Medications from the computer

Computer simulation helped inactivate the flu virus with mini-proteins

Daria Spasskaya, N+1

A high-performance approach has been developed to search for therapeutic mini-proteins against clinically significant targets. Using computer modeling, an international team of scientists tested the interaction of tens of thousands of small proteins with an average length of 40 amino acids with hemagglutinin of the influenza virus and botulinum toxin. By sequentially selecting molecules with the highest binding efficiency, scientists managed to find a mini-protein that effectively protected mice from the flu. The study is published in Nature.

One of the most effective therapeutic approaches in the treatment of certain diseases is the use of proteins that bind to important targets on the surface of viruses or cancer cells and inactivate them. For example, antibodies to these proteins are often used to "turn off" surface proteins. However, antibodies have a complex structure, so they are often unstable and expensive to produce. A more promising direction is the search for small protein inhibitors that are easy to produce, thermally stable and do not cause an immune response when ingested.

To optimize and scale the search for therapeutic mini-proteins, scientists from the University of Washington and colleagues combined several high—performance methods - computer modeling, synthesis of oligonucleotides (short DNA chains) on chips, yeast display and sequencing.

In their work, the scientists set the task of selecting proteins that bind to two clinically significant targets — influenza virus hemagglutinin (HA) and botulinum toxin B (BoNT). Influenza hemagglutinin is present on the surface of the viral particle and is necessary for its attachment to cells. Its "shutdown" with the help of an antibody or a mini-protein leads to a decrease in the pathogenicity of the virus. Botulinum toxin is a nerve-paralytic bacterial toxin, ingestion of which leads to the development of botulism. The binding of botulinum toxin prevents its penetration into neurons and the development of a toxic reaction.

To simulate the binding of mini-proteins to targets, the authors created "virtual libraries" of thousands of amino acid sequences with several conformation variants. For "ordinary" large proteins, it is impossible or very difficult to predict the stacking of the polypeptide chain due to the large number of possible variants, however, in this case, scientists worked with sequences of about 40 amino acids for which the secondary structure is well predicted. Taking advantage of the fact that both selected targets were previously crystallized in combination with inhibitors, the scientists selected key target sites, binding to which would inactivate them.

By selecting about ten thousand variants of inhibitors for HA and BoNT using modeling, the authors synthesized DNA sequences encoding these mini-proteins (modern technologies for the synthesis of oligonucleotides allow synthesizing up to ten thousand sequences of about one hundred nucleotides at a time). Then the resulting "genes" were injected into yeast cells. As a result, the cells on their surface contained all the variants of the selected mini-proteins (this is the essence of yeast display technology).

The cells were incubated with different concentrations of fluorescently labeled targets, and then, using a cell sorter, the variants that most effectively bind the targets in the smallest concentration were selected. On the second attempt, the scientists managed to create a library in which three percent (342 molecules) bound HA well.

Following the results of the next selection stages, the authors left eight anti-botulinum molecules and six anti-hemagglutinin molecules for testing on cells. The best of them (HB1.6928.2.3 for hemagglutinin and Bot.671.2.1 for botulinum toxin) were crystallized together with the targets and showed that they not only bind the previously identified key areas of the target, but also additionally interact with neighboring surfaces, enhancing the effect. The anti-influenza mini-protein was also compared with the existing antibody for hemagglutinin and found that its effective concentration is one hundred times less than for the antibody.

The molecule HB1.6928.2.3 was also tested on mice. The internasal administration (into the nose) of an inhibitor both before and after infection with a lethal dose of the virus led to the fact that not a single mouse died. In addition, the molecule turned out to be hyperstable and practically did not cause an immune response in the body of animals, so scientists suggested that it should work well as a preventive agent, preventing the development of the disease for a long time after application.

Portal "Eternal youth" http://vechnayamolodost.ru  29.09.2017