Skip to main content

DISRUPTIVE GPCR LEAD DISCOVERY PLATFORM DELIVERING NEW AND SAFER THERAPEUTICS

Periodic Reporting for period 1 - PICARD (DISRUPTIVE GPCR LEAD DISCOVERY PLATFORM DELIVERING NEW AND SAFER THERAPEUTICS)

Reporting period: 2019-10-01 to 2020-09-30

New drug development and clinical trials take about 15 years and cost on average EUR 2.3 billion with a success rate of around 12%. The long timeline and high financial cost are due to an inefficient drug selection process, as only 1 out of 5 drugs that enter long, and costly clinical trials gets approved. Current available technologies do not succeed in reducing the high drug failure rates. This is mainly due to the low predictability of how the drug candidates best performing during in vitro experiments will perform in in vivo (animal) models and then humans.

PICARD is a unique lead discovery platform for the design and selection of drug candidates targeting G protein coupled receptors (GPCRs) – the largest class of drug targets among all medicines on the market. PICARD combines mathematical modeling (systems biology) with experimental wet-lab measurements and uses machine learning algorithms to design novel compounds with the desired cell signaling properties. This holistic approach differentiates PICARD from technologies now used in the pharmaceutical industry. It can be applied ad hoc to customer needs, covering a multitude of drug discovery programs across all medical indications that are linked to GPCRs.

Our systems biology approach already provides PICARD with in vivo predictability, dramatically reducing the need for animal testing (60-80%) and increasing the probability of success of drug candidates in clinical trials. The PICARD project enables InterAx to implement AI and machine learning algorithms for GPCR drug design, which is both offered to pharmaceutical and biotech clients as a service and utilized for spearheading the in-house drug discovery programs of InterAx
During the first reporting period of PICARD (year 1), InterAx expanded its systems biology platform (PIC-Sys) to several additional GPCR targets. The focus was on one in-house drug discovery target (described in detail in WP1) and several additional GPCR targets commissioned by paying customers (early implementation of WP4), such as Lundbeck A/S and Boehringer Ingelheim. In addition, InterAx successfully implemented its unique AI-driven platform (PIC-VisAI), which now enables the fast design of GPCR drug molecules with custom-designed cellular signaling parameters (described in detail in WP2).

Milestone 1 - Expansion on PIC-Sys (WP1): Twenty novel compounds (Novel Chemical Entities, NCEs) were tested in the wet-lab using time-resolved signaling assays for an in-house GPCR drug target (=PIC-Say). In parallel, a mathematical model consisting of Ordinary Differential Equations (ODEs) was programmed for the new receptor and its downstream cellular signaling partners (=PIC-Sys). The time-resolved wet-lab data were analyzed using the novel GPCR-ODE-model and a set of distinct signaling kinetic parameters (so called parameter fingerprints) for each NCE was generated. The NCE fingerprints were then hierarchically clustered to assess their respective relationship to each other. Finally, these fingerprints were compared to both in vitro and in vivo (if available) properties of known drug candidates. In addition to the novel in-house GPCR ODE-model, two additional GPCR-ODE-models were commissioned by paying customers during the reporting period, and a third one is currently under development (early implementation of WP4). In achieving this significant milestone (and beyond), InterAx confirmed that the systems biology platform PIC-Sys is applicable to a broad range of GPCR targets.

Milestone 2 – Implementation of the AI-driven GPCR drug design platform – PIC-VisAI (WP2): This unique machine learning algorithm implemented by InterAx enables the design of GPCR drug molecules (=ligands) with a specifically desired cellular signaling outcome. The distinctive abilities of the platform are due to the training of the AI models on the unique InterAx systems biology datasets obtained with the PIC-Sys technology (see above). These datasets contain extended sets of GPCR signaling fingerprints for each ligand – for example i) the residence time of a given ligand on a GPCR, or ii) the strength of a given ligand-receptor-intracellular signaling partner complex. Such extended parameters reach beyond the standard descriptors of ligands, such as ligand affinity or potency values. Hence, with these datasets, PIC-Sys re-creates a part of the complex GPCR cellular signaling network and thereby better reproduces the in vivo behavior of ligands. By doing so, only a fraction of time and financial costs are accrued that are otherwise typically implicated in the GPCR drug discovery process.

In summary, within this reporting period, InterAx (1) established workflows for the generation of structural biology / chemistry datasets for novel GPCR ligands; (2) created the structural biology and systems biology training datasets using the beta-2-adrenergic receptor (B2AR) as a showcase receptor; and (3) established and validated core machine learning algorithms for the PIC-VisAI screening platform. The platform is now able to i) predict agonist molecules for the B2AR, and to ii) design ligands with at least two disctinct signaling properties, such as the ligand residence time and the complex stability of ligands on the B2AR. The PIC-VisAI platform can thus guide the design of ligands with desired signaling properties
The current screening process for a lead compound to be launched into the market begins with target and hit identification, resulting in the collection of 5,000 to 20,000 molecules that must be analyzed according to the probability to possess the desired biological activity. The screening process to select a suitable biological active compound (lead compound) can be either assay-based, structure-based or computational – however, each of these steps are prone to limitations, such as a high trial-and-error rate and/or little to non-existent in vivo predictability.

PICARD is a technology platform that combines mathematical modeling of biological processes with machine learning algorithms to custom design lead compounds with desired in vivo properties for pharmaceutical and biotech companies. PICARD remarkably increases the efficiency of the drug discovery process, offering customers economic savings of at least €150.7 million per developed drug and a reduction of time to market to 10 years.

PICARD focuses on G-Protein Coupled Receptors (GPCRs), which are the target of 40% of the developed drugs. However, >70% of the pharmacologically relevant GPCRs are yet to be exploited, which creates an untapped global market opportunity of €71.2 billion. We intend to acquire, in seven years, 1-2% of the European and US available markets (€940 million). To achieve this goal, we have designed a business model in which PICARD is offered at increasing levels of support, investment and risks for our customers: service projects, pilot projects and long term partnerships. This strategy was devised to lower market entry barriers, allowing us to start commercialization of our technology platform PICARD as early as year 2 of this project. In fact, we have outperformed our own goals by already entering the market in year 1 with currently three ongoing pilot projects commissioned by pharma and biotech companies