SCIENCE OBSERVER
Fat Enough for Two Belts
Christopher Brodie
It sounds like a joke by the late comedian Rodney Dangerfield.
"My uncle is so fat, he needs two
belts." But high-density lipoprotein or HDL, the
so-called "good" cholesterol, actually does consist of two
protein belts cinched around a tiny disk of fat, according to a
study published in the March 2007 issue of the Journal of
Structural Biology. Computer simulations described in the
paper predict that this structure forms spontaneously within
microseconds from a chaotic mix of protein and lipid molecules.
HDL acts like a molecular sheepdog in the body, rounding up
cholesterol from the blood and herding it to the liver for
breakdown. Despite its importance, fundamental questions about HDL,
such as, "how does it form?" and "what does it look
like?" remain.
The HDL particle begins as a pair of apolipoprotein A-I (apo A-I)
proteins, which collect lipids into a disk and, eventually, a
sphere. Thus, the size and shape of HDL varies with the number of
lipids it contains, making direct observation difficult. "This
is one of those opportunities where only the computer can show how
the protein does its job," says senior author Klaus Schulten of
the Beckman Institute, part of the University of Illinois at
Urbana-Champaign. But for a complex of this size, atom-by-atom
simulations are limited to a few nanoseconds—too short to see
the process of aggregation. Instead, Schulten's team used a
"coarse grained" computer model, which grouped atoms
together to ease the computational burden.
Snapshots from the simulation (above, right) show ropy apo
A-I proteins (blue and orange) and a spray of lipids
(gray). From an initially disordered state (a),
the molecules quickly form clumps to hide their hydrophobic parts
from the surrounding water (b). Within a microsecond, the
pieces unite to form a single lipid-protein aggregate (c);
the final structure (e) emerges after a period of
additional wriggling (d). Data from a paper now in
preparation suggest that this terminal, double-belted structure
takes about 10 microseconds to form.
In supporting the double-belted model of HDL, the simulation
disproved Schulten's own hypothesis, the "picket fence"
model. Even simulations that started with apo A-I making zigzag
pickets around the disk soon reverted to a double-belted shape.
The paper prompts two important follow-up questions. First, how
sensitive is the model to the starting concentration of lipids?
Second, given that these HDL particles are so difficult to observe
in real life, how can the model be tested?
In recent months, graduate student and first author Amy Shih has
answered both questions in papers awaiting publication. In a series
of new simulations, Shih describes changes in the HDL structure with
varying concentrations of lipids. She is also collaborating with
Stephen Sligar, another Beckman investigator, on experiments using
the Advanced Photon Source at Argonne National Laboratory. Their
small-angle x ray-scattering work illuminates HDL assembly and transformation.
As significant as this work is for the study of cholesterol, it may
be most notable for figuring out how such neat, discrete packages
self-assemble. That is valuable because, in the real world,
biophysicists are trying to use similarly shaped particles to study
the structures of physically isolated membrane proteins.