NIBR-LTSi is a selective LATS kinase inhibitor activating YAP signaling and expanding tissue stem cells in vitro and in vivo.

The YAP/Hippo pathway is an organ growth and size regulation rheostat safeguarding multiple tissue stem cell compartments. LATS kinases phosphorylate and thereby inactivate YAP, thus representing a potential direct drug target for promoting tissue regeneration. Here, we report the identification and characterization of the selective small-molecule LATS kinase inhibitor NIBR-LTSi. NIBR-LTSi activates YAP signaling, shows good oral bioavailability, and expands organoids derived from several mouse and human tissues. In tissue stem cells, NIBR-LTSi promotes proliferation, maintains stemness, and blocks differentiation in vitro and in vivo. NIBR-LTSi accelerates liver regeneration following extended hepatectomy in mice. However, increased proliferation and cell dedifferentiation in multiple organs prevent prolonged systemic LATS inhibition, thus limiting potential therapeutic benefit. Together, we report a selective LATS kinase inhibitor agonizing YAP signaling and promoting tissue regeneration in vitro and in vivo, enabling future research on the regenerative potential of the YAP/Hippo pathway.

Deep Learning Models Compared to Experimental Variability for the Prediction of CYP3A4 Time-Dependent Inhibition.

Most drugs are mainly metabolized by cytochrome P450 (CYP450), which can lead to drug-drug interactions (DDI). Specifically, time-dependent inhibition (TDI) of CYP3A4 isoenzyme has been associated with clinically relevant DDI. To overcome potential DDI issues, high-throughput in vitro assays were established to assess the TDI of CYP3A4 during the discovery and lead optimization phases. However, in silico machine learning models would enable an earlier and larger-scale assessment of TDI potential liabilities. For CYP inhibition, most modeling efforts have focused on highly imbalanced and small data sets. Moreover, assay variability is rarely considered, which is key to understand the model's quality and suitability for decision-making. In this work, machine learning models were built for the prediction of TDI of CYP3A4, evaluated prospectively, and compared to the variability of the experimental assay. Different modeling strategies were investigated to assess their influence on the model's performance. Through multitask learning, additional data sets were leveraged for model building, coming from public databases, in-house CYP-related assays, or other pharmaceutical companies (federated learning). Apart from the numerical prediction of inactivation rates of CYP3A4 TDI, three-class predictions were carried out, giving a negative (inactivation rate kobs < 0.01 min-1), weak positive (0.01 ≤ kobs ≤ 0.025 min-1), or positive (kobs > 0.025 min-1) output. The final multitask graph neural network model achieved misclassification rates of 8 and 7% for positive and negative TDI, respectively. Importantly, the presented deep learning-based predictions had a similar precision to the reproducibility of in vitro experiments and thus offered great opportunities for drug design, early derisk of DDI potential, and selection of experiments. To facilitate CYP inhibition modeling efforts in the public domain, the developed model was used to annotate ∼16 000 publicly available structures, and a surrogate data set is shared as Supporting Information.

Improving In Vitro-In Vivo Extrapolation of Clearance Using Rat Liver Microsomes for Highly Plasma Protein-Bound Molecules.

