460 nm to 560 nm.
5.2.4 Electron microscopy experiments
LEN amyloid fibrils were probed by Transmission Electron Microscopy (TEM).
A seeded 1 mg/mL sample, incubated for 48 hours in acidic pH at 37°C and 400 rpm was
used for the experiment. 5 μL aliquot of the fibril solution was applied to a 200 mesh Cu.
Formvar-Carbon grid (Canemco-Marivac) for a 2 minutes adsorption period. Excess
sample was blotted with filter paper and the grid was placed on 5 μL of freshly made 2%
102
(w/v) uranyl acetate for another 2 minutes. Excess staining solution was removed and the
grid was air-dried for 5 minutes. Electron micrographs were collected using a FEI Tecnai
G2 Spirit microscope with a voltage accelerator of 80 kV and a magnification of 120,000.
5.2.5 NMR spectroscopy
5.2.5.1 Assignments in DMSO
A uniformly 13
C, 15
N-labeled sample at a concentration of 1.5 mM was used for
the assignment of the protein in a DMSO buffer containing 95% (v/v) d6-dimethyl
sulfoxide, 4.5% (v/v) H2O, 0.5% (v/v) d2-dichloroacetic acid at pH 5.0 in a total volume
of 300 μL in a Shigemi microcell. Sequential backbone 1H
N,
15N,
13CO,
13Cα and
13Cβ
assignments were obtained using a suite of triple-resonance 3D HNCO, HNCA and
HN(CA)CB experiments based on the pulse schemes of Kay and co-workers.66
5.2.5.2 Fibril H/D exchange
Reference or H/D exchange samples containing 4 mg of lyophilized 15
N-labeled
LEN amyloid fibril were dissolved in 95% (v/v) d6-dimethyl sulfoxide, 4.5% (v/v) D2O
and 0.5% (v/v) d2-dichloroacetic acid, pD* 5.0. After each experiment, the pD* was
measured directly in the NMR tube using a specialized electrode (Hamilton, Reno,, NV)
and was found to be 5.0 ± 0.1 for each sample. After dissolution with DMSO-based
buffer, each sample was loaded into a Shigemi microcell to a volume of 280 μL and a
series of 60 1H-
15N Heteronuclear Single Quantum Correlation
(HSQC) spectra were
recorded over 8 hours. Each HSQC was acquired with 170 points in the F1 dimension,
1280 points in the direct dimension with a dwell time of 52.0 μs, 2 scans and an
103
experimental time of 8 minutes. The dead times for dissolution and loading of the
lyophilized samples as well as NMR parameters adjustment ranged from 15 to 20
minutes. In addition, a 1D 1H spectrum was acquired for each sample to assess the
reproducibility of the protein concentration.
All the NMR spectra were recorded at 25°C on a Bruker DRX-800 MHz
spectrometer equipped with a QXI cryogenic probe with z-axis gradients. Field frequency
lock was referenced to the signal from d6-DMSO. NMR spectra were processed using
NMRPipe67
and analyzed using NMRDraw.67
1H chemical shifts were referenced to DSS,
and 13
C and 15
N shifts were referenced indirectly.2
5.2.5.3 Native State H/D exchange
Two samples containing 280 µL of native state 15
N-LEN at 0.8 mM were
prepared in 20 mM Sodium Phosphate in H2O at pH 7.0. The first one was directly
loaded into a Shigemi microcell and a 15
N-1H HSQC was recorded as a reference. For the
second sample, the buffer was exchanged twice with 15 mL of 20 mM Sodium Phosphate
in D2O at pD* 7.0 and the volume was subsequently reduced to 280 µL by centrifugation.
After 3 hours of dead time, the first 15
N-1H HSQC was recorded for the H/D exchange
sample. A second 15
N-1H HSQC was recorded after 7 days of incubation in the Shigemi
tube kept at 25°C. Two 1D 1H NMR spectra of the methyl region were also acquired for
the reference and H/D exchange sample to correct for any differences in concentration.
For the Native State H/D exchange, all NMR experiments were recorded on a
Bruker DMX-600 spectrometer equipped with a room temperature TXI probe with triple
axis gradients. Each 15
N-1H HSQC was recorded with 240 points in the F1 dimension
with a dwell time of 52.0 μs, 8 scans and an experimental time of 40 minutes.
104
5.3 Results
5.3.1 Fibril formation monitored by Thioflavin T assay
As described in previous studies,63
LEN was shown to be a stable dimer under
physiological pH but can be converted to the amyloid state under destabilizing
conditions (e.g., in the presence of denaturant, (1-2 M guanidinium hydrochloride) or at
acidic pH, (pH 2.0) with strong agitation. In our study, a ThT fluorescence assay was
used to monitor the kinetics of LEN fibril formation at four different concentrations
(0.25, 0.5, 1 and 3 mg/mL). Figure 5.2A shows the initial absorbance spectra with only
ThT present (-) fibrils and the saturation absorbance profile after 48 hours of growth (+)
fibrils for the 1mg/mL sample. According to this assay, the fibrils-ThT fluorescence
showed a peak at 482 nm with an increasing intensity over time as more and more fibrils
are formed. Figure 5.2B summarizes the normalized kinetic results at 482 nm for
selected concentrations. For the 1 mg/mL concentration used in the H/D exchange
experiment, amyloid fibrils started to form after 20 hours of incubation and reached
saturation after 40 hours. Growth saturation was reached after only 20 hours for the 0.25
mg/mL and 0.5 mg/mL and 70 hours for the 3 mg/mL sample, confirming the inverse
concentration-dependent behavior of the LEN amyloid fibril formation established by
Fink and co-workers.103
5.3.2 Transmission electron microscopy
Prior to the H/D exchange experiment, fibrils from a 1 mg/mL 15
N-enriched LEN
sample were analyzed by Transmission Electronic Microscopy (TEM). Figure 5.3 shows
the TEM picture of this test sample, recorded after an incubation of 48 hours in pH 2.0
105
buffer, 37ºC and 400 rpm. Examination of the fibrillar precipitate by TEM revealed the
“needle-like” morphology characteristic of the LEN amyloid fibrils. The precipitate
contained mostly large aggregates of fibrils and individual, well isolated fibrils were
rarely seen. The electron microscopy images confirmed the amyloidogenic structures of
the fibrils that we used during the H/D exchange experiment and that their morphologies
were consistent with previous literature studies on the LEN amyloid fibrils.8
5.3.3 Assignment of the LEN backbone in DMSO
Figure 5.4 shows the assigned 2D 15
N-1H HSQC spectrum of the LEN protein
dissolved in 95% d6-DMSO, 4.5% H2O, 0.5% d2-dichloroacetic acid, pD* 5.0 recorded at
25°C and 800 MHz 1H frequency. The backbone amide
1H and
15N chemical shifts were
obtained for 97 of 108 non-proline residues. All backbone resonances were located in a
limited range of 1HN chemical shifts, from 7.6 to 8.5 ppm, suggesting that the protein was
completely denatured. Among the 11 missing assignments, ten residues (N22, C23, K24,
Y86, Y87, C88, Q89, Q90, Y91 and Y92) are located nearby the disulfide bridge (C23-
C88). The motion restriction induced by the disulfide bond probably leads to
conformational fluctuations on a millisecond time scale that would result in signal line
broadening. In addition, residue D1 could not be assigned. Although residues R18, Q37,
V58 and F98 could be assigned, they were found to be heavily overlapped with each
other and therefore could not be included in the H/D exchange calculations. Residues S76
and S77 were also fully overlapped but were including in the calculation and assumed to
have similar exchange profile due to their location. Altogether, 93 residues are selected
for the H/D exchange experiment.
106
Figure 5.2: (A) ThT fluorescence spectra of LEN amyloid fibril formation at t=0 and
after 48 hours of incubation for a protein concentration of 1 mg/mL. (B) Normalized
kinetic of fibril formation at 482 nm for selected protein concentrations.
107
Figure 5.3: Transmission Electron Microscopy (TEM) image of LEN amyloid fibrils after
an incubation of 48 hours at pH 2.0, 400 rpm and 37°C. The scale bar corresponds to
100 nm.
108
The protonated reference (in DMSO-H2O buffer) shown in Figure 5.4 is solely
used for assigning the 1H-
15N backbone amide resonances whereas a second reference
was made and dissolved with a DMSO-D2O buffer and used for the H/D exchange
calculations. Figure 5.5A shows the first HSQC corresponding to the experimental
reference, recorded after a dead time of 15 min. As the protein was unfolding in DMSO-
D2O buffer, the remaining hydrogens trapped in the fibril core started to exchange with
the deuterium atoms resulting in a slow decrease in peak intensity for many residues.
Figure 5.5B shows the first 2D 15
N-1H HSQC spectrum after 7days of H/D
exchange, dissolved in 95% d6-DMSO, 4.5% D2O, 0.5% d2-dichloroacetic acid, pD* 5.0.
