Adrenergic Receptor β1: Crucible Identifies Novel Scaffolds

Executive Summary

• β1-AR is implicated in a range of cardiovascular diseases, including hypertension, arrhythmias, and heart failure

• We identified a new structural class of β1 inhibitors that do not rely on structural motifs found in endogenous ligands such as epinephrine.

• Different structural motifs would suggest different side effect profiles allowing consumers to have better individual choices for their care

What is Adrenergic Receptor β1?

Figure 1: The protein structure of human adrenergic receptor β1 in complex with epinephrine (PDB ID: 7BTS).

The adrenergic receptor beta-1 (β1-AR) (Figure 1) is a G protein-coupled receptor primarily expressed in cardiac tissue, where it regulates heart rate, contractility, and conduction velocity in response to catecholamines like epinephrine and norepinephrine. Dysregulation of β1-AR signaling is implicated in a range of cardiovascular diseases, including hypertension, arrhythmias, and heart failure, making it a key therapeutic target. Nearly all existing β1-AR modulators—such as beta-blockers—are structurally derived from the HO–C–C–NH motif found in endogenous ligands, reflecting their roots in mimicking epinephrine’s pharmacophore. While these agents have been clinically effective, they exploit a very narrow chemical space. Structural optimization techniques, such as Crucible, offer the potential to explore novel scaffolds beyond the traditional framework, paving the way for innovative drug classes with improved selectivity, efficacy, and side effect profiles.

Lead Generation in Crucible

A cleaned version of the crystal structure of β1-AR was used as a disease target for Crucible to find potential drug leads. Similar to our first case study, we optimized 50 leads with crucible using a desktop computer (equipped with an Intel i9-14900K, one Nvidia RTX 3080, and 64 Gb of memory) totaling about 18 hours of compute time.

Interestingly, the HO-C-C-NH structural motif appeared more frequently in the log history of this computational run by about +50% in comparison to their prevalence in the same log history for our case study on dopamine receptor D2 (which has no particular preference for HO-C-C-NH). However, even though Crucible found these structures to be appealing early on, none of the final optimized leads contained this feature. This indicates that other molecular scaffolds may be even more effective inhibitors of β1-AR. An example of a few of the optimized leads are shown in Figure 2.

Figure 2: A selection of three Crucible optimized leads and one on-the-market beta-blocker (Propranolol, bottom right) for reference. It should be noted the HO-C-C-NH structure present in propranolol, but not in the other structures.

Validation of Crucible-generated Leads

Principle component analysis (PCA), a technique for clustering molecules based on structural similarities, was performed to assess similarities and differences between on-the-market beta-blockers and Crucible-generated leads. Extended connectivity fingerprinting was used in conjunction with PCA to generate Figure 3. On-the-market beta-blockers and novel Crucible lead clusterings occupy different space in the plot which indicates dissimilarities in their respective chemical scaffolds.

Figure 3: A comparison of the molecular fingerprints of on-the-market beta-blockers with Crucible generated optimized leads using PCA. The x- and y-axis are principle component 1 and 2, respectively. There is clear clustering between the two groups indicating significant structural differences between them.

Despite differences in structural features, the Crucible generated leads generally bind with higher affinity as compared to on-the-market beta-blockers based on AutoDock Vina docking scores (Figure 4).

Figure 4: Plot of AutoDock Vina scores of ligands against β1-AR. Vina scores correspond with an estimate of the ΔGbind in units of kcal/mol, therefore a larger negative value indicates stronger binding affinity. On-the-market beta-blockers are indicated by name on the x-axis, while Crucible leads are indicated by their smile string representation.

As the structures of the Crucible generated leads are significantly different than that of the on-the-market beta-blockers, it may be useful to visualize the interactions between the receptor and ligand. Figure 5A shows the interaction of propranolol with β1-AR, including the expected hydrogen bonding interactions with the HO-C-C-NH motif. In Figure 5B, hydrogen bonding interactions are present and appear to be in greater number than that of propranolol.

Figure 5: (top) A. Propranolol in complex with β1-AR. (bottom) B. Crucible optimized lead in complex with β1-AR. Two slightly different camera angles were used in A and B to better show atom proximity and therefore intermolecular interactions. Hydrogen bonding interactions are shown with a solid green line. In A, hydrogen bonding interactions using the ligand -OH and -NH- are present. In B, upon close inspection there are three hydrogen bonding interaction which might suggest significantly stronger binding.  

Conclusion

Crucible generated a novel class of therapeutic leads for β1-AR that do not rely on structural features of endogenous ligands. Docking scores based on AutoDock Vina indicate that these new compounds bind as well as or better than on-the-market beta-blockers. Having a unique set of scaffolds potentially allows for different side effect profiles which gives greater choice to consumers as to the quality of their care when taking these drugs. As Crucible is a disease-agnostic AI model, this same approach can be used on other disease targets for new lead generation.

Contact us to learn more about how we can help you rapidly identify new therapeutic leads for your disease areas of interest.