Coevolution, modularity and human disease

SEaCorAl: Identifying and contrasting the regulation-correlation bias in RNA-Seq paired expression data of patient groups

The Cancer Genome Atlas database offers the possibility of analyzing genome-wide expression RNA-Seq cancer data using paired counts, that is, studies where expression data are collected in pairs of normal and cancer cells, by taking samples from the same individual. Correlation of gene expression profiles is the most common analysis to study co-expression groups, which is used to find biological interpretation of -omics big data. The aim of the paper is threefold: firstly we show for the first time, the presence of a "regulation-correlation bias" in RNA-Seq paired expression data, that is an artifactual link between the expression status (up- or down-regulation) of a gene pair and the sign of the corresponding correlation coefficient. Secondly, we provide a statistical model able to theoretically explain the reasons for the presence of such a bias. Thirdly, we present a bias-removal algorithm, called SEaCorAl, able to effectively reduce bias effects and improve the biological significance of correlation analysis. Validation of the SEaCorAl algorithm is performed by showing a significant increase in the ability to detect biologically meaningful associations of positive correlations and a significant increase of the modularity of the resulting unbiased correlation network.

Evolutionary rate heterogeneity between multi- and single-interface hubs across human housekeeping and tissue-specific protein interaction network: Insights from proteins' and its partners' properties

Integrating gene expression into protein-protein interaction network (PPIN) leads to the construction of tissue-specific (TS) and housekeeping (HK) sub-networks, with distinctive TS- and HK-hubs. All such hub proteins are divided into multi-interface (MI) hubs and single-interface (SI) hubs, where MI hubs evolve slower than SI hubs. Here we explored the evolutionary rate difference between MI and SI proteins within TS- and HK-PPIN and observed that this difference is present only in TS, but not in HK-class. Next, we explored whether proteins' own properties or its partners' properties are more influential in such evolutionary discrepancy. Statistical analyses revealed that this evolutionary rate correlates negatively with protein's own properties like expression level, miRNA count, conformational diversity and functional properties and with its partners' properties like protein disorder and tissue expression similarity. Moreover, partial correlation and regression analysis revealed that both proteins' and its partners' properties have independent effects on protein evolutionary rate.

Cancer core modules identification through genomic and transcriptomic changes correlation detection at network level

BACKGROUND

Identification of driver mutations among numerous genomic alternations remains a critical challenge to the elucidation of the underlying mechanisms of cancer. Because driver mutations by definition are associated with a greater number of cancer phenotypes compared to other mutations, we hypothesized that driver mutations could more easily be identified once the genotype-phenotype correlations are detected across tumor samples.

RESULTS

In this study, we describe a novel network analysis to identify the driver mutation through integrating both cancer genomes and transcriptomes. Our method successfully identified a significant genotype-phenotype change correlation in all six solid tumor types and revealed core modules that contain both significantly enriched somatic mutations and aberrant expression changes specific to tumor development. Moreover, we found that the majority of these core modules contained well known cancer driver mutations, and that their mutated genes tended to occur at hub genes with central regulatory roles. In these mutated genes, the majority were cancer-type specific and exhibited a closer relationship within the same cancer type rather than across cancer types. The remaining mutated genes that exist in multiple cancer types led to two cancer type clusters, one cluster consisted of three neural derived or related cancer types, and the other cluster consisted of two adenoma cancer types.

CONCLUSIONS

Our approach can successfully identify the candidate drivers from the core modules. Comprehensive network analysis on the core modules potentially provides critical insights into convergent cancer development in different organs.

Protein localization as a principal feature of the etiology and comorbidity of genetic diseases

Proteins localized within the same subcellular compartment tend to be functionally associated. This study shows that subcellular localization and network distance between disease-associated proteins provide complementary information explaining patterns of disease comorbidity.

A positive correlation was found between subcellular localization of disease-associated protein pairs and measures of comorbidity.

A higher comorbidity tendency was found for disease-associated protein pairs that are positioned within a shorter distance in the protein interaction network.

The integration of subcellular localization information with protein interaction network sheds light onto the potential molecular connections underlying comorbidity patterns and will help to understand the mechanisms of human disease.

