TAMING THE GRIFFIN

Jobst Liebau

Taming the Griffin

Membrane interactions of peripheral and monotopic glycosyltransferases anddynamics of bacterial and plant lipids in bicelles

Jobst Liebau

©Jobst Liebau, Stockholm University 2017 ISBN print 978-91-7649-978-8ISBN PDF 978-91-7649-979-5 Cover art by Gefion Liebau: A griffin guarding a lipid membrane inside an NMR spectrometer. Printed in Sweden by Universitetsservice US-AB, Stockholm 2017Distributor: Department of Biochemistry and Biophysics, Stockholm University

Contents

List of Publications I

List of Additional Publications III

Abbreviations V

List of Figures VII

1 Introduction 11.1 Where does the journey lead us? . . . . . . . . . . . . . . . . 1

2 Lipid membranes in vivo and in vitro 52.1 General features of biological membranes . . . . . . . . . . . 6

2.2 Lipid dynamics . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.3 Biological membranes of plants and bacteria . . . . . . . . . . 9

2.3.1 E. coli inner membranes . . . . . . . . . . . . . . . . 9

2.3.2 Lipopolysaccharides . . . . . . . . . . . . . . . . . . 12

2.3.3 Chloroplast membranes . . . . . . . . . . . . . . . . 12

2.4 Membrane mimetics for biophysical applications . . . . . . . 13

2.4.1 Large unilamellar vesicles . . . . . . . . . . . . . . . 13

2.4.2 Micelles and bicelles . . . . . . . . . . . . . . . . . . 14

3 Glycosyltransferases 173.1 Membrane interaction of GT-B glycosyltransferases . . . . . . 18

3.2 The glycosyltransferase WaaG . . . . . . . . . . . . . . . . . 20

4 Methods 214.1 Circular dichroism . . . . . . . . . . . . . . . . . . . . . . . 22

4.2 Fluorescence spectroscopy . . . . . . . . . . . . . . . . . . . 22

4.2.1 Fluorescence of aromatic amino acids . . . . . . . . . 22

4.2.2 Quenching . . . . . . . . . . . . . . . . . . . . . . . 23

4.2.3 The parallax method . . . . . . . . . . . . . . . . . . 23

4.2.4 Anisotropy . . . . . . . . . . . . . . . . . . . . . . . 24

4.2.5 Stopped-flow fluorescence . . . . . . . . . . . . . . . 25

4.2.6 Measurements of the dissociation constant KD . . . . . 25

4.3 Nuclear magnetic resonance . . . . . . . . . . . . . . . . . . 26

4.3.1 Translational diffusion . . . . . . . . . . . . . . . . . 27

4.3.2 Structure determination . . . . . . . . . . . . . . . . . 28

4.3.3 Dynamics . . . . . . . . . . . . . . . . . . . . . . . . 29

4.3.4 Distance measurements using paramagnetic relaxation

enhancements . . . . . . . . . . . . . . . . . . . . . . 31

5 Results and Discussion 335.1 Mimicking plant and bacterial membranes . . . . . . . . . . . 33

5.1.1 Size of bicelles . . . . . . . . . . . . . . . . . . . . . 34

5.1.2 Lipid composition of bicelles made from E. coli phos-

pholipids . . . . . . . . . . . . . . . . . . . . . . . . 34

5.1.3 Rapid dynamics in bicelles . . . . . . . . . . . . . . . 35

5.1.4 Immersion depths of carbon atoms in membranes . . . 36

5.1.5 Bicelles as a plant and bacterial membrane mimetic . . 38

5.2 Monotopic and peripheral glycosyltransferases . . . . . . . . 39

5.2.1 Anchoring of WaaG to membranes . . . . . . . . . . . 39

5.2.2 Membrane interaction of full-length WaaG . . . . . . 40

6 Conclusions 43

Populärvetenskaplig Sammanfattning 45

Populärwissenschaftliche Zusammenfassung 49

Popular science summary 53

Acknowledgements 57

References 59

List of Publications

The following publications and manuscripts are included in this thesis and are

referred to by their roman numerals in the text. Reprints of the published

articles were made with permission from the publishers.

I New membrane mimetics with galactolipids: Lipid properties infast-tumbling bicellesWeihua Ye, Jobst Liebau, and Lena Mäler, The Journal of PhysicalChemistry B, 117(4), 1044−1050 (2013).

II Characterization of bicelles constructed from native Escherichia colilipidsJobst Liebau, Pontus Pettersson, Philipp Zuber, Candan Ariöz, and Lena

Mäler, Biochimica et Biophysica Acta – Biomembranes, 1858(9),

2097−2105 (2016).

III Immersion depths of lipid carbons in bicelles measured byparamagnetic relaxation enhancementJobst Liebau and Lena Mäler, The Journal of Physical Chemistry B,

121(32), 7660−7670 (2017).

IV Membrane interaction of the glycosyltransferase WaaGJobst Liebau, Pontus Pettersson, Scarlett Szpryngiel, and Lena Mäler,

Biophysical Journal, 109(3), 552−563 (2015).

V The glycosyltransferase WaaG: a peripheral membrane protein?Jobst Liebau, Biao Fu, Christian Brown, and Lena Mäler, submittedmanuscript.

List of Additional Publications

i Anionic lipid binding to the foreign protein MGS provides a tightcoupling between phospholipid synthesis and protein overexpressionin Escherichia coliCandan Ariöz, Weihua Ye, Amin Bakali, Changrong Ge, Jobst Liebau,

Hans-Jörg Götzke, Andreas Barth, Åke Wieslander, and Lena Mäler,

Biochemistry, 52(33), 5533−5544 (2013).

ii Effect of lipid bilayer properties on the photocycle of greenproteorhodopsinLjubica Lindholm, Candan Ariöz, Michael Jawurek, Jobst Liebau, Lena

Mäler, Åke Wieslander, Christoph von Ballmoos, and Andreas Barth,

Biochimica et Biophysica Acta – Bioenergetics, 1847(8), 698−708

(2015).

iii Characterization of fast-tumbling isotropic bicelles by PFG diffusionNMRJobst Liebau, Weihua Ye, and Lena Mäler, Magnetic Resonance inChemistry, 55(5), 395−404 (2017).

iv Substrate-mimicking domain of nucleotide-exchange factorFes1/HspBP1 ensures efficient release of persistent substrates fromHsp70Naveen Kumar Chandappa Gowda, Jayasankar Mohanakrishnan Kaimal,

Roman Kityk, Chammiran Daniel, Jobst Liebau, Marie Öhman, Matthias

P. Mayer, and Claes Andréasson, submitted manuscript.

Abbreviations

CD circular dichroism

CL cardiolipin

CMC critical micelle concentration

DGDG digalactosyl diacylglycerol

DHPC 1,2-dihexanoyl-sn-glycero-3-phosphatidylcholine

DMPC 1,2-dimyristoyl-sn-glycero-3-phosphatidylcholine

DMPE 1,2-dimyristoyl-sn-glycero-3-phosphatidylethanolamine

DPC dodecylphosphocholine

EPR electron paramagnetic resonance

GT glycosyltransferase

LPS lipopolysaccharide

LUV large unilamellar vesicle

MD molecular dynamics

MGDG monogalactosyl diacylglycerol

NOE nuclear Overhauser enhancement

PE phosphatidylethanolamine

PG phosphatidylglycerol

PRE paramagnetic relaxation enhancement

SANS small-angle neutron scattering

SQDG sulfoquinovosyl diacylglycerol

List of Figures

2.1 Representatives of different classes of membrane-associated

proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.2 Dynamics and their approximate correlation times in lipids and

membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.3 Lipids found in E. coli inner membranes . . . . . . . . . . . . 10

2.4 Glycolipids found in chloroplast membranes . . . . . . . . . . 10

2.5 Schematic structure of lipopolysaccharides . . . . . . . . . . . 11

2.6 Approximate diameters of LUVs, micelles, and bicelles . . . . 14

2.7 Idealized depiction of a small isotropic bicelle with q = 0.5 . . 15

3.1 GT-A, GT-B, and GT-C fold of GTs . . . . . . . . . . . . . . 18

3.2 Model of the membrane interaction of membrane-associated

GT-B GTs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

4.1 Pulse sequence for measuring translational diffusion . . . . . . 28

4.2 Illustration of dynamic parameters in lipids . . . . . . . . . . 32

5.1 Membrane interaction of the glycosyltransferase WaaG via its

anchor MIR-WaaG . . . . . . . . . . . . . . . . . . . . . . . 41

6.1 WaaG anchoring to a lipid membrane . . . . . . . . . . . . . 47

1. Introduction

We’re just about to walk off the map.—George Mallory, 1921 British Mt. Everest reconnaissance expedition

Venturing out into uncharted territory, never knowing which obstacle needs to

be negotiated next, or whether this is even going to be possible—this is the

challenge that adventurers and scientists face alike. While George Mallory

was climbing Mt. Everest with an obscure map in his left hand and an ice pick

in his right, life science very often starts with a pipette in one’s right hand and

a test tube in one’s left. The scientist’s endeavor is less dangerous, but in no

way less exciting or challenging than the mountaineer’s. Both the adventurer

and the scientist push to extend the limits of knowledge about the world in an

effort to remove the blank spots from their respective maps.

1.1 Where does the journey lead us?

This work is an endeavor to explore the smallest realms of life. The uncharted

territory consists of proteins and lipids and their mutual interactions. The ice

picks are biophysical methods, and instead of obscure maps, models or ideas

of proteins and lipids that scientists have developed guide the way.

Membranes are highly complex aggregates of lipids and proteins. Fre-

quently, lipids are considered to be the ‘matrix’ for membrane proteins. This is

evident in, for example, the simple depiction often found in text books of lipid

membranes as two parallel lines indicating the membrane surface, onto which

a complicated protein structure is drawn. However, such depictions neglect

the crucial functional importance of lipids and protein-lipid interactions for

many biological processes. The importance of membranes goes beyond mere

structural function, and an understanding of lipid properties in membranes and

their interaction with proteins is crucial for a comprehensive understanding of

membrane-related processes.

�

Lipid membranes have been extensively studied with biophysical tools. To

do so, native membranes very often need to be simplified or modified to ren-

der them amenable to experimentation. Membrane mimetics can be as simple

as an organic solvent, or as complex as a natural membrane extract, depend-

ing on the question that is being asked and the method used. The first focus

of this work is on the characterization by solution-state NMR of membrane

mimetics that mimic natural membranes more faithfully than those previously

employed. In particular, dynamics and localization of bacterial and plant lipids

inside membranes are investigated.

If lipids are important, so are proteins that synthesize lipids. A defec-

tive membrane reduces or abolishes the viability of organisms. It has been

shown, for instance, that defects in the structure of lipopolysaccharides (LPS),

lipids containing long polyglycan chains that are located in the outer leaflet of

the outer membrane in Gram-negative bacteria, destabilize the outer bacterial

membrane and make the bacterium more susceptible to antibiotics. This vul-

nerability can be exploited when searching for novel antibiotic targets. Multi-

resistant bacteria have become an imminent threat to society and it has recently

been estimated that by 2050 as many patients will die from infections caused

by bacteria resistant to all known antibiotics as will die of cancer [142]. More-

over, many forms of clinical treatment rely on the availability of functional

antibiotics. Even if the above prediction is not realized, antibiotic resistance

is already a threat today. Meanwhile, fewer and fewer new antibiotics are ap-

proved each year [187].

Proteins involved in the synthesis of bacterial lipids are thus promising

antibiotic targets. One such protein is the glycosyltransferase (GT) WaaG, a

membrane-associated protein that is involved in the synthesis of LPS. WaaG is

not only of interest because it is a promising antibiotic target but also because

of its peculiar function at the membrane: WaaG catalyzes the formation of a

glycosidic bond between a glucose sugar and a lipid acceptor. To do so, the

hydrophilic nucleotide sugar has to be brought into close proximity with a lipid

acceptor. At the same time, WaaG itself seems to be a hybrid between a soluble

and a membrane protein. Like the metaphor in the title of this thesis suggests,

WaaG is a microscopic analogy of a griffin. A griffin is a hybrid between lion

and eagle. WaaG appears to have properties that make it a hybrid creature of

its own kind. It can exist in both a soluble and a membrane-bound form. The

second focus of this thesis is thus on the investigation of the membrane interac-

tions of WaaG, and whether the observations made are in agreement with the

model of the membrane interaction of membrane-associated GTs previously

put forward.

The questions outlined above are addressed by a set of biophysical meth-

ods, including circular dichroism, fluorescence methods, and solution-state

�

NMR, with a focus on the latter. These biophysical methods put different

demands on the systems studied, but what they all have in common is that

both the membrane and the protein need to be simplified or modified for ob-

servations to be possible. Simplifications or modifications include the use of

membrane mimetic systems like vesicles or bicelles, the study of isolated seg-

ments of WaaG, and modifications of WaaG necessary to introduce reporter

probes.

The thesis is structured as follows: properties of plant and bacterial lipid

membranes and of membrane systems that can mimic such lipid bilayers are

discussed in chapter 2. A brief overview of GTs in general and membrane-

associated GTs in particular is given in chapter 3. This overview includes

background information on the GT WaaG. The biophysical methods used to

study both the membrane interaction of WaaG and lipid properties are outlined

in chapter 4. The two questions that are at the center of this thesis, how lipids

localize and move in ‘realistic’ membrane systems, and how WaaG interacts

with such or similar membrane systems, are discussed in chapter 5, and a per-

spective on possible future research avenues is outlined in chapter 6. The work

is concluded by a popular science summary.

�

�

2. Lipid membranes in vivo and

in vitro

God made the bulk; surfaces were invented by the devil.—Wolfgang Pauli

Biological membranes are highly complex aggregates consisting mainly of

lipids and proteins. Membranes compartmentalize cells, forming a semi-

permeable, protective barrier between the cell or compartment and the sur-

rounding environment, creating a confined space in which proteins and other

molecules exist at high concentrations and in which most processes required

to sustain life occur. Since membranes are the walls and gates of a cell, many

of those essential biological processes occur in or at membranes. Therefore,

apart from their protective and barrier functions, membranes also have other

functional roles, which comprise among others regulation and organization

of proteins, maintenance of charge and concentration gradients, information

transduction, and supply of lipid substrates for proteins [48; 125].

Biological membranes have to be simplified in order to make them

amenable to biophysical study. Membrane mimetics are of reduced complexity

compared to biological membranes and address method-specific limitations.

For instance, in solution-state NMR ‘large’ systems cannot be studied easily

and therefore membrane mimetics need to be ‘small’ (see also section 4.3).

In the following sections, general features of biological membranes are

outlined, dynamics that lipids undergo in membranes are discussed, and the

compositions of plant and bacterial membranes are described. At the end of

this chapter, a brief overview of membrane mimetics relevant to this work is

given.

�

2.1 General features of biological membranes

Despite the large variation in composition and the specialized functions of

some membranes and their components, some general properties of mem-

branes can be identified.

Membranes are formed from lipids which are amphiphile molecules con-

sisting of a hydrophilic headgroup and hydrophobic acyl chains (see Fig. 2.3

and 2.4). Driven by the hydrophobic effect, lipids spontaneously phase-separate

from aqueous solvent [125]. Depending on the lipid, the phase formed can be

lamellar, hexagonal, or cubic. The amphiphilic shape hypothesis provides a

means of predicting the phase properties of lipids [65]: if the headgroup’s

cross-sectional area is larger or smaller than the acyl chains’, lipids are of

‘conical’ shape and form a hexagonal or inverted hexagonal phase. Hexag-

onal phases are observed for lysolipids, which are lipids with only one acyl

chain. Inverted hexagonal phases are formed by lipids with small headgroups,

such as 1,2-dimyristoyl-sn-glycero-3-phosphatidylethanolamine (DMPE). If

the headgroup’s and the acyl chains’ cross sectional areas are similar in size,

the lipid is ‘cylindrical’ and prone to form bilayers. This is the case for 1,2-

dimyristoyl-sn-glycero-3-phosphatidylcholine (DMPC) or 1,2-dimyristoyl-sn-

glycero-3-phosphatidylglycerol (DMPG) [49]. A lamellar phase is character-

ized by a lipid bilayer, in which the polar headgroups of the lipids face the

aqueous environment and the hydrophobic acyl chains point towards the cen-

ter of the bilayer. The lipid composition of membranes needs to be such that

the membrane bilayer is stable, but at the same time some curvature stress is

required to provide the membrane with sufficient elasticity to enable processes

such as cell division, membrane fusion, and endo- or exocytosis. This means

that membranes have to consist of lamellar as well as non-lamellar lipids [64].

As described in the next section, biological membranes are in a liquid-

crystalline state providing them with the relevant flexibility to perform their

functions [164]. However, growing evidence suggests that the lateral organiza-

tion of lipids is not homogeneous. In eukaryotic membranes, this is manifested

by lipid rafts, typically enriched in sphingolipids and cholesterol. Rafts allow

for lateral sorting of membrane-associated proteins which means that certain

membrane proteins, mostly involved in signaling and trafficking, are enriched

in these rafts. Therefore, an important biological function is attributed to them

[53; 163]. In prokaryotes, a heterogeneous distribution of lipids has also been

observed, in particular clusters of anionic lipids [7; 20]. For instance, cell poles

in Escherichia coli are enriched in cardiolipin (CL) [134].

Membrane proteins are associated with or embedded into lipid bilayers.

Proteins can be categorized by their mode of membrane interaction; represen-

tatives of some important groups are shown in Fig. 2.1. Peripheral proteins

�

In order to maintain a liquid-crystalline phase, organisms adjust the mem-

brane composition to environmental conditions. It has, for instance, been

shown that E. coli alters its acyl chain composition in response to changes

in growth temperature such that the gel-to-liquid phase transition temperature

of the lipid mixture is about 10 ◦C below the growth temperature and non-

lamellar phases are observed approximately 20 ◦C above it [83; 135].

In the liquid-crystalline phase, lipids undergo a broad range of motions on

a wide range of timescales (see Fig. 2.2) [52; 56]. As explained in section

4.3.3, dynamics can be described by an amplitude of motion, represented by

an order parameter, and by a reorientation time of the motion, represented by

a correlation time [69; 119; 120; 196; 197]. Bond oscillations and gauche-

trans isomerizations have characteristic correlation times on the order of pi-

coseconds. These motions are particularly rapid and unrestricted in lipid acyl

chains. Consequently, the hydrophobic region of bilayers has liquid-like prop-

erties [186]. Lateral and rotational diffusion times are on the order of nanosec-

onds [52; 118; 186]. These diffusion processes give rise to the fluidity of the

membrane [125; 186]. While lateral dynamics are rapid, transversal flip-flop

motions, where a lipid flips from one leaflet of the bilayer to the other, have

very long correlation times of hours or days, such that an asymmetric lipid

distribution between the leaflets of a bilayer can be maintained [71; 133].

