Privacy-preserving techniques for decentralized and secure machine learning in drug discovery.

Data availability, data security, and privacy concerns often hamper optimal performance efficiency of machine learning (ML) techniques. Therefore, novel techniques for the utilization of private/sensitive data in the field of drug discovery have been proposed for ML model-building tasks. Some examples of the different techniques are secure multiparty computation, distributed deep learning, homomorphic encryption, blockchain-based peer-to-peer networking, differential privacy, and federated learning, as well as combinations of such techniques. In this paper, we present an overview of these techniques for decentralized ML to illustrate its benefits and drawbacks in the field of drug discovery.

Anti-cancer pro-inflammatory effects of an IgE antibody targeting the melanoma-associated antigen chondroitin sulfate proteoglycan 4.

Outcomes for half of patients with melanoma remain poor despite standard-of-care checkpoint inhibitor therapies. The prevalence of the melanoma-associated antigen chondroitin sulfate proteoglycan 4 (CSPG4) expression is ~70%, therefore effective immunotherapies directed at CSPG4 could benefit many patients. Since IgE exerts potent immune-activating functions in tissues, we engineer a monoclonal IgE antibody with human constant domains recognizing CSPG4 to target melanoma. CSPG4 IgE binds to human melanomas including metastases, mediates tumoricidal antibody-dependent cellular cytotoxicity and stimulates human IgE Fc-receptor-expressing monocytes towards pro-inflammatory phenotypes. IgE demonstrates anti-tumor activity in human melanoma xenograft models engrafted with human effector cells and is associated with enhanced macrophage infiltration, enriched monocyte and macrophage gene signatures and pro-inflammatory signaling pathways in the tumor microenvironment. IgE prolongs the survival of patient-derived xenograft-bearing mice reconstituted with autologous immune cells. No ex vivo activation of basophils in patient blood is measured in the presence of CSPG4 IgE. Our findings support a promising IgE-based immunotherapy for melanoma.

Ligand- and Structure-based Approaches for Transmembrane Transporter Modeling.

The study of transporter proteins is key to understanding the mechanism behind multi-drug resistance and drug-drug interactions causing severe side effects. While ATP-binding transporters are well-studied, solute carriers illustrate an understudied family with a high number of orphan proteins. To study these transporters, in silico methods can be used to shed light on the basic molecular machinery by studying protein-ligand interactions. Nowadays, computational methods are an integral part of the drug discovery and development process. In this short review, computational approaches, such as machine learning, are discussed, which try to tackle interactions between transport proteins and certain compounds to locate target proteins. Furthermore, a few cases of selected members of the ATP binding transporter and solute carrier family are covered, which are of high interest in clinical drug interaction studies, especially for regulatory agencies. The strengths and limitations of ligand-based and structure-based methods are discussed to highlight their applicability for different studies. Furthermore, the combination of multiple approaches can improve the information obtained to find crucial amino acids that explain important interactions of protein-ligand complexes in more detail. This allows the design of drug candidates with increased activity towards a target protein, which further helps to support future synthetic efforts.

Systematic screening identifies ABCG2 as critical factor underlying synergy of kinase inhibitors with transcriptional CDK inhibitors.

