Journal: International Journal of Molecular Sciences

Article Title: B Cell Lymphoma 6 (BCL6): A Conserved Regulator of Immunity and Beyond

doi: 10.3390/ijms252010968

Figure Lengend Snippet: Structure and function of the BCL6 protein. ( A ). Domains of the BCL6 protein: BTB (light blue), RD2/PEST (blue), and ZF (dark blue, with striping for fingers not involved in DNA binding). Above is a schematic representation of the structure of each domain along with their interacting proteins, including co-repressors (pink) and associated transcriptional regulators (orange) and DNA modifying proteins (green), with sites of ubiquitination (Ub), phosphorylation (P), and acetylation (Ac) indicated (brown). ( B ). Major molecular function(s) of each domain. ( C ). Biological roles mapped to the molecular functions, with proven connections shown as filled boxes and assumed ones as unfilled boxes. Abbreviations: BTB: broad complex/tram track/bric-a-brac; RD2: repression domain 2; PEST: proline–glutamic acid–serine–threonine; ZF: zinc finger.

Article Snippet: Clinical trials are currently underway to evaluate PROTAC-based BCL6 inhibitors in the context of relapsed/refractory NHL, such as ARV-393 (Arvinas, phase 1, NCT06393738) and BMS-986458 (Bristol-Myers-Squibb, phase 1/2, NCT06090539).

Techniques: Binding Assay, Ubiquitin Proteomics, Phospho-proteomics

Journal: Biochemical Society Transactions

Article Title: Discovery, development and application of drugs targeting BCL-2 pro-survival proteins in cancer

doi: 10.1042/BST20210749

Figure Lengend Snippet: Key BH3-mimetics leading to the clinical applications or as tool compounds

Article Snippet: Nevertheless, exciting new approaches such as PROTACs, ADCs and novel formulations show enormous potential for overcoming many of these issues, making it likely that BH3-mimetics will soon have significant impact for the treatment of a wider range of cancers in the future.

Techniques: Modification, Clinical Proteomics, In Vivo, Activity Assay, Solubility, High Throughput Screening Assay

Journal: ACS Omega

Article Title: Predictive Modeling of PROTAC Cell Permeability with Machine Learning

doi: 10.1021/acsomega.2c07717

Figure Lengend Snippet: PROTAC datasets and their characterization. (A) Overview of the structural composition of the PROTACs in the VHL ( n = 115) and CRBN ( n = 113) sets. (B) Distribution of the molecular descriptors of Lipinski’s and Veber’s guidelines for the two sets. Box plots show the 50 th percentiles as horizontal bars, the 25 th and 75 th percentiles as boxes, and the 25 th percentile minus 1.5 × the interquartile range and the 75 th percentile plus 1.5 × the interquartile range as whiskers. Outliers are shown both as red dots and as circles in the color of the appropriate descriptor. (C) Score plots of the first two principal components from principal component analyses (PCAs), which describe 71.5% of the variance for the VHL set and 74.9% of the variance for CRBN. The PCAs were based on the 17 descriptors calculated for each PROTAC, which were subsequently used for construction of the permeability models (cf. Figure A). Ellipses in green, yellow, and red shading show the 95% confidence intervals for highly, moderately, and lowly permeable compounds, respectively. The centroid of each permeability class is indicated with a large circle in the color of the respective class. The contributions of individual descriptors to the PCAs are indicated by arrows.

Article Snippet: All PROTACs were prepared at Bayer AG, and their structures were confirmed by high-resolution mass spectrometry and 1 H NMR spectroscopy.

Techniques: Permeability

Journal: ACS Omega

Article Title: Predictive Modeling of PROTAC Cell Permeability with Machine Learning

doi: 10.1021/acsomega.2c07717

Figure Lengend Snippet: (A) Principal component analysis comparing the chemical space of PROTACs in the public domain (red and cyan circles) to our in-house set (green circles). Public PROTACs that are within the applicability domain of the in-house set are in red, while those outside are in cyan. The centroids for each set are indicated with a large circle in the color of the respective set. (B) Examples of molecular structures of two PROTACs that reside outside the chemical space of the in-house set. The descriptors of the Lipinski and Veber guidelines are given below the structure of each PROTAC.

Article Snippet: All PROTACs were prepared at Bayer AG, and their structures were confirmed by high-resolution mass spectrometry and 1 H NMR spectroscopy.

Techniques:

Journal: ACS Omega

Article Title: Predictive Modeling of PROTAC Cell Permeability with Machine Learning

doi: 10.1021/acsomega.2c07717

Figure Lengend Snippet: Cohen’s kappa statistics for internal test set validation of different BCMs for three permeability scenarios of (A) CRBN and (B) VHL PROTACs. Box plots show the kappa values from 25 random seedlings, while the yellow circles show the kappa values from 10-fold cross validation. In the box plots, the 50 th percentiles are marked as horizontal bars, the 25 th and 75 th percentiles as boxes, and the 25 th percentile minus 1.5 × the interquartile range and the 75 th percentile plus 1.5 × the interquartile range as whiskers. Outliers are shown both as black dots and as circles in the color of the method used to build the model. DT: decision tree, kNN: kappa nearest neighbor, RF: random forest, and SVM: support vector machine. Classification models can be assessed using the following cut-offs for Cohen’s kappa: κ < 0: no agreement, 0–0.19: poor agreement, 0.20–0.39: fair agreement, 0.40–0.59: moderate agreement, and 0.60–0.79 and 0.80–1.00: substantial to perfect agreement.