This is for instance the case in plasma membranes of eukaryotes, or in the

outer membrane of Gram-negative bacteria [49; 184]. Lipid transport across

the membrane is facilitated by a protein class called flippases [213]. Finally,

membranes undergo undulations: thermally agitated, collective lipid motions

that occur on a broad range of timescales between microseconds and seconds.

Undulations relate to the elasticity of the membrane and are thus crucial for

processes that involve shape modifications of the membrane, e.g. endo- and

exocytosis [56].

In lipids, the glycerol backbone displays the highest rigidity, and gauche-

trans isomerizations are restricted in particular for the g1–g2 bond (see Fig. 2.3

for carbon atom nomenclature in lipids). Consequently, the glycerol backbone

undergoes only relatively slow dynamics, such as lipid rotational diffusion and

wobble [18; 52]. Phospholipid headgroups are mainly oriented parallel to the

membrane surface, however, the motional degrees of freedom about the P–O

bonds of the phosphate moiety and about the headgroup carbon-carbon bonds

are higher than for the glycerol carbons, such that other orientations of the

headgroup with respect to the membrane are also possible, albeit less likely

than the parallel orientation [160; 165; 212]. The highest flexibility is ob-

served for the terminal carbon atoms of the acyl chain, in particular the methyl

carbon [165]. Moreover, the probability distribution of the location of the

terminal carbons inside the membrane is broad since their motion is nearly

�

2.3.2 Lipopolysaccharides

The composition of the outer leaflet of the outer membrane of Gram-negative

bacteria differs from the other leaflets in that its main lipid component is

lipopolysaccharides (LPS). The outer membrane serves as a permeability bar-

rier for toxic compounds such as antibiotics, while nutrients and ions enter the

periplasmic space through pore proteins located inside the outer membrane

[141]. Moreover, LPS is required for the structural integrity of the outer mem-

brane [210]. For instance, ΔwaaG mutants can only produce a truncated ver-

sion of LPS which moreover lacks phosphorylations of the LPS core (see also

section 3.2). Those mutants are highly susceptible to hydrophobic antibiotics,

and are immobile due to their lack of flagella and pili [105; 145].

A schematic structure of LPS is shown in Fig. 2.5. LPS consists of a

hydrophobic moiety, called lipid A, the recognition of which by the toll-like

receptor 4 leads to the activation of the human immune system. A polysac-

charide connects to lipid A. This polysaccharide is subdivided into inner and

outer core and O-antigen. The inner core is highly conserved within a family,

whereas the outer core shows more variation. The O-antigen is a chain consist-

ing of repeating polyglycan units that show a large structural diversity between

different strains of the same species. LPS is assembled up to the outer core at

the inner leaflet of the inner membrane by gene products encoded by three

operons in the chromosomal waa region. The gmhD and waaA operons en-

code proteins involved in the inner core assembly, and the gene products of the

waaQ operon, among them WaaG, are responsible for the outer core assembly

and for core phosphorylations. Most GTs involved in the LPS core synthesis

are predicted, or have been shown to be peripheral membrane proteins located

at the inner leaflet of the inner membrane, and it has been suggested that they

form a coordinated complex. Following the core assembly, the deep-rough, i.e.

nascent, LPS is transferred to the outer leaflet of the inner membrane where the

O-antigen is attached. LPS is then transported across the periplasmic space and

to the outer leaflet of the outer membrane [72; 193].

2.3.3 Chloroplast membranes

Chloroplast membranes consist of phospholipids (~40 %) and glycolipids

(~60 %, see Fig. 2.4). More than 50 % of the chloroplast lipids are the galac-

tolipids monogalactosyl diacylglycerol (MGDG) and digalactosyl diacylglyc-

erol (DGDG) which contain one and two galactose moieties in the headgroup,

respectively [90]. MGDG and DGDG are thus believed to be the most abun-

dant lipids in nature [62]. Notably, the thylakoid membranes in chloroplasts

consist to 80 % of MGDG and DGDG, with smaller amounts of sulfoquinovo-

syl diacylglycerols (SQDG) and PG [90]. MGDG is a non-bilayer forming

��

lipid, whereas DGDG and SQDG are bilayer-forming lipids [47]. This means

that, similar to bacterial membranes, also chloroplast membranes are under

substantial curvature stress which is required for correct function of membrane

proteins and for membrane integrity [201]. Moreover, due to their high abun-

dance in thylakoid membranes, it is believed that galactolipids play a crucial

role in optimal function of photosynthesis [74]. Under stress conditions, in

particular when phosphate is scarce, synthesis of DGDG and SQDG is up-

regulated and the ratio of glycolipids in chloroplast membranes increases to

~80 %. Moreover, under phosphate deprivation, DGDG also replaces phos-

pholipids in extra-plastidial membranes, where it is not found otherwise. In

this way, phosphate is made available for other essential biological processes

[74; 188].

2.4 Membrane mimetics for biophysical applications

A broad range of membrane mimicking systems have been used for biophys-

ical investigations. They range from in vivo membranes to organic solvents.

The degree of complexity-reduction required depends on the phenomenon to

be investigated and on the method used. Within the scope of this work, relevant

membrane mimetics comprise large unilamellar vesicles (LUVs), micelles, and

bicelles. A size comparison is shown in Fig. 2.6.

2.4.1 Large unilamellar vesicles

LUVs are spherical membrane bilayers that have a typical diameter of 100 nm.

Among the mimicking systems used here, they emulate the size and (with im-

portant exceptions) the low curvature of natural membranes best. The mem-

brane composition can be controlled and a broad range of lipid mixtures can be

used to produce LUVs. Moreover, the size and lamellarity can be controlled,

such that it is also possible to make small, giant, and multilamellar vesicles. It

is also possible to produce vesicles with heterogeneous lipid distributions both

laterally, with asymmetric leaflet compositions, and transversal, with lipid do-

main formation [71]. Finally, LUVs provide a closed compartment which al-

lows for maintaining concentration or charge gradients across the membrane

[71; 127; 137]. However, LUVs cannot be used in NMR studies since the lipid

concentration is too low and the aggregates are too big. In solid-state NMR

applications, multilamellar vesicles have been used to study lipid or protein

properties [45]. However, for the purpose of solution-state NMR substantially

smaller membrane mimetics are required.

��

bate [8; 124; 183]. In 31P NMR studies of q = 0.5 bicelles, DHPC (detergent)

and DMPC (lipid) show distinct peaks despite identical headgroups. From that

it is concluded that detergent and lipid molecules experience different chem-

ical environments and that they sequester into different regions of the bicelle

[61]. Additional experimental evidence suggests that detergents are located in

a micelle-like environment and are thus found in the bicelle rim whereas lipids

are located in a central bilayer. However, it is unclear how strongly detergent

and lipids are separated [8; 61; 124]. Finally, SANS data are consistent with

the notion that small isotropic bicelles are oblate objects [124].

Since bicelles contain a lipid bilayer, they are better mimetics of natu-

ral membranes than micelles and are still amenable to solution-state NMR

experiments. For some proteins, it has been shown that bicelles are better

mimetics than micelles maintaining the structure and function of these pro-

teins [17; 40; 41]. However, there are also reports that protein structures can

be distorted in bicelles [54]. It can be speculated that this is due to interactions

between the protein and detergents in the bicelle rim. Thus, a frequent concern

when using bicelles as membrane mimetics is whether or not interactions of

the protein with the detergent molecules are significant.

��

3. Glycosyltransferases

Whether or not we find what we are seekingIs idle, biologically speaking.—Edna St. Vincent Millay

Glycosyltransferases (GTs) form a vast group of proteins that catalyze the for-

mation of glycosidic bonds between sugar moieties from an activated donor

molecule and a broad range of acceptor molecules [21; 102]. As of Septem-

ber 2017, about 350 000 GTs in 104 families are listed in the carbohydrate-

active enzymes database (www.cazy.org) [25; 38]. Families are defined by

sequence-similarity, and it is expected that all members of the same family

adopt the same three-dimensional fold. Despite the large number of GTs,

only three major folds have been observed, called GT-A, GT-B, and GT-C

fold [22; 34; 121; 122]. Only a handful of GTs that do not adopt any of these

folds have been described in the literature [102].

A crucial structural feature found in both the GT-A and the GT-B fold is the

Rossmann-like fold which is a structural motif that binds nucleotides. Thus,

all nucleotide sugar-dependent GTs adopt either a GT-A or a GT-B fold [179].

The Rossmann fold is a β -strand/α-helix/β -strand motif [70]. In GTs, the

motif is extended to several alternating β -strand and α-helices, giving rise to

a tertiary structure in which the β -strands form a β -sheet that is wrapped by

α-helices. Additionally, insertions of structural elements into the alternating

β/α/β motif are frequently observed [174].

Fig. 3.1 shows representatives of the three major folds observed in GTs.

GT-A GTs consist of two Rossmann-like domains that are tightly connected

to each other. The N-terminal domain binds the nucleotide sugar while the

C-terminal domain binds the acceptor molecule. The function of most GT-A

GTs is dependent on a divalent metal ion which is bound by a Asp-X-Asp mo-

tif. However, examples of GT-A GTs that do not contain the Asp-X-Asp motif

exist. In contrast, GT-B GTs consist of two Rossmann-like domains that are

connected by a flexible linker. The active site is located in the cleft between

��

3.2 The glycosyltransferase WaaG

The glycosyltransferase WaaG is involved in the synthesis of the core of LPS

molecules (see section 2.3.2 and Fig. 2.5), which are the main constituents of

the outer leaflet of the outer membrane in Gram-negative bacteria [131; 145].

WaaG has been identified as a potential antibiotic target, since WaaG-deficient

cell lines are immobile and hypersensitive to antibiotics [105]. This is because

cells with a deletion of the gene encoding WaaG (ΔwaaG cells) cannot produce

full-length LPS (see Fig. 2.5). Instead, they contain a truncated version, called

deep-rough LPS. Apart from lacking the outer core and O-antigen, ΔwaaGdeep-rough LPS also lacks phosphate moieties that are crucial for membrane

integrity. Therefore, the protective LPS leaflet is severely compromised in

ΔwaaG cells [210]. Moreover, ΔwaaG cells do not contain flagella and pili

and are thus immobile.

WaaG catalyzes the transfer of a glucose moiety from UDP-glucose to the

nascent chain of LPS. The stereochemistry of the sugar is maintained during

the reaction, i.e. WaaG is classified as a retaining GT [131]. WaaG belongs

to the GT4 family of GTs, which may be the most ancient family of retaining

GTs since it is the only family of retaining GTs found in primitive archaea

[38; 131]. WaaG has been classified as a monotopic GT [2]. This classifica-

tion is based on the subcellular localization of WaaG at the membrane which is

manifested by the need for detergent during purification. However, the crystal

structure of WaaG was obtained from samples that did not contain detergent

[131].

WaaG shows strong structural similarities to other membrane-associated

GT-B GTs, like MurG [75], WaaC [63], and WaaA [158]. It is therefore hy-

pothesized that the mode of membrane interaction of WaaG is in agreement

with the model outlined in section 3.1 [2].

��

4. Methods

The absence of evidence is not the evidence of absence.—Anon.

The size of lipids and proteins is on the order of nanometers. At the same

time, molecular processes may be very rapid—some processes, like gauche-

trans isomerizations, occur on the timescale of picoseconds, other processes,

like chemical exchange, are slower but still of biological relevance. These

properties of lipids and proteins determine the requirements that a biophysical

method must meet to yield insights into the features to be studied. The choice

of the method also depends on the question asked. At times, atomic resolution

is required to get an understanding of a property of interest; other times, molec-

ular resolution is sufficient. For some processes, fast or slow dynamics may be

of interest, in other instances a static picture provides sufficient information.

Each method has advantages, disadvantages, and limitations in terms of time

and spatial resolution and with respect to sample requirements. Thus, to ob-

tain a comprehensive answer to a biological question, ideally a set of methods

should be used. Three methods were employed in the framework of this thesis:

far-UV circular dichroism reveals the secondary structure of proteins and has

relatively low requirements on the sample. Fluorescence methods can be em-

ployed to measure nanosecond dynamics and nanometer distances; however,

they require the presence of a fluorophore. Finally, with solution-state NMR

a broad range of dynamics can be investigated and protein structures can be

determined. However, NMR puts very high requirements on the sample. For

instance, high concentrations are required, the molecule investigated must not

be too large, and it often has to be isotope-labeled.

In the following the methods relevant for this thesis are briefly discussed.

More comprehensive background information on theory and application of the

methods may be found in the references.

��

4.1 Circular dichroism

Far-UV circular dichroism (CD) of the peptide bond provides information on

the secondary structure content of a peptide or protein. CD signals arise due to

the differential absorption of left- and right-handed circularly polarized light

by a chromophore that either contains a chiral center, is covalently bound to a

chiral center, or is located in an asymmetric environment.

In proteins, the far-UV CD spectrum stems mainly from the π → π∗ tran-

sition at 190 nm and from the n→ π∗ transition at 220 nm of the peptide bond,

with minor contributions from aromatic amino acids. Secondary structure el-

ements give rise to different asymmetric environments of the peptide bond

which in turn leads to CD spectra that are characteristic for different types of

protein secondary structure. From a comparison with CD spectra of proteins

of known structure, the secondary structure of a protein can be inferred.

Quantum mechanically, CD is described by the rotational strength R:

Δε ∝ R = ℑ{⟨

Φ f∣∣m |Φi〉〈Φi|μμμ

∣∣Φ f⟩} �= 0 (4.1)

where Δε is the CD signal and mmm and μμμ are the magnetic and electric dipole

moment operators. A CD signal is observed if R is non-zero. This is the case

when mmm and μμμ are non-zero, i.e. when the molecule is a chromophore, and

when they are non-perpendicular as is the case for a chiral chromophore or for

an achiral chromophore in an asymmetric environment [88; 93; 94; 152; 203].

4.2 Fluorescence spectroscopy

Fluorescence spectroscopy is a versatile method to obtain structural and dy-

namic information for a fluorophore. Among others, it can be employed to

determine the rotational correlation time of a fluorophore, to assess whether a

fluorophore localizes in a lipophilic or hydrophilic environment, how deep it

is immersed into a lipid bilayer, and how strongly it interacts with a bilayer

[103]. Techniques to obtain such information are discussed in the following.

4.2.1 Fluorescence of aromatic amino acids

Upon absorption of a photon, an electron in a fluorophore is excited into a

higher electronic state. Within picoseconds it relaxes to the lowest vibrational

level of the first excited electronic state. Fluorescence occurs when the elec-

tron relaxes back into the electronic ground state upon emission of a photon.

The wavelength of the emitted photon is longer than the wavelength of the

absorbed photon which gives rise to the characteristic Stokes’ shift of fluores-

cence excitation and emission spectra. Fluorescence is only one pathway to

��

reach the electronic ground state. Other, non-radiative pathways exist and de-

crease the fluorescence lifetime and intensity [103].

Aromatic amino acids are fluorescent. However, because of their low quan-

tum yield, Phe residues are typically not observable if Tyr or Trp residues are

present [58]. The fluorescence lifetime of Tyr and Trp residues is on the order

of nanoseconds, i.e. molecular dynamics on that timescale can be observed

if the underlying processes affect fluorescence properties of the fluorophores

[103; 154]. Moreover, Trp and Tyr residues are enriched at membrane inter-

faces which is often exploited in fluorescence studies of protein-membrane

interactions [95].

4.2.2 Quenching

Quenching is non-radiative relaxation of an electron in an excited electronic

state to the electronic ground state. Quenching thus competes with radiative

processes, i.e. fluorescence. The relation between loss in intensity and con-

centration of a molecule that quenches fluorescence is described by the Stern-

Volmer equation [172]:

I0

I= 1+KSV [Q] (4.2)

where I is the fluorescence intensity at quencher concentration [Q], I0 is the

fluorescence intensity in the absence of quencher, and KSV is the Stern-Volmer

quenching constant.

This relation holds for collisional quenching, in which fluorescence is

quenched by collisions between the quencher and the fluorophore, and for

static quenching, where quencher and fluorophore form a non-fluorescent com-

plex. If both processes occur, the relation between [Q] and I0/I is not linear

[50].

A number of molecules quench fluorescence of aromatic amino acids

through collisions, among them the water-soluble acrylamide and amphiphilic

nitroxide-labeled lipids. Fluorescence quenching can thus be used to monitor

fluorophore accessibility to aqueous or lipid environments [101].

4.2.3 The parallax method

In the parallax method, the quenching efficiency, and thus accessibility of

nitroxide-labeled quenchers to a fluorophore, is exploited to obtain an immer-

sion depth estimate of the fluorophore inside a lipid bilayer, assuming that the

positions of the quenchers inside the bilayer are known [1; 28; 29]. The method

relies on using two lipids to which a quencher (a nitroxide moiety) is attached,

one with a shallow immersion depth of the quencher (1) and one with a deeper

��

immersion depth of the quencher (2). The distance of the fluorophore from the

bilayer center can be estimated by [28]:

z = Lc1 +

(−L2

21− ln(I1/I2)πC

2L21

)(4.3)

where Lc1 is the distance of the shallow quencher from the bilayer center, L21

is the distance between shallow and deep quencher, C is the quencher concen-

tration in the membrane plane, and I1 and I2 are the fluorescence intensities in

the presence of the shallow and deep quencher, respectively.

In Eq. 4.3 the immersion depth of the fluorophore is ‘triangulated’ from

the quenching efficiencies of the two quenchers, assuming that the quencher

positions are known. The fluorophore needs to be located between the deep

and the shallow quencher to obtain meaningful results and it must be located

within the quenching radius of the quenchers. It has been shown that quencher

or fluorophore dynamics do not affect the immersion depth estimate signifi-

cantly [1].

In order to measure fluorophore immersion depths inside lipid bilayers,

nitroxide-labeled lipids are commonly employed. Nitroxide moieties can be

located in the headgroup and at various positions of the acyl chain and there-

fore immersion depths can be analyzed for the entire width of the membrane.

The positions of the quenchers are known from molecular dynamics (MD)

simulations and various experimental methods, like fluorescence, NMR and

electron paramagnetic resonance (EPR) [101].

4.2.4 Anisotropy

The probability of absorption of plane polarized light by a fluorophore de-

pends on the orientation of the fluorophore’s absorption dipole moment with

respect to the plane of the electric field vector of the electromagnetic wave.

The absorption probability is zero if both are perpendicular to each other and

it is maximum if the electric field vector and absorption dipole moment are

parallel. Following absorption of plane polarized light, the fluorophore rotates

during the fluorescence lifetime and finally emits a photon preferentially along

its emission dipole moment. This leads to a depolarization of the polarized

excitation light. Three effects need to be distinguished: 1. Since absorption

and emission are statistical effects, depolarization always occurs when large

numbers of randomly oriented fluorophores are observed. In the absence of

other depolarization effects, the maximum anisotropy is 0.4. 2. Depolarization

occurs when absorption and emission dipole moments are not parallel. This

gives rise to the fundamental anisotropy r0 of a fluorophore which is smaller

than 0.4. 3. Depolarization occurs because of the rotational reorientation of the

��

fluorophore during its fluorescence lifetime leading to an observed anisotropy

which is lower than the fundamental anisotropy. As a consequence, anisotropy

measurements are sensitive to reorientational dynamics on the timescale of the

fluorescence lifetime of the fluorophore. Furthermore, resonance energy trans-

fer also leads to a decrease in anisotropy [103]. For instance, the fundamental

anisotropy of Tyr residues is approximately 0.3 [104]; however, it is decreased

to 0.19 if resonance energy transfer occurs between neighboring Tyr residues

[66].