It was shown that the emergence of phenotypically similar diseases are triggered as a result of molecular connections between disease-causing genes (Oti and Brunner, 2007; Zaghloul and Katsanis, 2010). From a genetics, perspective diseases are associated with certain genes (Goh et al, 2007; Feldman et al, 2008), whereas from a proteomics perspective phenotypically similar diseases are connected via biological modules such as protein–protein interactions (PPIs) or molecular pathways (Lage et al, 2007; Jiang et al, 2008; Wu et al, 2008; Linghu et al, 2009; Suthram et al, 2010). These molecular connections between diseases were observed on the population level as well: diseases connected through molecular connections such as shared genes, PPIs, and metabolic pathways tend to show elevated comorbidity (Rzhetsky et al, 2007; Lee et al, 2008; Zhernakova et al, 2009; Park et al, 2009a, 2009b). While these findings constitute a step toward improving our understanding of the mechanism of disease progression, there are still many more molecule-level connections between disease pairs that need to be explored in order to establish a firmer comorbidity association.

Subcellular localization provides spatial information of proteins in the cell; proteins target subcellular localizations to interact with appropriate partners and form functional complexes in signaling pathways and metabolic processes (Au et al, 2007). Abnormal protein localizations are known to lead to the loss of functional effects in diseases (Luheshi et al, 2008; Laurila and Vihinen, 2009). For example, mis-localizations of nuclear/cytoplasmic transport have been detected in many types of carcinoma cells (Kau et al, 2004). A proper identification of protein subcellular localization can hence be useful in discovering disease-associated proteins (Giallourakis et al, 2005; Calvo and Mootha, 2010). With this understanding, we postulate that disease-associated proteins connected by subcellular localizations could also explain the phenotypic similarities between diseases. Furthermore, such connections may also couple to disease progressions that contribute to multiple disease manifestation, that is, comorbidity.

Protein subcellular localization has been extensively studied through various methods to determine a variety of protein functions. To the best of our knowledge, the connection between diseases and subcellular localizations are yet to be studied systematically. To resolve this we constructed, for the first time, a human Disease-associated Protein and subcellular Localization (DPL) matrix (top panel in Box 1). Our DPL matrix provides the ‘cellular localization map of diseases' that represents the spatial index of diseases in the cell. We found that each disease shows unique characteristics of subcellular localization profile in the DPL matrix. We were interested in determining whether subsets of 1284 human diseases exhibit distinct enrichment profiles across subcellular localizations. We calculated pairwise correlations and performed a hierarchical clustering of the enrichments of the 1284 diseases across 10 different subcellular localizations.

Our DPL matrix revealed that 778 diseases (∼62%, P=1.40 × 10⁻³) are enriched in a single localization and 273 diseases (∼21%, P=3.45 × 10⁻³) are enriched in dual localizations. In the DPL matrix, certain disease-associated proteins are likely to be found in membrane-bounded organelles such as mitochondria, lysosome, and peroxisome, indicating that the mutations of proteins localized to these compartments are connected to the pathophysiological conditions of those organelles. Meanwhile, certain disease-associated proteins in the DPL matrix are enriched in dual localizations, such as extracellular/plasma membrane or endoplasmic reticulum/Golgi. Although these two pairs of subcellular localizations appear to be distinct compartments at first, they are functionally related compartments in close proximity during protein translocation process in the cell, and thus are likely to share interacting protein partners (Gandhi et al, 2006).

Comorbidity represents the co-occurrence of multiple diseases in the same individual (Lee et al, 2008; Hidalgo et al, 2009; Park et al, 2009a). Many comorbid disease pairs have been shown to share common genes in the human disease network. For example, Diabetes and Alzheimer's disease share a risk factor in angiotensin I converting enzyme, and frequently occur together in an individual. In such instances, comorbidity can be partially attributed to the disease connections on the molecular level. To explore the impact of protein subcellular localization on comorbidity, we hypothesized that certain disease pairs could also be connected via subcellular localization by the molecular connections between the disease-associated proteins (bottom panel in Box 1).

We found a positive correlation between subcellular localization similarity and relative risk (Figure 3B, Pearson's correlation coefficient between relative risk and subcellular localization similarity=0.81, P=2.96 × 10⁻⁵). The subcellular localization similarity represents the correlation of subcellular localization profiles between disease pairs. To our surprise, when we compared the relative risk of disease pairs linked via various molecular connections, we found that disease pairs connected by subcellular localization showed a near three-fold higher comorbidity tendency (with link distances equal to 2 or 3) when compared with random pairs (Figure 3E).

We then assessed quantitatively the impact of network distances and subcellular localizations on the comorbidity tendency of disease pairs. We expected the proteins associated with comorbid disease pairs to be located closely in the protein interaction network via fewer links compared with random disease pairs. Indeed, a higher comorbidity tendency was found when two disease-associated proteins were positioned within a shorter distance (gray plots in Figure 3F). Moreover, when subcellular localization information was combined with small network distances, the comorbidity tendency increased dramatically (orange plots in Figure 3F). It suggests that subcellular localization and close network distances, two conceptually distinct molecular connections, contributed synergistically to the comorbidity tendency.