Article Snippet: All PROTACs were prepared at Bayer AG, and their structures were confirmed by high-resolution mass spectrometry and 1 H NMR spectroscopy.

Techniques: Biomarker Discovery, Permeability, Plasmid Preparation

Journal: ACS Omega

Article Title: Predictive Modeling of PROTAC Cell Permeability with Machine Learning

doi: 10.1021/acsomega.2c07717

Figure Lengend Snippet: Probability distribution of true predictions for the random forest models built using the original VHL dataset. PROTACs having a probability smaller or larger than 0.5 were correctly classified as having low (orange) or high (green) permeability, respectively. A probability of 0.9–1.0 indicates that the compound was predicted to have a high permeability with >90% probability. Similarly, a probability of 0–0.1 indicates that the compound was predicted to have a low permeability with >90% probability.

Article Snippet: All PROTACs were prepared at Bayer AG, and their structures were confirmed by high-resolution mass spectrometry and 1 H NMR spectroscopy.

Techniques: Permeability

Journal: ACS Omega

Article Title: Predictive Modeling of PROTAC Cell Permeability with Machine Learning

doi: 10.1021/acsomega.2c07717

Figure Lengend Snippet: Number of compounds in the training sets of PROTACs used to construct BCMs (original and retrained set) and the datasets used as blinded test sets for validation of the models (blinded test sets 1 and 2). For each dataset, the distribution of compounds between VHL and CRBN PROTACs, as well as by permeability class, is given.

Article Snippet: All PROTACs were prepared at Bayer AG, and their structures were confirmed by high-resolution mass spectrometry and 1 H NMR spectroscopy.

Techniques: Construct, Biomarker Discovery, Permeability

Journal: ACS Omega

Article Title: Predictive Modeling of PROTAC Cell Permeability with Machine Learning

doi: 10.1021/acsomega.2c07717

Figure Lengend Snippet: Cohen’s kappa coefficient for prediction of the permeability of the VHL and CRBN PROTACs in the blinded test set 1. The kappa coefficient is given for the three permeability scenarios for models constructed using the DT, kNN, and RF methods based on the original dataset and its SMOTE versions. Kappa coefficients have been color-coded using red-orange-yellow-green for values ranging from −0.3 to 0.7.

Article Snippet: All PROTACs were prepared at Bayer AG, and their structures were confirmed by high-resolution mass spectrometry and 1 H NMR spectroscopy.

Techniques: Permeability, Construct

Journal: ACS Omega

Article Title: Predictive Modeling of PROTAC Cell Permeability with Machine Learning

doi: 10.1021/acsomega.2c07717

Figure Lengend Snippet: Cohen’s kappa statistics for internal validation of different retrained BCMs for three permeability scenarios of (A) CRBN and (B) VHL PROTACs in the retrained set. Box plots show the kappa values from 25 random seedlings, while the yellow circles show the kappa values from 10-fold cross validation. In the box plots, the 50 th percentiles are marked as horizontal bars, the 25 th and 75 th percentiles as boxes, and the 25 th percentile minus 1.5 × the interquartile range and the 75 th percentile plus 1.5 × the interquartile range as whiskers. Outliers are shown both as black dots and as circles in the color of the method used to build the model. DT: decision tree, kNN: kappa nearest neighbor, and RF: random forest. Classification models can be assessed using the following cut-offs for Cohen’s kappa: k < 0: no agreement, 0–0.19: poor agreement, 0.20–0.39: fair agreement, 0.40–0.59: moderate agreement, and 0.60–0.79 and 0.80–1.00: substantial to perfect agreement.

Article Snippet: All PROTACs were prepared at Bayer AG, and their structures were confirmed by high-resolution mass spectrometry and 1 H NMR spectroscopy.

Techniques: Biomarker Discovery, Permeability

Journal: ACS Omega

Article Title: Predictive Modeling of PROTAC Cell Permeability with Machine Learning

doi: 10.1021/acsomega.2c07717

Figure Lengend Snippet: Cohen’s kappa coefficient for prediction of the permeability of the VHL PROTACs in the blinded test set 2 using models constructed with the (A) original training set and the (B) retraining set. The kappa coefficient is given for the three permeability scenarios for models constructed using the DT, kNN, and RF methods. Kappa coefficients have been color-coded using red-orange-yellow-green for values ranging from −0.30 to 0.70.