By comparison with the deuterated reference, qualitative information about the fibril
structure can be estimated at a residue specific level. The intensities of residues S7, T69,
T94 and I106 were similar in the reference and 7 days exchange spectra, suggesting
protection from the solvent in the fibril structure whereas residues D9, S25, N28, Q38
and L46 showed a sharp decrease in intensity consistent with a solvent exposed part of
the fibril.
109
Figure 5.4: 800 MHz 15
N-1H HSQC spectrum of unfolded LEN in DMSO buffer showing
the backbone 1HN assignment of LEN dissolved in 95% d6-DMSO, 4.5% H2O, 0.5% d2-
dichloroacetic acid at pH 5.0 and 25°C.
110
Figure 5.5: 800 MHz 15
N-1H HSQC spectra of unfolded LEN amyloid fibrils: (A)
reference spectrum that were not subject to H/D exchange dissolved in 95% d6-DMSO,
4.5% D2O, 0.5% d2-dichloroacetic acid, at pD* 5.0. (B) Exchanged spectrum of LEN
amyloid fibrils for 7 days at pD* 7.0 and 25°C and dissolved in 95% d6-DMSO, 4.5%
D2O, 0.5% d2-dichloroacetic acid, at pD* 5.0.
111
5.3.4 H/D exchange on the LEN native state
To compare the H/D data of the fibrillar state, we have repeated the same H/D
exchange on the soluble dimeric native state at 25ºC and pD* 7.0. The full backbone
amide 1H and
15N chemical shifts of the LEN protein in H2O were described in a previous
publication6 and were obtained for 106 of the 108 non-proline residues (D1 and Y96
could not be assigned). Among the 106 residues available for the H/D exchange, L27c,
N31, V58, G68, E81 and F98 were located in unresolved clusters and were not used for
the calculations; four other residues, S27f, Q37, L47 and S77 were missing in the
Reference sample due to a lower concentration as compared to the assignment sample
and were also excluded. Altogether, 96 residues were selected for the H/D exchange
calculations of the LEN Native State. As expected, the native state structure was found to
be much less rigid and more solvent exposed than the fibril and 33 residues out of 90
were fully exchanged within the experimental dead time of three hours (data not shown).
Figure 5.6 shows the 15
N-1H HSQC spectra of the LEN protein acquired after 7 days of
incubation in 20 mM Sodium Phosphate in D2O at pD* 7.0. The 31 residues that
remained after 7 days of exchange were all located in β-sheets (from aa A19-S25, aa L33-
Q38, aa K45-Y49, aa F71-I75 and aa A84-Y92) where hydrogen bonds prevented the
amide 1H from exchanging with the solvent’s deuterons. By comparing the native state
data with the amyloid fibril data, information on the structure change during fibril
formation can be obtained.
112
Figure 5.6: 600 MHz 15
N-1H HSQC spectrum of the LEN protein after 7 days of
incubation in 20 mM Sodium Phosphate in D2O at pD* 7.0 and 25°C.
113
5.3.5 Post-trap exchange during NMR acquisition
To obtain reliable and accurate data for the protection level of each residue within
the fibril structure, the post-trap exchange in DMSO-D2O in the unfolded LEN structure
was followed through a series of 15
N-1H HSQC spectra recorded over 8 hours. As shown
in Figure 5.7, the post dissolution decays of the amide proton intensities were fitted to an
exponential function allowing, by extrapolation at time t=0, the calculation of the residue
intensities in the fibrillar state, i-e before dissolution. After extrapolation, intensities were
later corrected based on the differences in concentrations between the samples and the
reference. In all cases, a concentration correction of less than 10% was applied between
the six samples and the reference. The acquisition of multiple HSQCs had the other
advantage to remove the dependency of the peak intensities upon the various dead times
ranging from 15 to 20 minutes. Similarly, residue intensities of the protonated reference
in Figure 5.4 can be re-calculated by extrapolating at time zero the intensities
corresponding to the deuterated reference showed in Figure 5.5A.
5.3.6 H/D exchange on the LEN fibril
Figure 5.8 presents the H/D exchange comparison after 7 days of exchange
between the LEN amyloid fibril and the Native State. For each structure, the intensities
were normalized to the reference and plotted against the residue number. As mentioned
earlier, in the Native State, a set of ~30 residues, all located in the most highly structured
β-sheet region did not fully exchange within a period of 7 days (Figure 5.8A).
114
Figure 5.7: Relative Intensity decay curves for residue T5 obtained from multiple 1H-
15N
HSQCs acquired over 8 hours. The curves show the intensity decay for the reference
sample and 4 selected H/D exchange time points as a result of Hydrogen post-dissolution
exchange with the surrounding Deuterons. An extrapolation at t=0 using an exponential
fit enabled the calculation of the relative peak intensity for each time of exchange before
dissolution in DMSO.
115
Figure 5.8: Relative peak intensities in the 7 days H/D exchanged spectrum normalized to the reference and plotted against the residue
number for (A) LEN Amyloid fibril, (B) LEN Native State. Color coded three-dimensional structures of the LEN monomer mapping
the protection level of the (C) Amyloid fibril, (D) Native State. The thresholds for the degree of protection are blue I7days/Iref > 0.8,
green 0.8 > I7days/Iref > 0.5 and red I7days/Iref < 0.5. The residues that were not assigned or located in clusters are colored in grey.
116
In contrast, for the amyloid state, nearly half of the assigned residues showed
significant (> 80%) intensities relative to the reference sample for the 7 day incubation
time (Figure 5.8B). The most protected residues, which presumably make up the amyloid
core region of LEN, were found to be located in β-strands A and B (aa ~ 2-15),
complementary determining region CDR1 and strand D (aa ~27a-40) and a region
spanning aa ~70-105, encompassing strands G, H and I as well as CDR3 loop. On the
other hand, several residues, especially in the FR2 and CDR2 regions (aa ~ 36-56)
showed significant decrease in intensity suggesting that this part the fibril was more
solvent exposed.
The H/D exchange results for the 7 day incubation time were plotted on the 3D
structure of the LEN monomer for the amyloid fibril (Figure 5.8C) and Native State
(Figure 5.8D). The thresholds for the degree of protection were chosen as follow: blue
I7days/Iref > 0.8 for the protected residue that make up the amyloid core, green 0.8 >
I7days/Iref > 0.5 for the residues that displayed intermediate exchange and red I7days/Iref <
0.5 for the residues that showed significant decrease in intensity.
5.4 Discussion
The LEN protein fold consists of two anti-parallel β-sheets comprising the
framework region, and three hypervariable CDR loops located between strands C and D
(CDR1), E and F (CDR2) and H and I (CDR3). In the native state H/D exchange, data
were in good agreement with the crystal structure of the LEN protein and showed
protection from solvent for most amide protons located in β-strands. Protection from
solvent in the native structure was likely due to hydrogen bonds between the anti-parallel
117
strands and shielding of hydrophobic β-sheet residues. Surprisingly, the relative
intensities for the protected amide protons relative to the reference did not exceed 0.7 in
the native state H/D exchange even after concentration correction. A fast dissociation re-
formation of the LEN dimer or a partial unfolding re-folding of the protein upon mixing a
small amount of protein in a large excess of deuterated buffer could be responsible this
early, fast exchange phenomenon.
The amyloid fibril H/D exchange data suggested a highly ordered structure with
only one solvent exposed region. The framework regions (FR1 and FR4) forming the N
and C terminals of the protein appeared solvent exposed in the Native State but were
solvent protected in the amyloid fibril. Interestingly, the highly solvent exposed CDR1
loop in the Native State structure turned protected in the fibril, suggesting that this loop
may have become part of the β-sheet network of the amyloid core. Taken together, most
of amyloid structure was found to be solvent protected except the framework FR2 region
(from residue Q38 to Y49) that showed a sharp decrease in peak intensities after a week
of H/D exchange. The CDR2 region (from residue W50 to residue S56) and part of the
framework FR3 region (from residue G57 to residue D70) also appeared more solvent
exposed.
In order to obtain more quantitative data, the kinetic of the LEN fibril H/D
exchange was calculated for 6 different times of incubation in D2O: 10 min, 1 hour, 9
hours, 24 hours, 3 days and 7 days (Figure 5.9). Kinetic parameters were extracted by
fitting the data points with an exponential function:
(5.1)
118
Figure 5.9: Relative Intensity versus incubation time for selected residues during the
amyloid fibrils H/D exchange in D2O at pD* 7.0. Intensities are normalized to the
reference. (Left) No H/D exchange. (Center) Fast amide exchange rate; (Right) Slow
amide exchange rate
119
where t is the time of incubation in D2O, I∞ is the residue intensity at time infinity and kex
the observed exchange rate constant. Three different H/D exchange profiles were
observed: residues that appeared solvent protected and make-up the fibril core (Figure
5.9, Left), residues that exchanged rapidly with an average kex of ~ 10-2
min-1
(Figure
5.9, Center) and the ones that displayed a much slower exchange with an average kex of
~ 10-3
min-1
(Figure 5.9, Right). The latter exchange profile was only observed for 5
residues.
