Ian Morilla

Systematic computational prediction of protein interaction networks.

Determining the network of physical protein associations is an important first step in developing mechanistic evidence for elucidating biological pathways. Despite rapid advances in the field of high throughput experiments to determine protein interactions, the majority of associations remain unknown. Here we describe computational methods for significantly expanding protein association networks. We describe methods for integrating multiple independent sources of evidence to obtain higher quality predictions and we compare the major publicly available resources available for experimentalists to use.

Integration of biological data by kernels on graph nodes allows prediction of new genes involved in mitotic chromosome condensation.

A gene function prediction method suitable for the design of targeted RNAi libraries is described and used to predict chromosome condensation genes. Systematic experimental validation of candidate genes in a focused RNAi screen by automated microscopy and quantitative image analysis reveals many new chromosome condensation factors.

Finding the “Dark Matter” in Human and Yeast Protein Network Prediction and Modelling

Accurate modelling of biological systems requires a deeper and more complete knowledge about the molecular components and their functional associations than we currently have. Traditionally, new knowledge on protein associations generated by experiments has played a central role in systems modelling, in contrast to generally less trusted bio-computational predictions. However, we will not achieve realistic modelling of complex molecular systems if the current experimental designs lead to biased screenings of real protein networks and leave large, functionally important areas poorly characterised. To assess the likelihood of this, we have built comprehensive network models of the yeast and human proteomes by using a meta-statistical integration of diverse computationally predicted protein association datasets. We have compared these predicted networks against combined experimental datasets from seven biological resources at different level of statistical significance. These eukaryotic predicted networks resemble all the topological and noise features of the experimentally inferred networks in both species, and we also show that this observation is not due to random behaviour. In addition, the topology of the predicted networks contains information on true protein associations, beyond the constitutive first order binary predictions. We also observe that most of the reliable predicted protein associations are experimentally uncharacterised in our models, constituting the hidden or “dark matter” of networks by analogy to astronomical systems. Some of this dark matter shows enrichment of particular functions and contains key functional elements of protein networks, such as hubs associated with important functional areas like the regulation of Ras protein signal transduction in human cells. Thus, characterising this large and functionally important dark matter, elusive to established experimental designs, may be crucial for modelling biological systems. In any case, these predictions provide a valuable guide to these experimentally elusive regions.

Uncovering the Molecular Machinery of the Human Spindle-An Integration of Wet and Dry Systems Biology

The mitotic spindle is an essential molecular machine involved in cell division, whose composition has been studied extensively by detailed cellular biology, high-throughput proteomics, and RNA interference experiments. However, because of its dynamic organization and complex regulation it is difficult to obtain a complete description of its molecular composition. We have implemented an integrated computational approach to characterize novel human spindle components and have analysed in detail the individual candidates predicted to be spindle proteins, as well as the network of predicted relations connecting known and putative spindle proteins. The subsequent experimental validation of a number of predicted novel proteins confirmed not only their association with the spindle apparatus but also their role in mitosis. We found that 75% of our tested proteins are localizing to the spindle apparatus compared to a success rate of 35% when expert knowledge alone was used. We compare our results to the previously published MitoCheck study and see that our approach does validate some findings by this consortium. Further, we predict so-called “hidden spindle hub”, proteins whose network of interactions is still poorly characterised by experimental means and which are thought to influence the functionality of the mitotic spindle on a large scale. Our analyses suggest that we are still far from knowing the complete repertoire of functionally important components of the human spindle network. Combining integrated bio-computational approaches and single gene experimental follow-ups could be key to exploring the still hidden regions of the human spindle system.

Protein–Protein Interfaces from Cytochrome c Oxidase I Evolve Faster than Nonbinding Surfaces, yet Negative Selection Is the Driving Force

Respiratory complexes are encoded by two genomes (mitochondrial DNA [mtDNA] and nuclear DNA [nDNA]). Although the importance of intergenomic coadaptation is acknowledged, the forces and constraints shaping such coevolution are largely unknown. Previous works using cytochrome c oxidase (COX) as a model enzyme have led to the so-called “optimizing interaction” hypothesis. According to this view, mtDNA-encoded residues close to nDNA-encoded residues evolve faster than the rest of positions, favoring the optimization of protein–protein interfaces. Herein, using evolutionary data in combination with structural information of COX, we show that failing to discern the effects of interaction from other structural and functional effects can lead to deceptive conclusions such as the “optimizing hypothesis.” Once spurious factors have been accounted for, data analysis shows that mtDNA-encoded residues engaged in contacts are, in general, more constrained than their noncontact counterparts. Nevertheless, noncontact residues from the surface of COX I subunit are a remarkable exception, being subjected to an exceptionally high purifying selection that may be related to the maintenance of a suitable heme environment. We also report that mtDNA-encoded residues involved in contacts with other mtDNA-encoded subunits are more constrained than mtDNA-encoded residues interacting with nDNA-encoded polypeptides. This differential behavior cannot be explained on the basis of predicted thermodynamic stability, as interactions between mtDNA-encoded subunits contribute more weakly to the complex stability than those interactions between subunits encoded by different genomes. Therefore, the higher conservation observed among mtDNA-encoded residues involved in intragenome interactions is likely due to factors other than structural stability.

