Agnieszka Szwajda

Toward more realistic drug–target interaction predictions

A number of supervised machine learning models have recently been introduced for the prediction of drug–target interactions based on chemical structure and genomic sequence information. Although these models could offer improved means for many network pharmacology applications, such as repositioning of drugs for new therapeutic uses, the prediction models are often being constructed and evaluated under overly simplified settings that do not reflect the real-life problem in practical applications. Using quantitative drug–target bioactivity assays for kinase inhibitors, as well as a popular benchmarking data set of binary drug–target interactions for enzyme, ion channel, nuclear receptor and G protein-coupled receptor targets, we illustrate here the effects of four factors that may lead to dramatic differences in the prediction results: (i) problem formulation (standard binary classification or more realistic regression formulation), (ii) evaluation data set (drug and target families in the application use case), (iii) evaluation procedure (simple or nested cross-validation) and (iv) experimental setting (whether training and test sets share common drugs and targets, only drugs or targets or neither). Each of these factors should be taken into consideration to avoid reporting overoptimistic drug–target interaction prediction results. We also suggest guidelines on how to make the supervised drug–target interaction prediction studies more realistic in terms of such model formulations and evaluation setups that better address the inherent complexity of the prediction task in the practical applications, as well as novel benchmarking data sets that capture the continuous nature of the drug–target interactions for kinase inhibitors.

Quantitative scoring of differential drug sensitivity for individually optimized anticancer therapies

We developed a systematic algorithmic solution for quantitative drug sensitivity scoring (DSS), based on continuous modeling and integration of multiple dose-response relationships in high-throughput compound testing studies. Mathematical model estimation and continuous interpolation makes the scoring approach robust against sources of technical variability and widely applicable to various experimental settings, both in cancer cell line models and primary patient-derived cells. Here, we demonstrate its improved performance over other response parameters especially in a leukemia patient case study, where differential DSS between patient and control cells enabled identification of both cancer-selective drugs and drug-sensitive patient sub-groups, as well as dynamic monitoring of the response patterns and oncogenic driver signals during cancer progression and relapse in individual patient cells ex vivo. An open-source and easily extendable implementation of the DSS calculation is made freely available to support its tailored application to translating drug sensitivity testing results into clinically actionable treatment options.

Target Inhibition Networks: Predicting Selective Combinations of Druggable Targets to Block Cancer Survival Pathways

A recent trend in drug development is to identify drug combinations or multi-target agents that effectively modify multiple nodes of disease-associated networks. Such polypharmacological effects may reduce the risk of emerging drug resistance by means of attacking the disease networks through synergistic and synthetic lethal interactions. However, due to the exponentially increasing number of potential drug and target combinations, systematic approaches are needed for prioritizing the most potent multi-target alternatives on a global network level. We took a functional systems pharmacology approach toward the identification of selective target combinations for specific cancer cells by combining large-scale screening data on drug treatment efficacies and drug-target binding affinities. Our model-based prediction approach, named TIMMA, takes advantage of the polypharmacological effects of drugs and infers combinatorial drug efficacies through system-level target inhibition networks. Case studies in MCF-7 and MDA-MB-231 breast cancer and BxPC-3 pancreatic cancer cells demonstrated how the target inhibition modeling allows systematic exploration of functional interactions between drugs and their targets to maximally inhibit multiple survival pathways in a given cancer type. The TIMMA prediction results were experimentally validated by means of systematic siRNA-mediated silencing of the selected targets and their pairwise combinations, showing increased ability to identify not only such druggable kinase targets that are essential for cancer survival either individually or in combination, but also synergistic interactions indicative of non-additive drug efficacies. These system-level analyses were enabled by a novel model construction method utilizing maximization and minimization rules, as well as a model selection algorithm based on sequential forward floating search. Compared with an existing computational solution, TIMMA showed both enhanced prediction accuracies in cross validation as well as significant reduction in computation times. Such cost-effective computational-experimental design strategies have the potential to greatly speed-up the drug testing efforts by prioritizing those interventions and interactions warranting further study in individual cancer cases.

Chk1 targeting reactivates PP2A tumor suppressor activity in cancer cells

Checkpoint kinase Chk1 is constitutively active in many cancer cell types and new generation Chk1 inhibitors show marked antitumor activity as single agents. Here we present a hitherto unrecognized mechanism that contributes to the response of cancer cells to Chk1-targeted therapy. Inhibiting chronic Chk1 activity in cancer cells induced the tumor suppressor activity of protein phosphatase protein phosphatase 2A (PP2A), which by dephosphorylating MYC serine 62, inhibited MYC activity and impaired cancer cell survival. Mechanistic investigations revealed that Chk1 inhibition activated PP2A by decreasing the transcription of cancerous inhibitor of PP2A (CIP2A), a chief inhibitor of PP2A activity. Inhibition of cancer cell clonogenicity by Chk1 inhibition could be rescued in vitro either by exogenous expression of CIP2A or by blocking the CIP2A-regulated PP2A complex. Chk1-mediated CIP2A regulation was extended in tumor models dependent on either Chk1 or CIP2A. The clinical relevance of CIP2A as a Chk1 effector protein was validated in several human cancer types, including neuroblastoma, where CIP2A was identified as an NMYC-independent prognostic factor. Because the Chk1-CIP2A-PP2A pathway is driven by DNA-PK activity, functioning regardless of p53 or ATM/ATR status, our results offer explanative power for understanding how Chk1 inhibitors mediate single-agent anticancer efficacy. Furthermore, they define CIP2A-PP2A status in cancer cells as a pharmacodynamic marker for their response to Chk1-targeted therapy.