A number of residues (~aa 36-70) were characterized by a relatively rapid
decrease in intensity followed by a very slow phase represented by a plateau in the time
range of the experiment (Figure 5.9, Center and insets). The biphasic behavior of this
exchange kinetic in amyloid fibrils could be explained by taking into account that fibrils
are large molecular assemblies composed of huge numbers of protein molecules and each
protein molecule could display very heterogeneous structural properties like shape, length
and thickness as shown in the EM experiment (Figure 5.3).9,123
In order to calculate the protection factor (Table 5.1), the intrinsic exchange rates
kint, for unprotected amides at pH 7.0 and 25°C were obtained using the program
SPHERE (http://dino.fold.fccc.edu:8080/sphere.html).124,125
Typical values for kint were
on the magnitude of 10 - 100 min-1
. The ratio between the observed exchange rate kex and
the intrinsic exchange rate kint, which is expected it that site is freely exposed, is called
the protection factor and is defined by:
(5.2)
120
Table 5.1: H/D exchange kinetic results for the LEN amyloid fibril. The exchange
parameters were extracted from the fit I(t)=I∞+(1-I∞)exp(-kex.t) where I∞ corresponds
to the residue intensity at t=∞ and kex the experimental exchange rate constant. The ratio
between the experimental exchange rate (kex) and the intrinsic exchange rate (kint) is
defined as the protection factor. The intrinsic exchange rates (kint) were estimated using
the SPHERE software.124,125
(See text for details)
Residues I∞ kex kint Protection factor
I2* 1.02 ± 0.04 -7x10-6
± 8.00x10-6
2.11x103 3.02x10
8
V3 0.89 ± 0.02 2.16x10-2
± 1.95x10-2
4.80x101 2.22x10
3
M4 0.89 ± 0.01 2.89x10-2
± 1.74x10-2
2.91x102 1.01x10
4
T5 0.35 ± 0.05 2.15x10-3
± 6.96x10-4
4.53x102 2.11x10
5
Q6 0.65 ± 0.02 8.08x10-3
± 3.56x10-3
7.46x102 9.23x10
4
S7 0.91 ± 0.01 2.98x10-2
± 2.43x10-2
1.52x103 5.11x10
4
D9 0.32 ± 0.01 3.26x10-1
± 1.34x10-1
2.98x102 9.15x10
2
S10 0.95 ± 0.03 3.46x10-3
± 1.11x10-2
6.38x102 1.84x10
5
L11 0.85 ± 0.01 2.81x10-2
± 1.12x10-2
2.17x102 7.71x10
3
A12 0.93 ± 0.05 3.34x10-3
± 1.39x10-2
2.53x102 7.58x10
4
V13 0.92 ± 0.02 1.68x10-2
± 2.07x10-2
8.15x101 4.84x10
3
S14 0.85 ± 0.01 3.45x10-2
± 1.11x10-2
6.96x102 2.02x10
4
L15 0.87 ± 0.01 2.65x10-2
± 1.43x10-2
2.17x102 8.17x10
3
G16 0.78 ± 0.02 1.76x10-2
± 8.58x10-3
4.68x102 2.65x10
4
E17 0.77 ± 0.02 2.30x10-2
± 9.68x10-3
4.73x102 2.05x10
4
A19 0.43 ± 0.02 6.59x10-2
± 1.54x10-2
6.81x102 1.03x10
4
T20 0.87 ± 0.01 2.15x10-2
± 1.30x10-2
3.52x102 1.64x10
4
I21 0.94 ± 0.02 2.02x10-2
± 3.50x10-2
1.21x102 5.99x10
3
S25 0.63 ± 0.04 1.27x10-2
± 6.91x10-3
1.27x103 9.98x10
4
S26 0.48 ± 0.04 6.76x10-4
± 1.59x10-4
1.92x103 2.83x10
6
Q27 0.62 ± 0.04 4.52x10-4
± 1.41x10-4
9.38x102 2.08x10
6
S27a 0.93 ± 0.02 2.22x10-2
± 2.99x10-2
1.52x103 6.85x10
4
V27b 0.93 ± 0.02 1.49x10-2
± 2.04x10-2
1.62x102 1.09x10
4
L27c 1.06 ± 0.02 1.12x10-3
± 1.41x10-3
7.88x101 7.07x10
4
Y27d 0.96 ± 0.02 1.05x10-2
± 3.91x10-2
1.36x102 1.29x10
4
S27e 0.99 ± 0.04 2.84x10-3
± 8.25x10-2
1.08x103 3.80x10
5
S27f 0.97 ± 0.02 2.18x10-2
± 1.06x10-1
1.92x103 8.78x10
4
N28 0.35 ± 0.01 2.65x10-2
± 2.67x10-3
2.53x103 9.52x10
4
S29 0.63 ± 0.04 1.08x10-3
± 4.57x10-4
2.01x103 1.86x10
6
K30* 1.02 ± 0.02 2.10x10-5
± 6.00x10-6
7.45x102 3.55x10
7
N31 0.99 ± 0.02 1.76x10-2
± 1.28x10-1
1.67x103 9.48x10
4
Y32 0.89 ± 0.02 5.16x10-3
± 6.91x10-3
4.60x102 8.91x10
4
L33 0.92 ± 0.02 1.68x10-2
± 2.77x10-2
1.22x102 7.26x10
3
A34 0.89 ± 0.02 2.81x10-2
± 2.45x10-2
2.53x102 8.99x10
3
W35 0.83 ± 0.01 2.43x10-2
± 7.56x10-3
1.60x102 6.57x10
3
Y36 0.83 ± 0.02 1.18x10-2
± 8.74x10-3
1.71x102 1.45x10
4
Q38 0.42 ± 0.01 8.70x10-2
± 1.30x10-2
7.46x102 8.58x10
3
K39 0.51 ± 0.01 7.97x10-2
± 1.10x10-2
5.93x102 7.44x10
3
Continued
121
G41 0.71 ± 0.04 1.16x10-2
± 8.23x10-3
4.37x102 3.77x10
4
Q42 0.66 ± 0.02 8.21x10-2
± 3.24x10-2
6.96x102 8.48x10
3
K45 0.53 ± 0.02 5.83x10-2
± 1.57x10-2
2.15x102 3.69x10
3
L46 0.47 ± 0.02 6.53x10-2
± 1.53x10-2
1.43x102 2.20x10
3
L47 0.40 ± 0.03 7.95x10-2
± 2.50x10-2
6.71x101 8.44x10
2
I48 0.37 ± 0.01 4.30x10-2
± 5.19x10-3
4.71x101 1.10x10
3
Y49 0.35 ± 0.02 1.49x10-1
± 3.47x10-2
1.30x102 8.73x10
2
W50 0.40 ± 0.01 1.06x10-1
± 1.76x10-2
1.79x102 1.69x10
3
A51 0.82 ± 0.02 1.60x10-2
± 1.03x10-2
3.18x102 1.99x10
4
S52 0.58 ± 0.01 6.66x10-2
± 1.36x10-2
9.61x102 1.44x10
4
T53 0.65 ± 0.08 6.52x10-2
± 9.64x10-3
7.02x102 1.08x10
4
R54 0.66 ± 0.09 6.87x10-2
± 1.26x10-2
7.75x102 1.13x10
4
E55 0.68 ± 0.07 5.60x10-2
± 8.28x10-3
5.30x102 9.46x10
3
S56 0.69 ± 0.01 7.34x10-2
± 1.12x10-2
6.85x102 9.34x10
3
G57 0.69 ± 0.01 9.37x10-2
± 1.65x10-2
1.51x103 1.61x10
4
D60 0.38 ± 0.02 3.43x10-1
± 2.41x10-1
2.98x102 8.68x10
2
R61 0.59 ± 0.01 5.68x10-2
± 7.05x10-3
3.25x102 5.72x10
3
F62 0.51 ± 0.01 7.49x10-2
± 1.07x10-2
3.95x102 5.28x10
3
S63 0.62 ± 0.01 7.90x10-2
± 1.12x10-2
1.11x103 1.41x10
4
G64 0.76 ± 0.03 5.90x10-2
± 5.64x10-2
1.52x103 2.58x10
4
S65 0.71 ± 0.01 8.02x10-2
± 1.95x10-2
1.42x103 1.77x10
4
G66 0.69 ± 0.01 5.58x10-2
± 1.52x10-2
1.51x103 1.76x10
4
S67 0.78 ± 0.02 8.02x10-2
± 4.15x10-2
1.42x103 1.77x10
4
G68 0.80 ± 0.02 1.40x10-2
± 9.59x10-3
1.52x103 1.09x10
5
T69 0.61 ± 0.02 2.85x10-2
± 7.28x10-3
5.21x102 1.83x10
4
D70 0.38 ± 0.01 2.45x10-2
± 4.67x10-2
8.21x102 3.35x10
3
F71 0.84 ± 0.01 2.32x10-2
± 1.06x10-2
1.58x102 6.81x10
3
T72 0.89 ± 0.01 1.64x10-2
± 9.43x10-3
4.07x102 2.48x10
4
L73* 0.88 ± 0.03 -5.00x10-6
± 9.00x10-6
1.72x102 3.45x10
7
T74 0.92 ± 0.01 1.69x10-2
± 1.28x10-2
2.17x102 1.28x10
4
I75 0.92 ± 0.01 1.69x10-2
± 1.33x10-2
1.21x102 7.15x10
3
S76 0.90 ± 0.01 3.89x10-2
± 1.81x10-2
5.66x102 1.46x10
4
S77 0.90 ± 0.01 3.89x10-2
± 1.81x10-2
1.92x103 4.93x10
4
L78 0.95 ± 0.01 6.04x10-3
± 4.97x10-2
2.17x102 3.59x10
4
Q79 0.87 ± 0.01 3.72x10-2
± 1.23x10-2
2.90x102 7.81x10
3
A80 0.90 ± 0.01 3.25x10-2
± 2.74x10-2
6.50x102 2.00x10
4
E81 0.38 ± 0.01 7.06x10-4
± 2.07x10-4
3.19x102 4.53x10
5
D82 0.32 ± 0.03 1.93x10-1
± 8.18x10-3
1.46x102 7.57x10
2
V83 0.91 ± 0.01 1.92x10-2
± 1.52x10-2
5.41x101 2.81x10
3
A84 0.87 ± 0.01 4.73x10-2
± 2.08x10-2
2.97x102 6.28x10
3
V85 0.90 ± 0.01 1.98x10-2
± 1.10x10-2
8.15x101 4.11x10
3
S93 1.09 ± 0.02 4.36x10-2
± 6.59x10-2
1.08x103 2.47x10
4
T94 0.90 ± 0.01 2.43x10-2