Article Snippet: All PROTACs were prepared at Bayer AG, and their structures were confirmed by high-resolution mass spectrometry and 1 H NMR spectroscopy.

Techniques: Permeability, Construct

Journal: ACS Omega

Article Title: Predictive Modeling of PROTAC Cell Permeability with Machine Learning

doi: 10.1021/acsomega.2c07717

Figure Lengend Snippet: Contribution of the descriptors to the retrained RF models for prediction of the permeability of VHL PROTACs. The figure shows the mean values of the weight of each descriptor for permeability scenarios 1–3, with error bars indicating ± standard deviation. The weight of the contribution of each descriptor to the model was obtained from the 10-fold cross validation. The descriptors that contribute most to the model are indicated by the blue shading at a weight of ≥0.4. Color code: violet: countable descriptors, pink: chemical functionalities descriptors, and green: size and shape descriptors. Descriptor contributions for the individual models for scenarios 1–3 can be found in the Supporting Information, Figure S10B .

Article Snippet: All PROTACs were prepared at Bayer AG, and their structures were confirmed by high-resolution mass spectrometry and 1 H NMR spectroscopy.

Techniques: Permeability, Standard Deviation, Biomarker Discovery

Journal: ACS Omega

Article Title: Predictive Modeling of PROTAC Cell Permeability with Machine Learning

doi: 10.1021/acsomega.2c07717

Figure Lengend Snippet: Distribution of the molecular descriptors of Lipinski’s and Veber’s guidelines for the linker part ( n = 129) of the VHL PROTACs in the combined training set and blinded test set 1 ( n = 253). Distributions have been calculated for the linkers of the PROTACS in each of the three permeability classes. Box plots show the 50 th percentiles as horizontal bars, the 25 th and 75 th percentiles as boxes, and the 25 th percentile minus 1.5 × the interquartile range and the 75 th percentile plus 1.5 × the interquartile range as whiskers. Outliers are shown both as black dots. Statistical analysis was performed using Wilcoxon’s non-parametric test.

Article Snippet: All PROTACs were prepared at Bayer AG, and their structures were confirmed by high-resolution mass spectrometry and 1 H NMR spectroscopy.

Techniques: Permeability

Journal: ACS Omega

Article Title: Predictive Modeling of PROTAC Cell Permeability with Machine Learning

doi: 10.1021/acsomega.2c07717

Figure Lengend Snippet: Number of PROTACs Used for Data Analysis, Model Building, and Validation

Article Snippet: All PROTACs were prepared at Bayer AG, and their structures were confirmed by high-resolution mass spectrometry and 1 H NMR spectroscopy.

Techniques:

Journal: ACS Omega

Article Title: Predictive Modeling of PROTAC Cell Permeability with Machine Learning

doi: 10.1021/acsomega.2c07717

Figure Lengend Snippet: Overview of Purities of the PROTACs Included in the Training and Tests Sets

Article Snippet: All PROTACs were prepared at Bayer AG, and their structures were confirmed by high-resolution mass spectrometry and 1 H NMR spectroscopy.

Techniques: Standard Deviation

Journal: International Journal of Molecular Sciences

Article Title: Kinase Inhibitors and Kinase-Targeted Cancer Therapies: Recent Advances and Future Perspectives

doi: 10.3390/ijms25105489

Figure Lengend Snippet: Mechanism of action of kinase-targeted cancer therapies, including monoclonal antibodies and nanobodies, kinase degraders, and protein–kinase interaction inhibitors. ( a ) Monoclonal antibodies (mAbs) primarily inhibit signal transduction through the binding of the antibody’s Fab segment to the receptor’s extracellular domain. Nanobodies lack the Fc portion, also known as single-domain antibodies or VHH. ( b ) PROTACs are bifunctional molecules that induce the proximity of E3 ligases and the protein of interest (POI) by linking them via ligands. Molecular glue is a monovalent molecule that binds to the surface of E3 ligase receptors, achieving binding to the target protein through protein–protein interactions and leading to the degradation of the POI. ( c ) Protein–kinase interaction inhibitors (PKIIs), including small molecules and linear peptides, inhibit the interaction between kinases and substrates, being ideal candidates for inhibiting protein–protein interactions (PPIs).

Article Snippet: EGFR , HSK40118 , PROTACs , EGFR-positive NSCLC , Phase 1 Haisco Pharmaceutical Group Co., Ltd. (Chengdu, China). CTR20230926.

Techniques: Bioprocessing, Transduction, Binding Assay, Protein-Protein interactions

Journal: International Journal of Molecular Sciences

Article Title: Kinase Inhibitors and Kinase-Targeted Cancer Therapies: Recent Advances and Future Perspectives

doi: 10.3390/ijms25105489

Figure Lengend Snippet: New strategies for targeting kinases.

Article Snippet: EGFR , HSK40118 , PROTACs , EGFR-positive NSCLC , Phase 1 Haisco Pharmaceutical Group Co., Ltd. (Chengdu, China). CTR20230926.

Techniques: Mutagenesis