The anisotropy r is defined as [103]:

r =I‖ − I⊥

I‖+2I⊥(4.4)

where I‖ and I⊥ are the emitted fluorescence intensities when the emission

polarizer is oriented parallel or perpendicular with respect to the excitation

polarizer.

4.2.5 Stopped-flow fluorescence

In stopped-flow fluorescence experiments, two solutions are rapidly mixed and

the change in fluorescence of a reporter fluorophore is monitored. For sim-

ple binding processes, where one protein molecule binds to a patch of n lipid

molecules, a monoexponential change in fluorescence intensity is expected

with a characteristic rate constant kobs. kobs depends linearly on the lipid con-

centration [L] [139]:

kobs = kd + ka[L] (4.5)

where kd and ka are the dissociation and association rate constants of the bind-

ing. The dissociation constant KD is obtained as the ratio of kd and ka. Stopped-

flow fluorescence spectroscopy thus provides a means to quantify membrane

binding of proteins.

4.2.6 Measurements of the dissociation constant KD

The equilibrium dissociation constant KD between two molecules can be de-

termined by any experimental parameter that distinguishes the bound from the

unbound population. Examples for such parameters are the observed rate con-

stant kobs or the anisotropy, discussed in the previous sections. Moreover, for

Tyr and Trp residues the quantum yield and thus the fluorescence intensity in-

creases when they are moved from a polar to a non-polar environment [58].

The observed intensity is a weighted average between the fluorescence emis-

sion of fluorophores in the bound and unbound populations and can thus be

��

used to obtain KD. Assuming a single binding site, KD is obtained from:

I− I0

I0= Imax

[L]KD +[L]

(4.6)

where I0 is the fluorescence intensity of the unbound molecule, I is the fluores-

cence intensity at lipid concentration [L], and Imax is the fluorescence intensity

when all protein is bound [29; 108].

4.3 Nuclear magnetic resonance

Some atomic nuclei can absorb electromagnetic radiation when placed inside

an external magnetic field. This phenomenon is called nuclear magnetic res-

onance. Only a cursory overview of some aspects of NMR theory relevant to

this thesis can be given. More comprehensive information on basic NMR prin-

ciples, theory, and applications can be found in textbooks by e.g. Keeler [92],

Cavanagh et al. [26] and Kowalewski and Mäler [97].

Protons and neutrons possess an intrinsic quantum property called ‘spin’.

The spin property propagates to nuclei, which are composed of protons and

neutrons. The spin property is described by the spin quantum number S, which

can assume integer and half-integer values. Nuclei with S = 0, e.g. 12C, are not

NMR-active. For other nuclei, like 1H, 15N, 13C, and 31P, S = 1⁄2. Properties

of nuclei with higher spin quantum numbers, e.g. 2H or 14N, are not discussed

here.

Spin 1⁄2 nuclei have two degenerate energy levels, a ‘spin-up’ or α-state and

a ‘spin-down’ or β -state. If placed in an external magnetic field, the degener-

acy of the spin states is lifted and the separation of energy levels is proportional

to the magnetic field strength. The proportionality constant is called the gyro-

magnetic ratio, and it is nucleus-specific. At typical magnetic field strengths

of 9 – 20 T the separation of the energy levels for a proton is on the order of

several 100 MHz, i.e. radiofrequencies can be absorbed by the nuclei.

Radiofrequency pulses can be used to disturb the equilibrium spin state

thus creating non-equilibrium states that provide insights into structural and

dynamic properties of the molecule under investigation. NMR experiments

can provide structural information at atomic resolution and dynamic informa-

tion on a broad range of timescales. However, there are some limitations to the

investigation of biological samples by NMR. The energy difference between

the α- and β -state is low compared to thermal energy, therefore the states are

almost equally populated and thus NMR is a low-sensitivity method. There-

fore, micro- to millimolar sample concentrations are typically required. Fur-

thermore, isotope-labeling is required for many NMR applications, since the

natural abundance of the spin 1⁄2 isotopes 13C and 15N in e.g. organic molecules

��

such as proteins and lipids is 1 % and 0.4 %, respectively. Finally, ‘large’ bio-

logical molecules are a challenging to investigate by NMR. The large number

of atoms that give rise to observable signals leads to spectral crowding and

uninterpretable spectra. Even more importantly, large molecules exhibit broad

peaks due to their slow tumbling, which gives rise to high transverse relaxation

rates. This effect aggravates spectral crowding and may render peaks indistin-

guishable from noise. From a theoretical point of view, the latter problem is

the most serious one. However, experimental methods in combination with

site-specific labeling schemes can alleviate the adverse effects of line broaden-

ing.

In the following, NMR experiments that are relevant for this work are dis-

cussed. First, it is discussed how molecule size and binding can be analyzed

by translational diffusion experiments, then a brief overview of methods to

determine the three-dimensional structure of small peptides is given. Finally,

NMR relaxation methods for determining fast internal motions of molecules

and nanometer distances are presented.

4.3.1 Translational diffusion

Measurements of the translational diffusion coefficient Dt are carried out to

obtain information on the size and association of molecules [116].

The translational diffusion coefficient describes the stochastic thermal mo-

tion of molecules in solution for a system at equilibrium. The diffusion coeffi-

cient depends on the temperature of the system, its viscosity, and the morphol-

ogy of the diffusing particle. For a sphere, the translational diffusion coeffi-

cient is given by the Stokes-Einstein-Sutherland equation [51; 173]:

Dt =kBT

6πηrH(4.7)

where kB is the Boltzmann constant, T is the temperature, η is the viscosity,

and rH is the radius of the sphere, which is also called the hydrodynamic ra-

dius. Diffusion measurements can thus be used to determine particle sizes.

However, rH represents actual dimensions only for a spherical particle. De-

viations between the hydrodynamic radius and actual dimension arise due to

non-spherical shapes [5; 12; 31], due to obstruction of translational diffusion

by particles in solution [9; 89; 166], and because of hydration layers [42]. For

a non-spherical particle, rH differs from the particle’s actual dimensions, since

Dt depends on the morphology of the particle. Yet for particles that are, to a

good approximation, spherical (like micelles, small isotropic bicelles, or glob-

ular proteins) diffusion experiments can provide a good estimate of their size.

Moreover, association of molecules can be studied by diffusion experiments,

��

provide structural constraints for atoms in the molecule are performed. A suf-

ficient number of structural constraints limits the possible molecular confor-

mations to a small set of structures [26; 35; 207].

The most important structural constraints are obtained from nuclear Over-

hauser enhancements (NOEs), an effect that is based on through-space dipole-

dipole interactions between nuclei. If the energy levels of one nucleus are satu-

rated, magnetization is transferred through space via dipole-dipole interactions

to nearby nuclei. The distance dependence of the NOEs is r−6, and typically

distances up to 6 Å between nuclei can be observed. A quantitative interpre-

tation of NOEs is usually not reliable but a sufficiently large set of qualitative

spatial constraints is often sufficient for structure determination [35; 84].

The secondary structure element in which an amino acid is located af-

fects the magnetic environment and thus the chemical shifts of, e.g. Hα , Cα ,

and Cβ atoms. Secondary chemical shifts, i.e. deviations from the random

coil chemical shift for these nuclei in a specific amino acid, are characteristic

for α-helical or β -sheet environments so that the secondary structure element

in which an amino acid is located can be predicted from secondary chemical

shifts. This in turn constrains the rotamer space of the amino acids’ backbone,

giving rise to dihedral angle constraints [11; 37; 161; 202].

Other methods to obtain even more structural constraints from NMR ex-

periments exist. Examples are spatial constraints obtained from paramagnetic

relaxation enhancements (PREs) as described in section 4.3.4, orientational

constraints obtained from residual dipolar couplings [30], and hydrogen bond

constraints [26].

To obtain an NMR structure, all constraints need to be fulfilled simulta-

neously. Based on the amino acid sequence, polypeptide chains are generated

in silico, and the spatial arrangement of the amino acids is altered in an it-

erative way to fulfill the constraints. The quality of the obtained structures,

i.e. their ability to fulfill all constraints simultaneously, is evaluated by an en-

ergy function. Given that there are no major constraint violations and that the

three-dimensional structure is physically allowed, a small set of lowest energy

conformers represents the NMR structure [67; 68; 159; 159].

4.3.3 Dynamics

One particular strength of NMR is that dynamics on a broad range of timescales

can be probed. Timescales relevant for lipid motions range from very rapid

motions, like gauche-trans isomerizations in lipids, that occur on a picosecond

timescale, to very slow motions like spontaneous lipid flip-flops, that typically

happen on timescales on the order of hours (see section 2.2 and Fig. 2.2). Im-

portantly, dynamic events like protein folding, domain motions, and backbone

��

dynamics can be captured by NMR methods [81]. Here, the discussion is lim-

ited to NMR experiments that are employed to measure rapid lipid dynamics

that occur on a pico- to nanosecond timescale. Descriptions of methods that

are used to assess other types of motions are beyond the scope of this thesis.

In NMR, nano- to picosecond dynamics are investigated by measuring re-

laxation rates of non-equilibrium spin states [80; 97]. Relaxation of such states

is mediated by anisotropic interactions of spins which are in turn modulated

by molecular dynamics occurring at appropriate frequencies. For spin 1⁄2 nu-

clei the most important relaxation mechanisms are dipole-dipole interactions

between two spins, chemical shift anisotropy, which is due to the anisotropic

electronic environment at the location of the investigated spin, and paramag-

netic relaxation, which occurs for instance as an additional dipole-dipole relax-

ation mechanism between an unpaired electron and the investigated spin (see

section 4.3.4). Molecular dynamics give rise to random and local disturbances

of the magnetic field that occur at the frequency of the molecular motions.

These disturbances return non-equilibrium spin states to equilibrium and act

thus analogously to the intentional generation of non-equilibrium spin states

by radiofrequency pulses [92].

NMR relaxation parameters are linked to molecular dynamics via the spec-

tral density function J(ω). J(ω) is a probability function that describes the

distribution of motional frequencies ω in a molecule. At the same time, re-

laxation parameters can be expressed in terms of the spectral density function.

For systems with coupled spins, such as 13C–1H bonds in lipids or proteins,

the dominant relaxation mechanism is mediated through dipole-dipole interac-

tions. Then, the longitudinal relaxation rate R1 and the NOE are given by:

R1 =d2

CH10

[J(ωH −ωC)+3J(ωC)+6J(ωH +ωC)] (4.9)

NOE = 1+γH

γC

6J(ωH +ωC)− J(ωH −ωC)

J(ωH −ωC)+3J(ωC)+6J(ωH +ωC)(4.10)

where dCH =−μ0hγCγH/(8π2r3CH) and ωC and ωH are the Larmor frequencies

of 1H and 13C, respectively, μ0 is the vacuum permeability, h is Planck’s con-

stant, γH and γC are the gyromagnetic ratios of 1H and 13C, respectively, and

rCH is the 13C–1H bond length [80; 97].

Even though relaxation is driven by molecular motions, interpretation of

measured relaxation rates in terms of molecular motions is not straightforward

as long as the functional form of J(ω) is unknown. In its simplest form J(ω) is

expressed in terms of a single correlation time τc of a molecular motion, which

describes the rate of that motion:

J(ω) =τc

1+ω2τ2c

(4.11)

��

As discussed in the next section, this form of the spectral density function

can be used in paramagnetic systems to describe the reorientation of the vector

between the nucleus under investigation and a free electron.

The model-free approach allows for a more comprehensive description of

fast molecular motions. The motions are characterized in terms of an order

parameter S that describes the amplitude of motion, and by a correlation time

τ that describes the rate of motion. This approach was originally designed to

describe internal motions in proteins [119; 120] and detergents [69; 196; 197]

but it has been extended to characterize the motions that lipids undergo in

membranes [5; 52]. The extended spectral density function is given by:

J(ω) =

(S2

loc−S2locS2

lip

)τlip

1+ω2τ2lip

+

(1−S2

loc

)τT

1+ω2τ2T

(4.12)

where τT = τlipτloc/(τlip + τloc) and S2loc and τloc are interpreted as the am-

plitude and rate of motion of 13C–1H bond vectors, and S2lip and τlip refer to

the motions of the entire lipid. Both local and lipid motions occur on pico-

to nanosecond timescales and thus mediate spin relaxation. One central as-

sumption in the derivation of Eq. 4.12 is that local and lipid motions occur on

different timescales and are therefore independent of each other [119; 120].

An illustration of a possible interpretation of such motions in lipids is given

in Fig. 4.2. The rate of motion describes the reorientation time of the motional

vector, which is the 13C–1H bond vector for the local motion and the long axis

of the lipid for the lipid motion [97], and the amplitude of motion is described

by the aperture of a cone, in which the motional vector reorients.

4.3.4 Distance measurements using paramagnetic relaxation enhance-ments

In the presence of non-covalently bound unpaired electrons, paramagnetic re-

laxation enhancement (PRE) is a dipole-dipole relaxation mechanism that oc-

curs in addition to nuclear dipole-dipole relaxation when a free electron is

located in proximity of the observed nucleus. PREs are observed for both the

transverse and the longitudinal relaxation rates [36; 143]. The paramagnetic

relaxation enhancement Γ1 of the longitudinal relaxation rate R1 is described

by the Solomon-Bloembergen equation and is at sufficiently high magnetic

field given by [15; 16; 167; 168]:

Γ1 =2

5

( μ0

4π

)2

γ2Cg2

e μ2BS(S+1)J(ω)< r−6 > (4.13)

where μ0 is the magnetic permeability of vacuum, γC is the gyromagnetic ra-

tio of carbon, ge is the electron g-factor, μB is the Bohr magneton, and S is the

��

5. Results and Discussion

EYPHKA! num = Δ+Δ+Δ.—Carl F. Gauß

Membrane-associated GT-B GTs have the peculiar property that they catalyze

the formation of a glycosidic bond between a hydrophilic sugar donor sub-

strate and a lipid acceptor molecule. Moreover, peripheral GT-B GTs can ex-

change between a hydrophilic and a membrane environment. The GT WaaG is

a representative of this class of proteins. Since deletion of the gene encoding

WaaG has deleterious effects on the outer membrane of Gram-negative bacte-

ria, WaaG is considered to be a potential antibiotic target [2; 3; 105].

To study membrane interactions of proteins, the system under investiga-

tion needs to be amenable to the methods employed. Solution-state NMR in

particular puts high demands on the sample, one central limitation being that

larger objects give rise to broader linewidths leading to spectral overlap and

eventually to line broadening into noise. Therefore, membrane mimetics need

to be employed, which are amenable to NMR investigations, and which, at the

same time, faithfully mimic natural membranes.

Two central questions are addressed in the following: 1. Can faithful mem-

brane mimetics be produced that are useable in solution-state NMR studies and

what are the characteristics of such systems? 2. How do membrane-associated

GT-B GTs interact with membranes?

5.1 Mimicking plant and bacterial membranes

Membrane mimetics are necessarily compromises between the constraints of

the experimental method and the requirement for faithful emulation of physi-

ological conditions. It is expected that membrane interactions of proteins are

more native-like the closer a lipid bilayer mimics physiological membranes.

For solution-state NMR, small isotropic bicelles constitute such a compromise.

��

Bicelles are small enough to be amenable to solution-state NMR experiments,

but at the same time, the lipid composition can be modified [195].

In Papers I and II, bicelles that mimicked plant and bacterial membranes

were produced and their characteristics analyzed. These characteristics are:

1. size of the bicelle, 2. lipid composition, and 3. pico- to nanosecond lipid

dynamics. Furthermore, in Paper III a method was developed to determine the

immersion depths of lipid carbon atoms in membranes.

5.1.1 Size of bicelles

The bicelle size can be determined from the translational diffusion coefficient

of lipids which have a negligible monomer solubility in water. As discussed

in section 4.3.1 the hydrodynamic radius obtained from translational diffu-

sion measurements reflects the radius that a sphere with the observed diffusion

coefficient would have, i.e. diffusion measurements do not report on the mor-

phology of the investigated particle and dimensions are approximately accurate

only if it is roughly spherical in shape.

In Paper I, the diffusion coefficients of bicelles enriched in the galactolipids

MGDG and DGDG were determined. It was found that enrichment in both

MGDG and DGDG leads to an increase in bicelle size when compared to bi-

celles that only contain DMPC as the lipid component. The difference is at-

tributed to the bulkier headgroup size and the acyl chain composition (see Fig.

2.4). The dominant species of MGDG and DGDG contained 18:3-16:3 acyl

chains while DMPC acyl chains are saturated and 14 carbons long. It was also

observed that DGDG containing bicelles were on average smaller than MGDG

containing bicelles despite the bulkier headgroup of DGDG. The most signif-

icant difference between DGDG and MGDG is that the former is a bilayer

forming lipid while the latter is not [47]. Instead, MGDG forms an inverted

hexagonal phase. It is therefore speculated that the difference in hydrodynamic

radius actually reflects morphologic differences. Due to the bilayer forming

propensities of both DMPC and DGDG a lipid mixture of these two lipids

may still to a good approximation conform to the ideal bicelle model (see Fig.

2.7). On the other hand, the lipid bilayer in bicelles that contain MGDG might

be under considerable stress, leading to morphological changes of the bicelle

which result in smaller diffusion coefficients of MGDG containing bicelles

compared to DGDG containing bicelles.

5.1.2 Lipid composition of bicelles made from E. coli phospholipids

In Paper II, E. coli inner membranes were mimicked using E. coli phospho-

lipid extract as the lipid component in bicelles. The phospholipid composition

was analyzed by 31P-NMR, in which distinct peaks for all three phospholipids

��

that are found in E. coli (PE, PG, and CL) were observed. When cultures were

grown at 30 ◦C, a phospholipid composition was obtained that corresponds to

previously published data [200]. PE lipids are most abundant (~80 %) and the

amount was unchanged when cells were grown at 20 ◦C. In contrast, the com-

position of anionic lipids changed at lower growth temperature. This effect is,

however, not attributed to the difference in growth temperature but rather to the

difference in growth phase. Cells were harvested after 24 hours, at which point

cultures grown at 20 ◦C are in the exponential phase, whereas cultures grown

at 30 ◦C are in the stationary phase. In the stationary phase, CL is formed

at the expense of PG [73]. The observed increase in CL, which is accom-

panied by a decrease in PG in cultures grown at 30 ◦C compared to cultures

grown at 20 ◦C, is therefore a growth phase effect. At growth temperatures

above 30 ◦C, the phospholipid composition changed dramatically, an effect

that can be attributed to the temperature-dependent expression of the phos-

phatidylserine synthase PssA, an essential enzyme for the synthesis of PE. At

higher temperatures, the expression of PssA is impaired leading to a decrease

in PE content in the membranes. This effect is specific for the AD93 E. colistrain used in this study [44; 91].

