Here comes the good stuff.

By using freeze fracture techniques, the composition and structure of cell membranes can be determined.  This involves freezing the tissue and then fracturing it open with a knife blade.  This should create an opening between the phospholipid layers.  A metal deposit (Pt/C) will cover the exposed protoplasmic and ectoplasmic faces and create replica of what the insides of a membrane look like.

First of all, it can best be described as a sea of phospholipids with proteins swimming within.  It can be defined as a physical boundary of controlled composition that serves to create (allow) and exploit differences.  Examples of differences are pH or energy.

The basic structure of the plasma membrane is a combination of phospholipids that have polar heads (blue circles in the picture below) with two nonpolar tails (yellow lines) each.  The two layers are put together with the hydrocarbon tails on the inside.  For your informaion, a membrane made of 2 phospholipid molecules is 4 nm thick.

An important concept is that differences in membranes come from differences in membrane properties.  It may have a fluid physical nature as a liquid crystal or it could be in a gel state.  The physical property is determined by the lipid population.   Also, a population of proteins is responsible for specific functions of a given membrane (i.e., making a membrane selectively permeable).

An important characteristic (as you may already have deduced from the structure of membranes) is that they are amphipathic (or amphiphilic), which means having both hydrophilic and hydrophobic regions. With the glycerol (polar) head and the tail as the hydrophobic hydrocarbon chain, a membrane broken up into a single layer in water will make a micelle (below) to hide its nonpolar tails. 

The phospholipid bilayer has hydrogen bonds, ionic bonds, and Van der Waals forces in the outer head.  The tail region is what actually separates the outside medium from the cytosol since it is not water soluble.  Dr. Blanton describes the nonpolar chains as "a thin layer of grease separating the cell from its environment."   This inside portion contains hydrophobic and Van der Waals forces.  Movement among its carbon-hydrogen bonds creates kinetic energy.  Through all this, lipids are rafts in these seas of polar leaflets (heads).

Of course, different types of membranes will have different protein to lipid ratios.  It is 1:1 in the plasma membrane.  In the inner membrane of the mitochondria, there is a 3:1 ratio of proteins to lipids.  An additional observation is the asymmetric distribution of lipids from one leaflet to another.  Once again function can dictate the amount of exoplasmic and cytosolic lipids.  Membranes with much involvement in reception of extracellular ligands (molecules that can bind to receptors due to complementary structure), and membranes involved with the display of markers for attracting macrophages to itself tend to have more of the appropriate lipids on the outer leaflet.

Variation can be found in the head groups.  This is the source of steric and electrostatic interactions.  Variation in the tails comes in the form of length (# of carbon atoms) and fluidity which can be compared by their different phase transition temperature (amount of Van der Waals forces to overcome).  Another point is saturation.  An increase in the number of double bonds lowers the phase transition temperature.  This is because the kinks will disrupt the order among the tails.  Van der Waals forces are therefore weaker.  Plants are adapted to the harsh environment so they have more unsaturated tails in their cell membranes than animals do.  This allows for the plant membranes to retain fluidity at lower temperatures.  Cotton and blue-green algae can also use desaturases that introduce double bonds into phospholipid tails so that their cell membranes will stay fluid in colder environments. As you may have noticed, vegetable oils used for cooking are liquid while lard from animals is solid.  What about Crisco all-vegetable shortening? If you look on the label beneath the nutrition information, it will inform you that it has been "partially hydrogenated (double bonds removed) for freshness and consistency."  The result is better packing.  This is why it is solid at room temperature.

The word "protein" may have picked up some different connotations, but the word "cholesterol" has been bad mouthed by many.  Some of these multi-ring lipid molecules live in phospholipids, but they will argue (if they could) that they do good deeds as well.  The 1' OH polar group of cholesterol will obviously reside in the leaflet areas.  The rings that make it up contribute to their stiff, thick bodies.  As a result, the tails cannot be packed as tightly.  The phase transfer temperature range is broadened as a result.

1) Fusion of a mouse and a human cell allows for the monitoring of the proteins found on the new composite cell.  At the beginning, one half of the new fused cell contains membrane proteins from one organism while the other half has the other's.  Then protein specific antibodies for humans are colored differently than that of the mouse's.  When the antibodies are thrown into the mix to bind with the proteins, the two halves are colored differently as expected.  After 40 minutes or so, these colors have intermingled.  This displays the fluidity of proteins since they have been shown to move through the membrane.

2) An electric current can be applied to a membrane which will cause proteins to relocate as a response.

3) Fluorescence recovery after photobleaching (FRAP) is another technique that supports membrane fluidity.  A certain area of the membrane is bleached so that it will not reemit light at the same wavelength as the rest of its neighbors.  After time has elapsed, the color will return which hints at the movement of non-bleached portions into the previously bleached area.

Here are three classes of membrane proteins:

1) Extrinsic, Peripheral, Absorbed - these can be removed from the membrane by gentle methods (changing the pH or ionic conditions, etc.) for they are noncovalently bonded to the inner and outer leaflets.

2) Intrinsic, Integral, Transmembrane - these proteins cannot be removed by gentle methods.  Therefore, detergents (like Triton X-100) are used because they are amphipathic.  Integral proteins that penetrate the membrane once and therefore have 2 domains are called bitopic.  Their carboxyl (C) end will be outside the cell while the amino (N) end remains in the cytoplasm.  Polytopic integral proteins weave in and out of both sides of phospholipid bilayers many times so they will have multiple domains.  Their C and N termini can be on the same or different sides of the membrane.

3) Proteins covalently attached to membrane lipids - these can usually be removed by specific enzymes.  Two examples of this type include GPI-anchored and lipid-modified proteins.  They are found on both sides of the cell, but covalently bind to the membrane glycolipids or hydrophobic groups.

Let's look at #2, the integral proteins in more detail...  You may ask, how do their amino acids live in the nonpolar regions of the membrane.  After all, the R groups can be polar and the C and N groups are definitely polar.  Well, these proteins are in the form of alpha helices so the hydrogen bonding will take care of the carboxyl and amino groups in the amino acids.  Integral proteins end up containing nonpolar residues that are tightly sealed into the lipid layer so that the membrane retains its barrier status.  It takes about 20 - 22 amino acids to go through the
4 nm of the membrane.  Trying to identify a protein?  If it has 20 - 22 nonpolar amino acids, there's a good chance that it's a integral protein.  Computer-aided sequence analysis can help with interpretation of proteins.  For example, high and wide peaks on the positive side of a hydropathy plot will reveal hydrophobic amino acids.  A peak in the negative region of the graph would mean the residue is negatively charged.  A lack of strong peaks would hint at a cytosolic protein.