± 1.34x10-2
7.02x102 2.88x10
4
Y96 0.89 ± 0.01 2.41x10-2
± 1.40x10-2
1.27x102 5.25x10
3
S97 0.88 ± 0.01 2.90x10-2
± 1.31x10-2
1.08x103 3.71x10
4
G99 0.86 ± 0.02 1.68x10-2
± 1.09x10-2
8.77x102 5.22x10
4
Q100 0.81 ± 0.01 3.41x10-2
± 1.62x10-2
6.96x102 2.04x10
4
G101 0.56 ± 0.01 2.88x10-2
± 2.87x10-3
1.20x103 4.17x10
4
T102 0.80 ± 0.02 3.09x10-3
± 2.04x10-3
5.21x102 1.68x10
5
K103 0.71 ± 0.02 2.01x10-2
± 8.27x10-3
5.93x102 2.95x10
4
L104 0.52 ± 0.01 1.47x10-2
± 1.95x10-3
1.43x102 9.77x10
3
Table 5.1: continued
Continued
122
E105 0.73 ± 0.01 2.76x10-2
± 7.68x10-3
1.97x102 7.13x10
3
I106 0.76 ± 0.01 2.24x10-2
± 5.31x10-3
5.44x101 2.43x10
3
K107 0.42 ± 0.01 2.08x10-2
± 1.57x10-3
2.20x102 1.06x10
4
R108 0.62 ± 0.01 2.78x10-2
± 6.08x10-3
1.02x101 3.68x10
2
* Residues were fitted with a single exponential function
Table 5.1: continued
123
CHAPTER 6
6. ON GOING WORK: CHARACTERIZATION OF HUMAN, MOUSE AND
HAMSTER PrP AMYLOID FIBRILS BY TILTED-BEAM TRANSMISSION
ELECTRON MICROSCOPY
6.1 Introduction
6.1.1 The prion protein
The prion diseases, or transmissible spongiform encephalopathies (TSEs), are a
group of fatal neurodegenerative disorders that affect humans and animals. These
disorders are associated with the structural rearrangement of a largely α-helical cellular
brain protein, dubbed cellular prion protein PrPc, to highly organized, β-sheet rich
amyloid-like aggregate, called “scrapie” prion PrPSc
.126-130
The protein only hypothesis
asserts that the transmission of TSEs does not require nucleic acids and that PrPSc
itself is
the infectious agent by binding to cellular prion protein and catalyzing its conversion to
PrPSc
.131,132
While most recent evidence points to the ability of PrPSc
aggregates to adopt a
spectrum of molecular conformations as being a key determinant of prion propagation,
the molecular mechanism of prion to amyloid conversion remains poorly understand at
the atomic level.133-135
Previous studies by us and others have demonstrated that, in
contrast to the recombinant full length human prion protein, the Y145Stop mutant
124
(huPrP23-144), which is associated with a familial prion disease, undergoes an efficient
nucleation dependent conversion to the amyloid state in vitro.126-129
Although Pr23-144
fibrils have not been shown to be infectious in vivo, the simple system provides a
convenient model in vitro for studying some of the most fundamental aspects of
mammalian prion propagation including the phenomena of prion strains and barriers
species.130,131
Recombinant Pr23-144 of human, mouse and Syrian hamster are of
particular interest since the substitution of a single amino acid in a critical region
encompassing residues 138 and 139 (Figure 6.1) is sufficient to transform the properties
of these proteins to those of PrP23-144 from another species.48,136
Figure 6.1: Amino acid sequence comparison of human (hu), mouse (mo), and Syrian
hamster (Sha) PrP, highlighting amino acid differences in the 23-144 region.136
6.1.2 MPL determination by tilted beam transmission electron microscopy
The mass-per-length (MPL) of an amyloid fibril is an important constraint on its
molecular structure. Amyloid fibrils contain β-sheets with cross-β alignment relative to
the long fibril axis and the spacing between β-strands in β-sheets is always 0.47 ± 0.01
nm.24,80
The MPL can be expressed as:
(6.1)
125
where MW is the amyloid-forming polypeptide’s molecular weight and η is the number
of molecules in each β-sheet spacing. MPL data are usually quantitatively obtained by
scanning transmission electron microscopy (STEM). STEM images of unstained samples
are called dark field images in the sense that the background has low intensity from weak
electron scattering while the amyloid fibrils have higher intensity due to stronger electron
scattering.137-139
Dark field microscopy is simple yet effective technique and well suited
for uses involving unstained biological samples. The main limitation is the low light
levels seen in the final image. Traditionally, dark-field microscopes are equipped with a
specially sized disc blocks some light form the light source, leaving an outer ring of
illumination. The light is then focused by the condenser lens and enters the sample. The
scattered light enters the objective lens, while the direct transmitted light simply misses
the lens and is not collected due to a direct illumination block. Only the scattered light
goes on to produce the image while the directly transmitted light is omitted.
Dark field images of unstained samples can also be obtained with a conventional
transmission electron microscope (TEM) by tilting the incident electron beam by a small
angle so that it is blocked by the objective aperture after passing through the sample. This
method is called tilted-beam TEM (TB-TEM).10
The dark field image is then formed
from scattered electrons that can pass through the aperture, using the same optics as in
bright-field TEM. MPL values are then determined quantitatively by comparing the fibril
image intensities with image intensities of reference objects with known mass densities.
In this work, the MPL will be calculated for the human, mouse and Syrian hamster PrP
fibrils using a TB-TEM method and tobacco mosaic virus (TMV) as a reference.
126
6.2 Materials and methods
6.2.1 Protein expression and purification
The plasmid encoding huPrP23-144 with N-terminal linker containing His6-tag
and a thrombin cleavage site was previously described.140
Natural abundance huPrP23-
144, moPrP23-144 and ShaPrP23-144 were expressed in Escherichia coli BL21 STAR
(DE3) by using LB broth (Fisher Scientific) containing 100 μg/mL of Ampicillin. Protein
expression was induced at OD600 ~ 0.8 by the addition of isopropyl-β-D-thiogalactoside
to a final concentration of 1nM. Cells were grown for an additional 16 h and harvested by
centrifugation at 4,000 g and 4°C for 15 min. After extraction, the protein is purified by
using a nickel-nitrilotriacetic acid agarose resin.136
The His6-tag was cleaved by using
biotinylated thrombin (Novagen), the thrombin was sequestered by using streptavidin-
agarose beads, and the residual His6-tag was removed by dialysis against ultrapure water.