The acyl chain composition of the E. coli phospholipid extract was mon-

itored by 13C-NMR and it was found that the amount of unsaturated lipids

increases at lower growth temperature. E. coli bacteria regulate membrane flu-

idity via acyl chain modifications and thus maintain the membrane in a liquid-

crystalline phase [43; 110]. An increase in amount of unsaturations decreases

the phase transition temperature, i.e. the observed change in amount of un-

saturations is a temperature effect. Finally, PE can be depleted altogether in

membranes by using a strain that cannot express PssA [44].

5.1.3 Rapid dynamics in bicelles

In Papers I and II, lipid dynamics that occur on a pico- to nanosecond timescale

were analyzed for bicelles containing galactolipids, and bicelles consisting of

E. coli phospholipid extract, respectively. As described in section 4.3.3, the

model-free approach allows for obtaining an amplitude of motion, given by

the squared order parameter S2, and a rate of motion, given by the correlation

time τ , for 1H–13C bonds in lipids (S2loc and τloc) and for the entire lipid (S2

lipand τlip).

While S2loc parameters for MGDG and DGDG headgroup carbons were on

the same order as for glycerol carbons, S2loc parameters were somewhat lower

for headgroup carbons of PE and PG in E. coli bicelles. At the same time,

correlation times for galactolipid headgroup carbons were about one order of

magnitude larger than for headgroup carbons of E. coli phospholipids. Thus,

��

the rigidity of 1H–13C bonds in the headgroup depends on the size and struc-

ture of the headgroup. Larger headgroups show an increased rigidity and the

same is true for the rigid ring structure of galactose headgroups. The glycerol

region of the lipid is rigid for all lipid types and correlation times are several

hundred picoseconds. It should be noted that the dynamic parameters are not

entirely reliable for the glycerol and the galactolipid headgroup carbons since

the assumption of timescale separation (see section 4.3.3) between local and

lipid motions is violated. In consequence, the order parameters and correlation

times are over- and underestimated, respectively [119; 120].

The rigidity of the 1H–13C bonds in the upper acyl chain is comparable to

the carbons in smaller headgroups like PE or PG. The rigidity decreases down

the acyl chain to almost unrestricted motions of the terminal methyl carbons.

This trend is broken by acyl chain modifications. While a slight local increase

in rigidity is observed for unsaturated carbon atoms, S2loc is increased signifi-

cantly for carbons in cyclopropane rings.

The model-free approach can also provide order parameters and correla-

tion times that report on lipid dynamics. While these parameters were pre-

sumed and fixed in Paper I, they were treated as variables in Paper II. For lipids

in E. coli bicelles, S2lip was determined to be about 0.5 and the lipid correlation

time was found to be 800 ps. The rigidity is thus on the upper end of what

has been reported for other lipids [109], while the correlation time is com-

parable to correlation times found by EPR for lipids in the liquid-crystalline

phase [24; 130]. It can therefore be concluded that the bicelle membrane is in

a liquid-crystalline phase similar to the in vivo situation.

5.1.4 Immersion depths of carbon atoms in membranes

Immersion depths of molecules in membranes are an important parameter

for the understanding of biological processes that occur in or at membranes.

PREs can be employed to measure immersion depths since they are distance-

dependent, as described in section 4.3.4. They are particularly suited for mem-

brane immersion depth measurements since the distance over which PREs are

effective (up to 30 Å) [82] is on the same order as the thickness of a typical

membrane, which is about 40 Å [98–100; 146]. In Paper III a method was

developed to measure the immersion depths of lipid carbon atoms in bicelles

using PREs of the longitudinal relaxation rate R1.

Bicelles were doped with four different paramagnetically labeled lipids

that contained paramagnetic moieties in the headgroup and at various posi-

tions in the acyl chain, and PREs were measured for lipid carbon atoms. The

carbon atom PREs reflect the localization with respect to the paramagnetic

moieties inside the membrane and thus a qualitative estimate of the immersion

��

depths of the carbon atoms can be obtained [33; 128]. For example, when the

paramagnetic label was located at the far end of the acyl chain, the observed

PREs were strongest for the terminal acyl chain carbon atoms and weakest for

headgroup carbon atoms.

Carbon immersion depths were quantified employing immersion depths

and immersion depth distributions of the paramagnetic moieties, which are

known from MD simulations [101]. Experimental PREs were measured in the

presence of four different paramagnetically labeled lipids, which provided four

distance constraints for the immersion depth of each carbon atom. Theoretical

PREs were computed for a simple bicelle model where each lipid was rep-

resented by a cylinder and the detergent molecules in the rim were neglected.

Carbon atoms were placed at random immersion depths inside the center of the

lipid cylinder and one lipid was randomly chosen to carry the paramagnetic

label, located at a position obtained from the known immersion depths and

immersion depth distributions. From this model, theoretical PREs were cal-

culated using Eq. 4.13 by summing up all contributions from identical carbon

atoms. The theoretical PREs depend only on the carbon immersion depths1,

i.e. from a least square fit of the theoretical to the experimental PREs, the

carbon immersion depths were obtained. The fitting algorithm did not provide

accurate immersion depths for only one simulated bicelle, since the measured

PREs were ensemble averages, i.e., the conformational space of the possible

locations of the paramagnetic label had to be sampled. Therefore, the fitting

procedure was repeated for hundreds of bicelles. The theoretical PREs and fit-

ted carbon immersion depths were the means of the in silico bicelle ensemble.

Using this approach it was found that the headgroups in zwitterionic and E.coli bicelles are approximately 20 Å away from the bilayer center, the glycerol

carbons are roughly located at 16 Å from the bilayer center, and the first acyl

chain carbons are right below. The terminal acyl chain carbons are closest to

the bilayer center and acyl chain modifications (unsaturations and cyclopropa-

nations) in E. coli bicelles are about 8 Å away from the bilayer center.

Somewhat crude assumptions were made in the calculation of theoretical

PREs. For instance, a simple form of the spectral density function (see Eq.

4.13) was assumed, the bicelle model consisted of cylindrical lipids, and de-

tergents were neglected altogether. Nonetheless, the results are in good agree-

ment with results obtained by other methods [77; 99; 138; 144; 165].

The method presented in Paper III provides a means to quantitatively de-

termine immersion depths in membranes with an accuracy of 3 – 5 Å, and can

be extended to immersion depth measurements of membrane proteins or other

membrane-associated molecules.

1Eq. 4.13 has the form Γ1 =Cr−6 where C is a constant. Since not all parameters that constitute

C were known, C was treated as an additional, global variable in the fitting procedure.

��

5.1.5 Bicelles as a plant and bacterial membrane mimetic

Solution-state NMR, in particular, puts high demands on samples of biological

molecules, as the investigated molecules must not be too big. Since natural

membranes are large aggregates of proteins and lipids, membrane mimetics

that are sufficiently small are required to conduct NMR experiments.

Bicelles maintain a lipid bilayer which is dissolved in aqueous solution by

detergents located in the rim of the bicelles. In this work, it was shown that

the lipid composition can be adjusted in such a way that it faithfully mimics

plant or bacterial membranes. A major limitation for the production of bicelles

is that the lipid mixture must be prone to form bilayers. It was, for instance,

not possible to produce bicelles from synthetic lipids with the same phospho-

lipid composition as E. coli inner membranes. This indicates that a delicate

equilibrium between bilayer and non-bilayer forming lipids exists in natural

membranes. However, by choosing strain, growth temperature, and time point

of cell harvest, the final lipid composition of bicelles can be tuned to some

degree.

Properties of the bicelle, the bicelle lipids, and individual carbons of the

lipids can be investigated in some detail using NMR methods. Analysis of the

dynamics showed that the lipids are in a liquid-crystalline phase, and that lipid

dynamics display a large degree of heterogeneity along the membrane normal,

from very restricted local motions of the glycerol backbone to almost unre-

stricted motions at the terminus of the acyl chains. Moreover, from measure-

ments of carbon immersion depths, lipids could be localized inside the mem-

brane. As expected, their headgroup and glycerol carbons are located at or near

the surface of the membrane, and the acyl chains point towards the center of the

membrane. At the same time, the uncertainties in the immersion depths esti-

mates were high, especially for unsaturated and methyl carbons. This indicates

a broad distribution of these carbons inside the membrane. Taken together the

experimental evidence suggests that the lipids in bicelles behave similarly to

what can be expected for lipids in physiological membranes.

There are, however, drawbacks when using bicelles for protein-lipid in-

teraction studies. Firstly and most importantly, often, it cannot be ruled out

that the protein under investigation interacts with the detergent molecules in

the rim. Such interactions may, similar to interactions with detergent micelles,

distort the structure and impair the function of the protein. Secondly, the mor-

phology of the lipid bilayer may be affected when non-bilayer forming lipids

are introduced into the lipid mixture. Finally, it is still unclear to what degree

and under which conditions detergents and lipids segregate in bicelles into a

detergent rim and a lipid bilayer region.

��

5.2 Monotopic and peripheral glycosyltransferases

Based on studies of monotopic and peripheral GTs a model of their mem-

brane interaction has been proposed (see Fig. 3.2). In this model, the N-

terminal domain is anchored to the membrane while the C-terminal domain

binds less strongly to the membrane, undergoing an unspecified motion that

is required to accommodate the donor and acceptor molecules at the active

site in the cleft between the two domains [2; 57]. However, in membrane-

associated GTs this motion has not yet been observed. Moreover, it is not

known whether membrane-associated GTs bind certain lipids specifically and

membrane affinities are largely unknown.

Here, the applicability of this model to the GT WaaG was tested and the

open questions outlined above were addressed.

5.2.1 Anchoring of WaaG to membranes

To study the anchoring mechanism of WaaG to membranes, a potential mem-

brane interacting region of WaaG (MIR-WaaG) was identified and studied in

Paper IV. The selection of the segment was based on three properties: 1. it is

exposed in the crystal structure, 2. it contains positively charged residues, and

3. it contains Tyr residues that have previously been shown to be important for

membrane interaction [95].

It was found by fluorescence anisotropy and quenching studies that MIR-

WaaG interacts with vesicles enriched in anionic charge but does not bind

to zwitterionic vesicles. As evidenced by CD spectra, MIR-WaaG adopts an

α-helical structure in anionic vesicles, as expected from the crystal structure

[131], whereas it is random coil in buffer and in zwitterionic vesicles. More-

over, employing the parallax method (see section 4.2.3), the average immer-

sion depth of the Tyr residues could be determined to be ~16 Å. They are thus

located at a similar membrane immersion depth as the lipid glycerol backbone

[99].

To get a more detailed understanding of the structure of MIR-WaaG and

the localization of its residues within a membrane, the NMR structure was

determined in dodecylphosphocholine (DPC) micelles. The NMR structure

of MIR-WaaG agrees remarkably well with the equivalent segment in the

crystal structure. The localization of individual residues was determined

qualitatively by investigating the peak intensity reduction upon addition of

the water-soluble paramagnetic molecule Gd3+ diethylenetriamine pentaacetic

acid-bismethylamide (Gd(DTPA-BMA)) [10; 143; 211]. The intensity reduc-

tion is stronger for solvent exposed residues than for residues located inside the

DPC micelle. It was found that a central part of MIR-WaaG is located inside

��

the micelle while its N- and C-terminus are either solvent exposed or at the mi-

celle surface. These data were supported by hydrogen exchange experiments

[76; 107] that showed NOE cross-peaks between Tyr residues and water pro-

tons and hydrogen exchange between water and guanidinium protons in Arg

residues. From this information, the localization of MIR-WaaG in micelles

could be derived (see Fig. 5.1a). Based on the combined data, a model of the

membrane interaction of WaaG was proposed (see Fig. 5.1b), in which the N-

terminal domain anchors WaaG to the membrane via electrostatic interactions

between MIR-WaaG and anionic lipids. The interactions are potentially en-

hanced by Tyr residues located at the membrane’s interface and hydrophobic

residues pointing towards the center of the membrane.

5.2.2 Membrane interaction of full-length WaaG

WaaG has been suggested to be a monotopic membrane protein [2], and, as

discussed above, the N-terminal domain anchoring segment MIR-WaaG inter-

acts with lipids in a charge-dependent way. However, no information on the

membrane affinity of full-length WaaG was available. Therefore, these affini-

ties were determined in Paper V. Additionally, quantitative membrane-binding

studies could potentially reveal lipid-specificity of the binding as well as do-

main dynamics.

The interactions between full-length WaaG and LUVs of varying lipid

compositions were investigated by stopped-flow fluorescence. To be able to

use fluorescence spectroscopy, Trp residues were introduced into exposed re-

gions of either the N- or the C-terminal domain in positions likely to be affected

by membrane binding of WaaG. Upon interaction with lipids, the environment

of the Trp residues changed leading to a change in fluorescence intensity that

could be observed in real time. The change in fluorescence intensity is charac-

terized by an observed rate constant kobs, which depends on the lipid concen-

tration. From measurements of kobs at various lipid concentrations, association

and dissociation rate constants were obtained [139].

As expected from previous studies [136], WaaG does not interact with

zwitterionic vesicles, but it does interact with vesicles enriched with either

of the two anionic lipid species found in E. coli inner membranes, PG and CL.

However, WaaG does not preferentially bind either of these lipids; however,

the affinity is determined by the surface charge density of the vesicles. WaaG’s

affinity for vesicles made of E. coli phospholipid extract with 20 % anionically

charged lipids [114; 200], is lower than for vesicles containing 40 % synthetic

anionic lipids. Moreover, no difference in affinity was observed between WaaG

mutants that contained a Trp-mutation in either the N- or C-terminal domain.

In principle, different affinities of the flexibly linked domains, and thus mo-

��

tions between the domains or with respect to the membrane, could be observed

by this assay as long as the Trp environment is affected by the motions. This

would be the case if the C-terminal domain performs an ‘up-and-down’ move-

ment while the N-terminal domain is permanently anchored. However, equal

affinities, as observed here, do not necessarily mean that the two domains do

not move with respect to each other. For instance, it has been hypothesized

that the C-terminal domain performs a rolling motion [2; 57], which might not

change the environment of the Trp residue. Moreover, the C-terminal domain

might only move in the presence of the natural nascent LPS substrate.

The membrane affinities of very few membrane-associated GTs have thus

far been reported. Among them are the monotopic GT MGS which binds mem-

branes with nM – pM affinity [110] and the peripheral GT PimA which binds

membranes with μM affinity [153]. It is found for WaaG that its affinity for

anionic membranes is on the order of μM, i.e. WaaG seems to be a peripheral

membrane protein rather than a monotopic one, as previously suggested. It

is, however, unclear whether this observation is also true in vivo, since con-

vincing data on the in vivo concentration of WaaG have not been reported

[157; 169; 192].

��

6. Conclusions

The travails of the mountains lie behind us.Before us lie the travails of the plains.—Berthold Brecht

One of the aims of this work was to increase the understanding of the func-

tion of a peculiar class of proteins—membrane-associated GT-B GTs. The

interest in membrane-associated GTs has two aspects. Firstly, membrane-

associated GTs operate at the membrane surface and their function is to bring

two molecules with very different physical properties, a hydrophilic donor

sugar and a lipid acceptor molecule, into proximity so that a glycosidic bond

can be formed. At the same time, many of these GTs themselves seem to be

peripheral membrane proteins, i.e. hybrids between soluble and permanently

membrane-bound proteins. Secondly, some peripheral or monotopic GTs are

involved in the synthesis of glycolipids like LPS. Impairment of their func-

tion can be lethal to bacteria and these GTs are therefore promising antibiotic

targets. This demonstrates the relevance of another focus of this work that re-

volves around another peculiar class of biological molecules—lipids. The im-

portance of a properly ‘tuned’ and functioning membrane to a living organism

cannot be underestimated. Defects in the lipid composition of the membrane,

in the structure of the lipids, or of the lipid dynamics can lead to severe mal-

function of membrane-related biochemical processes.

The research described in this thesis shed light on the mode of membrane

interaction of a representative of the class of membrane-associated GT-B GTs,

WaaG. Electrostatic interactions play a key role in binding the N-terminal do-

main of WaaG to membranes. WaaG does not discriminate between anionic

lipids, but it senses the surface charge density of the membrane. The affinity

for membranes has been reported for very few GTs and here we find that the

membrane affinity of WaaG is rather weak; it may thus be categorized as a

peripheral membrane protein.

��

Lipids undergo a broad range of motions inside membranes. Rapid lipid

motions were probed here by NMR relaxation in bicelles that faithfully mim-

icked the membrane composition of bacterial and plant membranes. The glyc-

erol backbone is rather rigid and therefore mostly subjected to lipid wobbling

and rotational reorientation. Motions in the distal lipid regions, in particular at

the acyl chain terminus, are much more rapid and dominated by trans-gauche

isomerizations and bond reorientations. Lipid headgroup flexibility depends

on the size of the headgroup. Small headgroups undergo more rapid and less

restricted motions than larger headgroups. Furthermore, by employing PREs

an ‘atomic ruler’ was developed that allows the measurement of the immersion

depth of lipid carbon atoms in membranes.

The studies of lipid properties in realistic membrane mimetics and the in-

vestigations of the membrane interactions of WaaG were largely conducted

separately. However, a synthesis of the approaches would be desirable to get a

better understanding of WaaG’s membrane interaction in more realistic mem-

branes. In particular, it would be desirable to apply NMR techniques, which

are particularly suited to understand dynamic processes, to systems containing

WaaG and natural E. coli lipids. However, the resulting aggregates are very

large, even if small isotropic bicelles are employed so that less conventional

NMR techniques are required to conduct such investigations. One approach

would be to site-specifically label WaaG, and investigate properties of the la-

bel or the effects the label has on lipids in the membrane. Such studies could

lead to a more profound understanding of how peripheral GT-B GTs interact

with lipids and their substrates. On a larger scale, it is not understood if and

how different GTs act together at the membrane surface to synthesize LPS.

Moreover, some GT-B GTs seem to undergo large conformational changes

upon membrane and substrate binding, making this class of proteins a worth-

while target for further studies. Finally, multi-resistant bacteria have become

an imminent threat to society. Therefore, it is absolutely crucial to identify new

targets for antibiotics, to understand their function, and to impair it. WaaG and

other LPS-synthesizing proteins are such targets, and it is therefore worthwhile

to continue the effort to understand their mode of action at the membrane sur-

face.

As mentioned in the introduction, the scientist’s endeavor is not unlike the

mountaineer’s venture. At this point of the thesis a mountain range has been

negotiated and before us lie the plains that promise rest. Or do they? In the

distance, new undiscovered mountain ranges loom—and the grasslands are un-

charted territory. Some blank spots on the map of knowledge may have been

filled in during this endeavor. Yet, the mountaineer pushes on, and the scientist

asks new questions.

