Biophysical Methods

Lecture 3: Membrane Proteins

Modified October 1998

[Department of Chemistry and Biochemistry Home Page] [University of Guelph Home Page]

View of the membrane protein bacteriorhodopsin from the side, plane of membrane horizontal. The protein is shown in ribbon view, coloured by position in chain, N-terminal blue running through to red at the C-terminal.


Membrane Proteins

p>The organization of membrane proteins follows rules related to but distinct from those governing soluble globular proteins.

<

As recently as the mid-1960s, membrane protein was believed to form a beta sheet-like layer on the surfaces of the lipid bilayer. The Singer-Nicholson or Fluid Mosaic model was the first step towards our current understanding of the relationship between the lipid and protein components of membranes. This proposal first distinguished the roles of peripheral and integral proteins, the latter being globular folded proteins embedded in the bilayer by virtue of surface located non-polar amino acids which could interact with the fatty acid tails of phospholipids.

The original Singer-Nicholson model made no special correlation between alpha helix and membrane insertion, and presumed that membrane proteins would commonly be monotopic, i.e. exposed only to one side of the membrane. The idea that membrane proteins might traverse through to the opposite side of the membrane was considered rather daring.

In practice, monotopic proteins have turned out to be rare, and proteins associated with a single face are either peripheral or membrane-anchored. The latter are soluble globular proteins covalently bonded to a fatty acyl, polyprenyl or glycosylated phosphatidyl inositol (GPI) groups which are membrane embedded. The majority of true membrane proteins consist of one or more transmembrane alpha-helices, with helix axes normal or close to normal to the plane of the bilayer.

Bitopic proteins cross the membrane once and exposed separate domains on the extracellular and cytoplasmic faces. The extracellular domain may carry glycosyl side chains which may be N-linked (to Asn) or O-linked (to Ser, Thr or hydroxyLys). A typical role is to act as a cell surface marker, adhesion factor (such as integrin) or receptor. The cytoplasmic domain may play a role in cell signalling (tyrosine kinase or tyrosine kinase activator), or may connect to the cellular cytoskeleton.

A polytopic protein has many transmembrane segments, each alpha-helical. The helices arrange into a slightly splayed bundle. The splay arises because adjacent alpha-helices interlock side chains at certain favoured angles (20o or 50o). Polytopic helices are thus inclined at about 80o to the bilayer plane. Polytopic proteins include many hormone receptors, typically with 7 transmembrane segments, coupled to the heterotrimeric G-protein signalling pathway, and many channel or pore forming proteins involved in membrane transport. Both categories may be involved in cytoskeletal links.

For a review, see Singer, S.J. Ann. Rev. Cell Biol. 6: 247-296 (1990).

1) Hydropathy plots and the hydrophobic moment - identification of transmembrane segments of proteins.

Hydropathy: is a numeric representation of the tendency of each amino acid residue to enter a non-polar environment. A sequence is scanned in linearly and a moving average of hydropathy values is computed.

J. Kyte & R.F. Doolittle, J. Mol. Biol. 157, 105-132 (1982).

Experimental Thermodynamic scales:

e.g. measurement of free energy of transfer of amino acid from oil phase to water

Keq
A.A. (np) = A.A. (aq)

= -RT ln Keq (hydrophobic will give positive values);

Nozaki & Tanford, J.Biol. Chem. 246, 2211-2217 (1971).

This scale gives high value for Trp and Tyr due to large surface area, and a low value for Ala.

Statistical scales: e.g. Janin, Nature 277, 491 (1979); Sweet and Eisenberg, J. Mol. Biol 171 479 (1983); Engleman and Steitz, Ann. Rev. Biophys. & Biophys. Chem. 15 321 (1986). are based on mean % of surface area of each amino acid buried in protein interior (or bilayer interior) from compilations of existing structural data.

There is significant conflict between thermodynamic and statistical scales!