The purified 126-resdiue protein consisted of the PrP23-144 sequence with an N-terminal
Gly-Ser-Asp-Pro extension. The protein was stored as a lyophilized powder and had a
final purity of 95% as determined by SDS-PAGE. Molecular masses of the PrP23-144
variants were confirmed by MALDI mass spectrometry as shown in Figure 6.2 and
Figure 6.3. In all experiments, protein concentration was measured from absorbance at
280 nm using a molar extension coefficient of 42,970 M-1
cm-1
for all PrP23-144 variants.
6.2.2 Preparation of PrP23-144 amyloid fibrils
Lyophilized human, mouse and hamster PrP23-144 were dissolved in ultrapure
water at a concentration of 400 μM and fibrilization was initiated by the addition of
potassium phosphate buffer at pH 6.4 to a final concentration of 50 μM.136,141
127
Figure 6.2: MALDI-TOF mass spectrum of 20 μM natural abundance huPrP in pure
H2O. The theoretical molecular weight is 12,523 Da.
128
Figure 6.3: MALDI-TOF mass spectra of 20 μM natural abundance A) ShaPrP and B) moPrP in pure H2O. The theoretical
molecular weights are 12,658Da and 12,593Da respectively.
129
Fibrils were allowed to form at 25°C under no agitation for 48 hours.
6.2.3 Atomic force microscopy on the PrP23-144 amyloid fibrils
AFM imaging was performed on a Bruker Icon Dimension microscope. After
fibrilization, fibril samples were diluted 10 times with ultrapure water and 10 μL was
loaded onto a freshly cleaved and washed mica disc (Ted Pella, Inc) for 2 minutes. The
mica surface was then washed twice with 50 μL aliquots of ultrapure water and allowed
to dry in air for 2 hours before AFM analysis. Imaging was performed in tapping mode in
air using a 0.01-0.025 Ohm-cm Antimony (n) doped Si Cantilever. Data were acquired in
height, amplitude and phase modes. All images were processed by interactive plane
fitting and Gaussian low-pass filtering using the Nanoscope SPM software package.
Average fibril height, width and axial periodicity were measured from processed AFM
height images using the Nanoscope analysis software.
6.2.4 Transmission electron microscopy on the PrP23-144 amyloid fibrils
6.2.4.1 Sample preparation
For TB-TEM, fibrils were adsorbed onto Formvar carbon Films on 200 mesh
copper grids. Grids were glow-discharged immediately before use. The sample
preparation is described in reference.10
Fibril solutions were diluted 20 times with
ultrapure water to produce isolated specimen. A 5 μL diluted fibril aliquot and 1 μL of
TMV solution were mixed and applied simultaneously on the grid. The concentrations of
TMV (kindly provided by Dr Gerald Stubbs, Vanderbilt University) ranged from 0.5
mg/mL to 1 mg/mL. After a 5 min adsorption period, solutions were blotted, washed
130
three times with 5 μL aliquots of ultrapure water for 10 seconds each, blotted and dried in
air.
For the bright-field TEM images of negatively stained samples, grids were
prepared using the same procedure but without adsorption of TMV and with a 60 seconds
period of staining with 5 μL of 3% uranyl acetate before final blotting and drying.
6.2.4.2 Image acquisition and processing
Images were acquired with an FEI Tecnai G2 Spirit TEM, operating at 80 kV,
equipped with a side-mounted, 1 megapixel AMT Advantage HR CCD camera. High
resolution bright-field images were acquired between 120,000x and 150,000x
magnification. For the TB-TEM images, a 30,000x magnification was used with a beam
tilt angle of 0.8°, a 50 μm diameter objective aperture, and a 300 μm condenser aperture.
Other microscope settings included spot size 2, bias 3, and filament current in the 10-20
μA range. The sample grid was scan manually under bright-field condition for promising
areas at low magnification (1,250x) using the survey mode of the AMT software. Once
an area that contained both isolated TMV rods and fibrils was identified, the
magnification and beam intensity were increased and a bright field image was acquired
(bright-field of unstained sample). The dark-field mode was then switched on; the focus
adjusted to maximize the clarity of both TMV rods and fibrils and the corresponding TB-
TEM image was recorded.
Both bright-field and TB-TEM images were analyzed with ImageJ software
(http://rsbweb.nih.gov/ij/). For the dark-field, only fibrils that appeared to be single
filaments, rather than pairs or bundles, were selected for MPL measurements. The
rectangular areas were set to 100 nm in length and 16 nm in width (50x8 pixels) to fit the
131
fibril widths, to allow several MPL counts per images and to insure consistency and
reproducibility among the different samples.
6.3 Results and discussion
6.3.1 Structural information probed by atomic force microscopy
Species-specific characteristics of PrP23-144 amyloids were probed by atomic
force microscopy (AFM). After an incubation period of 48 hours, AFM images for all
three species showed an extensive network of amyloid fibrils as shown in Figure 6.4. In
most cases, these fibrils were very long, usually several micrometers. With low
magnification AFM, the fibrils appeared to be structurally similar for all three species.
However, amplitude-mode images, acquired under high magnification, reveal that PrP23-
144 amyloids corresponding to different species do not all share the same structural
architecture. Fibrils of human and mouse (Figure 6.5A and Figure 6.5B) always showed
marked periodicity and displayed a segmented “bead-like” morphology along the fibril
axis. In contrast, the hamster (Figure 6.5C) revealed a different surface with almost no
axial periodicity. In most cases, the morphology of the hamster fibrils displayed an
apparent helical twist along the axis. ShaPrP23-144 fibrils with an absolute smooth
surface like described in previous work48
were not observed.
To further confirmed the nanoscale differences in structure between the PrP23-
144 amyloid fibrils, longitudinal section analyses in height modes were performed.
132
Figure 6.4: PrP23-144 Amyloid fibrils as visualized by Atomic Force Microscopy. The fibrils shown in these low magnification
AFM height images were formed upon incubation of (A) huPrP23-144, (B) moPrP23-144 and (C) ShaPrP23-144 in 50 mM
phosphate buffer at pH 6.4 with a protein concentration of 400 μM. Scale bar corresponds to 500 nm (applies to all panels).
133
Figure 6.5: Species differences in PrP23-144 morphology as revealed by high magnification AFM amplitude-mode images of
typical amyloid fibrils formed by (A) huPrP23-144, (B) moPrP23-144, (C) ShaPrP23-144. Scale bars correspond to 100 nm.
134
Plots of height along the axis of huPrP23-144 and moPrP23-144 fibrils (Figure 6.6A and
Figure 6.6B) showed very clear and regular variations with a mean peak-to-peak repeat
distance of about 30 nm. On the other hand, sectioning through ShaPrP23-144 fibrils
(Figure 6.6C) revealed only small and highly irregular variation in heights. The three
PrP23-144 fibrils shared only one common structural feature, the relative height, between
4 to 4.5 nm. Altogether, the AFM structural characteristics presented in this work were
similar to those reported in previous studies.48
Interestingly, these architectural
characteristics strongly correlate with species-dependent differences in seeding
specificity. Human and mouse PrP23-144 appear to be “seeding compatible” whereas
human and Syrian hamster PrP23-144 display a strong seeding barrier.48,136
Most biological samples have structural features that are on the same size scale as
the probe tip used in AFM and which convolute with the shape of the tip during the
imaging process. As a result, widths of PrP23-144 fibrils calculated with AFM were
usually overestimated. Widths were then only characterized using EM techniques as
shown below.
6.3.2 Structural information probed by electron transmission microscopy
In order to probe the fibril widths, high resolution bright field TEM images of
human, mouse and hamster were acquired. All samples were negatively stained with
uranyl acetate and width measurements were obtained using the ImageJ software. Figure
6.7A and Figure 6.7B show an example of a single human fibril and double human fibril,
respectively. Double fibrils of human, mouse and hamster only represent about 5% of all
fibrils observed.
135
Figure 6.6: AFM height images reveal quantitative morphological difference in
periodicity between the PrP23-144 Amyloid Fibrils. (A) AFM height image of huPrP23-
144 fibril (top panel) and a plot of sample height along the line (bottom panel). This plot
reveals an axial periodicity with a mean to peak-to-peak repeat distance of 30 nm. (B)
Same as (A) but for moPrP23-144 fibril. The same peak-to-peak axial period of 30 nm is
observed. (C) Same as (A) but for ShaPrP23-144 fibril. The axial periodicity was found
to be different with height variations comparable to those of the background. This is
consistent with a difference in morphology observed for the ShaPrP23-144 fibrils. Scale
bar represents 50 nm (applies to all panels).