BACKGROUND: Triple-negative breast cancer (TNBC) is a subtype of breast cancer with limited treatment options and poor clinical prognosis. Inhibitors of transcriptional CDKs are currently under thorough investigation for application in the treatment of multiple cancer types, including breast cancer. These studies have raised interest in combining these inhibitors, including CDK12/13 inhibitor THZ531, with a variety of other anti-cancer agents. However, the full scope of these potential synergistic interactions of transcriptional CDK inhibitors with kinase inhibitors has not been systematically investigated. Moreover, the mechanisms behind these previously described synergistic interactions remain largely elusive. METHODS: Kinase inhibitor combination screenings were performed to identify kinase inhibitors that synergize with CDK7 inhibitor THZ1 and CDK12/13 inhibitor THZ531 in TNBC cell lines. CRISPR-Cas9 knockout screening and transcriptomic evaluation of resistant versus sensitive cell lines were performed to identify genes critical for THZ531 resistance. RNA sequencing analysis after treatment with individual and combined synergistic treatments was performed to gain further insights into the mechanism of this synergy. Kinase inhibitor screening in combination with visualization of ABCG2-substrate pheophorbide A was used to identify kinase inhibitors that inhibit ABCG2. Multiple transcriptional CDK inhibitors were evaluated to extend the significance of the found mechanism to other transcriptional CDK inhibitors. RESULTS: We show that a very high number of tyrosine kinase inhibitors synergize with the CDK12/13 inhibitor THZ531. Yet, we identified the multidrug transporter ABCG2 as key determinant of THZ531 resistance in TNBC cells. Mechanistically, we demonstrate that most synergistic kinase inhibitors block ABCG2 function, thereby sensitizing cells to transcriptional CDK inhibitors, including THZ531. Accordingly, these kinase inhibitors potentiate the effects of THZ531, disrupting gene expression and increasing intronic polyadenylation. CONCLUSION: Overall, this study demonstrates the critical role of ABCG2 in limiting the efficacy of transcriptional CDK inhibitors and identifies multiple kinase inhibitors that disrupt ABCG2 transporter function and thereby synergize with these CDK inhibitors. These findings therefore further facilitate the development of new (combination) therapies targeting transcriptional CDKs and highlight the importance of evaluating the role of ABC transporters in synergistic drug-drug interactions in general.

Improving the efficacy of plant-made anti-HIV monoclonal antibodies for clinical use.

Introduction: Broadly neutralising antibodies are promising candidates for preventing and treating Human Immunodeficiency Virus/Acquired Immunodeficiency Syndrome (HIV/AIDS), as an alternative to or in combination with antiretroviral therapy (ART). These mAbs bind to sites on the virus essential for virus attachment and entry, thereby inhibiting entry into the host cell. However, the cost and availability of monoclonal antibodies, especially combinations of antibodies, hampers implementation of anti-HIV bNAb therapies in low- to middle- income countries (LMICs) where HIV-1 prevalence is highest. Methods: We have produced three HIV broadly neutralizing antibodies (bNAbs), 10-1074, VRC01 and 3BNC117 in the Nicotiana benthamiana transient expression system. The impact of specific modifications to enhance potency and efficacy were assessed. To prolong half-life and increase bioavailability, a M252Y/S254T/T256E (YTE) or M428L/N434S (LS) mutation was introduced. To increase antibody dependent cellular cytotoxicity (ADCC), we expressed an afucosylated version of each antibody using a glycoengineered plant line. Results: The majority of bNAbs and their variants could be expressed at yields of up to 47 mg/kg. Neither the expression system nor the modifications impacted the neutralization potential of the bNAbs. Afucosylated bNAbs exhibit enhanced ability to bind to FcγRIIIa and trigger ADCC, regardless of the presence of Fc amino acid mutations. Lastly, we demonstrated that Fc-modified variants expressed in plants show enhanced binding to FcRn, which results in a favourable in vivo pharmacokinetic profile compared to their unmodified counterparts. Conclusion: Tobacco plants are suitable expression hosts for anti-HIV bNAbs with increased efficacy and an improved pharmacokinetic profile.

Folate receptor alpha in ovarian cancer tissue and patient serum is associated with disease burden and treatment outcomes.

BACKGROUND: Survival rates for ovarian cancer remain poor, and monitoring and prediction of therapeutic response may benefit from additional markers. Ovarian cancers frequently overexpress Folate Receptor alpha (FRα) and the soluble receptor (sFRα) is measurable in blood. Here we investigated sFRα as a potential biomarker. METHODS: We evaluated sFRα longitudinally, before and during neo-adjuvant, adjuvant and palliative therapies, and tumour FRα expression status by immunohistrochemistry. The impact of free FRα on the efficacy of anti-FRα treatments was evaluated by an antibody-dependent cellular cytotoxicity assay. RESULTS: Membrane and/or cytoplasmic FRα staining were observed in 52.7% tumours from 316 ovarian cancer patients with diverse histotypes. Circulating sFRα levels were significantly higher in patients, compared to healthy volunteers, specifically in patients sampled prior to neoadjuvant and palliative treatments. sFRα was associated with FRα cell membrane expression in the tumour. sFRα levels decreased alongside concurrent tumour burden in patients receiving standard therapies. High concentrations of sFRα partly reduced anti-FRα antibody tumour cell killing, an effect overcome by increased antibody doses. CONCLUSIONS: sFRα may present a non-invasive marker for tumour FRα expression, with the potential for monitoring patient response to treatment. Larger, prospective studies should evaluate FRα for assessing disease burden and response to systemic treatments.