Aggregate scales: Kyte & Doolittle (see above), and Eisenberg et al. J. Mol. Biol. 179, 125-142 (1984). average out many existing published scales. Criticism: this eliminates distinctive features of a particular scale that may be applicable and optimized for a specific circumstance.

Significance of "window" length in hydropathy plots:
The hydropathy plot is based on a moving average H(n) spanning a "window" of 2n + 1 amino acids. The original Kyte & Doolittle paper used a window of 7 (n 3), with the intention that the researcher use judgment.

However, a small window can be noisy and misleading. Currently it's accepted that a window of 7 to 9 detects hydrophobic helices in globular proteins, but a much larger window of 19 to 25 is appropriate for detecting transmembrane helices. The larger window has a) a better smoothing effect, and b) usually only transmembrane segments are that long.

Hydrophobic moment analysis

Eisenberg et al. J. Mol. Biol. 179, 125-142 (1984).

The orientation of a side chain is predictable based on standard geometries of the alpha-helix and beta-sheet:

alpha helix: orientation of ith side chain is i x 100o

extended strand (beta sheet): orientation of ith side chain is i x 180o

Hydrophobicity is then treated as a vector or a quantity with both magnitude and direction. Hydrophobic moment is the vector sum of the individual hydrophobicities.

If all side chains are hydrophobic, then vectors cancel, and the hydrophobic moment is low.

If one side is hydrophilic, as in an amphipathic helix, its vectors are negative in magnitude, and reinforce the positive hydrophobic vectors on the opposite side.

high H(9), low : fully hydrophobic, monotopic, membrane spanning helix.

medium H(9), high : amphipathic, polytopic membrane spanning helix.

medium H(4), high : amphipathic helix surface of globular protein.

low H, low : globular protein, solvent exposed helix.

The combination of these procedures makes it possible to infer the structural organization, particularly of membrane proteins, from the amino acid sequence (which is often readily available).

Although the majority of transmembrane proteins are believed to be alpha-helical, in most cases this is inferred from hydropathy plots, with perhaps some experimental evidence based on accessibility of the non-helical segments or on site directed mutagenesis. There are still relatively few instances of high resolution structures of membrane proteins. The porins represent one major exception to the rule suggesting that transmembrane proteins are alpha-helical. Porins form wide-bore low specificity channels in the outer membrane of gram negative bacteria, and of mitochondria in eukaryotes, and consist of antiparallel beta sheets of 14-18 strands.

There appears to be no a priori structural grounds why proteins should not be inherently stable in the polypeptide arrangements originally suggested in the Singer model, which merely require exposure of non-polar side chains on the protein surface in the regions contacting the bilayer core. The real question is how such a structure might achieve its folded state.

The observed membrane protein structures thus appear to be based on the mechanism by which they fold or insert into the membrane.

The signal/anchor mechanism of membrane insertion.

The signal sequence was first discovered in relation to proteins destined for secretion. In eukaryotes, such proteins are synthesized on ribosomes associated with rough endoplasmic reticulum (ER). If this mRNA is translated in vitro in the absence of ER, the resulting polypeptide is larger, and includes an N-terminal leader sequence absent from the secreted form of the protein. It was demonstrated that the leader sequence acted as a signal that directed the nascent polypeptide to the protein translocating machinery of the ER. Remarkably, the same leader sequence serves as a signal for membrane translocation in bacteria such as E. coli even though the components of the protein translocation system are totally different.

For secreted proteins, the leader sequence is removed by the action of signal peptidase once the polypeptide is inside the ER compartment. The polypeptide than folds by the usual mechanisms, and the contents of the ER are then packaged for secretion by the Golgi apparatus.

The same recognition and packaging system also serves to deliver membrane proteins to the plasma membrane. For bitopic transmembrane proteins, an additional internal sequence called the stop-transfer sequence causes the polypeptide to halt before translocation is complete, and the protein remains embedded in the membrane. Orientation will be C-terminal remaining on the cytoplasmic side.