136
Figure 6.7C shows a typical amyloid fibril observed for the moPrP23-144, whereas
Figure 6.7D and Figure 6.7E show a ShaPrP23-144 single fibril and a “double twisted”
fibril, respectively. For the Syrian hamster, the double twisted fibril is made of two fibrils
curled together. Fibril widths were found to be similar for the three species. Average
measurement for the huPrP23-144 is found to be 15.6 nm and 14.7 nm for moPrP23-144
and ShaPrP23-144.
For MPL calculation, dark-field images of unstained samples were used. Image
intensities were integrated over rectangular areas centered on the fibril axis (IF) and over
equal areas of background (IB1 and IB2). Similarly, image intensities were integrated over
rectangular areas centered on the TMV rods (ITMV) and over equal areas of background
on either side of each TMV segment (IB3 and IB4). The quantity <ITMV> was determined
as:10
(6.2)
MPL values were calculated as:
(6.3)
based on the known mass density of 131 kDa/nm for the TMV rods. All rectangles used
for the calculation were 16 nm in width and 100 nm in length (8x50 in pixels) to fit the
fibril segment.
137
Figure 6.7: High resolution bright field TEM images of negatively stained PrP amyloid fibrils. (A) huPrP23-144 “single”
fibrils. (B) huPrP23-144 “double” fibrils, (C) moPrP23-144 fibrils, (D) ShaPrP23-144 “single” fibrils, (E) ShaPrP23-144
“double twisted” fibrils. The pictures were acquired under a magnification of 120,000 for (C) and (E), 150,000 for (A), (B) and
(D). Scale bars represents 50 nm (applies to all panels).
138
In our experiment, all intensities (including TMV) used to measure a given MPL value,
were taken from the same image. A single dark field image typically yields around a
dozen MPL values.
Before calculating the MPL of human, mouse and Syrian Hamster Prion, a control
experiment was performed on a TMV only sample. A set of 10 dark field images were
acquired on a TMV sample (Figure 6.8A) and the corresponding MPL histogram is
shown in Figure 6.8B. The TMV histogram contains a total of 263 MPL values and was
fit to a single Gaussian function centered at 130.8 ± 2.2 kDa/nm and a full width at half
maximum (FWHM) of 26.01 ± 3.21 kDa/nm (Table 6.1). The broad base of the Gaussian
function confirms that the present method is relatively inaccurate and that a large number
of measurements are required to obtain a good estimation of the MPL value. A length
analysis was also performed (Figure 6.8C) and TMV rods were found to have a typical
length between 255 and 295 nm.
Figure 6.9 shows selected bright-field TEM images of unstained amyloid fibrils
(left column), the corresponding dark field TB-TEM images with MPL calculation
examples (center column) and MPL histograms (right column) for (A) huPrP23-144, (B)
moPrP23-144 and (C) ShaPrP23-144. The histograms were built with 555 MPL counts
for the human, 426 for the mouse and 390 for the Syrian hamster. Similarly to the TMV
experiment, MPL mean values were determined by fitting the histograms with a Gaussian
function:
(6.4)
139
Figure 6.8: (A) Dark field TB-TEM of a TMV only sample. The scale bar represents 200
nm. (B) Corresponding MPL histogram extracted from TB-TEM images and (C) TMV
rods length distribution.
140
Figure 6.9: Bright-field TEM images of unstained amyloid fibrils (left), corresponding
dark field TB-TEM images (middle) and MPL histogram (right). (A) huPrP23-144, (B)
moPrP23-144 and (C) ShaPrP23-144. For the dark field pictures, examples of MPL
values are shown along the fibrils axis. Double-headed arrows indicate the location of
the TMV rods. The scale bar represents 200 nm (applies to all panel).
141
where a is the amplitude, b the mean position in kDa/nm and c the FWHM in kDa/nm.
The fit parameters are summarized in Table 6.1. For the huPrP23-144, the MPL mean
value was found to be 50.8 ± 1.2 kDa/nm. For the moPrP23-144 and ShaPrP23-144, the
MPL means were respectively 54.3 ± 0.9 kDa/nm and 54.1 ± 0.6 kDa/nm.
Table 6.1: Gaussian fit parameters for the three variants: a the amplitude, b the mean
and c the full width at half maximum (FWHM). Errors were calculated with a 95%
confidence.
Using equation (6.1), the η number was found to be ~ 2 for all three variants,
exactly η = 1.9 for huPrP23-144, 2.0 for moPrP23-144 and ShaPrP23-144. These results
demonstrate that the human, mouse and Syrian hamster Prion fibrils are formed with two
molecules in each β-sheets spacing as shown in Figure 6.10.
Sample Amplitude Mean value FWHM
TMV 29.5 ± 3.2 130.8 ± 2.2 26.0 ± 3.2
Human 89.3 ± 7.3 50.8 ± 1.2 17.2 ± 1.6
Mouse 73.8 ± 5.2 54.3 ± 0.9 16.3 ± 1.3
Hamster 63.2 ± 2.7 54.1 ± 0.6 17.4 ± 0.9
142
Figure 6.10: Cartoon representation of a typical amyloid fibril with η = 2. Two
molecules (M and M’) are present in β-sheets spacing. The black arrow represents the
fibril axis. Note: This cartoon does not give any structural information.
143
LIST OF REFERENCES
(1) NMR of proteins and nucleic acids; Wiley: New York, 1986. (2) Cavanagh, J. Protein NMR spectroscopy : principles and practice; Academic Press: San Diego, 1996. (3) Uversky, V. N.; Fink, A. L. Protein misfolding, aggregation, and conformational diseases; Springer Science+Business Media: New York, N.Y., 2006. (4) Chiti, F.; Dobson, C. M. Annual review of biochemistry 2006, 75, 333. (5) Falk, R. H.; Comenzo, R. L.; Skinner, M. The New England journal of medicine 1997, 337, 898. (6) Mukherjee, S.; Pondaven, S. P.; Hofer, N.; Jaroniec, C. P. Biomolecular NMR assignments 2009, 3, 255. (7) Mukherjee, S.; Pondaven, S. P.; Jaroniec, C. P. Biochemistry 2011, 50, 5845. (8) Raffen, R.; Dieckman, L. J.; Szpunar, M.; Wunschl, C.; Pokkuluri, P. R.; Dave, P.; Wilkins Stevens, P.; Cai, X.; Schiffer, M.; Stevens, F. J. Protein science : a publication of the Protein Society 1999, 8, 509. (9) Hoshino, M.; Katou, H.; Yamaguchi, K.; Goto, Y. Biochimica et biophysica acta 2007, 1768, 1886. (10) Chen, B.; Thurber, K. R.; Shewmaker, F.; Wickner, R. B.; Tycko, R. Proceedings of the National Academy of Sciences of the United States of America 2009, 106, 14339. (11) Yonath, A.; Harms, J.; Hansen, H. A.; Bashan, A.; Schlunzen, F.; Levin, I.; Koelln, I.; Tocilj, A.; Agmon, I.; Peretz, M.; Bartels, H.; Bennett, W. S.; Krumbholz, S.; Janell, D.; Weinstein, S.; Auerbach, T.; Avila, H.; Piolleti, M.; Morlang, S.; Franceschi, F. Acta crystallographica. Section A, Foundations of crystallography 1998, 54, 945. (12) Ban, N.; Freeborn, B.; Nissen, P.; Penczek, P.; Grassucci, R. A.; Sweet, R.; Frank, J.; Moore, P. B.; Steitz, T. A. Cell 1998, 93, 1105. (13) Pervushin, K.; Riek, R.; Wider, G.; Wuthrich, K. Proceedings of the National Academy of Sciences of the United States of America 1997, 94, 12366. (14) Purcell, E. M.; Torrey, H. C.; Pound, R. V. Phys Rev 1946, 69, 37. (15) Bloch, F. Phys Rev 1946, 70, 460. (16) Ramsey, N. F. Phys Rev 1950, 78, 699. (17) Aue, W. P.; Bartholdi, E.; Ernst, R. R. J Chem Phys 1976, 64, 2229. (18) Ernst, R. R.; Bodenhausen, G.; Wokaun, A. Principles of nuclear magnetic resonance in one and two dimensions; Clarendon Press ; Oxford University Press: Oxford Oxfordshire New York, 1987. (19) Vuister, G. W.; Boelens, R.; Kaptein, R. J Magn Reson 1988, 80, 176. (20) Oschkinat, H.; Griesinger, C.; Kraulis, P. J.; Sorensen, O. W.; Ernst, R. R.; Gronenborn, A. M.; Clore, G. M. Nature 1988, 332, 374. (21) Bax, A.; Ikura, M.; Kay, L. E.; Barbato, G.; Spera, S. Ciba F Symp 1991, 161, 108. (22) Clore, G. M.; Gronenborn, A. M. Prog Nucl Mag Res Sp 1991, 23, 43.
