Ο σκοπός της παρούσας διδακτορικής διατριβής είναι να αποκρυπτογραφήσει μηχανισμούς σε ανθρώπινες ασθένειες, και συγκεκριμένα στην Οστεοαρθρίτιδα, στην μετάφραση βιολογικών δεδομένων από τον αρουραίο στον άνθρωπο, στην Σκλήρυνση κατά πλάκας, στον καρκίνο του ήπατος, στην καρδιομυοπάθεια που προκαλείται από φάρμακα και στην μη αλκοολική λιπώδη νόσο του ήπατος (ΜΑΛΝΗ). Στα παρακάτω κεφάλαια περιγράφονται τα ευρήματα σχετικά με την κάθε ανθρώπινη πάθηση και οι αντίστοιχες εργασίες.

Τα τελευταία 4 χρόνια, ο Δημήτρης Μεσσήνης έχει συγγράψει 6 δημοσιεύσεις σε επιστημονικά περιοδικά και 14 περιλήψεις - ανακοινώσεις σε διεθνή συνέδρια. Έχει μία δημοσίευση ως πρώτος συγγραφέας και έχει επιτύχει δείκτη h-index 6. Μέσω των προσπαθειών για την αποκρυπτογράφηση των μηχανισμών διαφορετικών ανθρώπινων ασθενειών, ο υποψήφιος διδάκτωρ έχει αποκτήσει ευρεία κατανόηση στη Συστημική μοντελοποίηση και ανάλυση μεγάλων δεδομένων, ανέπτυξε τις δικές του πειραματικές και υπολογιστικές μεθόδους και συνέβαλε σε αρκετά σημαντικά ευρήματα που περιγράφονται παραπάνω. Το έργο του στην Σκλήρυνση κατά πλάκας τον έκανε να επιλεχθεί μεταξύ 6 από 110 εργασιών για να δώσει μια σύντομη ομιλία στο International Conference on Systems Biology of Human Disease στο Harvard το 2014. Το 2016 συνεργάστηκε με το U.S. Food and Drug Administration για την ανάπτυξη μιας νέας υπολογιστικής μεθόδου για την πρόβλεψη της καρδιομυοπάθειας που προκαλείται από φάρμακα με ακρίβεια 88%, έρευνα που δημοσιεύθηκε στο CPT:​ Pharmacometrics & Systems Pharmacology. Η εργασία του πάνω στη ΜΑΛΝΗ αποκάλυψε έναν πολυπαραγοντικό μηχανισμό σηματοδότησης και κέρδισε το 1ο βραβείο από 56 εργασίες στο συνέδριο της Εταιρείας Ήπατος-ΠαγκρέατοςΧοληφόρων το 2017.

The purpose of this PhD Thesis is to decipher mechanisms in human disease, more specifically Osteoarthritis, Human/Rat Translational Research, Multiple Sclerosis, Hepatocellular Carcinoma, Drug-Induced Cardiomyopathy and Non-Alcoholic Fatty Liver Disease (NAFLD). Main findings regarding each human condition and project are discussed below.

Through research in Osteoarthritis it was demonstrated that healthy chondrocytes can have a strong inflammatory -rather than protective- response to various stimuli. In this manner, the generation of an inflammatory environment in the joint is sustained, because cartilage is an avascular tissue, lacking important antiinflammatory components of peripheral blood and this can eventually lead to the degradation of the tissue. For the first time, the signaling events that lead to the up-regulation of proinflammatory signals upon stimulation of the TLR were identified, uncovering two major inflammatory pathways:​ DEFB1 signals via its receptor to RAC1, to the MAPKs and ultimately activates HSP27. Flagellin signals through TLR5 to MYD88 and then merges with the IL1 pathway signaling through IRAK, TIFA, TRAF6 and activates the IKB, MAPK14 and HSP27 signals. Stimulation of chondrocytes with inflammatory mediators IL1B and Flagellin also leads to over-activation of growth-related signals CREB and MAP2K1 and the release of pro-growth cytokines, all connected to facilitating chondrocyte hypertrophy and bone ossification. Strong similarity between meniscus and cartilage cytokine releases was observed upon stimulation of those tissues with various stimuli, supporting the hypothesis that significant crosstalk between these two knee compartments exists and antiinflammatory therapies should take into consideration both tissues. Metabolomic analysis of human osteoarthritic synovial fluid was performed and metabolites connected to Osteoarthritis were identified.