It is common practice in drug discovery and development to predict in vivo hepatic clearance from in vitro incubations with liver microsomes or hepatocytes using the well-stirred model (WSM). When applying the WSM to a set of approximately 3000 Novartis research compounds, 73% of neutral and basic compounds (extended clearance classification system [ECCS] class 2) were well-predicted within 3-fold. In contrast, only 44% (ECCS class 1A) or 34% (ECCS class 1B) of acids were predicted within 3-fold. To explore the hypothesis whether the higher degree of plasma protein binding for acids contributes to the in vitro-in vivo correlation (IVIVC) disconnect, 68 proprietary compounds were incubated with rat liver microsomes in the presence and absence of 5% plasma. A minor impact of plasma on clearance IVIVC was found for moderately bound compounds (fraction unbound in plasma [fup] ≥1%). However, addition of plasma significantly improved the IVIVC for highly bound compounds (fup <1%) as indicated by an increase of the average fold error from 0.10 to 0.36. Correlating fup with the scaled unbound intrinsic clearance ratio in the presence or absence of plasma allowed the establishment of an empirical, nonlinear correction equation that depends on fup Taken together, estimation of the metabolic clearance of highly bound compounds was enhanced by the addition of plasma to microsomal incubations. For standard incubations in buffer only, application of an empirical correction provided improved clearance predictions. SIGNIFICANCE STATEMENT: Application of the well-stirred liver model for clearance in vitro-in vivo extrapolation (IVIVE) in rat generally underpredicts the clearance of acids and the strong protein binding of acids is suspected to be one responsible factor. Unbound intrinsic in vitro clearance (CLint,u) determinations using rat liver microsomes supplemented with 5% plasma resulted in an improved IVIVE. An empirical equation was derived that can be applied to correct CLint,u-values in dependance of fraction unbound in plasma (fup) and measured CLint in buffer.

Design of a Supersoft Topical JAK Inhibitor, Which Is Effective in Human Skin but Rapidly Deactivated in Blood.

We describe the discovery and characterization of the supersoft topical JAK inhibitor 3(R), which is potent in biochemical and cellular assays as well as in human skin models. In blood, the neutral ester 3(R) is rapidly hydrolyzed (t1/2 ∼ 6 min) to the corresponding charged carboxylic acid 4 exhibiting >30-fold reduced permeability. Consequently, acid 4 does not reach the intracellular JAK kinases and is inactive in cellular assays and in blood. Thus, hydrolysis by blood esterases leads to the rapid deactivation of topically active ester 3(R) at a rate beyond the maximal hepatic clearance.

Multispecies Machine Learning Predictions of In Vitro Intrinsic Clearance with Uncertainty Quantification Analyses.

In pharmaceutical research, compounds are optimized for metabolic stability to avoid a too fast elimination of the drug. Intrinsic clearance (CLint) measured in liver microsomes or hepatocytes is an important parameter during lead optimization. In this work, machine learning models were developed to relate the compound structure to microsomal metabolic stability and predict CLint for new compounds. A multitask (MT) learning architecture was introduced to model the CLint of six species simultaneously, giving as a result a multispecies machine learning model. MT graph neural network (MT-GNN) regression was identified as the top-performing method, and an ensemble of 10 MT-GNN models was evaluated prospectively. Geometric mean fold errors were consistently smaller than 2-fold. Moreover, high precision values were obtained in the prediction of "high" (>300 µL/min/mg) and "low" (<100 µL/min/mg) CLint compounds. Precision values ranged from 80 to 94% for low CLint predictions and from 75 to 97% for high CLint predictions, depending on the species. Uncertainty on experimental values and model predictions was systematically quantified. Experimental variability (aleatoric uncertainty) of all historical Novartis in vitro clearance experiments was analyzed. Interestingly, MT-GNN models' performance approached assays' experimental variability. Moreover, uncertainty estimation in predictions (epistemic uncertainty) enabled identifying predictions associated with lower and higher error. Taken together, our manuscript combines a multispecies deep learning model and large-scale uncertainty analyses to improve CLint predictions and facilitate early informed decisions for compound prioritization.

Evaluation of relative MS response factors of drug metabolites for semi-quantitative assessment of chemical liabilities in drug discovery.