FUN-L: gene prioritization for RNAi screens

MOTIVATION Most biological processes remain only partially characterized with many components still to be identified. Given that a whole genome can usually not be tested in a functional assay, identifying the genes most likely to be of interest is of critical importance to avoid wasting resources. RESULTS Given a set of known functionally related genes and using a state-of-the-art approach to data integration and mining, our Functional Lists (FUN-L) method provides a ranked list of candidate genes for testing. Validation of predictions from FUN-L with independent RNAi screens confirms that FUN-L-produced lists are enriched in genes with the expected phenotypes. In this article, we describe a website front end to FUN-L. AVAILABILITY AND IMPLEMENTATION The website is freely available to use at http://funl.org

Assessment of protein domain fusions in human protein interaction networks prediction Application to the human kinetochore model

In order to understand how biological systems function it is necessary to determine the interactions and associations between proteins. Some proteins, involved in a common biological process and encoded by separate genes in one organism, can be found fused within a single protein chain in other organisms. By detecting these triplets, a functional relationship can be established between the unfused proteins. Here we use a domain fusion prediction method to predict these protein interactions for the human interactome. We observed that gene fusion events are more related to physical interaction between proteins than to other weaker functional relationships such as participation in a common biological pathway. These results suggest that domain fusion is an appropriate method for predicting protein complexes. The most reliable fused domain predictions were used to build protein-protein interaction (PPI) networks. These predicted PPI network models showed the same topological features as real biological networks and different features from random behaviour. We built the PPI domain fusion sub-network model of the human kinetochore and observed that the majority of the predicted interactions have not yet been experimentally characterised in the publicly available PPI repositories. The study of the human kinetochore domain fusion sub-network reveals undiscovered kinetochore proteins with presumably relevant functions, such as hubs with many connections in the kinetochore sub-network. These results suggest that experimentally hidden regions in the predicted PPI networks contain key functional elements, associated with important functional areas, still undiscovered in the human interactome. Until novel experiments shed light on these hidden regions; domain fusion predictions provide a valuable approach for exploring them.

Exploring the interactions of the RAS family in the human protein network and their potential implications in RAS-directed therapies

RAS proteins are the founding members of the RAS superfamily of GTPases. They are involved in key signaling pathways regulating essential cellular functions such as cell growth and differentiation. As a result, their deregulation by inactivating mutations often results in aberrant cell proliferation and cancer. With the exception of the relatively well-known KRAS, HRAS and NRAS proteins, little is known about how the interactions of the other RAS human paralogs affect cancer evolution and response to treatment. In this study we performed a comprehensive analysis of the relationship between the phylogeny of RAS proteins and their location in the protein interaction network. This analysis was integrated with the structural analysis of conserved positions in available 3D structures of RAS complexes. Our results show that many RAS proteins with divergent sequences are found close together in the human interactome. We found specific conserved amino acid positions in this group that map to the binding sites of RAS with many of their signaling effectors, suggesting that these pairs could share interacting partners. These results underscore the potential relevance of cross-talking in the RAS signaling network, which should be taken into account when considering the inhibitory activity of drugs targeting specific RAS oncoproteins. This study broadens our understanding of the human RAS signaling network and stresses the importance of considering its potential cross-talk in future therapies.

Novel angiogenic functional targets predicted through dark matter assessment in protein networks

In order to model protein networks we must extend our knowledge of the protein associations occurring in molecular systems and their functional relationships. We have significantly increased the accuracy of protein association predictions by the meta-statistical integration of three computational methods specifically designed for eukaryotic proteomes. From this former work it was discovered that high-throughput experimental assays seem to perform biased screenings of the real protein networks and leave important areas poorly characterized. This finding supports the convenience to combine computational prediction approaches to model protein interaction networks. We address in this work the challenge of integrating context information, present in predicted and known protein network models, to functionally characterize novel proteins. We applied a random walk-with-restart kernel to our models aiming at fixing some poorly described or unknown proteins involve in angiogenesis. This approach reveals some novel key angiogenic components within the human interactome.

In silico prediction of targets for anti-angiogenesis and their in vitro evaluation confirm the involvement of SOD3 in angiogenesis

Biocomputational network approaches are being successfully applied to predict and extract previously unknown information of novel molecular components of biological systems. In the present work, we have used this approach to predict new potential targets of anti-angiogenic therapies. For experimental validation of predictions, we made use of two in vitro assays related to two key steps of the angiogenic process, namely, endothelial cell migration and formation of “tubular-like” structures on Matrigel. From 7 predicted candidates, experimental tests clearly show that superoxide dismutase 3 silencing or blocking with specific antibodies inhibit both key steps of angiogenesis. This experimental validation was further confirmed with additional in vitro assays showing that superoxide dismutase 3 blocking produces inhibitory effects on the capacity of endothelial cells to form “tubular-like” structure within type I collagen matrix, to adhere to elastin-coated plates and to invade a Matrigel layer. Furthermore, angiogenesis was also inhibited in the en vivo aortic ring assay and in the in vivo mouse Matrigel plug assay. Therefore, superoxide dismutase 3 is confirmed as a putative target for anti-angiogenic therapy.