In this issue of “Small Molecule Highlights” we bring you 5 new small molecules out of recent drug discovery journals. These molecules target a diverse selection of targets, including RXFP1, BRD4, eIF2B, and PKa. Enjoy!
AZ7976
AZ7976: Scientists at AstraZeneca and the Mitsubishi Tanabe Pharma Corporation have recently published a set of papers detailing their hit-to-lead and lead optimization efforts towards the development of AZ5462, a small molecule agonist of the relaxin family peptide receptor 1 (RXFP1). Their hit-to-lead work leading to AZ7976 will be discussed here. Agonism of RXFP1 has garnered a significant amount of attention over the past decade, especially for the treatment of heart failure. However, discovery of new agonists has been met with great difficulty. Both AstraZeneca and GSK (10.1038/s41598-017-10521-9) were recently unsuccessful in identifying chemical starting points via HTS of their internal compound collections. Furthermore, GPCRs, such as RXFP1, are notoriously flexible and contain multiple binding sites making structure-based drug design approaches challenging. Drug development thus far has relied on ML290, a small molecule allosteric agonist discovered by the NIH Chemical Genomics Center, as a starting point. The research team quickly identified issues with solubility, lipophilicity, and plasma protein binding (PPB) leading to terrible early dose to man (eD2M) predictions (37g/day/70 kg). With the goal of attaining a nanomolar agonist with an eD2M below 500 mg/kg, the team performed a comprehensive SAR study to optimize the parent scaffold. One of the most structurally conspicuous alterations was the incorporation of the central norbornane. This was done to break the planarity enforced by the previous phenyl substituent, allowing for greater solubility and lower clearance whilst maintaining activity. The terminal cyclohexyl carboxylic acid provided the largest impact across all the parameters considered (activity, solubility, lipophilicity, and clearance). As carboxylic acids are often susceptible to clearance via conjugation, a closely situated methyl group was installed to protect this functional group from metabolism. The pentafluorosulfur substituent offered a substantial boost in activity (more than 2-orders of magnitude), counterbalancing the poor metabolic profile of the compound and reducing eD2M to 60mg/day/70kg. In a receptor selectivity panel (Eurofins), AZ7976 exhibited exquisite selectivity for RXFP1 (> 5000-fold selectivity vs the rest of the panel). Interestingly, an enantiomer of the compound was poorly selective, owing to its greatly reduced activity against RXFP1 (10,000-fold less potent than AZ7976). To probe binding, a radiometric competition assay against relaxin-H2 provided further support for an allosteric binding mechanism leading to protein agonism. In an in vivo setting, target engagement and pharmacodynamic effect was demonstrated in anesthetized rats. Both AZ7976 and its inactive stereoisomer were benchmarked against relaxin-H2. In the study, AZ7976 exhibited durable modulation of heart rate and mean blood pressure, similar to the short-lived increases in both parameters observed in the relaxin-H2 treated cohort. As the inactive compound did not show any effect, direct agonism of RXFP1 by AZ7976 could be inferred. In scope, AZ7976 looks like a very promising lead optimization candidate. While activity and selectivity has been dialed in, more work must be done to address low solubility, high lipophilicity, and poor PK.