The N-terminal signal sequence is about 15-25 amino acids, usually with a positive charged lysine or arginine near the start, followed a leucine rich segment (which is likely to form an alpha helix). The stop transfer sequence consist of a long hydrophobic segment (19-25 amino acids) with some alpha-helical tendency. However, the bilayer environment appears to exert a significant bias towards helix formation. This shift in favour of alpha-helix formation can be duplicated in solution phase by adding trifluoroethanol to the aqueous solvent. For a bitopic protein, the hydrophobic segment will usually have a low hydrophobic moment.

The uniform distribution of non-polar side chains around the helix give it the characteristics of an inverted micelle, with the polar backbone C=O and N-H groups at the core.

Polytopic membrane proteins appear to be derived by a modification of this basic mechanism. Instead of the signal appearing as a single leader, multiple signal sequences occur internally in the polypeptide, alternating with stop transfer sequences. Unlike the N-terminal signal, the internal signal/anchor sequences are not cleaved by the signal peptidase that acts on N-terminal-signals. Internal signals are characterized by occurrence of charged amino acids at the helix boundary. The charges govern the orientation such that the end retained in the cytoplasm is positively charged, while the translocated end may be negative. Some bitopic membrane proteins lack N-terminal signal sequences, and may be inserted by the action of an internal signal/anchor sequence. This can account for bitopic transmembrane proteins with the wrong orientation, i.e. N-terminal on the cytoplasmic side.

The charge asymmetry provides a useful predictive rule for detecting and orienting transmembrane segments: L. Sipos and G. von Heijne, Eur. J. Biochem. 213: 1333-1340 (1993).

Polytopic transmembrane segments may be arranged to define a polar channel through the membrane. This can be determined by the presence of a distinct hydrophobic moment, or by the helical wheel method, which should demonstrate that the helices are amphipathic with clearly distingushed polar and non polar faces.

Some bitopic helices may also be amphipathic, which suggests that they form dimers or higher oligomers.

Bacteriorhodopsin as a model polytopic protein.

Because their structure depends being embedded in a bilayer, only a few membrane proteins have been crystallized, a necessary step for structure determination by diffraction methods. If prepared at high protein density, bacteriorhodopsin spontaneously forms crystalline 2-dimensional arrays, which have been analyzed by electron diffraction to about 3 angstroms resolution, which is sufficient to trace the polypeptide and locate amino acid side chains. More recently, a novel 3-dimensional cubic lipid phase has been exploited to obtain true 3-dimensional crystals, allowing structure determination to 2.5 angstroms by X-ray diffraction, E. Pebay-Peyroula et al. Science 277: 1676-1681 (1997).

Bacteriorhodopsin forms seven helical transmembrane segments A-G, enclosing the active chromophore retinal, which is covalently bonded to Lys 216. Helices ABC+EF define a polar channel which transfers H+ from cytoplasm to extracellular face in response to conformational changes induced by photoisomerization of the bound retinal. Cytoplasmic face on top, extracellular face below.

Several prolines (shown with rings coloured green) occur as expected at helix boundaries, but three are mid-helix. In addition, the helices are imperfect, with several interruptions in the regular H-bonding pattern (blue color; normal alpha-helix is shown in red.

The helix-like segment helix C from Trp 80-Val 101 is shown magnified, with hydrogen bonds indicated. Pro 91 is green. The proline ring breaks the chain of H-bonds, causing the helix to be slightly bent. This is a proline bulge. Non polar amino acid such as Leu and Ala face right indicating the periphery of the protein. The proline bulge enlarges the channel region in the vicinity of the retinal group. The bulge causes misalignment of several C=O groups both above and below the pro 91. Probably only the helix promoting property of the bilayer environment allows this segment to approximate helical with the proline in this position.

Pro 186 has a mid-helical position in helix F on the opposite side of the channel, causing a similar bulge there.

