Expression and structural analysis of membrane proteins

1.1 Membrane Proteins
Between one quarter and one third of all genes in eukaryotic and prokaryotic
organisms code for integral membrane proteins (IMPs) (Essen, 2002). These
proteins are essential parts of biological membranes and confer various functions,
such as energy conversion, transport, biosynthesis of lipids, signal transduction,
or cell recognition. The enormous economical potential of membrane proteins
is highlighted by the family of G-protein-coupled receptors (GPCRs), on which
approximately 60 % of all prescription drugs are targeted (Vassilatis et al., 2003).
Despite their eminent biological roles so far only 99 high-resolution structures of
IMPs have been solved by X-ray and electron crystallography to date
(http://blanco.biomol.uci.edu), mostly because of their amphiphilic character. IMPs
are uniquely adapted to the topological restraints of the lipid bilayer: hydrophobic
residues mediate the contact with the fatty acid tails of the lipids; hydrophilic
protein surfaces are exposed towards the aqueous bulk phase. The amphiphilic
nature poses a variety of technical problems, because membrane proteins have to
be kept in a solubilized state by detergents during the entire course of purification
and crystallization.
1.1.1 Assembly and Membrane Insertion
Membrane proteins use two different mechanisms to insert into the lipid bilayer,
termed the constitutive and the non-constitutive pathway (http://blanco.biomol.
uci.edu).
Constitutive membrane proteins are assembled via a complex mechanism: ribosomes,
that mediate the synthesis of membrane proteins, are transiently attached
to a protein complex referred to as the translocon. This translocon is located
within the cell membrane and provides a transmembrane ”tunnel” into which the
newly synthesized protein is injected. When protein synthesis has been completed,
the ribosome separates from the translocon and the protein is released into the
membrane bilayer where it assumes its final three-dimensional structure (Borel and
Simon, 1996; Do et al., 1996). Non-constitutive membrane proteins are usually
synthesized in a soluble monomeric form, which inserts before, upon, or after
oligomerization into the target membrane. Non-constitutive membrane proteins
generally belong to the class of toxins and antimicrobial peptides. A good example
is staphylococcal alpha-hemolysin whose crystallographic structure was solved by
Song et al., 1996.
The study of assembly and membrane insertion of another non-constitutive haemolytic
protein- ClyA- is shown in Chapter 3.
1.1.2 Structural Classes
IMPs mask the polar peptide bonds of their amino-acid backbone from energetically
unfavorable interactions with the hydrophobic core of the membrane by
forming hydrogen bonds. The two structural motifs used by IMPs for internal
hydrogen bond formation are the α-helix and the ß-barrel. Therefore, IMPs are
classified into α-helical and ß-barrel membrane proteins.
Receptors and ion channels that reside in the plasma membranes and in the membranes
of the endoplasmic reticulum are most often in the α-helical form, but
α helical motifs are also observed in pore-forming toxins (PFTs), like colicin
A (Parker et al., 1989). Initial information about the folded three-dimensional
structure of α-helical membrane proteins has largely been provided from studies of
bacteriorhodopsin (Henderson and Unwin, 1975).
The ß-barrel IMPs are found in the outer membranes of gram-negative bacteria,
mitochondria and chloroplasts (Wimley et al., 1998). Like the α-helical motive,
the ß-barrel is used by many bacteria to form cytotoxic transmembrane channels
(Parker and Feil, 2005).
For some IMPs it is difficult to predict the structural class merely by their origin and
amino acid sequence. An example herein is ClyA, a PFT from the outer membrane
vesicles of some Escherichia coli and Salmonella enterica strains. The structural
characterization of ClyA is described in Chapter 3.
1.1.3 Guideline for Expression and Purification
Membrane proteins carry out many essential functions and are thus of importance
in health and disease. Unfortunately many membrane proteins only occur in low
concentrations in native tissue. Therefore, it is necessary for most studies, to
develop and apply a practical expression and purification procedure.
Before initiating protein expression, one should consider the requirements of the
intended application. Five important factors need to be considered:
1. How much of the purified protein is required?
2. How homogeneous should the protein be?
3. Should the protein be fully active or is it possible to assess an inactive but
stable mutant?
4. What is the least expensive way to satisfy my requirements?
5. How much effort will it take to achieve those requirements?
Priorities need to be assigned with respect to the pursued goal. For example,
functional studies can be accomplished with protein of low purity and concentration,
whereas a highly pure and stable preparation is necessary for structural analyses
(see chapter 1.2). For other applications such as raising antibodies, the purified
protein is neither needed in large quantities nor in an active state or in its native
conformation. The goal of expression and purification is then to obtain a small
amount of the protein in a pure enough form to raise antibodies that are specific to
the protein of interest. For this type of application the material extracted from an
SDS-gel band does often suffice.
Different approaches to express and purify a range of membrane proteins for
structural determination are presented in chapter 2.
1.2 Structural Analysis of Membrane Proteins
The individual features of a protein are essentially determined by its threedimensional
structure. The three main methods of molecular structure determination
are X-ray crystallography, NMR spectroscopy and electron microscopy.
The basic principles as well as the advantages and disadvantages of each method,
with regards to structural analyses of membrane proteins in particular, are discussed
below.
1.2.1 X-ray Crystallography
The main technique that has been used to determine the 3D structure of molecules
and proteins at atomic resolution is x-ray crystallography. To date, almost 30’000
protein structures have been solved by x-ray crystallography, with a resolution
</= 1 °A for almost 1 % of all structures (http://www.pdb.org/).