��

Populärvetenskaplig Sammanfattning

Som när en Grip i mörkret genom öcknarMed vingadt lopp kring fält och sanka dalarFörföljer Arimaspen, som försåtligtHar hans förtrodda skatt från honom röfvat.—John Milton, Det förlorade paradiset

Man kan tänka sig proteiner som små, biologiska maskiner – de minsta maski-

nerna som finns, endast en miljondel av en millimeter i storlek. Tillsammans

med fetter och kolhydrater är de livets viktigaste byggstenar. Dessa små par-

tiklar kallas också för molekyler. Proteiner är mycket komplicerade molekyler,

som är sammansatta av tusentals atomer på ett komplicerat sätt till en välord-

nad struktur som katalyserar, det vill säga genomför, kemiska reaktioner med

en oerhörd precision. Ett protein kan katalysera en kemisk reaktion, som an-

nars skulle ta dagar eller veckor, på bråkdelen av en sekund. Något som är även

mer förvånande är att proteiner kontrollerar alla processer som sker i en cell,

så att livet blir möjligt. Utan proteiner finns det inget liv.

Men för att liv ska vara möjligt behövs det mer än ett par proteiner. En an-

nan viktigt sorts molekyler är lipider, som är en särskild grupp av fetter. Lipider

är den viktigaste beståndsdelen i membraner, som man kan föreställa sig som

husväggar. Så som husväggar skyddar människor från väder, omger membra-

ner celler och skyddar dem från deras omgivning. Membraner skapar en inne-

och en utemiljö. På cellens insida finns det livsviktiga proteiner och andra mo-

lekyler. Från utsidan får cellen näring, t. ex. kolhydrater, men där finns också

giftiga molekyler, som cellen måste skydda sig mot.

Membraner är uppbyggda av lipider. Lipiderna måste framställas och här

kommer proteiner in i bilden. Speciella proteiner finns för att framställa olika

lipider som är nödvändiga för att bygga upp ett membran. Ett av dessa prote-

iner har studerats i denna avhandling. Det kallas för WaaG och kommer ifrån

bakterien Escherichia coli. Vi undersökte hur proteinet utför sitt arbete i bak-

terien, det vill säga i en organism som består av en enda cell.

WaaG katalyserar ett enda, men mycket viktigt steg, i tillverkningen av en

speciell sorts lipider i bakterien. Detta är ett skäl till, varför vi vill förstå hur

��

proteinet fungerar. Som bekant finns bakterier som orsakar sjukdomar och mot

dessa finns antibiotika som dödar bakterien. Ett sätt att döda bakterierna är att

förstöra proteinerna som är involverade i uppbyggandet av membranen, det vill

säja t. ex. WaaG. Bakterierna kan inte leva med ett skadat eller förstört mem-

bran. För att kunna inhibera ett protein som t. ex. WaaG måste vi dock förstå

hur det fungerar. Men det finns ytterligare ett skäl till, varför vi är intresserade

av WaaG: Proteinet är inte bara inblandat i uppbyggandet av cellens membran,

det har egenskapen som gör att det liknar ett mytiskt fabeldjur: WaaG är en

grip.

Vem är gripen?

De antika grekerna berättar historier om en varelse, som har ett lejons kropp

och svans, och en örns huvud och vingar: en grip. Gripar är luftens och landets

kreatur. Tack vare den särskilda kombinationen av örn- och lejonegenskaper,

känner de sig hemma i båda omgivningarna. WaaG är en faktiskt existerande,

mikroskopisk analogi till en grip – ett kreatur av två världar. Tack vare dess

särskilda kombination av egenskaper känner proteinet sig hemma i två omgiv-

ningar som inte kunde vara mer olika varandra: vatten och lipider, som är en

undergrupp av fetter.1 Vatten och fett går inte att blanda. Det vet alla som har

någonsin försökt att diska en infettad stekpanna med vatten.

Hur kan ett protein trivas i båda omgivningarna? Detta är svårt att före-

ställa sig och därför undersökte vi WaaG noggrannare och hittade ett område

i proteinet, som kan docka som ett ankare till membranen (se figur 6.1). Detta

ankare orsakar inga problem när WaaG befinner sig i vatten, men så fort det

finns en lipidmembran i närheten dockar ankaret i WaaG till membranet.

Var lever gripen?

I den antika litteraturen finns det olika teorier om var gripar lever. Några för-

fattare tror att de vistas i fjärran länder i norden, andra misstänker att de bor

i fjärran östern. De flesta författare är dock överens om att griparna attraheras

starkt av guld. De påstås faktiskt vara väktare av otroliga skatter.

WaaG attraheras i sin egen del av en skatt: av vissa lipidmolekyler. Dessa

lipider måste vara negativt laddade, så att ankaret, som är positivt laddat, bin-

der till dessa lipider och på detta sätt till membranen. Ankaret fungerar inte

utan negativt laddade lipider.

1Ordet “lipid” härstammar från det grekiska ordet för fett: λ íπoς .

��

Existerar gripen på riktigt?

WaaG är alltså ett hybridprotein, som gillar både vatten och lipider, men det

måste befinna sig vid ett lipidmembran för att genomföra sitt jobb, vilket är

att bygga nya lipider. WaaG ankrar till negativt laddade lipider i membranen

genom ett positivt laddad ankare. Och medan detta händer, dansar lipiderna sin

dans och nya framställas om och om igen.

Alla dessa observationer och resultat bygger på undersökningar som gjor-

des i ett provrör. Men händer allt det här även i levande bakterier? Vi kan med

ganska stor säkerhet säga att några av iakttagelserna även försiggår i levande

celler. Lipider rör sig i alla celler. WaaG binder elektrostatiskt till membraner.

Men binder det jämnt och under alla förhållanden? Rör sig WaaG också? Det

vet vi inte än. Vi förstår nu bättre hur WaaG och lipiderna fungerar tillsam-

mans, men varje svar leder till nya frågor. Vetenskap handlar om att utvidga

kunskap och att få ett större underlag för en bättre förståelse av världen om-

kring oss. Med denna avhandling har kunskapen om en grupp proteiner som

beter sig mycket egendomligt utvidgats. Dessutom förstår vi bättre, hur lipi-

der rör sig och hur de interagerar med proteiner. Detta betyder att ännu en vit

fläck på kunskapens karta, som nämndes inledningsvis i denna avhandling, kan

suddas ut.

��

Populärwissenschaftliche

Zusammenfassung

Wie wenn ein Greif mit schnellbeschwingtem LaufBergauf, bergab, durch Wüste, Wald und MoorDen Arimasp, der ihm, ein schlauer Dieb,Sein streng bewachtes Gold entwandt, verfolgt.—John Milton, Das verlorene Paradies

Proteine sind Eiweiße und man kann sie sich wie kleine, biologische Maschi-

nen vorstellen – die kleinsten Maschinen, die es gibt, nur ein Millionstel ei-

nes Millimeters klein. Zusammen mit Fetten und Kohlenhydraten sind sie die

wichtigsten Bausteine des Lebens. Diese winzigen Teilchen nennt man auch

Moleküle. Proteine sind sehr komplizierte Moleküle, die sich auf eine kom-

plizierte Art und Weise aus Tausenden von Atomen zu einer Struktur zusam-

mensetzen, die mit großer Präzision chemische Reaktionen katalysieren, d.h.

durchführen, kann. Ein Protein kann eine chemische Reaktion, die ansonsten

Tage oder Wochen dauern würde, in Bruchteilen einer Sekunde katalysieren.

Mehr noch: Proteine kontrollieren alle Prozesse, die in einer Zelle vor sich ge-

hen, sodass Leben möglich wird. Ohne Proteine gäbe es kein Leben.

Allerdings wird zum Leben noch mehr benötigt als nur Proteine. Ein weite-

re wichtige Klasse von Molekülen sind Lipide, die eine besondere Gruppe von

Fetten sind. Lipide sind der wichtigste Bestandteil von Membranen, die man

sich wie Hauswände vorstellen kann. Genauso wie Hauswände Menschen vor

dem Wetter schützen, umgeben Membranen Zellen und schützen sie vor ihrer

Umgebung. Membranen schaffen ein „Innen“ und ein „Außen“. Im Inneren

der Zelle befinden sich lebenswichtige Proteine und andere Moleküle. Von au-

ßerhalb bezieht die Zelle Nahrung, z.B. Kohlenhydrate, aber dort befinden sich

auch giftige Moleküle, vor denen sich die Zelle schützen muss.

Membranen bestehen aus Lipiden. Lipide müssen gebaut werden und an

dieser Stelle kommen Proteine ins Spiel. Es gibt spezielle Proteine, die Lipide

herstellen. Eines dieser Proteine wurde hier untersucht. Es wird WaaG genannt

und findet sich in dem Bakterium Escherichia coli. Wir haben untersucht, wie

��

das Protein seine Arbeit in Bakterien, also Lebewesen, die aus nur einer einzi-

gen Zelle bestehen, verrichtet.

WaaG katalysiert einen einzigen, aber äußerst wichtigen Schritt in der

Produktion von bestimmten Lipiden in Bakterien. Das ist ein Grund, weswe-

gen wir verstehen wollen, wie es funktioniert. Manche Bakterien verursachen

ernsthafte Krankheiten. Antibiotika heilen solche Krankheiten, indem sie die

Bakterien töten. Eine Möglichkeit, dies zu tun, ist, die Proteine zu zerstören,

die mit dem Aufbau von Membranen zu tun haben, also zum Beispiel WaaG.

Mit einer beschädigten oder zerstörten Membran können Bakterien nicht le-

ben. Bevor wir die Funktion von WaaG unterbinden können, müssen wir je-

doch verstehen, wie es funktioniert. Es gibt aber noch einen weiteren Grund,

aus dem wir an WaaG interessiert sind: Es ist nicht nur in den Aufbau der

Zellmembran involviert, es hat gleichzeitig Eigenschaften, die es einem my-

thischen Fabelwesen ähneln lassen: Es ist ein Greif.

Wer ist der Greif?

Die alten Griechen erzählen Geschichten von einem Wesen, das den Körper

und Schwanz eines Löwen hat und den Kopf und die Flügel eines Adlers: der

Greif. Greife sind Kreaturen der Luft und des Landes. Dank einer speziellen

Kombination von Adler- und Löweneigenschaften, fühlen sie sich in beiden

Umgebungen zu Hause. WaaG ist eine tatsächlich existierende, mikroskopi-

sche Analogie des Greifs – eine Kreatur zweier Welten. Dank einer speziellen

Kombination von Eigenschaften fühlt es sich in zwei Umgebungen zu Hause,

die unterschiedlicher nicht sein könnten: Wasser und Lipide, die eine Unter-

gruppe der Fette sind.1 Wasser und Fett kann man nicht mischen. Das weiß

jeder, der schon einmal versucht hat, eine fettige Pfanne mit Wasser zu säu-

bern.

Wie kann sich ein Protein in beiden Umgebungen wohlfühlen? Dies ist

schwer vorstellbar und deshalb haben wir uns WaaG genauer angeschaut und

einen Bereich gefunden, der wie ein Anker an die Membran andocken kann

(siehe Abbildung 6.2). Dieser Anker verursacht keine Schwierigkeiten, wenn

WaaG sich in Wasser befindet, aber sobald eine Lipidmembran in der Nähe ist,

dockt er WaaG an die Membran an.

Wo lebt der Greif?

In der antiken Literatur gibt es verschiedene Theorien dazu, wo Greife leben.

Manche Autoren vermuten sie in fernen Ländern im Norden, andere spekulie-

1Das Wort „Lipid“ stammt von dem griechischen Wort für Fett: λ íπoς .

��

Einige der Atome sitzen sehr tief innerhalb der Membran und kommen nur

sehr selten an ihre Oberfläche. Andere befinden sich fast immer an der Ober-

fläche der Membran. Um das genauer zu messen, haben wir eine Methode ent-

wickelt, mit der man die Eintauchtiefe von Lipidatomen in die Membran mes-

sen kann. Die Methode kann man sich wie ein „atomares Lineal“ für sehr klei-

ne Abstände vorstellen – als würde man den Abstand zweier Millimeterstriche

auf einem Lineal noch 10 Millionen Mal unterteilen. Doch noch eindrucks-

voller ist, dass wir einen Tanz der Lipide, also ihre Bewegungen, beobachten

konnten. Einige Atome in den Lipiden bewegen sich sehr schnell, andere lang-

samer. Und so muss es sein. Die Membran ist keine steinharte Wand, sondern

weich und beweglich. Moleküle müssen sich bewegen, ansonsten können die

chemischen Reaktionen, die zu Leben führen, nicht vonstattengehen.

Gibt es den Greif tatsächlich?

WaaG ist also ein Zwitterprotein, das Wasser genauso wie Lipide mag, aber

es muss sich an der Lipidmembran befinden, um seine Arbeit, den Bau neu-

er Lipide, zu verrichten. WaaG bindet sich an negativ geladene Lipide in der

Membran mittels eines positiv geladenen Ankers. Und während dies geschieht,

tanzen die Lipide ihren Tanz und neue werden wieder und wieder hergestellt.

All diese Beobachtungen und Ergebnisse beruhen auf Untersuchungen, die

in einem Reagenzglas gemacht wurden. Aber geschieht all das auch in leben-

den Bakterien? Wir können mit großer Sicherheit sagen, dass einige der Beob-

achtungen auch in lebenden Zellen vor sich gehen. Lipide bewegen sich in al-

len Zellen. WaaG bindet elektrostatisch an Membranen. Aber bindet es immer

und unter allen Umständen? Bewegt es sich ebenfalls? Das wissen wir noch

nicht. Wir verstehen nun besser, wie WaaG und die Lipide gemeinsam funk-

tionieren, aber jede Antwort führt zu neuen Fragen. In der Wissenschaft geht

es darum Wissen zu erweitern und Grundlagen für ein besseres Verständnis der

Welt um uns herum zu legen. Mit dieser Arbeit wurde das Wissen über eine

Gruppe von Proteinen, die sich sehr merkwürdig verhält, erweitert. Außerdem

verstehen wir besser, wie sich Lipide bewegen und wie sie mit Proteinen inter-

agieren. Das bedeutet, dass wir einen weiteren weißen Fleck auf der Karte des

Wissens, von der auf den ersten Seiten dieser Arbeit die Rede war, beseitigen

können.

��

Popular science summary

As when a gryphon through the wildernessWith winged course, o’er hill or moory dale,Pursues the Arimaspian, who by stealthHad from his wakeful custody purloinedThe guarded gold.—John Milton, Paradise Lost

Proteins can be imagined as tiny, biological machines—the tiniest machines

that exist, just one millionth of a millimeter long. Together with fats and carbo-

hydrates they are the most important building blocks for life. These small parti-

cles are also called molecules. Proteins are immensely complicated molecules

that are made up of thousands of atoms in a complicated way that gives rise to

an ordered structure that catalyzes, i.e. performs, chemical reactions with ex-

tremely high precision. A protein can catalyze a chemical reaction in fractions

of a second that would otherwise take days or weeks to happen. Yet, what is

most astonishing is that proteins control all processes that happen in cells, so

that they may live. Without proteins, there can be no life.

However, more is needed for life than just a bunch of proteins. Another

important class of molecules are lipids, which are a specific group of fats.

Lipids are the most important constituents of membranes that can be thought

of as the walls of a house. Just like the walls protect people from the malev-

olence of the weather, membranes surround cells and protect them from the

malevolence of their environment. Membranes create an inside and an outside

environment. On the inside of a cell, vital proteins and other molecules are

found. On the outside of the cell there is nutrition, e.g. carbohydrates, but also

toxic molecules, from which the cell needs to protect itself.

Membranes consist of lipids. Lipids need to be manufactured by the cell

and that is where proteins enter the stage. Special proteins exist that manufac-

ture various types of lipids that are needed to build up a membrane. One of

these proteins was studied in this thesis. It is called WaaG and it is found in

the bacterium Escherichia coli. We investigated how WaaG functions inside

bacteria, which are organisms that consist of only one cell.

��

WaaG catalyzes one single but very important step in the construction of a

particular kind of lipid found in bacteria. This is one reason why we want to

understand how it works. Some bacteria cause serious diseases and such dis-

eases can be cured by killing the bacteria with antibiotics. One way of killing

them is by destroying proteins like WaaG that are involved in constructing

the bacterial membrane. Bacteria cannot survive with a damaged or destroyed

membrane. However, before we can impair the function of WaaG, we need to

understand how it works. But there is another reason why we are interested in

WaaG: Not only is it involved in the construction of the membrane, it also has

properties which makes it resemble a mythical creature: It is a griffin.

Who or what is the Griffin?

The ancient Greeks tell stories of a mythical creature that has the body and tail

of a lion, and the head and wings of an eagle: the griffin. Griffins are creatures

of the air and creatures of the land. Thanks to their peculiar combination of

eagle and lion characteristics, they are at home in both environments. WaaG is

a real-life, microscopic analogy to a griffin—a creature of two worlds. Thanks

to its peculiar combination of properties, this protein is at home in two envi-

ronments that couldn’t differ more: water, and lipids, which are very similar to

oil.1 Oil and water don’t mix. Everyone who tried to clean a greasy pan with

water knows that.

How can a molecule like both environments? This is hard to imagine and

this is why we had a closer look at WaaG. It turned out that it contains a re-

gion which can dock like an anchor to the membrane (see figure 6.3). Having

this anchor doesn’t cause WaaG any trouble in water, but once there is a lipid

membrane, it anchors WaaG to it.

Where does the Griffin live?

In ancient literature, several theories were put forward as to where griffins

might live. Some authors suspect that they reside in distant lands in the north;

others speculate they are found far to the east. Most of them agree, however,

that griffins are strongly attracted by gold. They are the guardians of unimag-

inable treasures.

WaaG is also attracted to its own kind of treasure: certain types of lipid

molecules. These lipids need to be negatively charged, so that the anchor,

which is positively charged, binds to the lipids and thus to the membrane. If

there are no negatively charged lipids, the anchor doesn’t work.

1The word ‘lipid’ originates from the Greek word for fat: λ íπoς .

��

Is the Griffin real?

So, WaaG is a hybrid protein that likes water and lipids alike but needs to be

at a lipid membrane to do its job, which is to build more lipids. It binds to

negatively charged lipids in the membrane using its positively charged anchor.

And, all the while, the lipids dance their dance and new ones are manufactured

over and over again.

However, all of our studies were done in a test tube. Is this all also happen-

ing in a real living cell? Some of the observations are most likely true for living

cells. Lipids move in all cells. WaaG binds electrostatically to membranes—

but does it always bind? Does it also move? This we don’t know yet. We have

gained a better understanding of how WaaG and the lipids work together, but

every answer gives rise to many new questions. Science is about extending

knowledge and laying foundations for a better understanding of the world that

surrounds us. We now have a better understanding about a very strangely be-

haved group of proteins. We also have a toolkit to look at the molecular dance

of lipids and proteins, and how they interact with each other. This means that

we can fill in another blank spot on the map of knowledge mentioned on the

first pages of this book.