Reference: https://doi.org/10.1021/acs.jmedchem.3c02183
AZD5462
AZD5462: Building on AZ7976 (vide supra), the AstraZeneca and Tanabe Pharma team aimed to improve the physicochemical properties and pharmacokinetic profile of this lead compound. As with most lead optimization endeavours, compromises had to be made to strike the right balance between PK and activity to enable clinical development. Fortuitously, AZ7976 was exceedingly potent (hRXFP1cAMP pEC50 = 9.4), allowing for much more rigorous exploration of chemical space to tune the desired properties. Overtly, the only change made was a swap of the pentafluorosulfur aniline to a cyclized neopentyl amine, however, every portion of the molecule, other than the norbornane core, was subject to optimization. For instance, the terminal carboxylic acid was replaced by a variety of mimics to circumvent racemization of the carbon attached to the group. Also, modifications to the fluorine substituent on the central phenyl ring were found to modulate activity. Interestingly, replacement with a cyano group increased activity by more than 10-fold. Circling back to replacement of the penatafluorosulfur aniline, this was a practical decision made out of necessity. The group was much too lipophilic and posed a potential genotoxicity risk (aniline group). The researchers settled on a substituent that was lipophilic enough to maintain binding interactions with the target whilst decreasing the planarity of the molecule. Pharmacokinetic characterization of AZD5462 revealed moderate oral bioavailability (F = 47%) and good plasma half-life (T1/2 = 4.6 h) in rat. Multiple assays assessing cardiac (hERG, Nav1.5 and Kv4.3 IC50 > 40 μM) and liver (BSEP IC50 = 7.1 μM) safety, as well as genotoxicity (negative Ames and A549 micronuclei tests), indicated no potential safety concerns. As was done with AZ7976, hemodynamic changes (heart rate and mean blood pressure) in rat were evaluated using AZD5462 and an inactive diastereomer. In this way, any effects observed could be ascribed to RXFP1 inhibition. In the study, AZD5462 activity surpassed that of the positive control (relaxin H2). Pharmacodynamic characterization in a disease relevant model (obese cynomolgus monkeys exhibiting heart failure) revealed statistically significant improvements in left ventricle ejection fraction (LVEF) that was sustained 13 weeks post administration in both dosing regimes tested (1 mg/kg q.d and 10 mg/kg b.i.d). With such promising results, AZD5462 has progressed into clinical evaluations in healthy volunteers (NCT04994106, NCT05512806, and NCT05395117) and has recently been linked to, although not recruiting, a Phase IIb study in patients with chronic heart failure (LUMINARA; NCT06299826).
Reference: https://doi.org/10.1021/acs.jmedchem.3c02184
CURE-PRO 55 + 63
CURE-PRO 55 + 63: A very interesting targeted protein degrader (TPD) development platform has recently been disclosed by researchers at Cornell University (USA). In an effort to resolve solubility and permeability issues often associated with PROTACs, the group leveraged their Coferon technology to develop self-assembling degraders. The principle is elegant and simple, use bio-orthogonal linker chemistries to unite protein of interest (POI) and E3-ligase recruiter ligands within the cell. The benefits are obvious; no “hook effect” at higher concentrations (significantly increases dosage window compared to PROTACs) and improved drug-like properties owing to smaller compound size. If that wasn’t enough to win over the most ardent ternary PROTAC supporters, a very clever acronym for the platform surely will; CURE-PROs (Combinatorial Ubiquitination REal-time PROteolysis). As a proof-of-concept, the medicinal chemistry team prepared a CURE-PRO targeting BRD4 using a catechol and boronic acid as the bio-orthogonal linker pair. Slight optimization of the catechol regiochemistry was required to fully dial in activity. Dose-dependant degradation of BRD4 was demonstrated in MCF7 cells (DC50 = 358 nM) and in a real-time kinetic degradation study using HiBiT-BRD4 KI HEK293 (LgBiT) cells. As was mentioned previously, no “hook effect” was observed at the highest concentration tested (10 μM) permitting durable degradation for sustained periods, even after wash-out. It bears mentioning that the stoichiometry of the POI ligand and E3-ligase recruiter doesn’t have to be 1:1 to enable protein degradation. Various stoichiometries of CURE-PRO ligand pairs were shown to be effective. Pharmacokinetic characterization of 55 and 63 revealed rapid absorption of both molecules (Tmax = 0.5 h), however, 63 was shown to be removed from circulation much more quickly than 55, potentially by extrahepatic clearance and renal filtration. Pharmacodynamic efficacy was demonstrated in an MV4-11 xenograft model in mice. In the study, significantly reduced BRD4 levels were observed in tumor samples across multiple time-points, aligning well with in vitro findings. The real beauty of the platform is its ability to discover novel degraders using combinatorial screening. In this approach, a variety of E3 recruiters with varying catechol linkers can be quickly assessed against POI ligands containing variations of the corresponding boronic acid partner, allowing for rapid identification of PROTAC starting points without protracted SAR studies/campaigns. Currently, the CURE-PRO platform has a variety of E3-recruiter fragments targeting VHL, CRBN, and even MDM2. It will be interesting to see how the platform evolves. We predict we will be seeing a covalent E3 recruiter CURE-PRO fragment in the near future.