Disease progression is not restricted to the mutation of disease-causing genes, but also affected by molecular connections in ‘disease modules,' resulting in comorbidity (Fraser, 2006; Lee et al, 2008). In this study, for the first time we applied subcellular localization information to elucidate the molecular connections between comorbid diseases. We believe that, based on our finding, our approach helps to define the boundaries of ‘disease modules.' Taken together, integration of diverse molecular connections should improve the molecular level understanding of hitherto unexplained comorbid disease pairs and help us in expanding the scope of our knowledge of the mechanism of human disease progression.

Proteins targeting the same subcellular localization tend to participate in mutual protein–protein interactions (PPIs) and are often functionally associated. Here, we investigated the relationship between disease-associated proteins and their subcellular localizations, based on the assumption that protein pairs associated with phenotypically similar diseases are more likely to be connected via subcellular localization. The spatial constraints from subcellular localization significantly strengthened the disease associations of the proteins connected by subcellular localizations. In particular, certain disease types were more prevalent in specific subcellular localizations. We analyzed the enrichment of disease phenotypes within subcellular localizations, and found that there exists a significant correlation between disease classes and subcellular localizations. Furthermore, we found that two diseases displayed high comorbidity when disease-associated proteins were connected via subcellular localization. We newly explained 7584 disease pairs by using the context of protein subcellular localization, which had not been identified using shared genes or PPIs only. Our result establishes a direct correlation between protein subcellular localization and disease association, and helps to understand the mechanism of human disease progression.

Genome-wide approaches to the study of adaptive gene expression evolution: systematic studies of evolutionary adaptations involving gene expression will allow many fundamental questions in evolutionary biology to be addressed

The role of gene expression in evolutionary adaptation has been a subject of debate for over 40 years. cis-regulation of transcription has been proposed to be the primary source of morphological novelty in evolution, though this is based on only a handful of examples. Recently the first genome-wide studies of gene expression adaptation have been published, giving us an initial global view of this process. Systematic studies such as these will allow a number of key questions currently facing the field of gene expression evolution to be addressed.

A systems biology approach for the investigation of the heparin/heparan sulfate interactome

A large body of evidence supports the involvement of heparan sulfate (HS) proteoglycans in physiological processes such as development and diseases including cancer and neurodegenerative disorders. The role of HS emerges from its ability to interact and regulate the activity of a vast number of extracellular proteins including growth factors and extracellular matrix components. A global view on how protein-HS interactions influence the extracellular proteome and, consequently, cell function is currently lacking. Here, we systematically investigate the functional and structural properties that characterize HS-interacting proteins and the network they form. We collected 435 human proteins interacting with HS or the structurally related heparin by integrating literature-derived and affinity proteomics data. We used this data set to identify the topological features that distinguish the heparin/HS-interacting network from the rest of the extracellular proteome and to analyze the enrichment of gene ontology terms, pathways, and domain families in heparin/HS-binding proteins. Our analysis revealed that heparin/HS-binding proteins form a highly interconnected network, which is functionally linked to physiological and pathological processes that are characteristic of higher organisms. Therefore, we then investigated the existence of a correlation between the expansion of domain families characteristic of the heparin/HS interactome and the increase in biological complexity in the metazoan lineage. A strong positive correlation between the expansion of the heparin/HS interactome and biosynthetic machinery and organism complexity emerged. The evolutionary role of HS was reinforced by the presence of a rudimentary HS biosynthetic machinery in a unicellular organism at the root of the metazoan lineage.

[Identifying candidate cancer genes based on co-evolving gene modules]

Data of somatic mutation screening of cancer genomes have provided us huge amounts of information for identifying new cancer genes. Current methods for identifying candidate cancer genes based on gene mutation frequencies tend to find cancer genes with high mutation frequencies. However, many genes with low mutation frequencies might also play important roles during tumorigenesis. Based on the assumption that genes with similar phylogenetic profiles and protein-protein interactions might have similar functions and their disruptions might lead to similar disease phenotypes, we proposed a new approach to find candidate cancer genes. First, we searched for protein-protein interaction subnetworks within which proteins have similar phylogenetic profiles, termed as co-evolving gene modules. Then, we identified genes that have at least one non-synonymous mutation in cancer genomes and directly interact with known cancer genes in the same co-evolving gene modules and predicted them as candidate cancer genes. In this way, we found 15 candidate cancer genes, among which only two genes had been identified previously as candidate cancer genes using the methods based on gene mutation frequencies. Thus, the candidate cancer genes with low mutation frequencies can be found by our method.

