Axial: https://linktr.ee/axialxyz
Axial partners with great founders and inventors. We invest in early-stage life sciences companies such as Appia Bio, Seranova Bio, Delix Therapeutics, Simcha Therapeutics, among others often when they are no more than an idea. We are fanatical about helping the rare inventor who is compelled to build their own enduring business. If you or someone you know has a great idea or company in life sciences, Axial would be excited to get to know you and possibly invest in your vision and company. We are excited to be in business with you — email us at info@axialvc.com
Medicinal chemistry has very little overlap with programming. The idea of AI automating, or even augmenting, medchem has been laughable given that the training data set is very complicated and validation requires multiple experiments: synthesis, screening, validation assays.
AI in small molecule drug discovery requires a hit-synthesis-lead optimization workflow. The process of discovering and developing a new drug is long, arduous, and expensive, often taking over a decade and costing over $1B. However, the early stages of hit identification and lead optimization are critical for laying the groundwork for an effective new therapeutic candidate. These stages bridge the gap between the initial discovery of a promising molecular hit compound and advancing that hit into a lead compound suitable for further preclinical and clinical development.
The hit identification stage begins with screening massive compound libraries, which can contain millions of discrete small molecules, to identify those that exhibit the desired biological activity against a particular therapeutic target, such as binding to a protein implicated in disease. This screening utilizes high-throughput assays designed to efficiently test each compound's ability to modulate the target.
Hits from these screens typically bind to the target with relatively low affinity, in the micromolar (10-6 M) range. They represent promising starting points, but require substantial optimization to increase their potency and improve other important drug-like properties. On average, only 1 in 5,000 hits will ultimately become an approved drug.
Before advancing to lead optimization, confirmed hits undergo extensive re-testing and evaluation to validate their activity and assess key properties:
- Confirmatory testing re-evaluates the activity to ensure it is reproducible.
- Dose-response curves determine the potency (IC50/EC50 value).
- Orthogonal assays test activity under distinct assay conditions.
- Secondary cellular assays assess functional activity.
- Physicochemical properties like solubility and lipophilicity are evaluated.
- Binding interactions are characterized using biophysical techniques like NMR and SPR.
- Synthetic tractability for analog synthesis is assessed.
- Patentability and freedom-to-operate are evaluated.
This rigorous testing whittles down the initial hits to a handful of the most promising chemical series that will proceed to hit-to-lead optimization.
Hit expansion marks the transition into the lead optimization phase, where focused medicinal chemistry efforts optimize the confirmed hit series' potency, selectivity, and drug-like properties. Ideal lead compounds exhibit low nanomolar potency, high selectivity over off-targets, favorable physicochemical properties, and suitable in vitro pharmacokinetic profiles.
Structure-activity relationship (SAR) studies systematically explore the effects of substituting different chemical moieties onto the core hit scaffold. This can involve:
- Purchasing and testing commercial compound libraries ("SAR by catalog")
- Parallel synthesis of focused analog libraries
- Classical step-wise organic synthesis of designed analogs
In addition to increasing potency, this process identifies the key molecular determinants of binding affinity. Where available, protein structural information can guide a structure-based drug design approach to rapidly develop the SAR.
In parallel, promising analogs undergo more extensive in vitro ADME (absorption, distribution, metabolism, excretion) profiling, such as:
- Metabolic stability in liver microsomes and hepatocytes
- CYP enzyme inhibition potential
- Plasma protein binding
- Permeability and solubility
Analogs with suitable preliminary PK properties may also undergo in vivo pharmacokinetic studies in animal models to evaluate oral bioavailability, half-life, clearance, and exposure.
Throughout this iterative optimization process, researchers aim to strike a balance between improving potency and maintaining acceptable drug-like properties. Medicinal chemistry efforts focus on the most promising few scaffolds that demonstrate a promising combination of potency, selectivity, and PK profiles to merit further advancement toward preclinical development.
The hit-to-lead stage is critical for transitioning hits into viable lead candidates. While hits from initial screens provide a key starting point, they almost invariably require substantial optimization to achieve the required potency and drug-like properties for further development.
Inadequately optimized hits are unlikely to succeed in subsequent stages. For example, insufficient potency can necessitate unacceptably high doses that increase off-target toxicities. Poor pharmacokinetics can limit oral bioavailability and exposure. And suboptimal selectivity profiles may trigger adverse effects by interaction with off-target proteins.
Extensive hit-to-lead optimization can overcome many of these potential liabilities. Studies estimate that up to 25% of lead optimization resources are devoted to improving suboptimal ADME properties from the hit stage. While challenging, this upfront investment provides a stronger foundation for advancing into costly preclinical studies by delivering lead compounds with significantly de-risked profiles.
By prioritizing chemically tractable, drug-like scaffolds from the outset, the hit-to-lead stage focuses resources on the most promising molecular starting points. Optimizing these hits into potent and selective lead series with acceptable in vitro and in vivo characteristics maximizes the chances of advancing a viable development candidate into further preclinical evaluation.
The hit-to-lead stage continues to evolve with innovations aiming to increase efficiency while prioritizing higher quality lead matter. For example, fragment-based lead discovery (FBLD) identifies very low molecular weight hit fragments that bind to the target, which are then elaborated into larger, more potent lead-like molecules. This approach can enable exploration of different binding sites compared to traditional HTS hits.
DNA-encoded chemical libraries allow ultra-large virtual libraries of billions of compounds to be efficiently screened for hit identification, while enabling rapid hit-to-lead optimization through iterative synthesis and screening of analog mixtures. Computational tools like free energy calculation methods estimate binding affinities to rapidly prioritize analogs and predict optimized structures, reducing resources spent on low-value synthesis. Machine learning and artificial intelligence algorithms can analyze public and proprietary data to identify patterns predictive of desired properties like potency and PK profiles to guide analog design.
Integrating these and other innovative technologies with robust medicinal chemistry approaches will be crucial for accelerating future hit-to-lead optimization campaigns in delivering higher quality lead matter for advancing drug discovery pipelines.
Hit synthesis and lead optimization represent an essential bridge in early drug discovery for transforming an initial molecular starting point into a viable lead compound suitable for later stage preclinical development. While hit identification provides the critical first step, extensive optimization is invariably required to increase potency, improve selectivity, and refine drug-like properties to de-risk candidates before more costly late lead optimization and preclinical development stages.
The hit-to-lead phase features an iterative medicinal chemistry campaign exploring analogs and back-ups of the most promising hit series to grow their structure-activity relationships. In parallel, rigorous in vitro ADME profiling and preliminary in vivo PK evaluation triage chemical scaffolds based on their overall profile of potency, selectivity, and acceptable drug-like properties.
This multidisciplinary hit-to-lead optimization delivers higher confidence lead compounds to increase the probability of success in subsequent stages of drug discovery. As technologies like computational modeling and DNA-encoded libraries continue advancing, the hit-to-lead phase will likely accelerate and become even more critically enabling for delivering higher quality lead matter while conserving resources across drug discovery pipelines.