User:Kinkreet/Protein Science/Working With Membrane Proteins

To find the structure of a membrane protein, it must first be produced (expression, solubilization, purification and crystallization) and then structural determination must be carried out (X-ray crystallography, EM and NMR)

As of Jun 2012, there are 82347 protein structures on the Protein DataBank, of which only 308 (as of 2011) are unique membrane proteins. The reason why membrane proteins, being in such large numbers in a cell, occupy such a small proportion of the proteins which structure have been determined, is largely due to the difficulty in both production of the protein as well as its structural determination. Rhodopsin is a GPCR that is found on highly folded membrane disks, thus it is in high quantities and easier to purify; however, most membrane proteins are expressed in smaller amounts and so harder to obtain pure samples.

To produce membrane proteins, like any proteins, we need to overexpress the protein in a host cell using recombinant systems. Expression of genes from the same species as the host will be best although not essential, but heterogenous expression do not always work. E. coli can express most bacterial proteins, yeast can express bacteria and a few mammalian proteins, although the most successful expression system for mammalian proteins are insects. Apart from diffuculties in expression, the yield is often limited and so a large culture is required to produce sufficient amount (this is a simpler and better alternative than to induce more production of the protein per cell). Often expression does not work, then another host have to be employed. Sometimes proteases degrade the protein of interest or parts of it, sometimes removing the tags used later for purification; in these instances, a strain which lack these proteases can be used). When highly-expressed, the cell might not have enough regulatory proteins to help the protein fold properly, and so the result might be a high level of misfolded proteins, forming aggregates and inclusion bodies. The proteins might fold properly, but are non-functional; this is usually due to a missing co-factor or lipid, remembering that the relationship between lipid and membrane proteins are not a passive one and that lipids often play a functional role.

To determine the structure of membrane proteins, the membrane protein must be in a lipid environment as it would be in its native form. Detergents are used to mimic the effect. Detergents are used because they can be washed away and substituted more easily than lipids.

All detergents have a hydrophilic head with a hydrophobic tail. At low concentrations, detergent exists as monomers, but to function it needs to form a micelle incorporating the protein. Micelles only form at a concentration called the critical micelle concentration, and so detergents are most effective at its critical micelle concentration. The detergent uses its hydrophobic tail to associate with the hydrophobic transmembrane domain, and displaces lipid molecules. The detergent micelle forms a detergent belt around the TM domain leaving hydrophilic surfaces which makes contact with other proteins to form a crystal.

First, the cell is broken (using French press, hand-held homogenizer etc) to give many membrane patches with embedded membrane proteins. Detergent is then added at concentrations above the critical micelle concentration, and then washed until the critical micelle concentration and incubate at 4°C for 30 - 60 minutes to separate proteins from bilayer. Debris (cell remains) are deposited using centrifugation; the supernatant will contain the detergent-protein, which can be purified.

There are many detergents available, and the selection is based on the harshness as well as the size of its tail relative to the transmembrane region of the protein. Often, trial and error is the only way to identify a good detergent.

SDS is a detergent used in SDS-PAGE, but it is much too harsh for integral membrane proteins, so others must be used.

After the protein is solubilized, it is purified using column chromatography at low salt concentrations. The methods used are similar to that for soluble proteins, but membrane proteins will associate more strongly with the ?media??, which means the yield will generally be lower than for soluble proteins. The purity of the samples are assessed by SDS-PAGE; a purity of >85% is deemed acceptable.

However, because SDS-PAGE is a denaturing gel, it might denature the membrane protein and so it will run at a different molecular weight to its real molecular weight. With membrane proteins, they tend to run at a smaller molecular weight than its actual (it runs faster), partly due to it binding to SDS (which migrates down) and that SDS cannot fully denature the protein.

Membrane proteins can only form crystals by making contacts through their hydrophilic faces; thus a highly-hydrophobic protein will aggregate together and not be able to form large crystals. A possible solution to this is to use a smaller detergent (e.g. from a C12 detergent Dodecyl-B-D-maltoside, to a C8 detergent Octyl-B-D-glucopyranoside), so that more of the protein will be exposed to form contacts. However, the cost of this is that the protein will become less stable, and so it is a balancing act between the two factors. An alternative to this is to use a mixture of detergents, so it'd provide both the stability as well as exposing more of the protein.

Detergents typically destabilize membrane proteins so they loses its conformation; some detergents such as MNG-3 are engineered to minimize this effect. MNG-3 contains a quaternary carbon atom, which makes it very rigid and thus a more stable micelle is formed, giving more stable membrane protein. This increase in stability is apparent when the protein is heated up, most of the proteins stay intact and functional after.

Sometimes, a protein is only stable in a select few detergents. When the detergent molecules are not small enough for the protein to form a crystal, the protein must be made bigger to suit the detergent. This can be done by adding antibody fragments to the protein and be crystallized together. It increases the hydrophilic face to allow for better contact, and because the structure of the fragment is already known, it can be used as a starting point when trying to elucidate the structure of the protein itself. The structures of KcsA (potassium channel) and ClC (chloride channel) were elucidated using this antibody-based method. However, a drawback is the cost and time requirements.

An alternative is to engineer the protein to be more soluble or be more suited to a detergent. It can also be made more stable and so be able to withstand even harsher detergents. One such mutagenesis technique is alanine scanning mutagenesis. The C-terminal tail (often flexible) can be removed to improve stability. A very stable protein such as T4 lysozyme, can be added to make it easier to crystallize.

Synchrotron radiation sources can be used to give a more power and concise X-ray beam. A final note on X-ray crystallography is that the beam should be shot at different spots along the length of the crystal, this is because the beam destroys that part of the crystal and thus different spots gives better quality data.

Often no single clear crystal was obtained, in these instances, it might be possible to merge the data from several crystals together to give an averaged structure. Electron microscope provides low resolution (even cryo-EM), but gives the highest success rates for membrane proteins. If we can capture a protein at different angles/orientations on an EM image, then these different images can be used to construct a predicted 3D structure of the protein. These structures are low in resolution but can give a general shape.

This model can be improved if we compare it with a structurally similar homologous protein, for example the model from EM analysis of V0V1 ATPase can be improved by comparing it with a well known structure of F0F1 ATPase. And of course, data from crystallography and EM can be used together to form a better model than they would individually.

There are two types of NMR: liquid-state and solid-state NMR. Liquid-state NMR can often only be used for small proteins, and thus structures of larger membrane proteins might not be accurately predicted. The structure cannot be accurately determined using NMR, because it simply tells you the environment it is in and so gives a good molecular skeleton, but a lot of extrapolation and interpretation is used in elucidating the structure.