Generation and Characterization of Native and Sialic Acid-Deficient IgE.

Efficient characterization of IgE antibodies and their glycan structures is required for understanding their function in allergy and in the emerging AllergoOncology field for antibody immunotherapy. We report the generation, glyco-profiling and functional analysis of native and sialic acid-deficient glyco-engineered human IgE. The antibodies produced from human embryonic kidney cells were purified via a human IgE class-specific affinity matrix and structural integrity was confirmed by SDS-PAGE and size-exclusion chromatography (SEC). Purified IgEs specific for the tumor-associated antigens Chondroitin Sulfate Proteoglycan 4 (CSPG4-IgE) and Human Epidermal Growth Factor Receptor 2 (HER2-IgE) were devoid of by-products such as free light chains. Using neuraminidase-A, we generated sialic acid-deficient CSPG4-IgE as example glyco-engineered antibody. Comparative glycan analyses of native and glyco-engineered IgEs by Hydrophilic interaction liquid chromatography (HILIC)-high performance liquid chromatography (HPLC) indicated loss of sialic acid terminal residues and differential glycan profiles. Native and glyco-engineered CSPG4-IgEs recognized Fc receptors on the surface of human FcεRI-expressing rat basophilic leukemia RBL-SX38 cells, and of CD23/FcεRII-expressing human RPMI-8866 B-lymphocytes and bound to CSPG4-expressing A2058 human melanoma cells, confirming Fab-mediated recognition. When cross-linked on the cell surface, both IgEs triggered RBL-SX38 degranulation. We demonstrate efficient generation and functional competence of recombinant native and sialic acid-deficient IgEs.

Engineering the N-glycosylation pathway of Nicotiana tabacum for molecular pharming using CRISPR/Cas9.

Molecular pharming in plants offers exciting possibilities to address global access to modern biologics. However, differences in the N-glycosylation pathway including the presence of ß(1,2)-xylose and core α(1,3)-fucose can affect activity, potency and immunogenicity of plant-derived proteins. Successful glycoengineering approaches toward human-like structures with no changes in plant phenotype, growth, or recombinant protein expression levels have been reported for Arabidopsis thaliana and Nicotiana benthamiana. Such engineering of N-glycosylation would also be desirable for Nicotiana tabacum, which remains the crop of choice for recombinant protein pharmaceuticals required at massive scale and for manufacturing technology transfer to less developed countries. Here, we generated N. tabacum cv. SR-1 ß(1,2)-xylosyltransferase (XylT) and α(1,3)-fucosyltransferase (FucT) knockout lines using CRISPR/Cas9 multiplex genome editing, targeting three conserved regions of the four FucT and two XylT genes. These two enzymes are responsible for generating non-human N-glycan structures. We confirmed full functional knockout of transformants by immunoblotting of total soluble protein by antibodies recognizing ß(1,2)-xylose and core α(1,3)-fucose, mass spectrometry analysis of recombinantly produced VRC01, a broadly neutralizing anti-HIV-1 hIgG1 antibody, and Sanger sequencing of targeted regions of the putative transformants. These data represent an important step toward establishing Nicotiana tabacum as a biologics platform for Global Health.

Using Jupyter Notebooks for re-training machine learning models.

Machine learning (ML) models require an extensive, user-driven selection of molecular descriptors in order to learn from chemical structures to predict actives and inactives with a high reliability. In addition, privacy concerns often restrict the access to sufficient data, leading to models with a narrow chemical space. Therefore, we propose a framework of re-trainable models that can be transferred from one local instance to another, and further allow a less extensive descriptor selection. The models are shared via a Jupyter Notebook, allowing the evaluation and implementation of a broader chemical space by keeping most of the tunable parameters pre-defined. This enables the models to be updated in a decentralized, facile, and fast manner. Herein, the method was evaluated with six transporter datasets (BCRP, BSEP, OATP1B1, OATP1B3, MRP3, P-gp), which revealed the general applicability of this approach.

