Fragment-based drug discovery is a method of drug lead identification wherein a battery of small molecule compounds are investigated for their affinity in binding to a particular therapeutic target.
A number of analytical methods are utilized to identify fragment-target affinity, and through successive generations of testing, the fragments are modified further to fine-tune favorable steric, electrostatic, and hydrophobic interactions, thereby improving drug efficacy. This stage of drug discovery is typically concerned with identifying molecules that will interact strongly with the molecular target, and affinity tests are performed in vitro. In vivo testing may then be performed later once-promising drug leads are identified.
Image Credit: Vectors Bang/Shutterstock.com
How are drug leads identified?
When a target biomolecule has been identified for further investigation by fragment-based drug discovery, a library of fragments is prepared that span a structurally diverse range of low molecular weight (<300 Da) compounds, many of which are sold commercially for testing and come pre-packaged, or may be specially tailored to a target with known vulnerabilities. Typically, fragments in the library carry three or fewer hydrogen bond donors or acceptors, respectively, and possess a partition coefficient (logP) of ≤3. This is known as the "rule of three" and aims to ensure that fragments maintain good oral bioavailability and pharmacokinetics following further modification, which may increase the number of hydrogen bond donors or acceptors, lipophilicity, molecular weight, and the number of rotatable bonds belonging to the fragment, breaking Lipinski's "rule of five" (≤500 Da, ≤5 hydrogen bond acceptors and donors, logP ≤5).
Initial affinity towards the target biomolecule is expected to be low prior to fragment optimization, so high concentrations are usually utilized, and false-positive hits are commonplace. Multiple chemical and physical screening methods are used to identify the most promising lead compounds, wherein the target molecule is conjugated to a surface and then exposed to the fragment under investigation, or less typically, the fragment is fixed to a surface and the target molecule applied.
Colorimetric methods of indicating successful bonding between a fragment and the target molecule are often utilized, frequently by the addition of fluorescent probes to the target molecule that emits or quench when a fragment successfully bonds. However, this method requires some alteration of the target and/or the fragment and may influence binding affinity and kinetics, also potentially producing false positives.
A label-free method of colorimetric detection employs gold or silver nanoparticles upon which the target is bound, which interact with light to strongly absorb at a particular wavelength in a process known as surface plasmon resonance. When a fragment bonds with the target, the refractive index of the nanoparticle's surroundings are altered slightly, causing a shift in surface plasmon resonance wavelength; this method allows for real-time observation of fragment bonding, indicating how quickly fragment-target bonds form and disassociate.
How are drug leads optimized?
It is important to the drug discovery process that the mechanism of bonding between the fragment and target is realized, and several methods are again utilized to understand what types of interactions are taking place, allowing medicinal chemists to optimize the lead compound. X-ray crystallography is used to discern the precise molecular structure of crystallized proteins and other molecules, and the target and bound fragment can be co-crystallized for detailed characterization of bonding mode.
In situ methods of analyzing bonding mode can also be achieved using various types of specialized nuclear magnetic resonance-based methods, wherein the chemical shift of atoms directly bound to a fragment is altered, also indicating the steric positioning of the bound fragment.
Once lead compounds are selected from the initial screening of the fragment library against the target. Then their chemical structure and bonding mode are considered for testing alterations that may enhance or reduce affinity towards the target, for example, introducing a hydroxyl group that assists in hydrogen bonding interactions with adjacent amino acids once bound or extending the length of an alkane chain to reach more deeply into a hydrophobic binding pocket.
Proposed improvements to fragment structure are then synthesized and applied to screening and examination processes against the target, with the process repeated over multiple successive generations. Interestingly, where two or more fragments are found to bond to the target in overlapping or spatially adjacent positions, lead compounds may be merged or linked to create larger molecules with greater overall affinity to the target, owing to more numerous binding sites.
Introducing long and flexible linking sections to the molecule may introduce a great degree of rotational freedom and make it more difficult for the bonding regions to align with the target correctly, though rational drug design can enhance positioning and allow multiple binding pockets to be simultaneously engaged.
In silico drug discovery
Modern pharmaceutical drug design and discovery heavily employ computational methods during fragment library assembly and initial screening, as well as in determining the most probable binding mode between fragment and target. Using in silico methods, millions of potential drug leads can be explored for affinity towards the target site and greatly refine the fragment library initially explored, and subsequent modifications can be similarly refined to the most promising lead compounds prior to their synthesis and analysis.
Deep learning methods are increasingly employed to suggest and optimize lead compounds with wide diversity, though they are still in the early stages of establishing structure-function relationships for more general applications.
For example, Stokes et al. (2020) demonstrated a deep-learning approach to antibiotic discovery that predicted antibiotic activity from a novel compound, Halicin, one of the few newly discovered antibiotics generated in the last several decades. The drug demonstrates a unique mechanism of action involving the sequestration of iron within bacterial cells that induces a pH imbalance and disrupts the electrochemical gradient of the cell membrane, inducing cell death by multiple potential routes, including interference with ATP production and transfer. It is thought that this mechanism of action should be difficult to develop resistance towards, and promisingly the drug has shown efficacy towards several drug-resistant strains of bacteria.
Sources:
- Kirsch, P., Hartman, A. M., Hirsch, A. K. H., & Empting, M. (2019). Concepts and Core Principles of Fragment-Based Drug Design. Molecules, 24(23), 4309. https://doi.org/10.3390/molecules24234309
- Smyth, M. S. (2000). X-ray crystallography. Molecular Pathology, 53(1), 8–14. https://doi.org/10.1136/mp.53.1.8
- Kim, J., Park, S., Min, D., & Kim, W. (2021). Comprehensive Survey of Recent Drug Discovery Using Deep Learning. International Journal of Molecular Sciences, 22(18), 9983. https://doi.org/10.3390/ijms22189983
- Stokes, J. M., Yang, K., Swanson, K., Jin, W., Cubillos-Ruiz, A., Donghia, N. M., Macnair, C. R., French, S., Carfrae, L. A., Bloom-Ackermann, Z., Tran, V. M., Chiappino-Pepe, A., Badran, A. H., Andrews, I. W., Chory, E. J., Church, G. M., Brown, E. D., Jaakkola, T. S., Barzilay, R., & Collins, J. J. (2020). A Deep Learning Approach to Antibiotic Discovery. Cell, 180(4), 688–702.e13. https://doi.org/10.1016/j.cell.2020.01.021
Further Reading