By
developing software that uses 3-D models of proteins involved in cystic
fibrosis, a team of scientists at Duke University has identified
several new molecules that may ease the symptoms of the disease.
Computer
algorithms created by the team predict how well a given molecular
structure will block a basic protein-protein interaction known to occur
in cystic fibrosis. To test the predictions, the scientists synthesized
the molecules and measured how well they attached to one of the proteins
in that interaction. The team then placed the best molecule into human
cells with the cystic fibrosis mutation in a laboratory dish and found
that their new drug blocked the protein-protein interaction and
increased the cells’ ability to balance salt and water levels.
The
results, which appear in the April 19 Public Library of Science
Computational Biology journal, suggest that computers could make drug design for
cystic fibrosis faster.
“We
have known the genetic cause of cystic fibrosis since 1985. Now, by
understanding its biology and chemistry, we can design and create
targeted drugs to correct for the genetic flaw,” said Bruce Donald, a
Duke computer scientist and biochemist who led the study.
Cystic
fibrosis, or CF, is a childhood disease causing the lungs and pancreas
to fill with mucus, making it hard to breathe and absorb nutrients from
food. The mucus builds in the organs as the levels of salt and water in
the cells become unbalanced because of a defective protein.
That
protein, called CFTR, the cystic fibrosis transmembrane conductance
regulator, regulates salt and water in the cell. In CF, it is defective
because the genes that generate it are mutated. CFTRs are routinely
rounded up for recycling in the cell by a protein called CAL that binds
to CFTR and hauls it away. But defective CFTR proteins in cystic
fibrosis patients send a signal that they are faulty, making their
recycling rate much higher.
Currently,
no treatments exist to target the genetic mutations that cause cystic
fibrosis. Scientists have discovered molecules that target CFTRs’
defects, such as incorrect folding and fast recycling, and there are a
few molecules that help correct how CFTR folds or slow down the CAL
recycling truck. These molecules help keep copies of CFTR functioning in
the cell membrane to maintain some balance between salt and water
levels.
Donald
and his graduate student Kyle Roberts thought that computer algorithms
based on the structure of CAL and similar proteins could quickly
generate several dozen more molecules for slowing recycling by CAL and
increase the pool of potential cystic fibrosis treatments.
“Research
shows that you only need a fraction of normal CFTR activity to
alleviate cystic fibrosis symptoms, so keeping CFTR in the membrane by
using our inhibitors could have a significant therapeutic effect,” said
Roberts, first author of the new study.
Donald
and Roberts’ algorithms searched several thousand potential inhibitors
and ranked them based on how strongly it predicted each would bind with
CAL. In collaboration with researchers at Dartmouth and in Germany, the
scientists synthesized 11 of the highest-ranked sequences and used
fluorescent light to measure each molecule’s attachment to CAL.
The
results show that many of the algorithm-generated molecules attach more
strongly to CAL than the connection between CAL and CFTR in nature. The
best computer-generated molecules also bind more efficiently to CAL
than any previously reported inhibitor.
In
a culture of human cells with the cystic fibrosis mutation, the best
algorithm-generated inhibitor increased CFTR activity by 12 percent.
Donald said the new molecule could be used in combination with another
molecule, which corrects how CFTR proteins fold and raises CFTR’s
activity by 15 percent. The two molecules should work together and could
increase CFTR’s activity by about 27 percent, he said.
He
cautioned that it could be several years before patients with the
disease could use the new molecular combination as treatment because the
molecules have not yet been tested in patients with the disease. The
team has made its software freely available, Donald said, so the
computer-design approach could quicken the pace at which molecules and
resulting cystic fibrosis therapies are developed.
The study was funded by the National Institutes of Health.
Computational Design of a PDZ Domain Peptide Inhibitor that Rescues CFTR Activity
Source: Duke University