Clinical and Translational Significance of Basophils in Patients with Cancer.

Despite comprising a very small proportion of circulating blood leukocytes, basophils are potent immune effector cells. The high-affinity receptor for IgE (FcɛRI) is expressed on the basophil cell surface and powerful inflammatory mediators such as histamine, granzyme B, and cytokines are stored in dense cytoplasmic granules, ready to be secreted in response to a range of immune stimuli. Basophils play key roles in eliciting potent effector functions in allergic diseases and type 1 hypersensitivity. Beyond allergies, basophils can be recruited to tissues in chronic and autoimmune inflammation, and in response to parasitic, bacterial, and viral infections. While their activation states and functions can be influenced by Th2-biased inflammatory signals, which are also known features of several tumor types, basophils have received little attention in cancer. Here, we discuss the presence and functional significance of basophils in the circulation of cancer patients and in the tumor microenvironment (TME). Interrogating publicly available datasets, we conduct gene expression analyses to explore basophil signatures and associations with clinical outcomes in several cancers. Furthermore, we assess how basophils can be harnessed to predict hypersensitivity to cancer treatments and to monitor the desensitization of patients to oncology drugs, using assays such as the basophil activation test (BAT).

The hepatocyte export carrier inhibition assay improves the separation of hepatotoxic from non-hepatotoxic compounds.

An in vitro/in silico method that determines the risk of human drug induced liver injury in relation to oral doses and blood concentrations of drugs was recently introduced. This method utilizes information on the maximal blood concentration (Cmax) for a specific dose of a test compound, which can be estimated using physiologically-based pharmacokinetic modelling, and a cytotoxicity test in cultured human hepatocytes. In the present study, we analyzed if the addition of an assay that measures the inhibition of bile acid export carriers, like BSEP and/or MRP2, to the existing method improves the differentiation of hepatotoxic and non-hepatotoxic compounds. Therefore, an export assay for 5-chloromethylfluorescein diacetate (CMFDA) was established. We tested 36 compounds in a concentration-dependent manner for which the risk of hepatotoxicity for specific oral doses and the capacity to inhibit hepatocyte export carriers are known. Compared to the CTB cytotoxicity test, substantially lower EC10 values were obtained using the CMFDA assay for several known BSEP and/or MRP2 inhibitors. To quantify if the addition of the CMFDA assay to our test system improves the overall separation of hepatotoxic from non-hepatotoxic compounds, the toxicity separation index (TSI) was calculated. We obtained a better TSI using the lower alert concentration from either the CMFDA or the CTB test (TSI: 0.886) compared to considering the CTB test alone (TSI: 0.775). In conclusion, the data show that integration of the CMFDA assay with an in vitro test battery improves the differentiation of hepatotoxic and non-hepatotoxic compounds in a set of compounds that includes bile acid export carrier inhibitors.

Insights from IgE Immune Surveillance in Allergy and Cancer for Anti-Tumour IgE Treatments.

IgE, the predominant antibody class of the allergic response, is known for its roles in protecting against parasites; however, a growing body of evidence indicates a significant role for IgE and its associated effector cells in tumour immunosurveillance, highlighted by the field of AllergoOncology and the successes of the first-in-class IgE cancer therapeutic MOv18. Supporting this concept, substantial epidemiological data ascribe potential roles for IgE, allergy, and atopy in protecting against specific tumour types, with a corresponding increased cancer risk associated with IgE immunodeficiency. Here, we consider how epidemiological data in combination with functional data reveals a complex interplay of IgE and allergy with cancer, which cannot be explained solely by one of the existing conventional hypotheses. We furthermore discuss how, in turn, such data may be used to inform future therapeutic approaches, including the clinical management of different patient groups. With epidemiological findings highlighting several high-risk cancer types protected against by high IgE levels, it is possible that use of IgE-based therapeutics for a range of malignant indications may offer efficacy to complement that of established IgG-class antibodies.