In practice, a three-dimensional protein crystal is exposed to monochromatic Xrays
and the resulting diffraction pattern is recorded by a suitable detector. If the
crystal is well ordered, the scattered waves interfere constructively at certain directions
according to Bragg’s law, and the intensities of these diffraction spots can be
used to reconstruct the three-dimensional image of the crystal structure, provided
the phases can be determined. Several methods have been developed to solve the
phase problem, such as multiwavelenth anomalous diffraction (MAD), multiple
isomorphous replacement (MIR) and molecular replacement. All of these methods
can be time consuming and delay the process of solving the crystal structure.
X-rays only interact with the electrons of the sample. An X-ray structure is therefore
represented by the electron density of the inspected sample (Voet, 1992).
The power of X-ray crystallography has increased tremendously in recent decades
thanks to the development of crystallization robots, the availability of synchrotron
light sources, and the processing software development by the strong X-ray crystallography
community. Crystallization robots can produce and screen up to 480
sample conditions per hour in parallel (Stevens, 2000), greatly facilitating the
laborious work of setting up crystallization trials. Synchrotron radiation, which is
emitted when charged particles are forced to move in a circular orbit close to the
speed of light, was originally regarded as an unwanted byproduct of high-energy
particle accelerators. Since the 1970’s, synchrotron radiation has been adapted
for use in X-ray crystallography because of its extraordinary brightness and wide
energy spectrum (Bai, 2005).
The prerequisite for achieving atomic scale resolution are well-purified protein samples
in milligram quantities and concentrations in the 10 mg/ml range that can grow
into high ordered crystals. The purification yield of membrane proteins, however,
often lies in the sub-milligram range. Another problem is that the detergent micelle
belts around solubilized membrane proteins have to be incorporated into crystal
lattices, which may impede the process of growing well ordered 3D-crystals from a
homogenous population of mixed membrane protein/detergent micelles.(Palanivelu
et al., 2006).
1.2.2 NMR
Developed in the 1940’s, nuclear magnetic resonance (NMR) spectroscopy relies
on a physical phenomenon based upon the magnetic property of an atom’s nucleus.
NMR spectroscopy studies the behavior of a magnetic spin, like that of a hydrogen
atom (H) or nitrogen (N) and sulfur (S) isotopes, by aligning it with a very
powerful external magnetic field and perturbing this alignment using an oscillating
electromagnetic field (Wuthrich, 2003). The response of the spins to this field is
what is exploited in NMR spectroscopy.
NMR is the only technique that can provide detailed information on the exact
three-dimensional structure of biological molecules in solution (Wuthrich, 1989).
However, resolution is lost as the size of a macromolecule increases and special
protocols are required to allow structures > 40 kDa to be determined (Patzelt et
al., 2002). To assess the structure of a protein, the sample needs to be labeled with
N- or S- isotopes during the course of expression. Unfortunately, isotropic labeling
is cost-intensive for higher eukaryotic expression systems and thus practically not
feasible (Wim de Grip; Lecture series E7 ’Large scale production of functional proteins’).
Additionally, NMR requires a protein concentration of ≥ 0.3 mM (Dotsch
et al., 1998), an amount too high to be achieved for most membrane proteins.
1.2.3 Electron Microscopy (EM)
With membrane protein yields routinely obtained at low concentration, only
material-saving techniques like 2D-crystallization and electron crystallography
allow atomic scale structural information to be derived (Essen, 2002). Unlike
NMR and X-ray crystallography, electron microscopy provides a direct image of
the sample, using electrons as the source of illumination. The small wavelength
of electron radiation helps to achieve a resolution limit close to 1 A° . However,
the energy transfer from impinging electrons to the sample leads to its destruction.
Therefore, two techniques are commonly used to address this problem: the
negative-stain technique and cryo-electron microscopy (Dubochet et al., 1988).
For negative-staining the sample, supported on a EM grid, is washed with a concentrated
solution of a heavy-metal salt such as uranyl acetate. The heavy-metal stain
surrounds the shape of the sample and enhances contrast by shielding the electron
beam. Thus, a negative image of the molecule is created. In cryo-EM, the liquid
sample is quick-frozen in liquid ethane to from amorphous ice. The resolution
achieved with cryo-EM is higher than with negative-staining, but it has to manage
with less contrast.
Electron-microscopic images recorded at low dose have a low signal to noise ratio
(SNR), making image processing a prerequisite for structural studies. To improve
the SNR and to reconstruct the 3D structure from projections, the information from
multiple images is combined with two main methods, single particle analysis and
2D-crystallography.
Single Particle Analysis
For single particle analysis, the purified protein sample is applied onto an electron
microscope grid, where it adsorbs in different orientations. Thousands of images
of the protein are then collected and subjected to alignment and classification
procedures. Aligned particles within structural classes can be averaged to reveal
structural details that were hidden by noise in the single images. Ultimately, a 3D
reconstruction can be calculated by merging the projections of differently oriented
particles.
The requirements for this technique are a homogeneous population of protein
particles with a minimal molecular size of approximately 500 kDa. The achievable
resolution lies between 10 - 30 A° .
In this work, single particle analysis has been applied to calculate projection maps
of four membrane proteins (Chapter 2). It was also used to solve the structure of
the ClyA pore complex at 12.4 °A resolution (see Chapter 3).
Electron Crystallography
When single or double-layered 2D crystals of a membrane protein can be assembled
by reconstitution in the presence of lipids, the alignment of the particles is no longer
required. In 2D crystals, membrane proteins are incorporated into lipid vesicles or
sheets, the natural environment of IMPs. Crystal images recorded in the electron
microscope are subjected to a crystallographic type of image processing, that
extracts valuable protein information from noise. As the direct images are used,
the phases are not lost.
The resolution of the method could be improved down to 1.9 °A (Gonen et al.,
2006), a limit where even lipid-protein interactions can be monitored.