AI-based design of pharmacologically relevant targets with target properties

Given: protein template in the form of a protein sequence or structure; target properties.

Predict: protein sequence that satisfies target properties and has minimal differences from the given template.

AI-based engineering of G protein-coupled receptors with enhanced stability

Molecular design is one of the most important and rapidly

developing fields in biotechnology. Optogenetics tools in

neurobiology, fluorescent proteins in cellular biology,

sequencing nanopores in molecular biology, drug discovery

in medicinal chemistry and many other examples in modern

biotechnology are based on protein engineering. With an

accumulation of biophysical data, AI-based approaches

become beneficial in protein design for biotechnology.

Typically, protein design starts with a template a protein

from a human or any other living organism and with a

target property, for example, protein stability or spectral

shift. Then, the goal is to modify the template to obtain an

engineered protein with the target property.

A particular case is to design stable forms of pharmacological

targets, such as G protein-coupled receptors (GPCRs).

Malfunctions of these receptors typically lead to various

diseases: neurodegenerative, oncological and cardiovascular

diseases, asthma, depression, obesity, drug dependence, etc.

GPCR receptors are one of the main targets for

pharmacological companies, and about 1/3 of all drugs

produced in the world are oriented on GPCRs. For the

development of more efficient and safer drugs, as well as

personalized drugs that take into account the characteristics

of the human genome (mutation), it is necessary to

understand how GPCRs work on a structural level. Obtaining

the spatial structure of a single receptor is an extremely

difficult and resource-intensive task. We developed an

innovative AI-based digital platform for GPCR design, which

allowed for a technological breakthrough in obtaining spatial

structures of GPCR ([259], [260]) . Thanks to the developed

technology over the last few years, spatial structures of ~10

GPCR receptors were determined, i.e. > 10 % of all receptors

with a known spatial structure to date. These include

relevant pharmacological targets, such as the human

cannabinoid receptor [261], the human serotonin receptor

([259], [264]), the human prostaglandin receptor [262], the

frizzled human receptor [263], the human adenosine

receptor [265], and the human cysteine receptors of types

one [266] and type two [267]. Structural analysis of each new

receptor has opened up opportunities for the rational

development of a new generation of drugs.