Characterisation of a highly potent and near pan-neutralising anti-HIV monoclonal antibody expressed in tobacco plants.

BACKGROUND: HIV remains one of the most important health issues worldwide, with almost 40 million people living with HIV. Although patients develop antibodies against the virus, its high mutation rate allows evasion of immune responses. Some patients, however, produce antibodies that are able to bind to, and neutralise different strains of HIV. One such 'broadly neutralising' antibody is 'N6'. Identified in 2016, N6 can neutralise 98% of HIV-1 isolates with a median IC50 of 0.066 µg/mL. This neutralisation breadth makes N6 a very promising therapeutic candidate. RESULTS: N6 was expressed in a glycoengineered line of N. benthamiana plants (pN6) and compared to the mammalian cell-expressed equivalent (mN6). Expression at 49 mg/kg (fresh leaf tissue) was achieved in plants, although extraction and purification are more challenging than for most plant-expressed antibodies. N-glycoanalysis demonstrated the absence of xylosylation and a reduction in α(1,3)-fucosylation that are typically found in plant glycoproteins. The N6 light chain contains a potential N-glycosylation site, which was modified and displayed more α(1,3)-fucose than the heavy chain. The binding kinetics of pN6 and mN6, measured by surface plasmon resonance, were similar for HIV gp120. pN6 had a tenfold higher affinity for FcγRIIIa, which was reflected in an antibody-dependent cellular cytotoxicity assay, where pN6 induced a more potent response from effector cells than that of mN6. pN6 demonstrated the same potency and breadth of neutralisation as mN6, against a panel of HIV strains. CONCLUSIONS: The successful expression of N6 in tobacco supports the prospect of developing a low-cost, low-tech production platform for a monoclonal antibody cocktail to control HIV in low-to middle income countries.

Immunotherapy using IgE or CAR T cells for cancers expressing the tumor antigen SLC3A2.

BACKGROUND: Cancer immunotherapy with monoclonal antibodies and chimeric antigen receptor (CAR) T cell therapies can benefit from selection of new targets with high levels of tumor specificity and from early assessments of efficacy and safety to derisk potential therapies. METHODS: Employing mass spectrometry, bioinformatics, immuno-mass spectrometry and CRISPR/Cas9 we identified the target of the tumor-specific SF-25 antibody. We engineered IgE and CAR T cell immunotherapies derived from the SF-25 clone and evaluated potential for cancer therapy. RESULTS: We identified the target of the SF-25 clone as the tumor-associated antigen SLC3A2, a cell surface protein with key roles in cancer metabolism. We generated IgE monoclonal antibody, and CAR T cell immunotherapies each recognizing SLC3A2. In concordance with preclinical and, more recently, clinical findings with the first-in-class IgE antibody MOv18 (recognizing the tumor-associated antigen Folate Receptor alpha), SF-25 IgE potentiated Fc-mediated effector functions against cancer cells in vitro and restricted human tumor xenograft growth in mice engrafted with human effector cells. The antibody did not trigger basophil activation in cancer patient blood ex vivo, suggesting failure to induce type I hypersensitivity, and supporting safe therapeutic administration. SLC3A2-specific CAR T cells demonstrated cytotoxicity against tumor cells, stimulated interferon-γ and interleukin-2 production in vitro. In vivo SLC3A2-specific CAR T cells significantly increased overall survival and reduced growth of subcutaneous PC3-LN3-luciferase xenografts. No weight loss, manifestations of cytokine release syndrome or graft-versus-host disease, were detected. CONCLUSIONS: These findings identify efficacious and potentially safe tumor-targeting of SLC3A2 with novel immune-activating antibody and genetically modified cell therapies.

Engineering the interactions between a plant-produced HIV antibody and human Fc receptors.

