Galapagos (France)

ABSTRACT High‐brightness, full‐color microdisplays with sub‐5 µm pixel pitch are key for next‐generation augmented and virtual reality applications. While blue InGaN/GaN µLEDs offer excellent luminance, the integration of efficient red and green emission remains a major challenge in the realization of full color microLED displays. Using color converters such as quantum dots or perovskites is a promising approach, but current pixelation methods for these nanomaterials often suffer from limited resolution, significant material waste, and optical degradation during or after the process. Herein, we introduce a scalable directed‐assembly method for fabricating red‐ and green‐converter 3D micro‐subpixels based on pure colloidal quantum nanoplatelets. This method, compatible with non‐polar solvent dispersions, enables the high‐resolution fabrication of micro‐subpixels of 3.5 µm (corresponding to ∼3000 PPI in RGB configuration) in width and over 1.3 µm in thickness, achieving more than 75% blue light absorption efficiency without requiring much thicker polymer matrices or nanocomposites. Integration on 8 µm × 8 µm InGaN/GaN µLED arrays demonstrates homogeneous large‐area coverage, high structural fidelity, and promising electro‐optical performance with external photoluminescent quantum yield of red micro‐subpixel reaching ∼71%. The process is compatible with 200 mm CMOS‐type processes and offers a compelling pathway toward next‐generation high‐brightness, high‐resolution full‐color microdisplays.

Computational hit finding, particularly virtual screening, has been a mainstay of drug discovery campaigns for decades, providing a cost-efficient complement to wet experiments. Innovation in this space slowed considerably as these approaches converged around mature software programs and stock chemical libraries up to ∼10 million in size. Recently, however, powered by massive increases in computational power, the emergence of ultralarge make-on-demand virtual libraries, the development of large capacity neural networks, the expansion of the domain of applicability of free energy calculations, and advances in protein structure prediction, the virtual screening field is currently seeing major change. We present a guide from industry practitioners summarizing key aspects on the changing computational hit finding landscape including practical recommendations for building a performant end-to-end screening workflow, strategies to mitigate risk by avoiding common pitfalls, determining success criteria, and a brief discussion of emerging technologies likely to impact drug discovery in the near future.

Janus kinase (JAK) inhibitors have gathered interest as treatments for several inflammatory and autoimmune diseases. The four first marketed inhibitors target JAK1, with varying selectivity towards other JAK family members, but none inhibit tyrosine kinase-2 (TYK2) at clinically relevant doses. TYK2 is required for the signaling of the interleukin (IL)-12 and IL-23 cytokines, which are key to the polarization of TH1 and TH17 cells, respectively; two cell subtypes that play major roles in inflammatory diseases. Herein, we report our effort towards the optimization of a potent and selective dual JAK1/TYK2 inhibitor series starting from a HTS hit. Structural information revealed vectors required to improve both JAK1 and TYK2 potency as well as selectivity towards JAK2. The potent inhibition of both JAK1 (3.5 nM) and TYK2 (5.7 nM) in biochemical assays by our optimized lead compound, as well as its notable selectivity against JAK2, were confirmed in cellular and whole blood assays. Inhibition of TYK2 by the lead compound was demonstrated by dose-dependent efficacy in an IL-23-induced psoriasis-like inflammation mouse model.

Where head-to-head trials are lacking, indirect comparative effectiveness can aid treatment decisions. We conducted matching-adjusted indirect comparisons of clinical outcomes with filgotinib vs recently approved comparators (vedolizumab, tofacitinib, ustekinumab) in patients with moderately to severely active ulcerative colitis (UC). Individual patient data from the SELECTION trial (NCT02914522) for filgotinib 200 mg were weighted to match average baseline characteristics of active treatment and placebo arms in comparator trials. Efficacy outcomes were compared for biologic-naive and biologic-experienced subgroups in induction and maintenance populations, if data were available. Safety and health-related quality of life outcomes were compared in the overall maintenance population. Filgotinib had a similar effect on efficacy outcomes compared with tofacitinib, ustekinumab, and subcutaneous vedolizumab in both the induction and maintenance populations. Filgotinib showed improved clinical response vs intravenous (IV) vedolizumab (odds ratio, 2.4; 95% confidence interval [CI], 1.0 to 5.5; P < .05) among the biologic-experienced induction population, and improved corticosteroid-free clinical remission (odds ratio, 15.2; 95% CI, 1.6 to 139.9; P < .05) among the biologic-naive maintenance population. Improved efficacy outcomes were reported with filgotinib compared with ustekinumab among the maintenance population. Higher estimates of serious adverse events were reported for filgotinib compared with vedolizumab IV 300 mg and tofacitinib 5 mg; however, imbalances were noted in their placebo groups. Health-related quality of life outcomes were similar between filgotinib and comparators. Matching-adjusted indirect comparison results suggest superiority of filgotinib 200 mg over vedolizumab IV in terms of clinical response and corticosteroid-free clinical remission in certain patient populations, noting small sample sizes and wide CIs, which may aid the selection of advanced therapies for moderately to severely active UC. A potential increased risk of serious adverse events was reported for filgotinib 200 mg vs vedolizumab IV and tofacitinib 5 mg, but findings should be interpreted with caution owing to underlying imbalances observed between the placebo groups of SELECTION and comparator trials.

