Nature 438, 569-570 (1 December 2005) | doi:10.1038/438569a
Cell biology: A greasy grip
Anthony G. Lee1
Top of page
Abstract
How do the lipids and proteins of the cell membrane interact to create a functioning barrier for the cell? A high-resolution structure of a membrane protein reveals intimate contacts with its lipid neighbours.
The design of a biological membrane is beautifully simple: a lipid bilayer provides the basic barrier, and into this are plugged a variety of membrane proteins. Each protein is designed to carry out some particular function for the cell — moving specific molecules in or out of the cell, say, or sensing the environment. The trick is to make sure that the lipid bilayer and the membrane proteins are mutually compatible: the proteins must be able to operate well in the environment provided by the lipid bilayer, and insertion of the proteins into the bilayer must not make it leaky, or the permeability barrier would be destroyed. In other words, the key to an effective membrane is to get the packing of the lipids and proteins right. On page 633 of this issue, Walz and colleagues1 present high-resolution structural data that clearly illustrate how this packing is achieved.
Thirty years ago, studies of biological membranes using electron spin resonance detected a population of lipid molecules whose fatty acyl chains were conformationally disordered2. These chains were thought to belong to the lipid molecules — called boundary or annular lipids — that contacted membrane proteins. But not everyone was convinced, hence the significance of Walz and colleagues' work; nothing is as compelling as actually seeing something with your own eyes.
The authors have used electron crystallography to determine the structure of a protein called lens-specific aquaporin-0 (AQP0) when it is immersed in an artificial lipid bilayer1. AQP0 is the most abundant protein in the plasma membranes of the fibre cells that make up the bulk of the lens in the human eye. Functionally, AQP0 mediates rapid movement of water into and out of the fibre cells, but it also has a structural role, forming membrane junctions between fibre cells. Under the right conditions (when at least a fraction of the AQP0 molecules are partially proteolytically cleaved) the AQP0–lipid-bilayer system forms two-dimensional crystals consisting of a pair of closely spaced membranes.
The structure that Walz and colleagues obtained from these crystals is a remarkable technical achievement: it is the highest-resolution structure ever produced from electron crystallography, and only the second to show the lipid environment of a membrane protein. The other structure showing this is of bacteriorhodopsin from the purple membranes of a photosynthetic bacterium3.
Walz and colleagues' structure shows the AQP0 molecules in their closed form, organized as complexes of four AQP0 units (tetramers) in each lipid bilayer. The flat extracellular surface of a tetramer in one bilayer contacts the extracellular surface of another AQP0 tetramer in the adjacent bilayer. This 'head-to-head' packing links the two adjacent bilayers to form a structure that looks very like an in vivo cell–cell junction4.
The artificial lipid bilayer is made of dimyristoylphosphatidylcholine, a molecule with a 'head' containing a negatively charged phosphate group and a positively charged choline group, and two 'tails' of fatty acyl chains, each with 14 carbons. In Walz and colleagues' structure, the lipids can be seen forming a shell around each AQP0 tetramer (Fig. 1). The tetramers are separated by a shell of lipid molecules, just one molecule thick. This makes the lipid molecules unusual in that most are in contact with two protein molecules, one from each of the adjacent tetramers.
Figure 1: Annular lipids.
A side view of the electron crystallographic structure reported by Walz and colleagues1 showing one face of the AQP0 tetramer with lipid molecules forming a bilayer shell around the protein. The AQP0 tetramer is shown as a surface plot (the lighter background molecule), with red representing regions of negative charge, blue, regions of positive charge, and grey, uncharged regions. The lipid molecules are shown in space-fill format (the molecules in the foreground). The charged lipid headgroups (oxygen, red; phosphorus, orange) and the lipid fatty acyl chains form a bilayer with almost uniform thickness around the protein. Presumably, in the membrane, lipid fatty acyl chains will cover the whole of the hydrophobic surface of the protein; only the most ordered of the lipid molecules will be resolved in the crystallographic structure.
High resolution image and legend (56K)
Perhaps this partly explains why so many of the lipid molecules are resolved in the structure. In the only other high-resolution structure showing a large number of lipid molecules, that of bacterio-rhodopsin, lipid molecules similarly mediate ordered packing of the protein molecules, in this case of trimers of bacteriorhodopsin3. The crystalline array of closely packed AQP0 tetramers seen in these reconstituted systems is very like that seen in the native lens fibre membrane5.
The striking thing about the structure of AQP0 is the excellent packing of the lipid fatty acyl chains against the rough surface of the AQP0 molecule. For some lipid molecules, this is achieved with almost straight (all-trans) fatty acyl chains, but for others considerable distortion of the chains is necessary for them to wrap around the bulky side-chains of the protein (Fig. 1). The lipid headgroup conformations also differ markedly between the various lipid molecules with some lipid headgroups being oriented almost parallel to the surface of the membrane, as occurs in crystals of phosphatidylcholines6, but with others being oriented almost vertically.