Plants can provide a cost-effective and scalable technology for production of therapeutic monoclonal antibodies, with the potential for precise engineering of glycosylation. Glycan structures in the antibody Fc region influence binding properties to Fc receptors, which opens opportunities for modulation of antibody effector functions. To test the impact of glycosylation in detail, on binding to human Fc receptors, different glycovariants of VRC01, a broadly neutralizing HIV monoclonal antibody, were generated in Nicotiana benthamiana and characterized. These include glycovariants lacking plant characteristic α1,3-fucose and ß1,2-xylose residues and glycans extended with terminal ß1,4-galactose. Surface plasmon resonance-based assays were established for kinetic/affinity evaluation of antibody-FcγR interactions, and revealed that antibodies with typical plant glycosylation have a limited capacity to engage FcγRI, FcγRIIa, FcγRIIb and FcγRIIIa; however, the binding characteristics can be restored and even improved with targeted glycoengineering. All plant-made glycovariants had a slightly reduced affinity to the neonatal Fc receptor (FcRn) compared with HEK cell-derived antibody. However, this was independent of plant glycosylation, but related to the oxidation status of two methionine residues in the Fc region. This points towards a need for process optimization to control oxidation levels and improve the quality of plant-produced antibodies.

Vienna LiverTox Workspace-A Set of Machine Learning Models for Prediction of Interactions Profiles of Small Molecules With Transporters Relevant for Regulatory Agencies.

Transporters expressed in the liver play a major role in drug pharmacokinetics and are a key component of the physiological bile flow. Inhibition of these transporters may lead to drug-drug interactions or even drug-induced liver injury. Therefore, predicting the interaction profile of small molecules with transporters expressed in the liver may help medicinal chemists and toxicologists to prioritize compounds in an early phase of the drug development process. Based on a comprehensive analysis of the data available in the public domain, we developed a set of classification models which allow to predict-for a small molecule-the inhibition of and transport by a set of liver transporters considered to be relevant by FDA, EMA, and the Japanese regulatory agency. The models were validated by cross-validation and external test sets and comprise cross validated balanced accuracies in the range of 0.64-0.88. Finally, models were implemented as an easy to use web-service which is freely available at https://livertox.univie.ac.at.

Interspecies comparison of putative ligand binding sites of human, rat and mouse P-glycoprotein.

Prior to the clinical phases of testing, safety, efficacy and pharmacokinetic profiles of lead compounds are evaluated in animal studies. These tests are primarily performed in rodents, such as mouse and rats. In order to reduce the number of animal experiments, computational models that predict the outcome of these studies and thus aid in prioritization of preclinical candidates are heavily needed. However, although computational models for human off-target interactions with decent quality are available, they cannot easily be transferred to rodents due to lack of respective data. In this study, we assess the transferability of human P-glycoprotein activity data for development of in silico models to predict in vivo effects in rats and mouse using a structure-based approach. P-glycoprotein (P-gp) is an ATP-dependent efflux transporter that transports xenobiotic compounds such as toxins and drugs out of cells and has a broad substrate and inhibitor specificity. Being mostly expressed at barriers, it influences the bioavailability of drugs and thus contributes also to toxicity. Comparison of the binding site interaction profiles of human, rat and mouse P-gp derived from docking studies with a set of common inhibitors suggests that the inhibitors share potentially similar binding modes. These findings encourage the use of in vitro human P-gp data for predicting in vivo effects in rodents and thus contributes to the 3Rs (Replace, Reduce and Refine) of animal experiments.

Ligand Desolvation Steers On-Rate and Impacts Drug Residence Time of Heat Shock Protein 90 (Hsp90) Inhibitors.

Residence time and more recently the association rate constant kon are increasingly acknowledged as important parameters for in vivo efficacy and safety of drugs. However, their broader consideration in drug development is limited by a lack of knowledge of how to optimize these parameters. In this study on a set of 176 heat shock protein 90 inhibitors, structure-kinetic relationships, X-ray crystallography, and molecular dynamics simulations were combined to retrieve a concrete scheme of how to rationally slow down on-rates. We discovered that an increased ligand desolvation barrier by introducing polar substituents resulted in a significant kon decrease. The slowdown was accomplished by introducing polar moieties to those parts of the ligand that point toward a hydrophobic cavity. We validated this scheme by increasing polarity of three Hsp90 inhibitors and observed a 9-, 13-, and 45-fold slowdown of on-rates and a 9-fold prolongation in residence time. This prolongation was driven by transition state destabilization rather than ground state stabilization.