Mika Gustafsson

Modules, networks and systems medicine for understanding disease and aiding diagnosis

Many common diseases, such as asthma, diabetes or obesity, involve altered interactions between thousands of genes. High-throughput techniques (omics) allow identification of such genes and their products, but functional understanding is a formidable challenge. Network-based analyses of omics data have identified modules of disease-associated genes that have been used to obtain both a systems level and a molecular understanding of disease mechanisms. For example, in allergy a module was used to find a novel candidate gene that was validated by functional and clinical studies. Such analyses play important roles in systems medicine. This is an emerging discipline that aims to gain a translational understanding of the complex mechanisms underlying common diseases. In this review, we will explain and provide examples of how network-based analyses of omics data, in combination with functional and clinical studies, are aiding our understanding of disease, as well as helping to prioritize diagnostic markers or therapeutic candidate genes. Such analyses involve significant problems and limitations, which will be discussed. We also highlight the steps needed for clinical implementation.

Comparison and validation of community structures in complex networks

The issue of partitioning a network into communities has attracted a great deal of attention recently. Most authors seem to equate this issue with the one of finding the maximum value of the modularity, as defined by Newman. Since the problem formulated this way is believed to be NP-hard, most effort has gone into the construction of search algorithms, and less to the question of other measures of community structures, similarities between various partitionings and the validation with respect to external information.

DNA Methylation Changes Separate Allergic Patients from Healthy Controls and May Reflect Altered CD4+ T-Cell Population Structure

Altered DNA methylation patterns in CD4+ T-cells indicate the importance of epigenetic mechanisms in inflammatory diseases. However, the identification of these alterations is complicated by the heterogeneity of most inflammatory diseases. Seasonal allergic rhinitis (SAR) is an optimal disease model for the study of DNA methylation because of its well-defined phenotype and etiology. We generated genome-wide DNA methylation (Npatients = 8, Ncontrols = 8) and gene expression (Npatients = 9, Ncontrols = 10) profiles of CD4+ T-cells from SAR patients and healthy controls using Illuminas HumanMethylation450 and HT-12 microarrays, respectively. DNA methylation profiles clearly and robustly distinguished SAR patients from controls, during and outside the pollen season. In agreement with previously published studies, gene expression profiles of the same samples failed to separate patients and controls. Separation by methylation (Npatients = 12, Ncontrols = 12), but not by gene expression (Npatients = 21, Ncontrols = 21) was also observed in an in vitro model system in which purified PBMCs from patients and healthy controls were challenged with allergen. We observed changes in the proportions of memory T-cell populations between patients (Npatients = 35) and controls (Ncontrols = 12), which could explain the observed difference in DNA methylation. Our data highlight the potential of epigenomics in the stratification of immune disease and represents the first successful molecular classification of SAR using CD4+ T cells.

Gene Expression Prediction by Soft Integration and the Elastic Net—Best Performance of the DREAM3 Gene Expression Challenge

Background To predict gene expressions is an important endeavour within computational systems biology. It can both be a way to explore how drugs affect the system, as well as providing a framework for finding which genes are interrelated in a certain process. A practical problem, however, is how to assess and discriminate among the various algorithms which have been developed for this purpose. Therefore, the DREAM project invited the year 2008 to a challenge for predicting gene expression values, and here we present the algorithm with best performance. Methodology/Principal Findings We develop an algorithm by exploring various regression schemes with different model selection procedures. It turns out that the most effective scheme is based on least squares, with a penalty term of a recently developed form called the “elastic net”. Key components in the algorithm are the integration of expression data from other experimental conditions than those presented for the challenge and the utilization of transcription factor binding data for guiding the inference process towards known interactions. Of importance is also a cross-validation procedure where each form of external data is used only to the extent it increases the expected performance. Conclusions/Significance Our algorithm proves both the possibility to extract information from large-scale expression data concerning prediction of gene levels, as well as the benefits of integrating different data sources for improving the inference. We believe the former is an important message to those still hesitating on the possibilities for computational approaches, while the latter is part of an important way forward for the future development of the field of computational systems biology.