144
(23) Kanelis, V.; Forman-Kay, J. D.; Kay, L. E. Iubmb Life 2001, 52, 291. (24) Sunde, M.; Blake, C. C. Quarterly reviews of biophysics 1998, 31, 1. (25) Kelly, J. W. Current opinion in structural biology 1998, 8, 101. (26) Dobson, C. M. Philosophical transactions of the Royal Society of London. Series B, Biological sciences 2001, 356, 133. (27) Buxbaum, J. Hematol Oncol Clin N 1992, 6, 323. (28) Gallo, G.; Goni, F.; Boctor, F.; Vidal, R.; Kumar, A.; Stevens, F. J.; Frangione, B.; Ghiso, J. The American journal of pathology 1996, 148, 1397. (29) Buxbaum, J.; Gallo, G. Hematology/oncology clinics of North America 1999, 13, 1235. (30) Uversky, V. N.; Fink, A. L. Biochimica et biophysica acta 2004, 1698, 131. (31) Lansbury, P. T.; Lashuel, H. A. Nature 2006, 443, 774. (32) Sipe, J. D. Annual review of biochemistry 1992, 61, 947. (33) Tan, S. Y.; Pepys, M. B. Histopathology 1994, 25, 403. (34) Westermark, P.; Benson, M. D.; Buxbaum, J. N.; Cohen, A. S.; Frangione, B.; Ikeda, S.; Masters, C. L.; Merlini, G.; Saraiva, M. J.; Sipe, J. D. Amyloid 2002, 9, 197. (35) Dinner, A. R.; Sali, A.; Smith, L. J.; Dobson, C. M.; Karplus, M. Trends in biochemical sciences 2000, 25, 331. (36) Tycko, R. Current opinion in structural biology 2004, 14, 96. (37) Serpell, L. C.; Sunde, M.; Benson, M. D.; Tennent, G. A.; Pepys, M. B.; Fraser, P. E. Journal of molecular biology 2000, 300, 1033. (38) Hoshino, M.; Katou, H.; Hagihara, Y.; Hasegawa, K.; Naiki, H.; Goto, Y. Nature structural biology 2002, 9, 332. (39) Carulla, N.; Zhou, M.; Giralt, E.; Robinson, C. V.; Dobson, C. M. Accounts of chemical research 2010, 43, 1072. (40) Petkova, A. T.; Ishii, Y.; Balbach, J. J.; Antzutkin, O. N.; Leapman, R. D.; Delaglio, F.; Tycko, R. Proceedings of the National Academy of Sciences of the United States of America 2002, 99, 16742. (41) Jaroniec, C. P.; MacPhee, C. E.; Astrof, N. S.; Dobson, C. M.; Griffin, R. G. Proceedings of the National Academy of Sciences of the United States of America 2002, 99, 16748. (42) Helmus, J. J.; Surewicz, K.; Nadaud, P. S.; Surewicz, W. K.; Jaroniec, C. P. Proceedings of the National Academy of Sciences of the United States of America 2008, 105, 6284. (43) Helmus, J. J.; Surewicz, K.; Surewicz, W. K.; Jaroniec, C. P. J Am Chem Soc 2010, 132, 2393. (44) Nadaud, P. S.; Helmus, J. J.; Sengupta, I.; Jaroniec, C. P. J Am Chem Soc 2010, 132, 9561. (45) Makin, O. S.; Atkins, E.; Sikorski, P.; Johansson, J.; Serpell, L. C. Proceedings of the National Academy of Sciences of the United States of America 2005, 102, 315. (46) Jimenez, J. L.; Nettleton, E. J.; Bouchard, M.; Robinson, C. V.; Dobson, C. M.; Saibil, H. R. Proceedings of the National Academy of Sciences of the United States of America 2002, 99, 9196. (47) Zandomeneghi, G.; Krebs, M. R.; McCammon, M. G.; Fandrich, M. Protein science : a publication of the Protein Society 2004, 13, 3314. (48) Jones, E. M.; Surewicz, W. K. Cell 2005, 121, 63. (49) Rochet, J. C.; Lansbury, P. T., Jr. Current opinion in structural biology 2000, 10, 60.
145
(50) Solomon, A.; Weiss, D. T. Amyloid 1995, 2, 269. (51) Kyle, R. A.; Gertz, M. A. Seminars in hematology 1995, 32, 45. (52) Hurle, M. R.; Helms, L. R.; Li, L.; Chan, W.; Wetzel, R. Proceedings of the National Academy of Sciences of the United States of America 1994, 91, 5446. (53) Kim, Y.; Wall, J. S.; Meyer, J.; Murphy, C.; Randolph, T. W.; Manning, M. C.; Solomon, A.; Carpenter, J. F. The Journal of biological chemistry 2000, 275, 1570. (54) Epp, O.; Lattman, E. E.; Schiffer, M.; Huber, R.; Palm, W. Biochemistry 1975, 14, 4943. (55) Huang, D. B.; Chang, C. H.; Ainsworth, C.; Johnson, G.; Solomon, A.; Stevens, F. J.; Schiffer, M. Molecular immunology 1997, 34, 1291. (56) Bourne, P. C.; Ramsland, P. A.; Shan, L.; Fan, Z. C.; DeWitt, C. R.; Shultz, B. B.; Terzyan, S. S.; Moomaw, C. R.; Slaughter, C. A.; Guddat, L. W.; Edmundson, A. B. Acta crystallographica. Section D, Biological crystallography 2002, 58, 815. (57) Wall, J. S.; Gupta, V.; Wilkerson, M.; Schell, M.; Loris, R.; Adams, P.; Solomon, A.; Stevens, F.; Dealwis, C. Journal of molecular recognition : JMR 2004, 17, 323. (58) Baden, E. M.; Owen, B. A.; Peterson, F. C.; Volkman, B. F.; Ramirez-Alvarado, M.; Thompson, J. R. The Journal of biological chemistry 2008, 283, 15853. (59) Stevens, P. W.; Raffen, R.; Hanson, D. K.; Deng, Y. L.; Berrios-Hammond, M.; Westholm, F. A.; Murphy, C.; Eulitz, M.; Wetzel, R.; Solomon, A.; et al. Protein science : a publication of the Protein Society 1995, 4, 421. (60) Kabat, E. A.; Wu, T. T. J Immunol 1991, 147, 1709. (61) Khurana, R.; Gillespie, J. R.; Talapatra, A.; Minert, L. J.; Ionescu-Zanetti, C.; Millett, I.; Fink, A. L. Biochemistry 2001, 40, 3525. (62) Souillac, P. O.; Uversky, V. N.; Millett, I. S.; Khurana, R.; Doniach, S.; Fink, A. L. The Journal of biological chemistry 2002, 277, 12657. (63) Souillac, P. O.; Uversky, V. N.; Fink, A. L. Biochemistry 2003, 42, 8094. (64) Gutierrez-Gonzalez, L. H.; Muresanu, L.; del Pozo-Yauner, L.; Sanchez, R.; Guereca, L.; Becerril, B.; Lucke, C. Biomolecular NMR assignments 2007, 1, 159. (65) Cai, M.; Huang, Y.; Sakaguchi, K.; Clore, G. M.; Gronenborn, A. M.; Craigie, R. Journal of biomolecular NMR 1998, 11, 97. (66) Yamazaki, T.; Lee, W.; Arrowsmith, C. H.; Muhandiram, D. R.; Kay, L. E. J Am Chem Soc 1994, 116, 11655. (67) Delaglio, F.; Grzesiek, S.; Vuister, G. W.; Zhu, G.; Pfeifer, J.; Bax, A. Journal of biomolecular NMR 1995, 6, 277. (68) Farrow, N. A.; Muhandiram, R.; Singer, A. U.; Pascal, S. M.; Kay, C. M.; Gish, G.; Shoelson, S. E.; Pawson, T.; Forman-Kay, J. D.; Kay, L. E. Biochemistry 1994, 33, 5984. (69) Palmer, A. G.; Rance, M.; Wright, P. E. J Am Chem Soc 1991, 113, 4371. (70) Palmer, A. G., 3rd Chemical reviews 2004, 104, 3623. (71) Jarymowycz, V. A.; Stone, M. J. Chemical reviews 2006, 106, 1624. (72) Ishima, R.; Torchia, D. A. Nature structural biology 2000, 7, 740. (73) Henzler-Wildman, K.; Kern, D. Nature 2007, 450, 964. (74) Baldwin, A. J.; Kay, L. E. Nature chemical biology 2009, 5, 808. (75) Liu, K.; Cho, H. S.; Lashuel, H. A.; Kelly, J. W.; Wemmer, D. E. Nature structural biology 2000, 7, 754. (76) Liu, K.; Kelly, J. W.; Wemmer, D. E. Journal of molecular biology 2002, 320, 821. (77) Dumoulin, M.; Canet, D.; Last, A. M.; Pardon, E.; Archer, D. B.; Muyldermans, S.; Wyns, L.; Matagne, A.; Robinson, C. V.; Redfield, C.; Dobson, C. M. Journal of molecular biology 2005, 346, 773.