Drafting the CLN3 protein interactome in SH-SY5Y human neuroblastoma cells: a label-free quantitative proteomics approach.

Neuronal ceroid lipofuscinoses (NCL) are the most common inherited progressive encephalopathies of childhood. One of the most prevalent forms of NCL, Juvenile neuronal ceroid lipofuscinosis (JNCL) or CLN3 disease (OMIM: 204200), is caused by mutations in the CLN3 gene on chromosome 16p12.1. Despite progress in the NCL field, the primary function of ceroid-lipofuscinosis neuronal protein 3 (CLN3) remains elusive. In this study, we aimed to clarify the role of human CLN3 in the brain by identifying CLN3-associated proteins using a Tandem Affinity Purification coupled to Mass Spectrometry (TAP-MS) strategy combined with Significance Analysis of Interactome (SAINT). Human SH-SY5Y-NTAP-CLN3 stable cells were used to isolate native protein complexes for subsequent TAP-MS. Bioinformatic analyses of isolated complexes yielded 58 CLN3 interacting partners (IP) including 42 novel CLN3 IP, as well as 16 CLN3 high confidence interacting partners (HCIP) previously identified in another high-throughput study by Behrends et al., 2010. Moreover, 31 IP of ceroid-lipofuscinosis neuronal protein 5 (CLN5) were identified (18 of which were in common with the CLN3 bait). Our findings support previously suggested involvement of CLN3 in transmembrane transport, lipid homeostasis and neuronal excitability, as well as link it to G-protein signaling and protein folding/sorting in the ER.

Proteomic analysis of the palmitoyl protein thioesterase 1 interactome in SH-SY5Y human neuroblastoma cells

UNLABELLED Neuronal ceroid lipofuscinoses (NCL) are a group of inherited progressive childhood disorders, characterized by early accumulation of autofluorescent storage material in lysosomes of neurons or other cells. Clinical symptoms of NCL include: progressive loss of vision, mental and motor deterioration, epileptic seizures and premature death. CLN1 disease (MIM#256730) is caused by mutations in the CLN1 gene, which encodes palmitoyl protein thioesterase 1 (PPT1). In this study, we utilised single step affinity purification coupled to mass spectrometry (AP-MS) to unravel the in vivo substrates of human PPT1 in the brain neuronal cells. Protein complexes were isolated from human PPT1 expressing SH-SY5Y stable cells, subjected to filter-aided sample preparation (FASP) and analysed on a Q Exactive Hybrid Quadrupole-Orbitrap mass spectrometer. A total of 23 PPT1 interacting partners (IP) were identified from label free quantitation of the MS data by SAINT platform. Three of the identified PPT1 IP, namely CRMP1, DBH, and MAP1B are predicted to be palmitoylated. Our proteomic analysis confirmed previously suggested roles of PPT1 in axon guidance and lipid metabolism, yet implicates the enzyme in novel roles including: involvement in neuronal migration and dopamine receptor mediated signalling pathway. BIOLOGICAL SIGNIFICANCE The significance of this work lies in the unravelling of putative in vivo substrates of human CLN1 or PPT1 in brain neuronal cells. Moreover, the PPT1 IP implicate the enzyme in novel roles including: involvement in neuronal migration and dopamine receptor mediated signalling pathway.

Naturally Occurring Human Phosphorylcholine Antibodies Are Predominantly Products of Affinity-Matured B Cells in the Adult

Phosphorylcholine (PC) is a classic T-independent Ag that is exposed on apoptotic cells, oxidized phospholipids, and bacterial polysaccharides. Experimental as well as epidemiological studies have over the past decade implicated Abs against PC (anti-PC) as anti-inflammatory and a strong protective factor in cardiovascular disease. Although clinically important, little is known about the development of anti-PC in humans. This study was conceived to dissect the human anti-PC repertoire and generate human mAbs. We designed a PC-specific probe to identify, isolate, and characterize PC-reactive B cells from 10 healthy individuals. The donors had all mounted somatically mutated Abs toward PC using a broad variety of Ig genes. PC-reactive B cells were primarily found in the IgM+ memory subset, although significant numbers also were detected among naive, IgG+, and CD27+CD43+ B cells. Abs from these subsets were clonally related, suggesting a common origin. mAbs derived from the same donors exhibited equivalent or higher affinity for PC than the well-characterized murine T-15 clone. These results provide novel insights into the cellular and molecular ontogeny of atheroprotective PC Abs, thereby offering new opportunities for Ab-based therapeutic interventions.