Reverse Engineering of Gene Networks with LASSO and Nonlinear Basis Functions

The quest to determine cause from effect is often referred to as reverse engineering in the context of cellular networks. Here we propose and evaluate an algorithm for reverse engineering a gene regulatory network from time‐series and steady‐state data. Our algorithmic pipeline, which is rather standard in its parts but not in its integrative composition, combines ordinary differential equations, parameter estimations by least angle regression, and cross‐validation procedures for determining the in‐degrees and selection of nonlinear transfer functions. The result of the algorithm is a complete directed network, in which each edge has been assigned a score from a bootstrap procedure. To evaluate the performance, we submitted the outcome of the algorithm to the reverse engineering assessment competition DREAM2, where we used the data corresponding to the InSilico1 and InSilico2 networks as input. Our algorithm outperformed all other algorithms when inferring one of the directed gene‐to‐gene networks.

Targeted omics and systems medicine: personalising care

A key problem in the management of respiratory diseases is that subsets of patients do not respond to available treatments. Ideally, clinicians should have access to diagnostic markers to personalise drugs for patients with respiratory diseases before starting treatment. Although such markers do not exist in clinical settings, some markers for personalised medicine could reach the clinic in the near future. For example, in asthma, biomarkers such as airway eosinophilia, high exhaled nitric oxide, and transcriptional signatures from bronchial brushings might potentially stratify patients for available and novel biological drugs. However, highthroughput (omics technologies) studies of patients with seasonal allergic rhinitis show the complexity of identification of biomarkers for treatment response. Thousands of genes change expression in nasal fluids, nasal fluid cells, nasal mucosa, and allergen-challenged blood T cells in patients with seasonal allergic rhinitis compared with controls. The involvement of different cell types and external triggers suggests that similar numbers of genes might change in expression in airways diseases such as asthma and chronic obstructive pulmonary disease (COPD). This complexity indicates that single biomarkers might not suffice to stratify accurately patients for personalised medicine. Instead, a change of scale to omics-based diagnostics might be needed. Diagnostic kits that measure transcriptional signatures are now used to stratify patients with breast cancer. But will this technique be achievable in the respiratory clinic? Achievement of this goal can be broken into three parts—one, measurement and validation of human genes, gene products, and metabolites on a genome-wide scale; two, extraction of parts of the genome-wide data with potential diagnostic value, ideally based on functional understanding or socalled targeted omics; and third, such biomarkers being clinically useful at a reasonable cost. Omics technologies allow analyses of most types of potentially relevant molecules, such as genes, transcripts, or proteins on a genome-wide scale. These have been reviewed recently, leading to stringent criteria for the clinical use and validation of omics. The potential of targeted omics has already been shown in breast cancer. However, the selection of potential biomarkers for targeted omics is a formidable challenge because of the large number of genes, gene products, and metabolites identified. Systems medicine is an emerging discipline that aims to find novel diagnostic markers and therapeutic targets by combining omics with bioinformatics, as well as functional and clinical studies. One important principle for how to understand and prioritise potential biomarkers is to map genes or gene products on the corresponding proteins in protein-interaction networks. The most relevant genes or gene products tend to co-localise and form highly interconnected modules in such networks. Such modules can be further analysed to detect pathways and individual genes that might represent diagnostic or therapeutic candidates. Investigators showed translational feasibility of module-based analysis in a study of seasonal allergic rhinitis. The study spanned from a genome-wide analysis of gene expression to high-throughput knockdowns of candidate genes, and diagnostic and therapeutic studies in an animal model of allergy and samples from patients. An important limitation was that it was based on only mRNA and protein expression. Studies of allergic rhinitis and asthma have shown the importance of other genomic layers, such as genetic variants, DNA methylation, and non-coding RNAs. These layers only partially correlate, and given the complexity of respiratory diseases, combinations of targeted omics analyses from different layers might be needed for personalisation. To bring such analyses to the clinic, many challenges would need to be addressed—first, clinical research is needed to determine and validate what targeted omics analyses are needed for diagnosis. Recent studies of asthma and seasonal allergic rhinitis support the feasibility of targeted transcriptomics to stratify patients for treatment response. However, larger studies that include other omics analyses are needed. Second, laboratory technology for targeted omics must be developed in clinical settings. An initial solution could be that commercial providers of omics-based diagnostics do the analyses, as in the case of breast cancer. Third, software is needed for diagnostic classification based on the omics data. In the simplest case, patients could be stratified into high-risk and low-risk groups, again as in breast cancer. However, user-friendly software that can potentially be customised to allow clinicians to make diagnostic classifications in any disease based on Pa sie ka /S cie nc e P ho to Li br ar y