For the purpose of Human/Rat translational research, a multi-layer systems biology dataset was generated that was comprised of phosphoproteomics, transcriptomics and cytokine data derived from normal human and rat bronchial epithelial cells exposed in parallel to more than 50 different stimuli under identical conditions. Major signaling pathways were conserved between human and rat species with only isolated components diverging. An exception was the targets of transcription factors, which seemed more difficult to predict. Transcription factor CREB1, showed the best consensus for the edges upstream of it but the connection from RSK1 was present only in the human consensus network, which might be explained by the fact that human isoforms of RSK1 have functional redundancy (i.e. RSK2, RSK3, RSK4). In contrast, this is most likely not the case in rodents;​ Zeniou et al. reported that the mouse RSK1 and RSK3 genes may not be able to fully compensate for the lack of RSK2 function.

A huge dataset was created to study Multiple Sclerosis (MS). 250 donors were recruited (190 MS, 60 Healthy) and PBMCs were isolated from them, then treated with 20 stimuli including 4 MS drugs and the response of 17 phosphoproteins (5’ and 25’) and 22 secreted cytokines (24h) was measured. This approach allowed to characterize the signaling networks in a patient-specific manner and to predict new targets for combination therapy for MS. The combination of fingolimod with either a TAK1 inhibitor or EGCG was also validated in an animal model.

Analysis using 3 Hepatocellular Carcinoma (HCC) cell lines presented new mechanistic insights into the targeted anti‐inflammatory actions of three promising nutraceuticals, epigallocatechin gallate (EGCG), fisetin (FIS), and eriodictyol (ERI). EGCG was the most effective modulator of inflammatory cytokine secretion (followed by FIS and ERI) and HEP3B cells were the best responders. Despite previous extensive literature, this was the first study showing the outstanding capability of this compound to concurrently reduce a wide range of HCC‐secreted cytokines.

With an application on Drug-Induced Cardiomyopathy, it was shown that constructing specific signaling pathways can computationally capture a drug’s mode of action and increase cardiomyopathy prediction accuracy from 79% to 88%, compared to just using the transcriptomic data at hand. Using Elastic Net regularization, 33 protein/gene predictors were extracted, that best predict the toxicity classification of drug-induced cardiotoxicity. The microRNAs that reportedly regulate expression of the 6 top predictors are of diagnostic value for natural heart failure or doxorubicin-induced cardiomyopathy. Among them, miR193-3p and miR26b-5p reportedly regulated 4 and 3 predictors, respectively, therefore it might be worthy of clinical studies to determine whether those micro RNAs are useful in vivo biomarkers for drug-induced cardiomyopathy.

It was observed that Non-Alcoholic Fatty Liver Disease (NAFLD) has a multifactorial nature and there is no single treatment for all subtypes of NAFLD, highlighting the need for a systemic approach and personalised therapeutic interventions to better understand and treat NAFLD. This was the first time that a study aims to understand the multifactorial nature of NAFLD at the signaling level by studying 5 NAFLD induction models in primary human hepatocytes. The results confirmed a large body of literature findings for NAFLD signaling mechanisms. Furthermore, CHK2 and EPOR have emerged as potential NAFLD players that may be interesting to study further since they are important factors in liver regeneration.

Over the last 4 years, Dimitris Messinis has co-authored 6 peer-reviewed publications and 14 international conference abstracts. He has one first-author publication and has attained an h-index of 6. Through the efforts to decipher the mechanisms of different human conditions, the PhD candidate has gained a broad understanding in Systems Biology and Big Data Modeling, developed his own experimental and computational methods and contributed to several important findings described above. His work in Multiple Sclerosis got him invited among 6 out of 110 abstracts to give a lightning talk at the International Conference on Systems Biology of Human Disease at Harvard in 2014. In 2016, he worked with the U.S. Food and Drug Administration developing a novel computational method to predict drug-induced Cardiomyopathy with 88% accuracy published in CPT:​ Pharmacometrics & Systems Pharmacology. His Cue-signal-response analysis in NAFLD revealed a multifactorial signaling mechanism and won the 1st prize among 56 submissions at the Panhellenic conference of Hepato-Pancreato-Biliary Association in 2017.