��

Acknowledgements

It takes two to tango.—Pearl Bailey

Even though I wrote large parts of this book sitting by myself on a balcony

with a glimpse of Täby’s central park to my left and a forest of succulents to

the right, this thesis is not and cannot be the product of a solitary endeavor. On

the contrary, many people have contributed to it in one way or another. I’m

very grateful to all of them for sharing their knowledge and experience with

me, for discussions and advice, for criticism and support.

I’m grateful to Lena Mäler for supervising my work, for giving me freedom

to try out my ideas, and for coming up with new thoughts when I thought I

reached a dead end, for guidance to wrap up a story and for forcing me to

present it to others.

I like to thank Andreas Barth who undertook the task of scrutinizing my work

during all these years.

Special thanks go out to present, former, and adopted group members, in alpha-

betical order: Ane Metola, Axel Abelein, Biao Fu, Christian Brown, JoanPatrick, Johannes Björnerås, Montse Tersa, Pontus Pettersson, ScarlettSzpryngiel, Sofia Unnerståle, and Weihua Ye. I highly appreciate your ad-

vice that helps me in solving simple and complicated problems, I thank you for

climbing real and imaginary mountains with me and for laying science aside,

too.

I’m thankful for collaborations with Candan Ariöz, Claudio Muheim, andSusan Black and her group in Aberdeen who let me in on the secrets of protein

and lipid purification.

��

Creating a pleasant atmosphere during work, lunch, and fika, when traveling

to courses and conferences, and helping out with the small things in life is

extremely important. Thank you for doing this, Anna Wahlström, AstridGräslund, Cecilia Wallin, Cesare Baronio, Chenge Li, Christoph Loderer,Eloy Vallina, Erik Sjölund, Fan Yang, Fatemeh Madani, Göran Eriks-son, Huabing Wang, Ida Lindström, Inna Grinberg, Jakob Dogan, JensDanielsson, Johan Nordholm, Jüri Jarvet, Maurizio Baldassarre, NadjaEremina, Nicklas Österlund, Sarah Leeb, Sebastian Wärmländer, SeongilChoi, Syed Haq, Therese Grönlund, and Xin Mu.

Without technical and administrative support, nothing would work. Thank you

to: Britt-Marie Olsson, Haidi Astlind, Matthew Bennett, and TorbjörnAstlind.

Zum Schluss möchte ich meiner Familie danken. Uta und Till, Gefion undUrs, Nico und Johannes, Christian, Albrecht und Meng: Ihr habt mich bei

meinen Entscheidungen unterstützt und beraten, auch wenn das oft zur Folge

hatte, dass ich in Schweden war und nicht an unseren Familientreffen teil-

nehmen konnte. Ihr habt mir den Weg gewiesen und euch immer wieder auf

den Weg gemacht. Danke!

Having written this, I get the impression that Pearl Bailey was not quite right:

Sometimes, it takes more than two to tango.

��

References

I can afford not to get cited.—Wolfgang Pauli

[1] F. S. ABRAMS AND E. LONDON. Extension of the parallax analysis ofmembrane penetration depth to the polar region of model membranes:use of fluorescence quenching by a spin-label attached to the phospholipidpolar headgroup. Biochemistry, 32(40):10826–10831, 1993. 23, 24

[2] D. ALBESA-JOVÉ, D. GIGANTI, M. JACKSON, P. M. ALZARI, AND M. E.

GUERIN. Structure-function relationships of membrane-associated GT-Bglycosyltransferases. Glycobiology, 24(2):108–124, 2014. 18, 19, 20, 33, 39,

40, 42

[3] D. ALBESA-JOVÉ AND M. E. GUERIN. The conformational plasticity ofglycosyltransferases. Current Opinion in Structural Biology, 40:23–32, 2016.

18, 33

[4] M. ALWARAWRAH, F. HUSSAIN, AND J. HUANG. Alteration of lipid mem-brane structure and dynamics by diacylglycerols with unsaturated chains.

Biochimica et Biophysica Acta – Biomembranes, 1858(2):253–263, 2016. 9

[5] A. ANDERSSON AND L. MÄLER. Magnetic resonance investigations of lipidmotion in isotropic bicelles. Langmuir, 21(17):7702–7709, 2005. 9, 27, 31

[6] C. ARIÖZ, W. YE, A. BAKALI, C. GE, J. LIEBAU, H. GÖTZKE, A. BARTH,

Å. WIESLANDER, AND L. MÄLER. Anionic lipid binding to the foreignprotein MGS provides a tight coupling between phospholipid synthesis andprotein overexpression in Escherichia coli. Biochemistry, 52(33):5533–5544,

2013. 15

[7] I. BARÁK AND K. MUCHOVÁ. The role of lipid domains in bacterial cellprocesses. International Journal of Molecular Sciences, 14(2):4050–4065,

2013. 6

[8] M. BEAUGRAND, A. A. ARNOLD, J. HENIN, D. E. WARSCHAWSKI, P. T.

WILLIAMSON, AND I. MARCOTTE. Lipid concentration and molar ratioboundaries for the use of isotropic bicelles. Langmuir, 30(21):6162–6170,

2014. 16

��

[9] G. M. BELL. Self-diffusion of ions in the electric fields of spherical parti-cles. Transactions of the Faraday Society, 60:1752–1759, 1964. 27

[10] A. BERNINI, V. VENDITTI, O. SPIGA, AND N. NICCOLAI. Probing proteinsurface accessibility with solvent and paramagnetic molecules. Progress inNuclear Magnetic Resonance Spectroscopy, 54(3):278–289, 2009. 32, 39

[11] A. BHATTACHARYA, R. TEJERO, AND G. T. MONTELIONE. Evaluatingprotein structures determined by structural genomics consortia. Proteins,

66(4):778–795, 2007. 29

[12] H. BIVERSTÅHL, A. ANDERSSON, A. GRÄSLUND, AND L. MÄLER. NMRsolution structure and membrane interaction of the N-terminal sequence(1-30) of the bovine prion protein. Biochemistry, 43(47):14940–14947, 2004.

27

[13] J. BJÖRNERÅS, M. NILSSON, AND L. MÄLER. Analysing DHPC/DMPCbicelles by diffusion NMR and multivariate decomposition. Biochimica etBiophysica Acta – Biomembranes, 1848(11):2910–2917, 2015. 28

[14] G. BLOBEL. Intracellular protein topogenesis. Proceedings of the NationalAcademy of Sciences of the United States of America, 77(3):1496–1500, 1980.

7

[15] N. BLOEMBERGEN. Proton relaxation times in paramagnetic solutions. TheJournal of Chemical Physics, 27(2):572–573, 1957. 31

[16] N. BLOEMBERGEN AND L. O. MORGAN. Proton relaxation times in param-agnetic solutions. Effects of electron spin relaxation. The Journal of Chem-ical Physics, 34(3):842–850, 1961. 31

[17] P. J. BOOTH, M. L. RILEY, S. L. FLITSCH, R. H. TEMPLER, A. FA-

ROOQ, A. R. CURRAN, N. CHADBORN, AND P. WRIGHT. Evidence thatbilayer bending rigidity affects membrane protein folding. Biochemistry,

36(1):197–203, 1997. 14, 16

[18] V. L. BRAACH-MAKSVYTIS AND B. A. CORNELL. Chemical shiftanisotropies obtained from aligned egg yolk phosphatidylcholine by solid-state 13C nuclear magnetic resonance. Biophysical Journal, 53(5):839–843,

1988. 8

[19] M. H. BRACEY, B. F. CRAVATT, AND R. C. STEVENS. Structural common-alities among integral membrane enzymes. FEBS letters, 567(2-3):159–165,

2004. 7

[20] M. BRAMKAMP AND D. LOPEZ. Exploring the existence of lipid rafts inbacteria. Microbiology and Molecular Biology Reviews, 79(1):81–100, 2015.

6

��

[21] C. BRETON, S. FOURNEL-GIGLEUX, AND M. M. PALCIC. Recent struc-tures, evolution and mechanisms of glycosyltransferases. Current Opinionin Structural Biology, 22(5):540–549, 2012. 17

[22] C. BRETON, L. SNAJDROVA, C. JEANNEAU, J. KOCA, AND A. IM-

BERTY. Structures and mechanisms of glycosyltransferases. Glycobiology,

16(2):29R–37R, 2006. 17

[23] B. R. BROOKS, C. L. III BROOKS, A. D. MACKERELL JR, L. NILS-

SON, R. J. PETRELLA, B. ROUX, Y. WON, G. ARCHONTIS, C. BARTELS,

S. BORESCH, A. CAFLISCH, L. CAVES, Q. CUI, A. R. DINNER, M. FEIG,

S. FISCHER, J. GAO, M. HODOSCEK, W. IM, K. KUCZERA, T. LAZARIDIS,

J. MA, V. OVCHINNIKOV, E. PACI, R. W. PASTOR, C. B. POST, J. Z. PU,

M. SCHAEFER, B. TIDOR, R. M. VENABLE, H. L. WOODCOCK, X. WU,

W. YANG, D. M. YORK, AND M. KARPLUS. CHARMM: the biomolecu-lar simulation program. Journal of Computational Chemistry, 30(10):1545–

1614, 2009. 7, 18, 41

[24] H. SILVA CAMARGOS AND A. ALONSO. Electron paramagnetic resonance(EPR) spectral components of spin-labeled lipids in saturated phospho-lipid bilayers: effect of cholesterol. Química Nova, 36(6):815–821, 2013.

36

[25] J. A. CAMPBELL, G. J. DAVIES, V. BULONE, AND B. HENRISSAT. Aclassification of nucleotide-diphospho-sugar glycosyltransferases based onamino acid sequence similarities. The Biochemical Journal, 326(Pt 3):929–

939, 1997. 17

[26] J. CAVANAGH, W. J. FAIRBROTHER, A. G. PALMER III, AND N. J. SKEL-

TON. Protein NMR spectroscopy: principles and practice. Academic Press,

Burlington, 2007. 26, 29

[27] S. J. CHARNOCK AND G. J. DAVIES. Structure of the nucleotide-diphospho-sugar transferase, SpsA from Bacillus subtilis, in native andnucleotide-complexed forms. Biochemistry, 38(20):6380–6385, 1999. 18

[28] A. CHATTOPADHYAY AND E. LONDON. Parallax method for direct mea-surement of membrane penetration depth utilizing fluorescence quenchingby spin-labeled phospholipids. Biochemistry, 26(1):39–45, 1987. 23, 24

[29] A. CHAUDHURI AND A. CHATTOPADHYAY. Lipid binding specificity ofbovine α-lactalbumin: a multidimensional approach. Biochimica et Bio-physica Acta – Biomembranes, 1838(8):2078–2086, 2014. 23, 26

[30] K. CHEN AND N. TJANDRA. The use of residual dipolar coupling in studyingproteins by NMR, pages 47–67. NMR of Proteins and Small Biomolecules.

Springer, Heidelberg, 2011. 29

��

[31] J. J. CHOU, J. L. BABER, AND A. BAX. Characterization of phospholipidmixed micelles by translational diffusion. Journal of Biomolecular NMR,

29(3):299–308, 2004. 15, 27

[32] J. J. CHOU, J. D. KAUFMAN, S. J. STAHL, P. T. WINGFIELD, AND A. BAX.

Micelle-induced curvature in a water-insoluble HIV-1 Env peptide re-vealed by NMR dipolar coupling measurement in stretched polyacry-lamide gel. Journal of the American Chemical Society, 124(11):2450–2451,

2002. 14

[33] S. CHU, S. MALTSEV, A. EMWAS, AND G. A. LORIGAN. Solid-state NMRparamagnetic relaxation enhancement immersion depth studies in phos-pholipid bilayers. Journal of Magnetic Resonance, 207(1):89–94, 2010. 37

[34] B. C. CHUNG, J. ZHAO, R. A. GILLESPIE, D. Y. KWON, Z. GUAN,

J. HONG, P. ZHOU, AND S. Y. LEE. Crystal structure of MraY, anessential membrane enzyme for bacterial cell wall synthesis. Science,

341(6149):1012–1016, 2013. 17

[35] G. M. CLORE AND A. M. GRONENBORN. Multidimensional heteronuclearnuclear magnetic resonance of proteins. Methods in Enzymology, 239:349–

362, 1994. 29

[36] G. M. CLORE AND J. IWAHARA. Theory, practice, and applications ofparamagnetic relaxation enhancement for the characterization of tran-sient low-population states of biological macromolecules and their com-plexes. Chemical Reviews, 109(9):4108–4139, 2009. 31

[37] G. CORNILESCU, F. DELAGLIO, AND A. BAX. Protein backbone anglerestraints from searching a database for chemical shift and sequence ho-mology. Journal of Biomolecular NMR, 13(3):289–302, 1999. 29

[38] P. M. COUTINHO, E. DELEURY, G. J. DAVIES, AND B. HENRISSAT. Anevolving hierarchical family classification for glycosyltransferases. Journalof Molecular Biology, 328(2):307–317, 2003. 17, 20

[39] J. E. CRONAN. Bacterial membrane lipids: where do we stand? AnnualReview of Microbiology, 57:203–224, 2003. 11

[40] T. A. CROSS, M. SHARMA, M. YI, AND H. X. ZHOU. Influence of solubiliz-ing environments on membrane protein structures. Trends in BiochemicalSciences, 36(2):117–125, 2011. 14, 16

[41] L. CZERSKI AND C. R. SANDERS. Functionality of a membrane protein inbicelles. Analytical Biochemistry, 284(2):327–333, 2000. 16

[42] J. GARCIA DE LA TORRE. Hydration from hydrodynamics. General con-siderations and applications of bead modelling to globular proteins. Bio-physical Chemistry, 93(2-3):159–170, 2001. 27

��

[43] D. DE MENDOZA AND J. E. CRONAN. Thermal regulation of membranelipid fluidity in bacteria. Trends in Biochemical Sciences, 8(2):49–52, 1983.

35

[44] A. DECHAVIGNY, P. N. HEACOCK, AND W. DOWHAN. Sequence and in-activation of the pss gene of Escherichia coli. Phosphatidylethanolaminemay not be essential for cell viability. The Journal of Biological Chemistry,

266(8):5323–5332, 1991. 35

[45] A. P. G. DINGELDEIN, Š. POKORNÁ, M. LIDMAN, T. SPARRMAN,

R. ŠACHL, M. HOF, AND G. GRÖBNER. Apoptotic Bax at oxidativelystressed mitochondrial membranes: lipid dynamics and permeabilization.

Biophysical Journal, 112(10):2147–2158, 2017. 13

[46] A. M. DIXON AND C. K. LARIVE. NMR spectroscopy with spectral editingfor the analysis of complex mixtures. Applied Spectroscopy, 53(11):426A–

440A, 1999. 28

[47] P. DÖRMANN AND C. BENNING. Galactolipids rule in seed plants. Trendsin Plant Science, 7(3):112–118, 2002. 13, 34

[48] W. DOWHAN. Molecular basis for membrane phospholipid diversity: whyare there so many lipids? Annual Review of Biochemistry, 66:199–232, 1997.

5

[49] W. DOWHAN AND M. BOGDANOV. Functional roles of lipids in membranes,

pages 1–35. Biochemistry of Lipids, Lipoproteins and Membranes. Elsevier,

Amsterdam, New Comprehensive Biochemistry edition, 2002. 6, 8, 9

[50] M. R. EFTINK AND C. A. GHIRON. Fluorescence quenching studies withproteins. Analytical Biochemistry, 114(2):199–227, 1981. 23

[51] A. EINSTEIN. On the motion of small particles suspended in liquids atrest required by the molecular-kinetic theory of heat. Annalen der Physik,

17:549–560, 1905. 27

[52] J. F. ELLENA, L. S. LEPORE, AND D. S. CAFISO. Estimating lipid lateraldiffusion in phospholipid vesicles from carbon-13 spin-spin relaxation. TheJournal of Physical Chemistry, 97(12):2952–2957, 1993. 8, 31

[53] D. M. ENGELMAN. Membranes are more mosaic than fluid. Nature,

438(7068):578–580, 2005. 6, 7

[54] G. E. FANUCCI, J. Y. LEE, AND D. S. CAFISO. Membrane mimetic en-vironments alter the conformation of the outer membrane protein BtuB.

Journal of the American Chemical Society, 125(46):13932–13933, 2003. 16

[55] S. GAEMERS AND A. BAX. Morphology of three lyotropic liquid crystallinebiological NMR media studied by translational diffusion anisotropy. Jour-nal of the American Chemical Society, 123(49):12343–12352, 2001. 15

��

[56] K. GAWRISCH. The Dynamics of Membrane Lipids, pages 147–171. The

structure of biological membranes. CRC press, Boca Raton, 2004. 7, 8, 9

[57] C. GE, A. GEORGIEV, A. ÖHMAN, Å. WIESLANDER, AND A. A. KELLY.

Tryptophan residues promote membrane association for a plant lipid gly-cosyltransferase involved in phosphate stress. The Journal of BiologicalChemistry, 286(8):6669–6684, 2011. 19, 39, 42

[58] A. B. T. GHISAIDOOBE AND S. J. CHUNG. Intrinsic tryptophan fluores-cence in the detection and analysis of proteins: a focus on Förster reso-nance energy transfer techniques. International Journal of Molecular Sci-ences, 15(12):22518–22538, 2014. 23, 25

[59] S. J. GIBBS AND C. S. JOHNSON. A PFG NMR experiment for accuratediffusion and flow studies in the presence of eddy currents. Journal of Mag-netic Resonance (1969), 93(2):395–402, 1991. 28

[60] D. GIGANTI, D. ALBESA-JOVÉ, S. URRESTI, A. RODRIGO-UNZUETA,

M. A. MARTÍNEZ, N. COMINO, N. BARILONE, M. BELLINZONI,

A. CHENAL, AND M. E. GUERIN. Secondary structure reshuffling mod-ulates glycosyltransferase function at the membrane. Nature Chemical Bi-ology, 11(1):16–18, 2015. 18

[61] K. J. GLOVER, J. A. WHILES, G. WU, N. YU, R. DEEMS, J. O. STRUPPE,

R. E. STARK, E. A. KOMIVES, AND R. R. VOLD. Structural evaluationof phospholipid bicelles for solution-state studies of membrane-associatedbiomolecules. Biophysical Journal, 81(4):2163–2171, 2001. 14, 15, 16

[62] K. GOUNARIS AND J. BARBER. Monogalactosyldiacylglycerol: The mostabundant polar lipid in nature. Trends in Biochemical Sciences, 8(10):378–

381, 1983. 12

[63] S. GRIZOT, M. SALEM, V. VONGSOUTHI, L. DURAND, F. MOREAU,

H. DOHI, S. VINCENT, S. ESCAICH, AND A. DUCRUIX. Structure of the Es-cherichia coli heptosyltransferase WaaC: binary complexes with ADP andADP-2-deoxy-2-fluoro heptose. Journal of Molecular Biology, 363(2):383–

394, 2006. 20

[64] S. M. GRUNER. Intrinsic curvature hypothesis for biomembrane lipidcomposition: a role for nonbilayer lipids. Proceedings of the NationalAcademy of Sciences of the United States of America, 82(11):3665–3669, 1985.

