Martin Eberhardt

Model-based genotype-phenotype mapping used to investigate gene signatures of immune sensitivity and resistance in melanoma micrometastasis

In this paper, we combine kinetic modelling and patient gene expression data analysis to elucidate biological mechanisms by which melanoma becomes resistant to the immune system and to immunotherapy. To this end, we systematically perturbed the parameters in a kinetic model and performed a mathematical analysis of their impact, thereby obtaining signatures associated with the emergence of phenotypes of melanoma immune sensitivity and resistance. Our phenotypic signatures were compared with published clinical data on pretreatment tumor gene expression in patients subjected to immunotherapy against metastatic melanoma. To this end, the differentially expressed genes were annotated with standard gene ontology terms and aggregated into metagenes. Our method sheds light on putative mechanisms by which melanoma may develop immunoresistance. Precisely, our results and the clinical data point to the existence of a signature of intermediate expression levels for genes related to antigen presentation that constitutes an intriguing resistance mechanism, whereby micrometastases are able to minimize the combined anti-tumor activity of complementary responses mediated by cytotoxic T cells and natural killer cells, respectively. Finally, we computationally explored the efficacy of cytokines used as low-dose co-adjuvants for the therapeutic anticancer vaccine to overcome tumor immunoresistance.

An Interactive Macrophage Signal Transduction Map Facilitates Comparative Analyses of High-Throughput Data

Macrophages (Mϕs) are key players in the coordination of the lifesaving or detrimental immune response against infections. The mechanistic understanding of the functional modulation of Mϕs by pathogens and pharmaceutical interventions at the signal transduction level is still far from complete. The complexity of pathways and their cross-talk benefits from holistic computational approaches. In the present study, we reconstructed a comprehensive, validated, and annotated map of signal transduction pathways in inflammatory Mϕs based on the current literature. In a second step, we selectively expanded this curated map with database knowledge. We provide both versions to the scientific community via a Web platform that is designed to facilitate exploration and analysis of high-throughput data. The platform comes preloaded with logarithmic fold changes from 44 data sets on Mϕ stimulation. We exploited three of these data sets—human primary Mϕs infected with the common lung pathogens Streptococcus pneumoniae, Legionella pneumophila, or Mycobacterium tuberculosis—in a case study to show how our map can be customized with expression data to pinpoint regulated subnetworks and druggable molecules. From the three infection scenarios, we extracted a regulatory core of 41 factors, including TNF, CCL5, CXCL10, IL-18, and IL-12 p40, and identified 140 drugs targeting 16 of them. Our approach promotes a comprehensive systems biology strategy for the exploitation of high-throughput data in the context of Mϕ signal transduction. In conclusion, we provide a set of tools to help scientists unravel details of Mϕ signaling. The interactive version of our Mϕ signal transduction map is accessible online at https://vcells.net/macrophage.

MicroRNAs Constitute a Negative Feedback Loop in Streptococcus pneumoniae–Induced Macrophage Activation

Streptococcus pneumoniae causes high mortality as a major pneumonia-inducing pathogen. In pneumonia, control of innate immunity is necessary to prevent organ damage. We assessed the role of microRNAs (miRNAs) as regulators in pneumococcal infection of human macrophages. Exposure of primary blood-derived human macrophages with pneumococci resulted in transcriptional changes in several gene clusters and a significant deregulation of 10 microRNAs. Computational network analysis retrieved miRNA-146a as one putatively important regulator of pneumococci-induced host cell activation. Its induction depended on bacterial structural integrity and was completely inhibited by blocking Toll-like receptor 2 (TLR-2) or depleting its mediator MyD88. Furthermore, induction of miRNA-146a release did not require the autocrine feedback of interleukin 1β and tumor necrosis factor α released from infected macrophages, and it repressed the TLR-2 downstream mediators IRAK-1 and TRAF-6, as well as the inflammatory factors cyclooxygenase 2 and interleukin 1β. In summary, pneumococci recognition induces a negative feedback loop, preventing excessive inflammation via miR-146a and potentially other miRNAs.

A web platform for the network analysis of high-throughput data in melanoma and its use to investigate mechanisms of resistance to anti-PD1 immunotherapy

Cellular phenotypes are established and controlled by complex and precisely orchestrated molecular networks. In cancer, mutations and dysregulations of multiple molecular factors perturb the regulation of these networks and lead to malignant transformation. High-throughput technologies are a valuable source of information to establish the complex molecular relationships behind the emergence of malignancy, but full exploitation of this massive amount of data requires bioinformatics tools that rely on network-based analyses. In this report we present the Virtual Melanoma Cell, an online tool developed to facilitate the mining and interpretation of high-throughput data on melanoma by biomedical researches. The platform is based on a comprehensive, manually generated and expert-validated regulatory map composed of signaling pathways important in malignant melanoma. The Virtual Melanoma Cell is a tool designed to accept, visualize and analyze user-generated datasets. It is available at: https://www.vcells.net/melanoma. To illustrate the utilization of the web platform and the regulatory map, we have analyzed a large publicly available dataset accounting for anti-PD1 immunotherapy treatment of malignant melanoma patients.