Reference: https://doi.org/10.1021/acs.jmedchem.3c02097
DNL343
DNL343: The integrated stress response (ISR) is an adaptive signalling pathway regulating protein homeostasis and promoting overall survival in response to cellular insult(s). While the pathway’s cytoprotective role is very much context specific (agonism and antagonism of the pathway has been shown modulate survival), a growing body of evidence has implicated chronic activation of ISR in a range of neurodegenerative diseases, such as vanishing white matter (VWM) disease and amyotrophic lateral sclerosis (ALS). The drug development team at Denali Therapeutics have targeted a crucial nucleotide exchange factor (eIF2B) in the pathway. To support structure-based drug design endeavors, the team built a homology model of human eIF2B using a homologous protein found in yeast. Further scanning mutagenesis studies were used to identify the putative binding site with the help of ISRIB (a previously published inhibitor of the eIF2B). Discovery efforts were also supported by a substantial HTS screening campaign of approximately 400,000 compounds. Lead development hinged on the optimization of properties promoting CNS penetration. Towards this end, an oxadiazole substituent replaced a secondary amide group to lessen active transport of the compound out of the CNS via Pgp mediated efflux. Also, the sp3 character and solubility was greatly increased by limiting the arene content of previous congeners by incorporating a cyclobutyltrifluoromethyl ether in its place. Pharmacokinetic characterization of DNL343 revealed impressive oral bioavailability (F(%) = 65/>99/>99 in rat/dog/monkey) and plasma half-life (T1/2(h)= 12.5/7.4/7.6 in rat/dog/monkey) across species. Also, passage into the CNS was relatively high with brain-to-plasma ratios of 0.8 and 0.9 being observed in rat and monkey, respectively. No risks associated with drug-drug interactions (DDI), genotoxicity/mutagenicity (negative AMES and micronuclei tests), and cardiac toxicity were flagged in follow up safety studies. In vitro characterization of the compound in an H4 cellular model of ALS confirmed potent suppression of sodium arsenite mediated stress granule formation via the ISR pathway (IC50 = 13 nM). Pharmacokinetic efficacy was evaluated in a mouse model of ISR driven neurodegenerative disease (Eif2b5 R191H mice). In the study, DNL343 was found to significantly decrease ISR transcript markers in the brain. DNL343 has made its way into phase I and Ib clinical trials in healthy (NCT04581772) and ALS stricken individuals (NCT05006352), respectively. Results from the trial indicated fairly consistent PK with predictable dose/exposure correlations. Multicenter phase II and III studies are currently recruiting/enrolling participants (NCT05842941 and NCT04297683).
Reference: https://doi.org/10.1021/acs.jmedchem.3c02422
Compound 20
Compound 20: Researchers at KalVista Pharmaceuticals and the University of Nottingham have recently disclosed their preliminary medicinal chemistry efforts to develop a covalent inhibitor for plasma kallikrein (PKa) protein. PKa is a serine protease involved in the bradykinin (BK) signaling which impacts the homeostasis of blood vessels. Dysregulation of BK can lead to increased vasodilation and blood vessel leakage, with severe cases resulting in systemic angioedema. While a variety of small molecule treatment options (berotralstat and sebetralstat) are available, none are irreversible, possibly pointing to difficulties in developing a covalent inhibitor for this particular target. Using a structure-based drug design approach, the team was able to identify a scaffold and potential attachment points for a covalent warhead (WH) targeting Ser195. The chosen warhead, a pinacol-boronic ester, comes as no surprise as boronic acids are ideally suited for engaging serine’s hydroxy sidechain as a result of its Lewis acidic nature. Iterative rounds of SAR led to compound 20 which demonstrated superb time-dependant inhibition of PKa (IC50 = 66/6.9/0.3 nM @ 1/10/60 min) and exquisite selectivity over closely related proteases (>1000-fold selectivity). The activity of 20 was curious as molecular docking studies had failed to orient the rather large pinacol boronate within the Ser195 sub-pocket. Fortunately, some key observations made during the synthesis of related compounds were able to shed some light on the mystery; the pinacol boronate was prone to hydrolysis. As such, the active compound was likely a boronic acid or the corresponding cyclization product with the adjacent amide oxygen (oxaborolane). NMR studies carried out in PBS buffer confirmed cleavage of the pinacol to afford the corresponding boronic acid and oxaborolane which exist in equilibrium. While no further ADME, PK, or PD studies were carried out, presumably due to the labile nature of the WH contributing to synthesis and purification difficulties, compound 20 adds valuable insights into the use of amidoboronates as covalent WHs, especially in the serine protease inhibitor space.