What the structure makes clear is that the AQP0 surface is not covered by a set of uniform binding sites for phospholipid molecules. The binding of lipids to AQP0 is therefore very unlike the interactions between many other phospholipids and proteins, where particular lipids bind to unique binding sites on the surface of the protein, a classic example being the highly specific binding of phosphatidylinositols to protein domains called PH domains. Rather, the picture presented is of a number of phospholipid molecules, each with their fatty acyl chains interacting with the large, rough, hydrophobic surface of the protein; their charged headgroups interact with the charged residues flanking the hydrophobic surface. Another interesting feature of the structure is that the thickness of the lipid bilayer around the AQP0 tetramer is rather uniform, despite the fact that the structures adopted by the individual lipid molecules are very different. The lipid bilayer observed within the recently published structure of another membrane protein similarly seems to be of constant thickness7.
The picture given here is, of course, of a membrane frozen in time — the structure was determined at low temperature. At normal temperatures the lipid bilayer would be more fluid; lipid molecules would probably be rapidly entering and leaving the annular shells around the protein molecules, and the encounter between a particular lipid molecule and a particular protein molecule would be brief8. Nevertheless, the structures adopted by the lipid molecules when they are on the protein surface would be much like those pictured here, and fast swapping of lipid molecules between the annular shell and the bulk lipid bilayer would not change the environment 'experienced' by the protein. The protein would always experience lipid molecules in the disordered states shown in the structure of Walz and colleagues1.
Crystalline membranes of AQP0 can be formed with a variety of lipids other than dimyristoylphosphatidylcholine1. It might therefore be possible to use this system to answer several questions exercising the minds of membranologists. If AQP0 were reconstituted with a longer chain lipid, how would the long fatty acyl chains of the lipid molecules distort to ensure that the hydrophobic thickness of the lipid bilayer matched the hydrophobic thickness of the protein? How would lipids such as phosphatidylethanolamine that prefer to form non-bilayer structures interact with AQP0? Would changing the structure of the lipid result in any change in the structure of the AQP0 molecules? With answers to these and similar questions, we can start to understand how lipid and protein molecules coevolved to form membranes that are fit for their purpose.
Cell biology: A greasy grip
Anthony G. Lee1
Top of page
Abstract
How do the lipids and proteins of the cell membrane interact to create a functioning barrier for the cell? A high-resolution structure of a membrane protein reveals intimate contacts with its lipid neighbours.
The design of a biological membrane is beautifully simple: a lipid bilayer provides the basic barrier, and into this are plugged a variety of membrane proteins. Each protein is designed to carry out some particular function for the cell — moving specific molecules in or out of the cell, say, or sensing the environment. The trick is to make sure that the lipid bilayer and the membrane proteins are mutually compatible: the proteins must be able to operate well in the environment provided by the lipid bilayer, and insertion of the proteins into the bilayer must not make it leaky, or the permeability barrier would be destroyed. In other words, the key to an effective membrane is to get the packing of the lipids and proteins right. On page 633 of this issue, Walz and colleagues1 present high-resolution structural data that clearly illustrate how this packing is achieved.
Thirty years ago, studies of biological membranes using electron spin resonance detected a population of lipid molecules whose fatty acyl chains were conformationally disordered2. These chains were thought to belong to the lipid molecules — called boundary or annular lipids — that contacted membrane proteins. But not everyone was convinced, hence the significance of Walz and colleagues' work; nothing is as compelling as actually seeing something with your own eyes.
The authors have used electron crystallography to determine the structure of a protein called lens-specific aquaporin-0 (AQP0) when it is immersed in an artificial lipid bilayer1. AQP0 is the most abundant protein in the plasma membranes of the fibre cells that make up the bulk of the lens in the human eye. Functionally, AQP0 mediates rapid movement of water into and out of the fibre cells, but it also has a structural role, forming membrane junctions between fibre cells. Under the right conditions (when at least a fraction of the AQP0 molecules are partially proteolytically cleaved) the AQP0–lipid-bilayer system forms two-dimensional crystals consisting of a pair of closely spaced membranes.
The structure that Walz and colleagues obtained from these crystals is a remarkable technical achievement: it is the highest-resolution structure ever produced from electron crystallography, and only the second to show the lipid environment of a membrane protein. The other structure showing this is of bacteriorhodopsin from the purple membranes of a photosynthetic bacterium3.