Functional evolution of the microfibril-associated glycoproteins

The microfibril-associated glycoproteins (MAGPs) are cysteine-rich low molecular weight components of the fibrillin-based microfibrillar complex. MAGPs are evolutionarily conserved in vertebrates and have important roles in microfibril and elastic fiber structure, homeostasis, and vascular development. Two MAGPs, designated MAGP1 and MAGP2, are encoded in the mammalian genome. Although MAGP sequences have been identified in several vertebrate species, the extent of conservation and evolutionary history of the MAGPs in vertebrates is unknown. Sequence similarity searches of nucleotide and protein databases identified the first homologs of MAGP1 in monotremes, birds, elasmobranchs and agnathans, and the first MAGP2 genes in marsupials, birds and teleosts. A model for MAGP evolution is presented. Phylogenetic analysis identified the ancient origin of MAGP1 and the evolution of MAGP2 from a gene duplication event early in vertebrate evolution. Phylogenomic analysis shows conservation of synteny between teleosts and tetrapods and suggests a multigene duplication event. The MAGP2 gene has evolved rapidly as an innovation in the bony vertebrate lineage. Estimates of functional divergence and complex nucleotide substitution models suggest that the divergence of MAGP2 took place by relaxation of selective constraints; and that MAGP1 has consistently been constrained by strong purifying selection. Correlated evolution between MAGP1 and the developmental regulator, Notch1, may explain some of the selective forces acting on MAGP2.

Variations in the progranulin gene affect global gene expression in frontotemporal lobar degeneration

Frontotemporal lobar degeneration is a fatal neurodegenerative disease that results in progressive decline in behavior, executive function and sometimes language. Disease mechanisms remain poorly understood. Recently, however, the DNA- and RNA-binding protein TDP-43 has been identified as the major protein present in the hallmark inclusion bodies of frontotemporal lobar degeneration with ubiquitinated inclusions (FTLD-U), suggesting a role for transcriptional dysregulation in FTLD-U pathophysiology. Using the Affymetrix U133A microarray platform, we profiled global gene expression in both histopathologically affected and unaffected areas of human FTLD-U brains. We then characterized differential gene expression with biological pathway analyses, cluster and principal component analyses, and subgroup analyses based on brain region and progranulin (GRN) gene status. Comparing 17 FTLD-U brains to 11 controls, we identified 414 upregulated and 210 downregulated genes in frontal cortex (P-value < 0.001). Moreover, cluster and principal component analyses revealed that samples with mutations or possibly pathogenic variations in the GRN gene (GRN+, 7/17) had an expression signature that was distinct from both normal controls and FTLD-U samples lacking GRN gene variations (GRN-, 10/17). Within the subgroup of GRN+ FTLD-U, we found >1300 dysregulated genes in frontal cortex (P-value < 0.001), many participating in pathways uniquely dysregulated in the GRN+ cases. Our findings demonstrate a distinct molecular phenotype for GRN+ FTLD-U, not readily apparent on clinical or histopathological examination, suggesting distinct pathophysiological mechanisms for GRN+ and GRN- subtypes of FTLD-U. In addition, these data from a large number of human brains provide a valuable resource for future testing of disease hypotheses.

Evolution of the type III secretion system and its effectors in plant-microbe interactions

Many bacterial plant pathogens require the type III secretion system (T3SS) and its effector proteins (T3SEs) to invade and extract nutrients from their hosts successfully. While the molecular function of this system is being studied intensively, we know comparatively little about the evolutionary and ecological pressures governing its fate over time, and even less about the detailed mechanisms underlying and driving complex T3SS-mediated coevolutionary dynamics. In this review we summarize our current understanding of how host-pathogen interactions evolve, with a particular focus on the T3SS of bacterial plant pathogens. We explore the evolutionary origins of the T3SS relative to the closely related flagellar system, and investigate the evolutionary pressures on this secretion and translocation apparatus. We examine the evolutionary forces acting on T3SEs, and compare the support for vertical descent with modification of these virulence-associated systems (pathoadaptation) vs horizontal gene transfer. We address the evolutionary origins of T3SEs from the perspective of both the evolutionary mechanisms that generate new effectors, and the mobile elements that may be the source of novel genetic material. Finally, we propose a number of questions raised by these studies, which may serve to guide our thinking about these complex processes.