146
(78) Serag, A. A.; Altenbach, C.; Gingery, M.; Hubbell, W. L.; Yeates, T. O. Nature structural biology 2002, 9, 734. (79) Margittai, M.; Langen, R. Quarterly reviews of biophysics 2008, 41, 265. (80) Tycko, R. Quarterly reviews of biophysics 2006, 39, 1. (81) Heise, H. Chembiochem : a European journal of chemical biology 2008, 9, 179. (82) Cavanagh, J.; Palmer, A. G.; Wright, P. E.; Rance, M. J Magn Reson 1991, 91, 429. (83) Kay, L. E.; Keifer, P.; Saarinen, T. J Am Chem Soc 1992, 114, 10663. (84) Wang, A. C.; Bax, A. Journal of biomolecular NMR 1993, 3, 715. (85) Tjandra, N.; Szabo, A.; Bax, A. J Am Chem Soc 1996, 118, 6986. (86) Kroenke, C. D.; Loria, J. P.; Lee, L. K.; Rance, M.; Palmer, A. G. J Am Chem Soc 1998, 120, 7905. (87) Palmer, A. G.; Kroenke, C. D.; Loria, J. P. Nuclear Magnetic Resonance of Biological Macromolecules, Pt B 2001, 339, 204. (88) Wang, C. Y.; Palmer, A. G. Magn Reson Chem 2003, 41, 866. (89) Hansen, D. F.; Vallurupalli, P.; Kay, L. E. J Phys Chem B 2008, 112, 5898. (90) Tollinger, M.; Skrynnikov, N. R.; Mulder, F. A. A.; Forman-Kay, J. D.; Kay, L. E. J Am Chem Soc 2001, 123, 11341. (91) Korzhnev, D. M.; Salvatella, X.; Vendruscolo, M.; Di Nardo, A. A.; Davidson, A. R.; Dobson, C. M.; Kay, L. E. Nature 2004, 430, 586. (92) Kay, L. E.; Torchia, D. A.; Bax, A. Biochemistry 1989, 28, 8972. (93) http://www.palmer.hs.columbia.edu/software.html. (94) Tjandra, N.; Feller, S. E.; Pastor, R. W.; Bax, A. J Am Chem Soc 1995, 117, 12562. (95) Lipari, G.; Szabo, A. J Am Chem Soc 1982, 104, 4546. (96) Lipari, G.; Szabo, A. J Am Chem Soc 1982, 104, 4559. (97) Cole, R.; Loria, J. P. Journal of biomolecular NMR 2003, 26, 203. (98) Mandel, A. M.; Akke, M.; Palmer, A. G. Journal of molecular biology 1995, 246, 144. (99) Butterwick, J. A.; Loria, J. P.; Astrof, N. S.; Kroenke, C. D.; Cole, R.; Rance, M.; Palmer, A. G. Journal of molecular biology 2004, 339, 855. (100) Carver, J. P.; Richards, R. E. J Magn Reson 1972, 6, 89. (101) http://abragam.med.utoronto.ca/software.html. (102) Skrynnikov, N. R.; Dahlquist, F. W.; Kay, L. E. J Am Chem Soc 2002, 124, 12352. (103) Souillac, P. O.; Uversky, V. N.; Millett, I. S.; Khurana, R.; Doniach, S.; Fink, A. L. The Journal of biological chemistry 2002, 277, 12666. (104) Khurana, R.; Souillac, P. O.; Coats, A. C.; Minert, L.; Ionescu-Zanetti, C.; Carter, S. A.; Solomon, A.; Fink, A. L. Amyloid 2003, 10, 97. (105) Spera, S.; Bax, A. J Am Chem Soc 1991, 113, 5490. (106) Chiti, F.; Dobson, C. M. Nature chemical biology 2009, 5, 15. (107) Clore, G. M.; Szabo, A.; Bax, A.; Kay, L. E.; Driscoll, P. C.; Gronenborn, A. M. J Am Chem Soc 1990, 112, 4989. (108) http://www.bmrb.wisc.edu. (109) Pokkuluri, P. R.; Huang, D. B.; Raffen, R.; Cai, X.; Johnson, G.; Stevens, P. W.; Stevens, F. J.; Schiffer, M. Struct Fold Des 1998, 6, 1067. (110) Sluzky, V.; Tamada, J. A.; Klibanov, A. M.; Langer, R. Proceedings of the National Academy of Sciences of the United States of America 1991, 88, 9377. (111) Mahler, H. C.; Muller, R.; Friess, W.; Delille, A.; Matheus, S. European journal of pharmaceutics and biopharmaceutics : official journal of Arbeitsgemeinschaft fur Pharmazeutische Verfahrenstechnik e.V 2005, 59, 407.
147
(112) Solomon, A. Method Enzymol 1985, 116, 101. (113) Johnson, S. M.; Wiseman, R. L.; Sekijima, Y.; Green, N. S.; Adamski-Werner, S. L.; Kelly, J. W. Accounts of chemical research 2005, 38, 911. (114) Ray, S. S.; Nowak, R. J.; Brown, R. H.; Lansbury, P. T. Proceedings of the National Academy of Sciences of the United States of America 2005, 102, 3639. (115) Merlini, G.; Bellotti, V. N Engl J Med 2003, 349, 583. (116) Stevens, F. J.; Myatt, E. A.; Chang, C. H.; Westholm, F. A.; Eulitz, M.; Weiss, D. T.; Murphy, C.; Solomon, A.; Schiffer, M. Biochemistry 1995, 34, 10697. (117) Chamberlain, A. K.; Marqusee, S. Biochemistry 1998, 37, 1736. (118) Hosszu, L. L.; Baxter, N. J.; Jackson, G. S.; Power, A.; Clarke, A. R.; Waltho, J. P.; Craven, C. J.; Collinge, J. Nat Struct Biol 1999, 6, 740. (119) Katou, H.; Hoshino, M.; Kamikubo, H.; Batt, C. A.; Goto, Y. J Mol Biol 2001, 310, 471. (120) Kuwata, K.; Shastry, R.; Cheng, H.; Hoshino, M.; Batt, C. A.; Goto, Y.; Roder, H. Nat Struct Biol 2001, 8, 151. (121) Luhrs, T.; Ritter, C.; Adrian, M.; Riek-Loher, D.; Bohrmann, B.; Dobeli, H.; Schubert, D.; Riek, R. Proc Natl Acad Sci U S A 2005, 102, 17342. (122) Alexandrescu, A. T.; Rathgeb-Szabo, K. J Mol Biol 1999, 291, 1191. (123) Olofsson, A.; Ippel, J. H.; Wijmenga, S. S.; Lundgren, E.; Ohman, A. The Journal of biological chemistry 2004, 279, 5699. (124) Bai, Y.; Milne, J. S.; Mayne, L.; Englander, S. W. Proteins 1993, 17, 75. (125) Connelly, G. P.; Bai, Y.; Jeng, M. F.; Englander, S. W. Proteins 1993, 17, 87. (126) Prusiner, S. B. Proceedings of the National Academy of Sciences of the United States of America 1998, 95, 13363. (127) Collinge, J. Annual review of neuroscience 2001, 24, 519. (128) Caughey, B.; Chesebro, B. Advances in virus research 2001, 56, 277. (129) Aguzzi, A.; Polymenidou, M. Cell 2004, 116, 313. (130) Weissmann, C. Nature reviews. Microbiology 2004, 2, 861. (131) Griffith, J. S. Nature 1967, 215, 1043. (132) Prusiner, S. B. Science 1982, 216, 136. (133) Collinge, J.; Clarke, A. R. Science 2007, 318, 930. (134) Aguzzi, A.; Sigurdson, C.; Heikenwaelder, M. Annual review of pathology 2008, 3, 11. (135) Cobb, N. J.; Surewicz, W. K. Biochemistry 2009, 48, 2574. (136) Vanik, D. L.; Surewicz, K. A.; Surewicz, W. K. Molecular cell 2004, 14, 139. (137) Engel, A. Ultramicroscopy 1978, 3, 273. (138) Engel, A.; Baumeister, W.; Saxton, W. O. P Natl Acad Sci-Biol 1982, 79, 4050. (139) Zeitler, E.; Bahr, G. F. J Appl Phys 1962, 33, 847. (140) Morillas, M.; Swietnicki, W.; Gambetti, P.; Surewicz, W. K. The Journal of biological chemistry 1999, 274, 36859. (141) Kundu, B.; Maiti, N. R.; Jones, E. M.; Surewicz, K. A.; Vanik, D. L.; Surewicz, W. K. Proceedings of the National Academy of Sciences of the United States of America 2003, 100, 12069.