6

[65] S. M. GRUNER, P. R. CULLIS, M. J. HOPE, AND C. P. S. TILCOCK. Lipidpolymorphism: the molecular basis of nonbilayer phases. Annual Reviewof Biophysics and Biophysical Chemistry, 14(1):211–238, 1985. 6

[66] I. GRYCZYNSKI, R. F. STEINER, AND J. R. LAKOWICZ. Intensity andanisotropy decays of the tyrosine calmodulin proteolytic fragments, as

��

studied by GHz frequency-domain fluorescence. Biophysical Chemistry,

39(1):69–78, 1991. 25

[67] P. GÜNTERT. Automated NMR structure calculation with CYANA, pages 353–

378. Protein NMR Techniques. Springer, New York, Methods in Molecular

Biology edition, 2004. 29

[68] P. GÜNTERT AND L. BUCHNER. Combined automated NOE assignmentand structure calculation with CYANA. Journal of Biomolecular NMR,

62(4):453–471, 2015. 29

[69] B. HALLE AND H. WENNERSTRÖM. Interpretation of magnetic resonancedata from water nuclei in heterogeneous systems. The Journal of ChemicalPhysics, 75(4):1928–1943, 1981. 8, 31

[70] I. HANUKOGLU. Proteopedia: Rossmann fold: A beta-alpha-beta fold atdinucleotide binding sites. Biochemistry and Molecular Biology Education,

43(3):206–209, 2015. 17

[71] F. A. HEBERLE AND G. PABST. Complex biomembrane mimetics on thesub-nanometer scale. Biophysical Reviews, pages 1–21, 2017. 8, 13

[72] D. E. HEINRICHS, M. A. VALVANO, AND C. WHITFIELD. Biosynthesis andgenetics of lipopolysaccharide core, pages 305–330. Endotoxin in health and

disease. CRC Press, New York, 1999. 12

[73] S. HIRAOKA, H. MATSUZAKI, AND I. SHIBUYA. Active increase in cardi-olipin synthesis in the stationary growth phase and its physiological signif-icance in Escherichia coli. FEBS Letters, 336(2):221–224, 1993. 35

[74] G. HOLZL AND P. DÖRMANN. Structure and function of glycoglycerolipidsin plants and bacteria. Progress in Lipid Research, 46(5):225–243, 2007. 13

[75] Y. HU, L. CHEN, S. HA, B. GROSS, B. FALCONE, D. WALKER,

M. MOKHTARZADEH, AND S. WALKER. Crystal structure of theMurG:UDP-GlcNAc complex reveals common structural principles of asuperfamily of glycosyltransferases. Proceedings of the National Academyof Sciences of the United States of America, 100(3):845–849, 2003. 7, 18, 20

[76] T. HWANG, S. MORI, A. J. SHAKA, AND P. C. M. VAN ZIJL. Application ofphase-modulated CLEAN chemical EXchange spectroscopy (CLEANEX-PM) to detect water-protein proton exchange and intermolecular NOEs.

Journal of the American Chemical Society, 119(26):6203–6204, 1997. 40

[77] M. T. HYVÖNEN, T. T. RANTALA, AND M. ALA-KORPELA. Structure anddynamic properties of diunsaturated 1-palmitoyl-2-linoleoyl-sn-glycero-3-phosphatidylcholine lipid bilayer from molecular dynamics simulation.

Biophysical Journal, 73(6):2907–2923, 1997. 37

��

[78] C. R. SANDERS II AND J. H. PRESTEGARD. Magnetically orientablephospholipid bilayers containing small amounts of a bile salt analogue,CHAPSO. Biophysical Journal, 58(2):447–460, 1990. 14

[79] C. R. SANDERS II AND J. P. SCHWONEK. Characterization of magnet-ically orientable bilayers in mixtures of dihexanoylphosphatidylcholineand dimyristoylphosphatidylcholine by solid-state NMR. Biochemistry,

31(37):8898–8905, 1992. 14

[80] A. G. PALMER III. NMR probes of molecular dynamics: overviewand comparison with other techniques. Annual Review of Biophysics andBiomolecular Structure, 30(1):129–155, 2001. 30

[81] A. G. PALMER III. NMR characterization of the dynamics of biomacro-molecules. Chemical Reviews, 104(8):3623–3640, 2004. 30

[82] J. IWAHARA, C. D. SCHWIETERS, AND G. M. CLORE. Ensemble approachfor NMR structure refinement against 1H paramagnetic relaxation en-hancement data arising from a flexible paramagnetic group attached toa macromolecule. Journal of the American Chemical Society, 126(18):5879–

5896, 2004. 32, 36

[83] M. B. JACKSON AND J. E. CRONAN JR. An estimate of the minimumamount of fluid lipid required for the growth of Escherichia coli. Biochim-ica et Biophysica Acta, 512(3):472–479, 1978. 8

[84] J. M. JEENER, B. H. MEIER, P. BACHMANN, AND R. R. ERNST. Investi-gation of exchange processes by two-dimensional NMR spectroscopy. TheJournal of Chemical Physics, 71(11):4546–4553, 1979. 29

[85] S. JO, T. KIM, AND W. IM. Automated builder and database of pro-tein/membrane complexes for molecular dynamics simulations. PloS One,

2(9):e880, 2007. 7, 18, 41

[86] S. JO, T. KIM, V. G. IYER, AND W. IM. CHARMM-GUI: a web-basedgraphical user interface for CHARMM. Journal of Computational Chem-istry, 29(11):1859–1865, 2008.

[87] S. JO, J. B. LIM, J. B. KLAUDA, AND W. IM. CHARMM-GUI MembraneBuilder for mixed bilayers and its application to yeast membranes. Bio-physical Journal, 97(1):50–58, 2009. 7, 18, 41

[88] W. C. JOHNSON, N. BEROVA, K. NAKANISHI, AND R. W. WOODY. Circulardichroism: principles and applications. Wiley, New York, 2000. 22

[89] B. JÖNSSON, H. WENNERSTRÖM, P. G. NILSSON, AND P. LINSE. Self-diffusion of small molecules in colloidal systems. Colloid and Polymer Sci-ence, 264(1):77–88, 1986. 27

��

[90] J. JOYARD, E. MARÉCHAL, C. MIÈGE, M. A. BLOCK, A. DORNE, AND

R. DOUCE. Structure, distribution and biosynthesis of glycerolipids fromhigher plant chloroplasts, pages 21–52. Lipids in photosynthesis: Structure,

function and genetics. Springer, New York, 1998. 12

[91] J. E. CRONAN JR. Phospholipid alterations during growth of Escherichiacoli. Journal of Bacteriology, 95(6):2054–2061, 1968. 35

[92] J. KEELER. Understanding NMR spectroscopy. John Wiley & Sons, Chich-

ester, 2011. 26, 30

[93] S. M. KELLY, T. J. JESS, AND N. C. PRICE. How to study proteins bycircular dichroism. Biochimica et Biophysica Acta – Proteins and Proteomics,

1751(2):119–139, 2005. 22

[94] S. M. KELLY AND N. C. PRICE. Circular dichroism to study protein inter-actions. Current Protocols in Protein Science, Chapter 20:Unit 20.10, 2006.

22

[95] J. A. KILLIAN AND G. VON HEIJNE. How proteins adapt to a membrane-water interface. Trends in Biochemical Sciences, 25(9):429–434, 2000. 23,

39

[96] J. L. KITCHEN, Z. LI, AND E. CROOKE. Electrostatic interactions duringacidic phospholipid reactivation of DnaA protein, the Escherichia coli ini-tiator of chromosomal replication. Biochemistry, 38(19):6213–6221, 1999.

11

[97] J. KOWALEWSKI AND L. MÄLER. Nuclear spin relaxation in liquids: theory,experiments, and applications. CRC Press, Boca Raton, 2006. 26, 30, 31

[98] N. KUCERKA, M. A. KISELEV, AND P. BALGAVÝ. Determination of bilayerthickness and lipid surface area in unilamellar dimyristoylphosphatidyl-choline vesicles from small-angle neutron scattering curves: a comparisonof evaluation methods. European Biophysics Journal, 33(4):328–334, 2004.

36

[99] N. KUCERKA, M. NIEH, AND J. KATSARAS. Fluid phase lipid ar-eas and bilayer thicknesses of commonly used phosphatidylcholines as afunction of temperature. Biochimica et Biophysica Acta – Biomembranes,

1808(11):2761–2771, 2011. 37, 39

[100] N. KUCERKA, S. TRISTRAM-NAGLE, AND J. F. NAGLE. Structure of fullyhydrated fluid phase lipid bilayers with monounsaturated chains. The Jour-nal of Membrane Biology, 208(3):193–202, 2006. 36

[101] A. KYRYCHENKO AND A. S. LADOKHIN. Molecular dynamics simulationsof depth distribution of spin-labeled phospholipids within lipid bilayer. TheJournal of Physical Chemistry B, 117(19):5875–5885, 2013. 9, 23, 24, 37

��

[102] L. L. LAIRSON, B. HENRISSAT, G. J. DAVIES, AND S. G. WITHERS. Gly-cosyltransferases: structures, functions, and mechanisms. Annual Reviewof Biochemistry, 77:521–555, 2008. 17

[103] J. R. LAKOWICZ. Principles of fluorescence spectroscopy. Springer, New

York, 2006. 22, 23, 25

[104] J. R. LAKOWICZ, B. KIERDASZUK, P. CALLIS, H. MALAK, AND

I. GRYCZYNSKI. Fluorescence anisotropy of tyrosine using one- and two-photon excitation. Biophysical Chemistry, 56(3):263–271, 1995. 25

[105] J. LANDSTRÖM, K. PERSSON, C. RADEMACHER, M. LUNDBORG,

W. WAKARCHUK, T. PETERS, AND G. WIDMALM. Small molecules con-taining hetero-bicyclic ring systems compete with UDP-Glc for binding toWaaG glycosyltransferase. Glycoconjugate Journal, 29(7):491–502, 2012.

12, 20, 33

[106] T. L. LAU, C. KIM, M. H. GINSBERG, AND T. S. ULMER. The structure ofthe integrin αIIbβ3 transmembrane complex explains integrin transmem-brane signalling. The EMBO Journal, 28(9):1351–1361, 2009. 14

[107] Y. LEE, E. LEE, Y. SEO, Y. KANG, J. HA, H. KIM, AND J. LEE. Mea-surement of hydrogen exchange times of the RNA imino protons using byphase-modulated CLEAN chemical exchange spectroscopy. Bulletin of theKorean Chemical Society, 30:2197–2198, 2009. 40

[108] S. T. LEFURGY AND T. S. LEYH. Analytical expressions for the homotropicbinding of ligand to protein dimers and trimers. Analytical Biochemistry,

421(2):433–438, 2012. 26

[109] L. S. LEPORE, J. F. ELLENA, AND D. S. CAFISO. Comparison of the lipidacyl chain dynamics between small and large unilamellar vesicles. Bio-physical Journal, 61(3):767–775, 1992. 36

[110] L. LI, P. STORM, O. P. KARLSSON, S. BERG, AND Å. WIESLAN-

DER. Irreversible binding and activity control of the 1,2-diacylglycerol3-glucosyltransferase from Acholeplasma laidlawii at an anionic lipid bi-layer surface. Biochemistry, 42(32):9677–9686, 2003. 19, 35, 42

[111] J. LIEBAU. Membrane mimetics for studying glycosyltransferase – lipid inter-actions. Master’s thesis, Stockholm University, 2012. 32

[112] J. LIEBAU. Membrane interactions of glycosyltransferases. Licentiate thesis,

Stockholm University, 2015. 7

[113] J. LIEBAU AND L. MÄLER. Immersion depths of lipid carbons in bicellesmeasured by paramagnetic relaxation enhancement. The Journal of Physi-cal Chemistry B, 121(32):7660–7670, 2017. 10

��

[114] J. LIEBAU, P. PETTERSSON, C. ARIÖZ, AND L. MÄLER. Fast-tumblingbicelles constructed from native Escherichia coli lipids. Biochimica et Bio-physica Acta – Biomembranes, 1858(9):2097–2105, 2016. 9, 15, 40

[115] J. LIEBAU, P. PETTERSSON, S. SZPRYNGIEL, AND L. MÄLER. Mem-brane interaction of the glycosyltransferase WaaG. Biophysical Journal,109(3):552–563, 2015. 41, 47, 51, 55

[116] J. LIEBAU, Y. WEIHUA, AND L. MÄLER. Characterization of fast-tumblingisotropic bicelles by PFG diffusion NMR. Magnetic Resonance in Chemistry,

55(5):395–404, 2017. 14, 15, 27

[117] J. LIND, J. NORDIN, AND L. MÄLER. Lipid dynamics in fast-tumblingbicelles with varying bilayer thickness: effect of model transmembranepeptides. Biochimica et Biophysica Acta – Biomembranes, 1778(11):2526–

2534, 2008. 15

[118] G. LINDBLOM AND G. ORÄDD. NMR studies of translational diffusion inlyotropic liquid crystals and lipid membranes. Progress in Nuclear MagneticResonance Spectroscopy, 26:483–515, 1994. 8

[119] G. LIPARI AND A. SZABO. Model-free approach to the interpretation ofnuclear magnetic relaxation in macromolecules. 2. Analysis of experimen-tal results. Journal of the American Chemical Society, 104:4559–4570, 1982.

8, 31, 36

[120] G. LIPARI AND A. SZABO. Model-free approach to the interpretation ofnuclear magnetic resonance relaxation in macromolecules. 1. Theory andrange of validity. Journal of the American Chemical Society, 104(17):4546–

4559, 1982. 8, 31, 36

[121] J. LIU AND A. MUSHEGIAN. Three monophyletic superfamilies accountfor the majority of the known glycosyltransferases. Protein Science,

12(7):1418–1431, 2003. 17

[122] C. LIZAK, S. GERBER, S. NUMAO, M. AEBI, AND K. P. LOCHER. X-raystructure of a bacterial oligosaccharyltransferase. Nature, 474(7351):350–

355, 2011. 17, 18

[123] Z. LU, W. D. VAN HORN, J. CHEN, S. MATHEW, R. ZENT, AND C. R.

SANDERS. Bicelles at low concentrations. Molecular Pharmaceutics,

9(4):752–761, 2012. 15

[124] P. A. LUCHETTE, T. N. VETMAN, R. S. PROSSER, R. E. HANCOCK, M. P.

NIEH, C. J. GLINKA, S. KRUEGER, AND J. KATSARAS. Morphology offast-tumbling bicelles: a small angle neutron scattering and NMR study.

Biochimica et Biophysica Acta – Biomembranes, 1513(2):83–94, 2001. 14, 15,

16

��

[125] M. LUCKEY. Membrane structural biology. Cambridge University Press, New

York, 2008. 5, 6, 7, 8

[126] L. MÄLER. Solution NMR studies of peptide-lipid interactions in modelmembranes. Molecular Membrane Biology, 29(5):155–176, 2012. 14

[127] L. MÄLER AND A. GRÄSLUND. Artificial membrane models for the studyof macromolecular delivery, pages 129–139. Macromolecular Drug Delivery.

Springer, New York, Methods in Molecular Biology edition, 2009. 13

[128] S. MALTSEV, S. M. HUDSON, I. D. SAHU, L. LIU, AND G. A. LORIGAN.

Solid-state NMR 31P paramagnetic relaxation enhancement membraneprotein immersion depth measurements. The Journal of Physical ChemistryB, 118(16):4370–4377, 2014. 37

[129] A. G. MARR AND J. L. INGRAHAM. Effect of temperature on the composi-tion of fatty acids in Escherichia coli. Journal of Bacteriology, 84(6):1260–

1267, 1962. 11

[130] D. MARSH. Molecular motion in phospholipid bilayers in the gel phase:long axis rotation. Biochemistry, 19(8):1632–1637, 1980. 36

[131] C. MARTINEZ-FLEITES, M. PROCTOR, S. ROBERTS, D. N. BOLAM, H. J.

GILBERT, AND G. J. DAVIES. Insights into the synthesis of lipopolysaccha-ride and antibiotics through the structures of two retaining glycosyltrans-ferases from family GT4. Chemistry & Biology, 13(11):1143–1152, 2006. 20,

39, 41

[132] Y. MATSUURA, T. TAKANO, AND R. E. DICKERSON. Structure of cy-tochrome c551 from Pseudomonas aeruginosa refined at 1.6 Å resolutionand comparison of the two redox forms. Journal of Molecular Biology,

156(2):389–409, 1982. 7

[133] H. M. MCCONNELL AND R. D. KORNBERG. Inside-outside transitions ofphospholipids in vesicle membranes. Biochemistry, 10(7):1111–1120, 1971.

8

[134] E. MILEYKOVSKAYA AND W. DOWHAN. Visualization of phospholipid do-mains in Escherichia coli by using the cardiolipin-specific fluorescent dye10-N-nonyl acridine orange. Journal of Bacteriology, 182(4):1172–1175,

2000. 6

[135] S. MOREIN, A. ANDERSSON, L. RILFORS, AND G. LINDBLOM. Wild-typeEscherichia coli cells regulate the membrane lipid composition in a "win-dow" between gel and non-lamellar structures. The Journal of BiologicalChemistry, 271(12):6801–6809, 1996. 8, 11

[136] C. MUHEIM, A. BAKALI, O. ENGSTRÖM, ÅKE WIESLANDER, D. O. DA-

LEY, AND G. WIDMALM. Identification of a fragment-based scaffold thatinhibits the glycosyltransferase WaaG from Escherichia coli. Antibiotics,

5(1):10, 2016. 40

��

[137] B. MUI, L. CHOW, AND M. J. HOPE. Extrusion technique to generateliposomes of defined size. Methods in Enzymology, 367:3–14, 2003. 13

[138] K. MURZYN, T. RÓG, AND M. PASENKIEWICZ-GIERULA.

Phosphatidylethanolamine-phosphatidylglycerol bilayer as a modelof the inner bacterial membrane. Biophysical Journal, 88(2):1091–1103,

2005. 37

[139] E. A. NALEFSK AND A. C. NEWTON. Use of stopped-flow fluorescence spec-troscopy to measure rapid membrane binding by protein kinase C, pages 115–

128. Protein Kinase C protocols. Humana Press, Totowa, Methods in Molecular

Biology edition, 2003. 25, 40

[140] M. NIEH, C. J. GLINKA, S. KRUEGER, S. R. PROSSER, AND J. KATSARAS.

SANS study of the structural phases of magnetically alignable lanthanide-doped phospholipid mixtures. Langmuir, 17(9):2629–2638, 2001. 15

[141] H. NIKAIDO. Molecular basis of bacterial outer membrane permeabilityrevisited. Microbiology and Molecular Biology Reviews, 67(4):593–656, 2003.