Genome-wide Chromatin Profiling of Legionella pneumophila-Infected Human Macrophages Reveals Activation of the Probacterial Host Factor TNFAIP2.

BACKGROUND Legionella pneumophila is a causative agent of severe pneumonia. Infection leads to a broad host cell response, as evident, for example, on the transcriptional level. Chromatin modifications, which control gene expression, play a central role in the transcriptional response to L. pneumophila METHODS We infected human-blood-derived macrophages (BDMs) with L. pneumophila and used chromatin immunoprecipitation followed by sequencing to screen for gene promoters with the activating histone 4 acetylation mark. RESULTS We found the promoter of tumor necrosis factor α-induced protein 2 (TNFAIP2) to be acetylated at histone H4. This factor has not been characterized in the pathology of L. pneumophila TNFAIP2 messenger RNA and protein were upregulated in response to L. pneumophila infection of human-BDMs and human alveolar epithelial (A549) cells. We showed that L. pneumophila-induced TNFAIP2 expression is dependent on the NF-κB transcription factor. Importantly, knock down of TNFAIP2 led to reduced intracellular replication of L. pneumophila Corby in A549 cells. CONCLUSIONS Taken together, genome-wide chromatin analysis of L. pneumophila-infected macrophages demonstrated induction of TNFAIP2, a NF-κB-dependent factor relevant for bacterial replication.

An information-theoretic classification of amino acids for the assessment of interfaces in protein–protein docking

Docking represents a versatile and powerful method to predict the geometry of protein–protein complexes. However, despite significant methodical advances, the identification of good docking solutions among a large number of false solutions still remains a difficult task. We have previously demonstrated that the formalism of mutual information (MI) from information theory can be adapted to protein docking, and we have now extended this approach to enhance its robustness and applicability. A large dataset consisting of 22,934 docking decoys derived from 203 different protein–protein complexes was used for an MI-based optimization of reduced amino acid alphabets representing the protein–protein interfaces. This optimization relied on a clustering analysis that allows one to estimate the mutual information of whole amino acid alphabets by considering all structural features simultaneously, rather than by treating them individually. This clustering approach is fast and can be applied in a similar fashion to the generation of reduced alphabets for other biological problems like fold recognition, sequence data mining, or secondary structure prediction. The reduced alphabets derived from the present work were converted into a scoring function for the evaluation of docking solutions, which is available for public use via the web service score-MI: http://score-MI.biochem.uni-erlangen.de

Bacterial Adherence and Dwelling Probability: Two Drivers of Early Alveolar Infection by Streptococcus pneumoniae Identified in Multi-Level Mathematical Modeling

Pneumococcal infection is the most frequent cause of pneumonia, and one of the most prevalent diseases worldwide. The population groups at high risk of death from bacterial pneumonia are infants, elderly and immunosuppressed people. These groups are more vulnerable because they have immature or impaired immune systems, the efficacy of their response to vaccines is lower, and antibiotic treatment often does not take place until the inflammatory response triggered is already overwhelming. The immune response to bacterial lung infections involves dynamic interactions between several types of cells whose activation is driven by intracellular molecular networks. A feasible approach to the integration of knowledge and data linking tissue, cellular and intracellular events and the construction of hypotheses in this area is the use of mathematical modeling. For this paper, we used a multi-level computational model to analyse the role of cellular and molecular interactions during the first 10 h after alveolar invasion of Streptococcus pneumoniae bacteria. By “multi-level” we mean that we simulated the interplay between different temporal and spatial scales in a single computational model. In this instance, we included the intracellular scale of processes driving lung epithelial cell activation together with the scale of cell-to-cell interactions at the alveolar tissue. In our analysis, we combined systematic model simulations with logistic regression analysis and decision trees to find genotypic-phenotypic signatures that explain differences in bacteria strain infectivity. According to our simulations, pneumococci benefit from a high dwelling probability and a high proliferation rate during the first stages of infection. In addition to this, the model predicts that during the very early phases of infection the bacterial capsule could be an impediment to the establishment of the alveolar infection because it impairs bacterial colonization.