The road to modularity

A network of interactions is called modular if it is subdivided into relatively autonomous, internally highly connected components. Modularity has emerged as a rallying point for research in developmental and evolutionary biology (and specifically evo-devo), as well as in molecular systems biology. Here we review the evidence for modularity and models about its origin. Although there is an emerging agreement that organisms have a modular organization, the main open problem is the question of whether modules arise through the action of natural selection or because of biased mutational mechanisms.

Coadaptation and malaria control

Malaria emerges from a disequilibrium of the system 'human-plasmodium-mosquito' (HPM). If the equilibrium is maintained, malaria does not ensue and the result is asymptomatic plasmodium infection. The relationships among the components of the system involve coadaptive linkages that lead to equilibrium. A vast body of evidence supports this assumption, including the strategies involved in the relationships between plasmodium and human and mosquito immune systems, and the emergence of resistance of plasmodia to antimalarial drugs and of mosquitoes to insecticides. Coadaptive strategies for malaria control are based on the following principles: (1) the system HPM is composed of three highly complex and dynamic components, whose interplay involves coadaptive linkages that tend to maintain the equilibrium of the system; (2) human and mosquito immune systems play a central role in the coadaptive interplay with plasmodium, and hence, in the maintenance of the system's equilibrium; the under- or overfunction of human immune system may result in malaria and influence its severity; (3) coadaptation depends on genetic and epigenetic phenomena occurring at the interfaces of the components of the system, and may involve exchange of infectrons (genes or gene fragments) between the partners; (4) plasmodia and mosquitoes have been submitted to selective pressures, leading to adaptation, for an extremely long while and are, therefore, endowed with the capacity to circumvent both natural (immunity) and artificial (drugs, insecticides, vaccines) measures aiming at destroying them; (5) since malaria represents disequilibrium of the system HPM, its control should aim at maintaining or restoring this equilibrium; (6) the disequilibrium of integrated systems involves the disequilibrium of their components, therefore the maintenance or restoration of the system's equilibrium depend on the adoption of integrated and coordinated measures acting on all components, that means, panadaptive strategies. Coadaptive strategies for malaria control should consider that: (1) host immune response has to be induced, since without it, no coadaptation is attained; (2) the immune response has to be sustained and efficient enough to avoid plasmodium overgrowth; (3) the immune response should not destroy all parasites; (4) the immune response has to be well controlled in order to not harm the host. These conditions are mostly influenced by antimalarial drugs, and should also be taken into account for the development of coadaptive malaria vaccines.

A simple model of co-evolutionary dynamics caused by epistatic selection

Epistasis is the dependency of the effect of a mutation on the genetic background in which it occurs. Epistasis has been widely documented and implicated in the evolution of species barriers and the evolution of genetic architecture. Here we propose a simple model to formalize the idea that epistasis can also lead to co-evolutionary patterns in molecular evolution of interacting genes. This model epistasis is represented by the influence of one gene substitution on the fitness rank of the resident allele at another locus. We assume that increasing or decreasing fitness rank occur equally likely. In simulations we show that this form of epistasis leads to co-evolution in the sense that the length of an adaptive walk between interacting loci is highly correlated. This effect is caused by episodes of elevated rate of evolution in both loci simultaneously. We find that the influence of epistasis on these measures of co-evolutionary dynamics is relatively robust to the details of the model. The main factor influencing the correlation in evolutionary rates is the probability that a substitution will have an epistatic effect, but the strength of epistasis or the asymmetry of the initial fitness ranks of the alleles have only a minor effect. We suggest that covariance in rates of evolution among loci could be used to detect epistasis among loci.

Modular co-evolution of metabolic networks

Background

The architecture of biological networks has been reported to exhibit high level of modularity, and to some extent, topological modules of networks overlap with known functional modules. However, how the modular topology of the molecular network affects the evolution of its member proteins remains unclear.

Results

In this work, the functional and evolutionary modularity of Homo sapiens (H. sapiens) metabolic network were investigated from a topological point of view. Network decomposition shows that the metabolic network is organized in a highly modular core-periphery way, in which the core modules are tightly linked together and perform basic metabolism functions, whereas the periphery modules only interact with few modules and accomplish relatively independent and specialized functions. Moreover, over half of the modules exhibit co-evolutionary feature and belong to specific evolutionary ages. Peripheral modules tend to evolve more cohesively and faster than core modules do.

Conclusion

The correlation between functional, evolutionary and topological modularity suggests that the evolutionary history and functional requirements of metabolic systems have been imprinted in the architecture of metabolic networks. Such systems level analysis could demonstrate how the evolution of genes may be placed in a genome-scale network context, giving a novel perspective on molecular evolution.