When preparing training, validation and test sets for machine learning on molecular datasets, it is desirable to combine two requirements: 1) robustness, i.e. making a test set that is chemically dissimilar from the training set; 2) data balance, i.e. ensuring that the proportion of data points and the distribution of data labels (categorical) / data values (continuous) are as homogeneous as possible among the sets, for each individual property to model, while partitioning the overall set of compounds as required. Recent literature shows that meeting both these requirements simultaneously is sometimes very difficult. This is especially true for multi-task learning, but also for single-task learning if one aims to balance the distribution of data labels or values, too. In this work we present a method that resolves this issue by first carrying out a chemistry-guided clustering of the initial dataset to ensure the separation of chemical matter, and subsequently applying linear programming to select the lists of clusters that – once assembled into the final sets – result in the best possible data balance.

Introduction Filgotinib (FIL) is a Janus kinase 1 inhibitor being investigated for ulcerative colitis (UC).

Mounting evidence from the literature suggests that blocking S1P2 receptor (S1PR2) signaling could be effective for the treatment of idiopathic pulmonary fibrosis (IPF). However, only a few antagonists have been so far disclosed. A chemical enablement strategy led to the discovery of a pyridine series with good antagonist activity. A pyridazine series with improved lipophilic efficiency and with no CYP inhibition liability was identified by scaffold hopping. Further optimization led to the discovery of 40 (GLPG2938), a compound with exquisite potency on a phenotypic IL8 release assay, good pharmacokinetics, and good activity in a bleomycin-induced model of pulmonary fibrosis.

ObjectiveJanus kinase inhibitors (JAKinibs) are efficacious in rheumatoid arthritis (RA) with variable reported rates of adverse events, potentially related to differential JAK family member selectivity. Filgotinib was compared with baricitinib, tofacitinib and upadacitinib to elucidate the pharmacological basis underlying its clinical efficacy and safety.MethodsIn vitro JAKinib inhibition of signal transducer and activator of transcription phosphorylation (pSTAT) was measured by flow cytometry in peripheral blood mononuclear cells and whole blood from healthy donors and patients with RA following cytokine stimulation of distinct JAK/STAT pathways. The average daily pSTAT and time above 50% inhibition were calculated at clinical plasma drug exposures in immune cells. The translation of these measures was evaluated in ex vivo-stimulated assays in phase 1 healthy volunteers.ResultsJAKinib potencies depended on cytokine stimulus, pSTAT readout and cell type. JAK1-dependent pathways (interferon (IFN)α/pSTAT5, interleukin (IL)-6/pSTAT1) were among the most potently inhibited by all JAKinibs in healthy and RA blood, with filgotinib exhibiting the greatest selectivity for JAK1 pathways. Filgotinib (200 mg once daily) had calculated average daily target inhibition for IFNα/pSTAT5 and IL-6/pSTAT1 that was equivalent to tofacitinib (5 mg two times per day), upadacitinib (15 mg once daily) and baricitinib (4 mg once daily), with the least average daily inhibition for the JAK2-dependent and JAK3-dependent pathways including IL-2, IL-15, IL-4 (JAK1/JAK3), IFNγ (JAK1/JAK2), granulocyte colony stimulating factor, IL-12, IL-23 (JAK2/tyrosine kinase 2) and granulocyte-macrophage colony-stimulating factor (JAK2/JAK2). Ex vivo pharmacodynamic data from phase 1 healthy volunteers clinically confirmed JAK1 selectivity of filgotinib.ConclusionFilgotinib inhibited JAK1-mediated signalling similarly to other JAKinibs, but with less inhibition of JAK2-dependent and JAK3-dependent pathways, providing a mechanistic rationale for its apparently differentiated efficacy:safety profile.

This study assesses whether currently available acute oral toxicity (AOT) in silico models, provided by the widely employed Leadscope software, are fit-for-purpose for categorization and labelling of chemicals. As part of this study, a large data set of proprietary and marketed compounds from multiple companies (pharmaceutical, plant protection products, and other chemical industries) was assembled to assess the models’ performance. The absolute percentage of correct or more conservative predictions, based on a comparison of experimental and predicted GHS categories, was approximately 95%, after excluding a small percentage of inconclusive (indeterminate or out of domain) predictions. Since the frequency distribution across the experimental categories is skewed towards low toxicity chemicals, a balanced assessment was also performed. Across all compounds which could be assigned to a well-defined experimental category, the average percentage of correct or more conservative predictions was around 80%. These results indicate the potential for reliable and broad application of these models across different industrial sectors. This manuscript describes the evaluation of these models, highlights the importance of an expert review, and provides guidance on the use of AOT models to fulfill testing requirements, GHS classification/labelling, and transportation needs.