This is a great question and one that many people in the field of membrane proteins / crystallography will also ask, quite frequently. It's possible to crystallise membrane proteins in detergent and this has been done on quite a number of occasions, with good structures obtained in some. As you correctly infer, there is some reasonable debate as to whether or not these proteins, in detergent rather than the lipid bilayer of a cell, are in their native conformations. Clearly the way a protein interacts with lipid is going to be different to the way it interacts with detergent. Furthermore, we know that in some cases, interaction between a membrane protein and specific phospholipids can modulate function. So it's reasonable to speculate that this might be due to a conformational change (and that such conformational change might also occur in a detergent micelle). If you would like to read up on one example, I would point you in the direction of Abcb1a (the murine homologue of MDR1/ABCB1), which has been crystallised in detergent and, to some people, the solved structure looks a bit... odd... perhaps? ABCB1 is also known to be functionally influence by the lipid environment (PS flopping by P2X7 is inhibitory... there is a paper by a guy called James Elliott on this).

One can also crystallise proteins in lipids but, since the goal is to achieve a homogeneous crystal, one would presumably elect to use purified (reconstituted) phospholipids which, again, are not identical to the lipid bilayer of a plasma membrane.

This is not my opinion, of course (regarding the crystal structure), I am just reflecting some of the general opinions in our field and it might well be that the structure presented is perfectly valid. Or it could be that, if the structure contains artefacts, they are due to non-physiolocal, intermolecular, interactions that alter conformation.

Some people get around these problems by expressing and crystallising truncated forms of their favourite membrane protein, without the transmembrane domains. Be that NBDs alone, in solution (for ABC transporters) or extracellular domains in the case of some receptors which have minimal transmembrane domains. Here you will have proteins crystallised in more physiological conditions.

Finally, it is possible in some cases to do experiments on cells expressing your membrane protein of interest in order to test the veracity of a crystal structure. For example, one could use a range of cross linking reagents (on an appropriately engineered protein, with carefully placed cysteins) to infer intramolecular distances under various conditions (at different stages of a transport cycle, for example). Alternatively, one could use EPR (again, with an appropriately engineered protein) under a rang of conditions. The list goes on but I think you get the point!

I didn't start out with the intention of writing an essay... Whoops!

This is a great question and one that many people in the field of membrane proteins / crystallography will also ask, quite frequently. It's possible to crystallise membrane proteins in detergent and this has been done on quite a number of occasions, with good structures obtained in some. As you correctly infer, there is some reasonable debate as to whether or not these proteins, in detergent rather than the lipid bilayer of a cell, are in their native conformations. Clearly the way a protein interacts with lipid is going to be different to the way it interacts with detergent. Furthermore, we know that in some cases, interaction between a membrane protein and specific phospholipids can modulate function. So it's reasonable to speculate that this might be due to a conformational change (and that such conformational change might also occur in a detergent micelle). If you would like to read up on one example, I would point you in the direction of Abcb1a (the murine homologue of MDR1/ABCB1), which has been crystallised in detergent and, to some people, the solved structure looks a bit... odd... perhaps? ABCB1 is also known to be functionally influence by the lipid environment (PS flopping by P2X7 is inhibitory... there is a paper by a guy called James Elliott on this).

One can also crystallise proteins in lipids but, since the goal is to achieve a homogeneous crystal, one would presumably elect to use purified (reconstituted) phospholipids which, again, are not identical to the lipid bilayer of a plasma membrane.

This is not my opinion, of course (regarding the crystal structure), I am just reflecting some of the general opinions in our field and it might well be that the structure presented is perfectly valid. Or it could be that, if the structure contains artefacts, they are due to non-physiolocal, intermolecular, interactions that alter conformation.

Some people get around these problems by expressing and crystallising truncated forms of their favourite membrane protein, without the transmembrane domains. Be that NBDs alone, in solution (for ABC transporters) or extracellular domains in the case of some receptors which have minimal transmembrane domains. Here you will have proteins crystallised in more physiological conditions.

Finally, it is possible in some cases to do experiments on cells expressing your membrane protein of interest in order to test the veracity of a crystal structure. For example, one could use a range of cross linking reagents (on an appropriately engineered protein, with carefully placed cysteins) to infer intramolecular distances under various conditions (at different stages of a transport cycle, for example). Alternatively, one could use EPR (again, with an appropriately engineered protein) under a rang of conditions. The list goes on but I think you get the point!

I didn't start out with the intention of writing an essay... Whoops!

Dear Edward, it was a nice answer indeed... you did clear my doubts... thanx for that... again one more question, Dont people do some computational calculations in order to confirm that whatever structure they got after solving the crystal is what they should get????

Edward very nicely outlined the possibilities of membrane protein crystallization. If I may add, some people have managed to successfully crystallize membrane proteins using nanodiscs (the protein has been embedded/reconstituted on a pool of lipids which are surrounded by scaffold proteins), which presumably ideally deliver a situation that corresponds to a real lipid bilayer. Of course it all comes down to the massive variety of lipids that occur in vivo. Nanodiscs can also be used to simplify purification and even biochemical studies.

Argh! How could I forget nanodiscs?! :)

With regards to your follow up question Nishant, people will use various forms of computational analysis to test their structures in silico. Perhaps a bone fide crystallographer will appear here and shed some more light on what exactly those processes are. One suspects that, on a very simplistic level, it will involve trying to achieve the lowest energy state, satisfying hydrophobic domains (such that they are not exposed to water)... among other things. These, however, do not fully 'confirm' that their structure is correct but may be used to rule out structures that are definitely incorrect.

Again, referring back to ABCB1 (because that's what most closely relates to my work, as you may have guessed / read), the controversial thing about this structure is that the nucleotide binding domains (NBDs) are very far apart. I can't remember how many Angstroms, exactly, but we have written a review which features this stuff... see my very short list of publications (it's just easier for me to refer you to my work than search pubmed for something else!). Anyway, the point is that, unless I am mistaken, computational analysis alone would not be able to test whether the reported degree of NBD separation genuinely reflects what happens in cells, or is an artefact of crystallisation.

Thanks a lot Edward and Arne for this fruitful discussion.. :)

You're most welcome :)

Nature. 2006 Nov 9;444(7116):179-80. Epub 2006 Oct 11.

Crystal structure of a rhomboid family intramembrane protease.

Wang Y, Zhang Y, Ha Y.

Source

Department of Pharmacology, Yale School of Medicine, 333 Cedar Street, New Haven, Connecticut 06520, USA.

Abstract

Escherichia coli GlpG is an integral membrane protein that belongs to the widespread rhomboid protease family. Rhomboid proteases, like site-2 protease (S2P) and gamma-secretase, are unique in that they cleave the transmembrane domain of other membrane proteins. Here we describe the 2.1 A resolution crystal structure of the GlpG core domain. This structure contains six transmembrane segments. Residues previously shown to be involved in catalysis, including a Ser-His dyad, and several water molecules are found at the protein interior at a depth below the membrane surface. This putative active site is accessible by substrate through a large 'V-shaped' opening that faces laterally towards the lipid, but is blocked by a half-submerged loop structure. These observations indicate that, in intramembrane proteolysis, the scission of peptide bonds takes place within the hydrophobic environment of the membrane bilayer. The crystal structure also suggests a gating mechanism for GlpG that controls substrate access to its hydrophilic active site.

Comment in

Structural biology: enzyme theory holds water. [Nature. 2006]

PMID: 17051161 [PubMed - indexed for MEDLINE]

Publication Types, MeSH Terms, Substances, Secondary Source ID

LinkOut - more resourc