Walz and colleagues' structure shows the AQP0 molecules in their closed form, organized as complexes of four AQP0 units (tetramers) in each lipid bilayer. The flat extracellular surface of a tetramer in one bilayer contacts the extracellular surface of another AQP0 tetramer in the adjacent bilayer. This 'head-to-head' packing links the two adjacent bilayers to form a structure that looks very like an in vivo cell–cell junction4.
The artificial lipid bilayer is made of dimyristoylphosphatidylcholine, a molecule with a 'head' containing a negatively charged phosphate group and a positively charged choline group, and two 'tails' of fatty acyl chains, each with 14 carbons. In Walz and colleagues' structure, the lipids can be seen forming a shell around each AQP0 tetramer (Fig. 1). The tetramers are separated by a shell of lipid molecules, just one molecule thick. This makes the lipid molecules unusual in that most are in contact with two protein molecules, one from each of the adjacent tetramers.
Figure 1: Annular lipids.
A side view of the electron crystallographic structure reported by Walz and colleagues1 showing one face of the AQP0 tetramer with lipid molecules forming a bilayer shell around the protein. The AQP0 tetramer is shown as a surface plot (the lighter background molecule), with red representing regions of negative charge, blue, regions of positive charge, and grey, uncharged regions. The lipid molecules are shown in space-fill format (the molecules in the foreground). The charged lipid headgroups (oxygen, red; phosphorus, orange) and the lipid fatty acyl chains form a bilayer with almost uniform thickness around the protein. Presumably, in the membrane, lipid fatty acyl chains will cover the whole of the hydrophobic surface of the protein; only the most ordered of the lipid molecules will be resolved in the crystallographic structure.
High resolution image and legend (56K)
Perhaps this partly explains why so many of the lipid molecules are resolved in the structure. In the only other high-resolution structure showing a large number of lipid molecules, that of bacterio-rhodopsin, lipid molecules similarly mediate ordered packing of the protein molecules, in this case of trimers of bacteriorhodopsin3. The crystalline array of closely packed AQP0 tetramers seen in these reconstituted systems is very like that seen in the native lens fibre membrane5.
The striking thing about the structure of AQP0 is the excellent packing of the lipid fatty acyl chains against the rough surface of the AQP0 molecule. For some lipid molecules, this is achieved with almost straight (all-trans) fatty acyl chains, but for others considerable distortion of the chains is necessary for them to wrap around the bulky side-chains of the protein (Fig. 1). The lipid headgroup conformations also differ markedly between the various lipid molecules with some lipid headgroups being oriented almost parallel to the surface of the membrane, as occurs in crystals of phosphatidylcholines6, but with others being oriented almost vertically.
What the structure makes clear is that the AQP0 surface is not covered by a set of uniform binding sites for phospholipid molecules. The binding of lipids to AQP0 is therefore very unlike the interactions between many other phospholipids and proteins, where particular lipids bind to unique binding sites on the surface of the protein, a classic example being the highly specific binding of phosphatidylinositols to protein domains called PH domains. Rather, the picture presented is of a number of phospholipid molecules, each with their fatty acyl chains interacting with the large, rough, hydrophobic surface of the protein; their charged headgroups interact with the charged residues flanking the hydrophobic surface. Another interesting feature of the structure is that the thickness of the lipid bilayer around the AQP0 tetramer is rather uniform, despite the fact that the structures adopted by the individual lipid molecules are very different. The lipid bilayer observed within the recently published structure of another membrane protein similarly seems to be of constant thickness7.
The picture given here is, of course, of a membrane frozen in time — the structure was determined at low temperature. At normal temperatures the lipid bilayer would be more fluid; lipid molecules would probably be rapidly entering and leaving the annular shells around the protein molecules, and the encounter between a particular lipid molecule and a particular protein molecule would be brief8. Nevertheless, the structures adopted by the lipid molecules when they are on the protein surface would be much like those pictured here, and fast swapping of lipid molecules between the annular shell and the bulk lipid bilayer would not change the environment 'experienced' by the protein. The protein would always experience lipid molecules in the disordered states shown in the structure of Walz and colleagues1.
Crystalline membranes of AQP0 can be formed with a variety of lipids other than dimyristoylphosphatidylcholine1. It might therefore be possible to use this system to answer several questions exercising the minds of membranologists. If AQP0 were reconstituted with a longer chain lipid, how would the long fatty acyl chains of the lipid molecules distort to ensure that the hydrophobic thickness of the lipid bilayer matched the hydrophobic thickness of the protein? How would lipids such as phosphatidylethanolamine that prefer to form non-bilayer structures interact with AQP0? Would changing the structure of the lipid result in any change in the structure of the AQP0 molecules? With answers to these and similar questions, we can start to understand how lipid and protein molecules coevolved to form membranes that are fit for their purpose.