A Generally Applicable Translational Strategy Identifies S100A4 as a Candidate Gene in Allergy

A module-based strategy identified and validated an important diagnostic and therapeutic candidate gene in allergy. Modular Approach Nothing to Sneeze At If genome-wide association studies have taught us anything, it’s that paring down genes potentially involved in specific diseases into viable therapeutic candidates can be an overwhelming task. Thousands of genes can be implicated, which then must be functionally validated. Now, Bruhn et al. take a modular approach to this problem in the context of allergy. They focus on genes co-regulated with the proallergic gene interleukin-13 (IL13). The authors identified a T helper 2 (TH2) cell module using small interfering RNA–mediated knockdown of putative IL13-regulating transcription factors. They then further validated one of these genes, S100A4, which has previously been shown to be involved in inflammatory cell recruitment. Loss of S100A4 resulted in decreased allergy-associated immune responses both in vitro and in a mouse model. Moreover, blocking S100A4 with an antibody decreased allergic response both in their mouse model and in cells derived from allergic patients, which suggests that S100A4 may be a new therapeutic target for allergy. The identification of diagnostic markers and therapeutic candidate genes in common diseases is complicated by the involvement of thousands of genes. We hypothesized that genes co-regulated with a key gene in allergy, IL13, would form a module that could help to identify candidate genes. We identified a T helper 2 (TH2) cell module by small interfering RNA–mediated knockdown of 25 putative IL13-regulating transcription factors followed by expression profiling. The module contained candidate genes whose diagnostic potential was supported by clinical studies. Functional studies of human TH2 cells as well as mouse models of allergy showed that deletion of one of the genes, S100A4, resulted in decreased signs of allergy including TH2 cell activation, humoral immunity, and infiltration of effector cells. Specifically, dendritic cells required S100A4 for activating T cells. Treatment with an anti-S100A4 antibody resulted in decreased signs of allergy in the mouse model as well as in allergen-challenged T cells from allergic patients. This strategy, which may be generally applicable to complex diseases, identified and validated an important diagnostic and therapeutic candidate gene in allergy.

A validated gene regulatory network and GWAS identifies early regulators of T cell–associated diseases

Combining a gene regulatory network and disease association data identified early regulators of T cell–associated diseases. Identifying disease before it starts Diseases may be easier to treat if caught early. However, means of identifying early disease—especially before symptoms appear—are in short supply. Now, Gustafsson et al. identify early regulators of T cell–mediated disease by finding transcription factors involved in T cell differentiation that are enriched in disease-associated polymorphisms. Three such experimentally validated transcription factors—GATA3, MAF, and MYB—and their targets were found to be differentially expressed in asymptomatic stages of two different T cell–mediated diseases—multiple sclerosis and seasonal allergic rhinitis. These data not only provide potential markers of disease development but also shed light on the mechanistic underpinning of T cell–mediated disease. Early regulators of disease may increase understanding of disease mechanisms and serve as markers for presymptomatic diagnosis and treatment. However, early regulators are difficult to identify because patients generally present after they are symptomatic. We hypothesized that early regulators of T cell–associated diseases could be found by identifying upstream transcription factors (TFs) in T cell differentiation and by prioritizing hub TFs that were enriched for disease-associated polymorphisms. A gene regulatory network (GRN) was constructed by time series profiling of the transcriptomes and methylomes of human CD4+ T cells during in vitro differentiation into four helper T cell lineages, in combination with sequence-based TF binding predictions. The TFs GATA3, MAF, and MYB were identified as early regulators and validated by ChIP-seq (chromatin immunoprecipitation sequencing) and small interfering RNA knockdowns. Differential mRNA expression of the TFs and their targets in T cell–associated diseases supports their clinical relevance. To directly test if the TFs were altered early in disease, T cells from patients with two T cell–mediated diseases, multiple sclerosis and seasonal allergic rhinitis, were analyzed. Strikingly, the TFs were differentially expressed during asymptomatic stages of both diseases, whereas their targets showed altered expression during symptomatic stages. This analytical strategy to identify early regulators of disease by combining GRNs with genome-wide association studies may be generally applicable for functional and clinical studies of early disease development.