Drug metabolism studies are performed in drug discovery to identify metabolic soft spots, detect potentially toxic or reactive metabolites and provide an early insight into potential species differences. The relative peak area approach is often used to semi-quantitatively estimate the abundance of metabolites. Differences in the liquid chromatography-mass spectrometry responses result in an underestimation or overestimation of the metabolite and misinterpretation of results. The relative MS response factors (RF) of 132 structurally diverse drug candidates and their 233 corresponding metabolites were evaluated using a capillary-liquid chromatography/high-resolution mass spectrometry system. All of the synthesized metabolites discussed here were previously identified as key biotransformation products in discovery investigations or predicted to be formed. The most commonly occurring biotransformation mechanisms such as oxygenation, dealkylation and amide cleavage are represented within this dataset. However, relatively few phase II metabolites were evaluated because of the limited availability of authentic standards. Approximately 85% of these metabolites had a relative RF in the range between 0.2 (fivefold under-prediction) and 2.0 (twofold over-prediction), and the median MS RF was 0.6. Exceptions to this included very small metabolites that were hardly detectable. Additional experiments performed to understand the impact of the MS platform, flow rate and concentration suggested that these parameters do not have a significant impact on the RF of the compounds tested. This indicates that the use of relative peak areas to semi-quantitatively estimate the abundance of metabolites is justified in the drug discovery setting in order to guide medicinal chemistry efforts. Copyright © 2017 John Wiley & Sons, Ltd.

An integrated approach for identification and target validation of antifungal compounds active against Erg11p.

Systemic life-threatening fungal infections represent a significant unmet medical need. Cell-based, phenotypic screening can be an effective means of discovering potential novel antifungal compounds, but it does not address target identification, normally required for compound optimization by medicinal chemistry. Here, we demonstrate a combination of screening, genetic, and biochemical approaches to identify and characterize novel antifungal compounds. We isolated a set of novel non-azole antifungal compounds for which no target or mechanism of action is known, using a screen for inhibition of Saccharomyces cerevisiae proliferation. Haploinsufficiency profiling of these compounds in S. cerevisiae suggests that they target Erg11p, a cytochrome P450 family member, which is the target of azoles. Consistent with this, metabolic profiling in S. cerevisiae revealed a buildup of the metabolic intermediates prior to Erg11p activity, following compound treatment. Further, human cytochrome P450 is also inhibited in in vitro assays by these compounds. We modeled the Erg11p protein based on the human CYP51 crystal structure, and in silico docking of these compounds suggests that they interact with the heme center in a manner similar to that of azoles. Consistent with these docking observations, Candida strains carrying azole-resistant alleles of ERG11 are also resistant to the compounds in this study. Thus, we have identified non-azole Erg11p inhibitors, using a systematic approach for ligand and target characterization.

CYP3A time-dependent inhibition risk assessment validated with 400 reference drugs.

Although reversible CYP3A inhibition testing is well established for predicting the drug-drug interaction potential of clinical candidates, time-dependent inhibition (TDI) has become the focus of drug designers only recently. Failure of several late-stage clinical candidates has been attributed to TDI, and this mechanism is also suspected to play a role in liver toxicities often observed in preclinical species. Measurement of enzyme inactivation rates (k(inact) and K(I)) is technically challenging, and a great deal of variability can be found in the literature. In this article, we have evaluated the TDI potential for 400 registered drugs using a high-throughput assay format based on determination of the inactivation rate (k(obs)) at a single concentration of test compound (10 µM). The advantages of this new assay format are highlighted by comparison with data generated using the IC50 shift assay, a current standard approach for preliminary assessment of TDI. With use of an empirically defined positive/negative k(obs) bin of 0.02 min⁻¹, only 4% of registered drugs were found to be positive. This proportion increased to more than 20% when in-house lead optimization molecules were considered, emphasizing the importance of identifying this property in selection of promising drug candidates. Finally, it is suggested that the data and technology described here may be a good basis for building structure-activity relationships and in silico modeling.

Metabolic soft spot identification and compound optimization in early discovery phases using MetaSite and LC-MS/MS validation.

Metabolic stability is a key property to enable drugs to reach therapeutic concentrations. Microsomal clearance assays are used to dial out labile compounds in early discovery phases. However, because they do not provide any information on soft spots, the rational design of more stable compounds remains challenging. A robust soft spot identification procedure combining in silico prediction ranking using MetaSite and mass-spectrometric confirmation is described. MetaSite's first rank order predictions were experimentally confirmed for only about 55% of the compounds. For another 29% of the compounds, the second (20%) or the third (9%) rank order predictions were detected. This automatic and high-throughput reprioritization of a likely soft-spot increases the likelihood of working on the right soft spot from about 50% to more than 80%. With this information, the structure-metabolism relationships are likely to be understood faster and earlier in drug discovery.