12

[142] J. O’NEILL. Tackling drug-resistant infections globally: final report andrecommendations. 2016, https://amr-review.org/Publications.html, accessed

31.08.2017. 2

[143] G. OTTING. Protein NMR using paramagnetic ions. Annual Review of Bio-physics, 39:387–405, 2010. 31, 39

[144] J. PAN, F. A. HEBERLE, S. TRISTRAM-NAGLE, M. SZYMANSKI,

M. KOEPFINGER, J. KATSARAS, AND N. KUCERKA. Molecular structuresof fluid phase phosphatidylglycerol bilayers as determined by small angleneutron and X-ray scattering. Biochimica et Biophysica Acta – Biomem-branes, 1818(9):2135–2148, 2012. 37

[145] C. T. PARKER, A. W. KLOSER, C. A. SCHNAITMAN, M. A. STEIN,

S. GOTTESMAN, AND B. W. GIBSON. Role of the rfaG and rfaP genes indetermining the lipopolysaccharide core structure and cell surface prop-erties of Escherichia coli K-12. Journal of Bacteriology, 174(8):2525–2538,

1992. 12, 20

[146] J. PENCER, M. NIEH, T. A. HARROUN, S. KRUEGER, C. ADAMS, AND

J. KATSARAS. Bilayer thickness and thermal response of dimyristoylphos-phatidylcholine unilamellar vesicles containing cholesterol, ergosterol andlanosterol: a small-angle neutron scattering study. Biochimica et Biophys-ica Acta – Biomembranes, 1720(1):84–91, 2005. 36

[147] W. S. PRICE. Pulsed-field gradient nuclear magnetic resonance as a tool forstudying translational diffusion: Part 1. Basic theory. Concepts in MagneticResonance, 9(5):299–336, 1997. 28

��

[148] W. S. PRICE. Pulsed-field gradient nuclear magnetic resonance as a tool forstudying translational diffusion: Part II. Experimental aspects. Conceptsin Magnetic Resonance, 10(4):197–237, 1998. 28

[149] J. QIAN, T. A. GARRETT, AND C. R. H. RAETZ. In vitro assembly ofthe outer core of the lipopolysaccharide from Escherichia coli K-12 andSalmonella typhimurium. Biochemistry, 53(8):1250–1262, 2014. 11

[150] C. R. H. RAETZ AND C. WHITFIELD. Lipopolysaccharide endotoxins. An-nual Review of Biochemistry, 71(1):635–700, 2002. 11

[151] S. RECKEL, D. GOTTSTEIN, J. STEHLE, F. LÖHR, M. VERHOE-

FEN, M. TAKEDA, R. SILVERS, M. KAINOSHO, C. GLAUBITZ, AND

J. WACHTVEITL. Solution NMR structure of proteorhodopsin. AngewandteChemie International Edition, 50(50):11942–11946, 2011. 7

[152] A. RODGER AND B. NORDÉN. Circular dichroism and linear dichroism. Ox-

ford University Press, Oxford, 1997. 22

[153] A. RODRIGO-UNZUETA, M. A. MARTINEZ, N. COMINO, P. M. ALZARI,

A. CHENAL, AND M. E. GUERIN. Molecular basis of membrane associa-tion by the phosphatidylinositol mannosyltransferase PimA enzyme frommycobacteria. The Journal of Biological Chemistry, 291(27):13955–13963,

2016. 19, 42

[154] J. B. A. ROSS, W. R. LAWS, K. W. ROUSSLANG, AND H. R. WYSSBROD.

Tyrosine fluorescence and phosphorescence from proteins and polypeptides,

pages 1–64. Topics in fluorescence spectroscopy. Springer, New York, 2002.

23

[155] C. R. SANDERS, B. J. HARE, K. P. HOWARD, AND J. H. PRESTEGARD.

Magnetically-oriented phospholipid micelles as a tool for the study ofmembrane-associated molecules. Progress in Nuclear Magnetic ResonanceSpectroscopy, 26:421–444, 1994. 14

[156] C. R. SANDERS AND R. SCOTT PROSSER. Bicelles: a model membranesystem for all seasons? Structure, 6(10):1227–1234, 1998. 15

[157] A. SCHMIDT, K. KOCHANOWSKI, S. VEDELAAR, E. AHRNÉ, B. VOLKMER,

L. CALLIPO, K. KNOOPS, M. BAUER, R. AEBERSOLD, AND M. HEINE-

MANN. The quantitative and condition-dependent Escherichia coli pro-teome. Nature Biotechnology, 34(1):104–110, 2016. 42

[158] H. SCHMIDT, G. HANSEN, S. SINGH, A. HANUSZKIEWICZ, B. LINDNER,

K. FUKASE, R. W. WOODARD, O. HOLST, R. HILGENFELD, U. MA-

MAT, AND J. R. MESTERS. Structural and mechanistic analysis of themembrane-embedded glycosyltransferase WaaA required for lipopolysac-charide synthesis. Proceedings of the National Academy of Sciences of theUnited States of America, 109(16):6253–6258, 2012. 19, 20

��

[159] C. D. SCHWIETERS, J. J. KUSZEWSKI, AND G. M. CLORE. Using Xplor-NIH for NMR molecular structure determination. Progress in NuclearMagnetic Resonance Spectroscopy, 48(1):47–62, 2006. 29

[160] J. SEELIG, H. GALLY, AND R. WOHLGEMUTH. Orientation and flexibilityof the choline head group in phosphatidylcholine bilayers. Biochimica etBiophysica Acta – Biomembranes, 467(2):109–119, 1977. 8

[161] Y. SHEN, F. DELAGLIO, G. CORNILESCU, AND A. BAX. TALOS+: a hy-brid method for predicting protein backbone torsion angles from NMRchemical shifts. Journal of Biomolecular NMR, 44(4):213–223, 2009. 29

[162] A. SHOKRI AND G. LARSSON. Characterisation of the Escherichia colimembrane structure and function during fedbatch cultivation. MicrobialCell Factories, 3(1):9, 2004. 9

[163] K. SIMONS AND E. IKONEN. Functional rafts in cell membranes. Nature,

387(6633):569–572, 1997. 6, 7

[164] S. J. SINGER AND G. L. NICOLSON. The fluid mosaic model of the struc-ture of cell membranes. Science, 175(4023):720–731, 1972. 6, 7

[165] A. M. SMONDYREV AND M. L. BERKOWITZ. Structure of dipalmitoylphos-phatidylcholine/cholesterol bilayer at low and high cholesterol concentra-tions: molecular dynamics simulation. Biophysical Journal, 77(4):2075–

2089, 1999. 8, 37

[166] O. SÖDERMAN, P. STILBS, AND W. S. PRICE. NMR studies of surfactants.

Concepts in Magnetic Resonance Part A, 23(2):121–135, 2004. 27

[167] I. SOLOMON. Relaxation processes in a system of two spins. Physical Re-view, 99(2):559, 1955. 31

[168] I. SOLOMON AND N. BLOEMBERGEN. Nuclear magnetic interactions in theHF molecule. The Journal of Chemical Physics, 25(2):261–266, 1956. 31

[169] B. SOUFI, K. KRUG, A. HARST, AND B. MACEK. Characterization of theE. coli proteome and its modifications during growth and ethanol stress.

Frontiers in Microbiology, 6:103, 2015. 42

[170] T. STAMS, S. K. NAIR, T. OKUYAMA, A. WAHEED, W. S. SLY, AND D. W.

CHRISTIANSON. Crystal structure of the secretory form of membrane-associated human carbonic anhydrase IV at 2.8 Å resolution. Proceed-ings of the National Academy of Sciences of the United States of America,

93(24):13589–13594, 1996. 7

[171] E. O. STEJSKAL AND J. E. TANNER. Spin diffusion measurements: spinechoes in the presence of a time-dependent field gradient. The Journal ofChemical Physics, 42(1):288–292, 1965. 28

��

[172] O. STERN AND M. VOLMER. On the quenching-time of fluorescence.

Physikalische Zeitschrift, 20:183–188, 1919. 23

[173] W. SUTHERLAND. LXXV. A dynamical theory of diffusion for non-electrolytes and the molecular mass of albumin. Philosophical Magazine,

9(54):781–785, 1905. 27

[174] S. SZPRYNGIEL. Structure and lipid interactions of membrane-associated gly-cosyltransferases – Cationic patches and anionic lipids regulate biomembranebinding of both GT-A and GT-B enzymes. PhD thesis, Stockholm University,

2016. 17

[175] H. KEAT TAM, J. HÄRLE, S. GERHARDT, J. ROHR, G. WANG, J. S. THOR-

SON, A. BIGOT, M. LUTTERBECK, W. SEICHE, AND B. BREIT. Structuralcharacterization of O- and C-glycosylating variants of the landomycinglycosyltransferase LanGT2. Angewandte Chemie International Edition,

54(9):2811–2815, 2015. 19

[176] J. E. TANNER. Use of the stimulated echo in NMR diffusion studies. TheJournal of Chemical Physics, 52(5):2523–2526, 1970. 28

[177] N. TJANDRA AND A. BAX. Direct measurement of distances and anglesin biomolecules by NMR in a dilute liquid crystalline medium. Science,

278(5340):1111–1114, 1997. 15

[178] E. K. TUOMINEN, C. J. WALLACE, AND P. K. KINNUNEN. Phospholipid-cytochrome c interaction: evidence for the extended lipid anchorage. TheJournal of Biological Chemistry, 277(11):8822–8826, 2002. 7

[179] U. M. ÜNLIGIL AND J. M. RINI. Glycosyltransferase structure and mech-anism. Current Opinion in Structural Biology, 10(5):510–517, 2000. 17

[180] S. UNNERSTÅLE AND L. MÄLER. pH-dependent interaction between C-peptide and phospholipid bicelles. Journal of Biophysics, 2012:185907,

2012. 15

[181] S. UNNERSTÅLE, L. MÄLER, AND R. R. DRAHEIM. Structural character-ization of AS1-membrane interactions from a subset of HAMP domains.

Biochimica et Biophysica Acta, 1808(10):2403–2412, 2011. 15

[182] J. VALERO-GONZÁLEZ, C. LEONHARD-MELIEF, E. LIRA-NAVARRETE,

G. JIMÉNEZ-OSÉS, C. HERNÁNDEZ-RUIZ, M. CARMEN PALLARÉS,

I. YRUELA, D. VASUDEVAN, A. LOSTAO, AND F. CORZANA. A proac-tive role of water molecules in acceptor recognition by protein O-fucosyltransferase 2. Nature Chemical Biology, 12(4):240–246, 2016. 19

[183] L. VAN DAM, G. KARLSSON, AND K. EDWARDS. Direct observation andcharacterization of DMPC/DHPC aggregates under conditions relevantfor biological solution NMR. Biochimica et Biophysica Acta – Biomembranes,

1664(2):241–256, 2004. 16

��

[184] G. VAN MEER, D. R. VOELKER, AND G. W. FEIGENSON. Membrane lipids:where they are and how they behave. Nature Reviews Molecular Cell Biol-ogy, 9(2):112–124, 2008. 8, 9

[185] I. VASILENKO, B. DE KRUIJFF, AND A. J. VERKLEIJ. Polymorphic phasebehaviour of cardiolipin from bovine heart and from Bacillus subtilis as de-tected by 31P-NMR and freeze-fracture techniques. Effects of Ca2+, Mg2+,Ba2+ and temperature. Biochimica et Biophysica Acta, 684(2):282–286,

1982. 11

[186] R. M. VENABLE, Y. ZHANG, B. J. HARDY, AND R. W. PASTOR. Moleculardynamics simulations of a lipid bilayer and of hexadecane: an investigationof membrane fluidity. Science, 262:223–223, 1993. 8

[187] C. L. VENTOLA. The antibiotic resistance crisis: part 1: causes andthreats. Pharmacy and Therapeutics, 40(4):277–283, 2015. 2

[188] S. VIKSTRÖM, L. LI, O. P. KARLSSON, AND Å. WIESLANDER. Key roleof the diglucosyldiacylglycerol synthase for the nonbilayer-bilayer lipidbalance of Acholeplasma laidlawii membranes. Biochemistry, 38(17):5511–

5520, 1999. 13

[189] J. VOGT AND G. E. SCHULZ. The structure of the outer membrane pro-tein OmpX from Escherichia coli reveals possible mechanisms of virulence.

Structure, 7(10):1301–1309, 1999. 7

[190] R. R. VOLD, R. S. PROSSER, AND A. J. DEESE. Isotropic solutions ofphospholipid bicelles: a new membrane mimetic for high-resolution NMRstudies of polypeptides. Journal of Biomolecular NMR, 9(3):329–335, 1997.

14

[191] J. WANG, J. R. SCHNELL, AND J. J. CHOU. Amantadine partition and lo-calization in phospholipid membrane: a solution NMR study. Biochemicaland Biophysical Research Communications, 324(1):212–217, 2004. 28

[192] M. WANG, C. J. HERRMANN, M. SIMONOVIC, D. SZKLARCZYK, AND

C. MERING. Version 4.0 of PaxDb: protein abundance data, integratedacross model organisms, tissues, and cell-lines. Proteomics, 15(18):3163–

3168, 2015. 42

[193] X. WANG AND P. J. QUINN. Endotoxins: lipopolysaccharides of gram-negative bacteria, pages 3–25. Endotoxins: Structure, Function and Recog-

nition. Springer, Dordrecht, 2010. 12

[194] D. E. WARSCHAWSKI. Membrane proteins of known structure determinedby NMR. http://www.drorlist.com/nmr/MPNMR.html, accessed 31.08.2017.

14, 28

��

[195] D. E. WARSCHAWSKI, A. A. ARNOLD, M. BEAUGRAND, A. GRAVEL,

E. CHARTRAND, AND I. MARCOTTE. Choosing membrane mimetics forNMR structural studies of transmembrane proteins. Biochimica et Bio-physica Acta – Biomembranes, 1808(8):1957–1974, 2011. 14, 34

[196] H. WENNERSTRÖM, G. LINDBLOM, AND B. LINDMAN. Theoretical aspectson NMR of quadrupolar ionic nuclei in micellar solutions and amphiphilicliquid-crystals. Chemica Scripta, 6(3):97–103, 1974. 8, 31

[197] H. WENNERSTRÖM, B. LINDMAN, O. SÖDERMAN, T. DRAKENBERG, AND

J. B. ROSENHOLM. 13C magnetic relaxation in micellar solutions. Influ-ence of aggregate motion on T1. Journal of the American Chemical Society,

101(23):6860–6864, 1979. 8, 31

[198] S. H. WHITE AND W. C. WIMLEY. Membrane protein folding and stability:physical principles. Annual Review of Biophysics and Biomolecular Structure,

28(1):319–365, 1999. 7

[199] G. WIDER, V. DOTSCH, AND K. WÜTHRICH. Self-compensating pulsedmagnetic-field gradients for short recovery times. Journal of Magnetic Res-onance, Series A, 108(2):255–258, 1994. 28

[200] M. WIKSTRÖM, A. A. KELLY, A. GEORGIEV, H. M. ERIKSSON, M. R.

KLEMENT, M. BOGDANOV, W. DOWHAN, AND Å. WIESLANDER. Lipid-engineered Escherichia coli membranes reveal critical lipid headgroup sizefor protein function. The Journal of Biological Chemistry, 284(2):954–965,

2009. 11, 35, 40

[201] W. P. WILLIAMS. The physical properties of thylakoid membrane lipids andtheir relation to photosynthesis, pages 103–118. Lipids in Photosynthesis:

Structure, Function and Genetics. Springer, New York, 1998. 13

[202] D. S. WISHART, B. D. SYKES, AND F. M. RICHARDS. The chemicalshift index: a fast and simple method for the assignment of protein sec-ondary structure through NMR spectroscopy. Biochemistry, 31(6):1647–

1651, 1992. 29

[203] R. W. WOODY. Theory of circular dichroism of proteins, pages 25–68. Circular

Dichroism and the Conformational Analysis of Biomolecules. Springer, New

York, 1996. 22

[204] D. H. WU, A. D. CHEN, AND C. S. JOHNSON. An improved diffusion-ordered spectroscopy experiment incorporating bipolar-gradient pulses.

Journal of Magnetic Resonance, Series A, 115(2):260–264, 1995. 28

[205] E. L. WU, X. CHENG, S. JO, H. RUI, K. C. SONG, E. M. DAVILA-

CONTRERAS, Y. QI, J. LEE, V. MONJE-GALVAN, R. M. VENABLE, J. B.

KLAUDA, AND W. IM. CHARMM-GUI Membrane Builder toward realis-tic biological membrane simulations. Journal of Computational Chemistry,

35(27):1997–2004, 2014. 7, 18, 41

��

[206] H. WU, K. SU, X. GUAN, M. E. SUBLETTE, AND R. E. STARK. As-sessing the size, stability, and utility of isotropically tumbling bicelle sys-tems for structural biology. Biochimica et Biophysica Acta – Biomembranes,

1798(3):482–488, 2010. 15

[207] K. WÜTHRICH. Determination of three-dimensional protein structures insolution by nuclear magnetic resonance: An overview. Methods in Enzy-mology, 177:125–131, 1989. 29

[208] W. YE, J. LIEBAU, AND L. MÄLER. New membrane mimetics with galac-tolipids: lipid properties in fast-tumbling bicelles. The Journal of PhysicalChemistry B, 117(4):1044–1050, 2013. 9, 15

[209] W. YE, J. LIND, J. ERIKSSON, AND L. MÄLER. Characterization of themorphology of fast-tumbling bicelles with varying composition. Langmuir,

30(19):5488–5496, 2014. 15

[210] J. A. YETHON, E. VINOGRADOV, M. B. PERRY, AND C. WHITFIELD. Mu-tation of the lipopolysaccharide core glycosyltransferase encoded by waaGdestabilizes the outer membrane of Escherichia coli by interfering withcore phosphorylation. Journal of Bacteriology, 182(19):5620–5623, 2000.

12, 20

[211] K. ZANGGER, M. RESPONDEK, C. GOBL, W. HOHLWEG, K. RASMUSSEN,

G. GRAMPP, AND T. MADL. Positioning of micelle-bound peptides by para-magnetic relaxation enhancements. The Journal of Physical Chemistry B,

113(13):4400–4406, 2009. 39

[212] W. ZHAO, T. RÓG, A. A. GURTOVENKO, I. VATTULAINEN, AND M. KART-

TUNEN. Atomic-scale structure and electrostatics of anionic palmitoy-loleoylphosphatidylglycerol lipid bilayers with Na+ counterions. Biophysi-cal Journal, 92(4):1114–1124, 2007. 8

[213] L. ZHENG, Y. LIN, S. LU, J. ZHANG, AND M. BOGDANOV. Biogenesis,transport and remodeling of lysophospholipids in Gram-negative bacte-ria. Biochimica et Biophysica Acta – Molecular and Cell Biology of Lipids,

1862(11):1404–1413, 2016. 8

��