Global Analysis of S-nitrosylation Sites in the Wild Type (APP) Transgenic Mouse Brain-Clues for Synaptic Pathology

Alzheimers disease (AD) is characterized by an early synaptic loss, which strongly correlates with the severity of dementia. The pathogenesis and causes of characteristic AD symptoms are not fully understood. Defects in various cellular cascades were suggested, including the imbalance in production of reactive oxygen and nitrogen species. Alterations in S-nitrosylation of several proteins were previously demonstrated in various AD animal models and patients. In this work, using combined biotin-switch affinity/nano-LC-MS/MS and bioinformatic approaches we profiled endogenous S-nitrosylation of brain synaptosomal proteins from wild type and transgenic mice overexpressing mutated human Amyloid Precursor Protein (hAPP). Our data suggest involvement of S-nitrosylation in the regulation of 138 synaptic proteins, including MAGUK, CamkII, or synaptotagmins. Thirty-eight proteins were differentially S-nitrosylated in hAPP mice only. Ninety-five S-nitrosylated peptides were identified for the first time (40% of total, including 33 peptides exclusively in hAPP synaptosomes). We verified differential S-nitrosylation of 10 (26% of all identified) synaptosomal proteins from hAPP mice, by Western blotting with specific antibodies. Functional enrichment analysis linked S-nitrosylated proteins to various cellular pathways, including: glycolysis, gluconeogenesis, calcium homeostasis, ion, and vesicle transport, suggesting a basic role of this post-translational modification in the regulation of synapses. The linkage of SNO-proteins to axonal guidance and other processes related to APP metabolism exclusively in the hAPP brain, implicates S-nitrosylation in the pathogenesis of Alzheimers disease.

Identification of selective cytotoxic and synthetic lethal drug responses in triple negative breast cancer cells

BackgroundTriple negative breast cancer (TNBC) is a highly heterogeneous and aggressive type of cancer that lacks effective targeted therapy. Despite detailed molecular profiling, no targeted therapy has been established. Hence, with the aim of gaining deeper understanding of the functional differences of TNBC subtypes and how that may relate to potential novel therapeutic strategies, we studied comprehensive anticancer-agent responses among a panel of TNBC cell lines.MethodThe responses of 301 approved and investigational oncology compounds were measured in 16 TNBC cell lines applying a functional profiling approach. To go beyond the standard drug viability effect profiling, which has been used in most chemosensitivity studies, we utilized a multiplexed readout for both cell viability and cytotoxicity, allowing us to differentiate between cytostatic and cytotoxic responses.ResultsOur approach revealed that most single-agent anti-cancer compounds that showed activity for the viability readout had no or little cytotoxic effects. Major compound classes that exhibited this type of response included anti-mitotics, mTOR, CDK, and metabolic inhibitors, as well as many agents selectively inhibiting oncogene-activated pathways. However, within the broad viability-acting classes of compounds, there were often subsets of cell lines that responded by cell death, suggesting that these cells are particularly vulnerable to the tested substance. In those cases we could identify differential levels of protein markers associated with cytotoxic responses. For example, PAI-1, MAPK phosphatase and Notch-3 levels associated with cytotoxic responses to mitotic and proteasome inhibitors, suggesting that these might serve as markers of response also in clinical settings. Furthermore, the cytotoxicity readout highlighted selective synergistic and synthetic lethal drug combinations that were missed by the cell viability readouts. For instance, the MEK inhibitor trametinib synergized with PARP inhibitors. Similarly, combination of two non-cytotoxic compounds, the rapamycin analog everolimus and an ATP-competitive mTOR inhibitor dactolisib, showed synthetic lethality in several mTOR-addicted cell lines.ConclusionsTaken together, by studying the combination of cytotoxic and cytostatic drug responses, we identified a deeper spectrum of cellular responses both to single agents and combinations that may be highly relevant for identifying precision medicine approaches in TNBC as well as in other types of cancers.

Quantitative analysis of PPT1 interactome in human neuroblastoma cells

Mutations in the CLN1 gene that encodes Palmitoyl protein thioesterase 1 (PPT1) or CLN1, cause Infantile NCL (INCL, MIM#256730). PPT1 removes long fatty acid chains such as palmitate from modified cysteine residues of proteins. The data shown here result from isolated protein complexes from PPT1-expressing SH-SY5Y stable cells that were subjected to single step affinity purification coupled to mass spectrometry (AP-MS). Prior to the MS analysis, we utilised a modified filter-aided sample preparation (FASP) protocol. Based on label free quantitative analysis of the data by SAINT, 23 PPT1 interacting partners (IP) were identified. A dense connectivity in PPT1 network was further revealed by functional coupling and extended network analyses, linking it to mitochondrial ATP synthesis coupled protein transport and thioester biosynthetic process. Moreover, the terms: inhibition of organismal death, movement disorders and concentration of lipid were predicted to be altered in the PPT1 network. Data presented here are related to Scifo et al. (J. Proteomics, 123 (2015) 42–53).