Cytostar-T plates--a valid alternative for microplate scintillation counting of low radioactivity in combination with high-performance liquid chromatography in drug metabolism studies?

The determination of radioactivity from metabolite patterns in ADME studies in a low radioactivity/residue situation is a very challenging process requiring special technologies. The recently introduced TopCount technology uses LumaPlates for the collection of the column effluent after HPLC separation to subsequently determine radioactivity for the generation of the metabolite profile. Samples from drug metabolism studies were used to compare the performance of the widely used LumaPlates with Cytostar-T plates regarding sensitivity and recovery of metabolites for structure elucidation by MS. Optimized counting parameters were investigated for the Cytostar-T plates. This had led to higher sensitivity and therefore to a preferential signal to noise ratio. Metabolites which were collected into Cytostar-T instead of LumaPlates could be easily recovered and directly used for structure elucidation by MS. The full scan mass spectra of recovered metabolites showed higher quality allowing the characterization of metabolites without any further sample pre-treatment. This is a major advantage which could further speed-up the structure elucidation process of metabolites in complex biological matrices.

Two-dimensional liquid chromatography/mass spectrometry set-up for structural elucidation of metabolites in complex biological matrices.

For absorption, distribution, metabolism and excretion (ADME) studies of drug candidates, mass spectrometry (MS) has become an indispensable tool for the characterization of biotransformation pathways. Samples from in vivo animal studies such as plasma, tissue extracts or excreta contain vast amounts of endogenous compounds. Therefore, the generation of metabolite patterns requires dedicated sample pre-treatment and sophisticated separation methods. Methodologies used for metabolite separation are often inappropriate for structure elucidation. Therefore, a two-dimensional liquid chromatography (LC) approach in combination with MS was developed. Study samples were analyzed using high-performance liquid chromatography (HPLC) for the generation of a qualitative and quantitative metabolite pattern (first dimension) with high reproducibility and recovery without extensive sample pre-treatment. Selected radioactive metabolite fractions were then applied to micro-HPLC with off-line radioactivity monitoring and subsequent MS detection (second dimension). Applying the two-dimensional HPLC/MS approach not only major metabolites could be identified, even minor and trace metabolites were characterized. The usage of sampled metabolite fractions allowed also the re-analysis of specific metabolites for additional investigations (e.g. H/D exchange experiments or product ion scanning experiments). It could be clearly shown that the two-dimensional HPLC/MS approach showed mass spectra with higher sensitivity and selectivity significantly improving the characterization of minor and trace metabolites in in vivo ADME studies.

Comparison of a two-dimensional liquid chromatography/mass spectrometry approach with a chip-based nanoelectrospray device for structural elucidation of metabolites in a human ADME study using a quadrupole time-of-flight mass spectrometer.

The study of the metabolic fate of drugs is essential for the safety assessment of new compounds in the drug development process. However, the characterization and structural elucidation of metabolites from in vivo experiments is still a very challenging task. In this paper, we compare a two-dimensional liquid chromatography/mass spectrometry (LC/MS) approach using either a capillary LC/MS system or the recently introduced chip-based nanoelectrospray/MS system (Nanomate) as the second dimension for structural elucidation of metabolites by MS. More than 30 radioactive fractions of a chromatographic separation from a human urine sample were analyzed and 54 metabolites could be identified. The long persisting and stable nanoelectrospray enabled the search for unknown metabolites by precursor-ion scanning experiments followed by product-ion scanning experiments of potential metabolites using a quadrupole time-of-flight (qTOF) mass spectrometer. The number of fragments produced by nanoelectrospray with product-ion scanning was significantly higher compared to LC/MS experiments with in-source fragmentation. Therefore, the assignment of possible modifications in metabolites to certain moieties of the drug could be investigated with higher accuracy. The capillary LC/MS system for the second dimension was more sensitive in the case of low abundant metabolites. These metabolites could not be detected by direct nanoelectrospray infusion, which limits the application of the Nanomate for trace metabolites.