Designing Focused Chemical Libraries for High-throughput Screening in Pharmaceutical Research – Ash Ermish

Legend: Focused library selection and design is an essential step in any drug development screen. What is included in the library can affect how many hits and leads are obtained from the screening process. Computational technology has greatly improved quality of libraries by looking at 3D structural interactions, but it is limited to compounds that have a known structure. This technology is only one, or a part of, many construction methods, but it has resulted in the most improvement in recent years.

Introduction:

The development of high-throughput technology revolutionized the field of drug discovery by allowing researchers to screen thousands of potential drug compounds in one day using a miniature assay format and automation. Leads for potential drugs were able to be identified much quicker, yet there are increasingly fewer drugs being released on the market per year3,4. Part of the issue lies in the limitations encountered when trying to develop new and innovative chemical libraries that are diverse and meet the standards set by the medical industry. To understand what challenges scientists are facing, it is important to first know what kind of chemical libraries are being used in research today and how they are constructed.

Summary:

Chemical libraries, also known as compound libraries, are a critical step to any assay set-up in high-throughput screening. Due to the high numbers of compounds in the libraries, automation is often used to screen the chemicals at a high-throughput pace. These molecules used for screening are often formed from combinational synthesis or gene mutations, and represent a fraction of the theoretical compounds that have not yet been made4. Depending on the purpose of the research study, the type of compounds in the library will change and be more focused on what kind of reaction you are looking for. In drug discovery research the libraries are modeled after the concept of drug-likeness. By using this rule of thumb, compounds are filtered out that have poor absorption, distribution, excitation, metabolism, and toxicity, any traits that would make a poor drug quality4. Computational chemistry is sometimes used to screen virtually for structures that will fit together or predict how a compound would change in response to a reagent.

When constructing a new target-based library there are a variety of methods that can be used to create a compilation that will complement the experimental goal, although each has its limitations. Within the past 5-8 years a popular method of refining libraries is by studying protein-protein interactions between a target and screening compounds using 3D models on computers4. This approach requires extensive knowledge of the structural chemistry of the target, but if the structure is known, it can be used to accurately predict how the proteins will react upon contact with each other. There is a diverse database of knowledge on G-protein-coupled-receptors, kinases, and nuclear receptors because these are the classes of therapeutic targets that have resulted in the most approved drugs. This allows chemists to select compounds from this database for use in their focused library by analyzing the protein-protein interactions.

Ligand-based design uses properties of known active compounds and does not need structures to operate but has many complications for membrane surface proteins because the natural environment needs to be maintained1. As a partial solution, DNA-encoded chemical libraries (DEL) have emerged, but the technology is limited to purified proteins. In December 2020, a study was published that gave a solution to this challenge by designing a method that labeled protein membranes with DNA tags, enabling “target-specific DEL selections against endogenous membrane proteins on live cells without overexpression or any other genetic manipulation.”1 The development of this method opens potential for drug discovery with high-throughput approaches that focuses on membrane proteins.

Another popular method of library design is descriptor-based. This method follows the assumption that molecules with similar descriptors have similar properties and will then have similar activities in a screen4. Descriptors are parameters that describe the physiochemical or structural properties of the molecules in 1D, 2D, or 3D dimensions. Pharmacophores are an example of 3D descriptors. Selection for the library is based on compounds with similar descriptors and is often accomplished by computer-assisted data mining4.

Dereplicated phytochemical libraries are important in the drug-development industry because they offer more chemical diversity than synthetic libraries, but it takes a lot of time and resources to create a library of high quality2. Computational technology has reduced some of the time, but the process is still cumbersome so there is room for improvement. If the process can be streamlined, it may result in the development of more approved drugs because they can be used for a variety of targets2.

There are more approaches to design than were discussed in this paper, and each has its own unique set of challenges. There is the challenge of finding compounds that keep protein structure throughout varied environments and will deliver to the proper system. New combinations of compounds with varied structures must be generated for library use if new treatments are going to be discovered. Computational technology has allowed chemical libraries to become more focused but comes with a high cost and takes time. If a library is not designed properly, then there will not be as many hits in the high-throughput screening, and the time and resources put into the research will not be as rewarding. All of these are things that researchers must think about before the design of the focused library even begins.

References:

  1. Huang, Y., Meng, L., Nie, Q., Zhou, Y., Chen, L., Yang, S., … Li, X. Selection of DNA-encoded chemical libraries against endogenous membrane proteins on live cells. Nature Chemistry. 2020. 13(1), 77–88. https://doi.org/10.1038/s41557-020-00605-x
  2. Nahar, L, Sarker, S.D., Chapter 5 – Application of Computation in Building Dereplicated Phytochemical Libraries. Computational Phytochemistry. Elsevier. 2018. https://doi.org/10.1016/B978-0-12-812364-5.00005-5
  3. Sipos, A., Pató, J., Székely, R., Hartkoorn, R. C., Kékesi, L., Őrfi, L., … Kéri, G. Lead selection and characterization of antitubercular compounds using the Nested Chemical Library. Tuberculosis. 2015. 95. https://doi.org/10.1016/j.tube.2015.02.028
  4. Zhang, X, Stephane, B, Morelli, X, Roche, P. Focused chemical libraries – design and enrichment: an example of protein–protein interaction chemical space. Future Med. Chem. 6(11), 1291–1307. DOI: 10.4155/FMC.14.57.