Close up of proline 91 in spacefilling display, adjacent to what would be its hydrogen bond acceptor, the C=O group of Leu 87. The C=O group is closely juxtaposed to non H-bonding -CH2- groups of the proline ring. There is probably some degree of van der Waals repulsion, forcing the C=O out of normal alignment. This affects the alignment of the C=O group of Phe 88 as well.

This image of helices A + EFG on the front face of the molecule has Asp + Glu in red, Lys + Arg in blue. The cytoplasmic side of the molecule faces the top of the image, and the correlation between helix orientation and charge (positive towards cytoplasm) can be clearly seen.

Glu 161 and Glu 166 at the top of the structure lie outside the helix boundaries, so do not conflict with the helix orientation rules.

Description of the bacteriorhodopsin channel.

Helices ABC+FG are lined on their inside faces by polar groups, forming a channel. Selective placement of the weak acid aspartate provides a series of stepping places for transit of H+ through the channel. In the middle is Lys 216, which forms a Schiff base with the chromophore, retinal. This lysine (4-atom chain in red) acts as the gate in the channel.

The ground state of the system has the retinal in trans stereochemistry. Photoisomerization to cis stereochemistry causes the Schiff base to reorient, so the proton is presented to Asp 85 (red) below it. The neighbouring Asp 212 (red) acts as a stabilizing counterion, H-bonding to the protonated Asp 85. The mutation of D85N causes 100% loss of activity, whereas D212N causes 30% reduction of activity. Further down towards the aqueous exterior, Arg 82 provides a complementary positive ion that induces rapid proton release from Asp 85 to the widening extracellular vestibule of the channel. The separation between Schiff base and Asp 85 is 4.1 angstroms, within range for direct transfer. However the rate of transfer suggests mediation through bound H2O.

Meanwhile, the deprotonated retinal-Schiff base reverts to the thermodynamically more stable trans state, presenting a vacant lone pair to the channel facing up to the cytoplasm. Asp 96 (blue), at the top of the channel is also susceptible to mutation suggesting that this is the proton donor. However the distance between Asp 96 and Lys 216 suggests that H+ transfer occurs via a chain of intermediate H2O molecules. The Lys-retinal Schiff base deprotonates an H2O, likely H-bonded to a variety of polar groups in this region (Thr 46,47,170), which then obtains the H+ from the cytoplasmic medium via Asp 96.

Neutron diffraction studies reveal the presence of at least 7 H2O molecules in the channel, and four of these are confirmed by the X-ray diffraction data. Working from the cytoplasmic face outwards, H2O molecules have been placed at Asp 96 (#5), between Asp 96 and Schiff base (#1), adjacent to Schiff base on cytoplasmic side (#6), between Schiff base and Asp 85 (#7), and two adjacent to Asp 85/Asp212/Arg82 complex (#3,4).

The bacteriorhoposin channel, viewed from the cytoplasmic face. Helices run from the top clockwise A-B-C-D (exiting bottom right), F-G on the left. Helix E is out of view at the bottom.

Helix C carries Asp 96 (near plane), Asp 85 and Arg 82 behind. Helix G carries Lys 216 and Asp 212. The distal end of the retinal is feld between bulky Trp 86, and Trp 182 and 189 on Helix F. This limits movement of the rigid retinal so that the photoinduced cis-trans isomerization is forced to reorient the Schiff base N+-H.

Spacefilling view of bacteriorhodopsin in cutaway section. Non polar amino acids in red, polar in light gray. However, Rasmol considers Tyr to be polar despite its benzene ring; several of the peripheral "polar" groups are actually Tyr. The core of the molecule encloses a polar channel. At right: A section near the cytoplasmic face shows Asp 96 and polar neighbours Thr 46, Thr 47.

Below left:A section near the midline shows Asp 85 and Asp 212 just outwards from the retinal. A slight cavity exists between these two amino acids.

Below right:A section near the extracellular face shows Glu 204 and Arg 82 in the central cavity. The cavity enlarges before opening to the extracellular surface.