Integrated genomic and prospective clinical studies show the importance of modular pleiotropy for disease susceptibility, diagnosis and treatment.

BackgroundTranslational research typically aims to identify and functionally validate individual, disease-specific genes. However, reaching this aim is complicated by the involvement of thousands of genes in common diseases, and that many of those genes are pleiotropic, that is, shared by several diseases.MethodsWe integrated genomic meta-analyses with prospective clinical studies to systematically investigate the pathogenic, diagnostic and therapeutic roles of pleiotropic genes. In a novel approach, we first used pathway analysis of all published genome-wide association studies (GWAS) to find a cell type common to many diseases.ResultsThe analysis showed over-representation of the T helper cell differentiation pathway, which is expressed in T cells. This led us to focus on expression profiling of CD4+ T cells from highly diverse inflammatory and malignant diseases. We found that pleiotropic genes were highly interconnected and formed a pleiotropic module, which was enriched for inflammatory, metabolic and proliferative pathways. The general relevance of this module was supported by highly significant enrichment of genetic variants identified by all GWAS and cancer studies, as well as known diagnostic and therapeutic targets. Prospective clinical studies of multiple sclerosis and allergy showed the importance of both pleiotropic and disease specific modules for clinical stratification.ConclusionsIn summary, this translational genomics study identified a pleiotropic module, which has key pathogenic, diagnostic and therapeutic roles.

Combined Multivariate and Pathway Analyses Show That Allergen-Induced Gene Expression Changes in CD4+ T Cells Are Reversed by Glucocorticoids

Background Glucocorticoids (GCs) play a key role in the treatment of allergy. However, the genome-wide effects of GCs on gene expression in allergen-challenged CD4+ T cells have not been described. The aim of this study was to perform a genome-wide analysis to investigate whether allergen-induced gene expression changes in CD4+ T cells could be reversed by GCs. Methodology/Principal Findings Gene expression microarray analysis was performed to profile gene expression in diluent- (D), allergen- (A), and allergen + hydrocortisone- (T) challenged CD4+ T cells from patients with seasonal allergic rhinitis. Principal component analysis (PCA) showed good separation of the three groups. To identify the correlation between changes in gene expression in allergen-challenged CD4+ T cells before and after GC treatment, we performed orthogonal partial least squares discriminant analysis (OPLS-DA) followed by Pearson correlation analysis. This revealed that allergen-induced genes were widely reversed by GC treatment (r = −0.77, P<0.0001). We extracted 547 genes reversed by GC treatment from OPLS-DA models based on their high contribution to the discrimination and found that those genes belonged to several different inflammatory pathways including TNFR2 Signalling, Interferon Signalling, Glucocorticoid Receptor Signalling and T Helper Cell Differentiation. The results were supported by gene expression microarray analyses of two independent materials. Conclusions/Significance Allergen-induced gene expression changes in CD4+ T cells were reversed by treatment with glucocorticoids. The top allergen-induced genes that reversed by GC treatment belonged to several inflammatory pathways and genes of known or potential relevance for allergy.