Gene network inference using continuous time Bayesian networks: a comparative study and application to Th17 cell differentiation

Background

Dynamic aspects of gene regulatory networks are typically investigated by measuring system variables at multiple time points. Current state-of-the-art computational approaches for reconstructing gene networks directly build on such data, making a strong assumption that the system evolves in a synchronous fashion at fixed points in time. However, nowadays omics data are being generated with increasing time course granularity. Thus, modellers now have the possibility to represent the system as evolving in continuous time and to improve the models’ expressiveness.

Results

Continuous time Bayesian networks are proposed as a new approach for gene network reconstruction from time course expression data. Their performance was compared to two state-of-the-art methods: dynamic Bayesian networks and Granger causality analysis. On simulated data, the methods comparison was carried out for networks of increasing size, for measurements taken at different time granularity densities and for measurements unevenly spaced over time. Continuous time Bayesian networks outperformed the other methods in terms of the accuracy of regulatory interactions learnt from data for all network sizes. Furthermore, their performance degraded smoothly as the size of the network increased. Continuous time Bayesian networks were significantly better than dynamic Bayesian networks for all time granularities tested and better than Granger causality for dense time series. Both continuous time Bayesian networks and Granger causality performed robustly for unevenly spaced time series, with no significant loss of performance compared to the evenly spaced case, while the same did not hold true for dynamic Bayesian networks. The comparison included the IRMA experimental datasets which confirmed the effectiveness of the proposed method. Continuous time Bayesian networks were then applied to elucidate the regulatory mechanisms controlling murine T helper 17 (Th17) cell differentiation and were found to be effective in discovering well-known regulatory mechanisms, as well as new plausible biological insights.

Conclusions

Continuous time Bayesian networks were effective on networks of both small and large size and were particularly feasible when the measurements were not evenly distributed over time. Reconstruction of the murine Th17 cell differentiation network using continuous time Bayesian networks revealed several autocrine loops, suggesting that Th17 cells may be auto regulating their own differentiation process.     

Th1/Th2 differentiation and cross-regulation

The T helper (Th) phenotypes, Th1/Th2, are acquired upon interaction of a naive T helper cell and an antigen presenting cell (APC). Naive T helper cells may differentiate into either phenotype, and the actual outcome is determined by the density and avidity of the antigenic determinants presented by the APC, and the APCs inherent costimulatory properties. Until recently it was thought that differentiation is further affected by cytokines. However, Murphy et al. (1996, J. Exp. Med. 183, 901) have demonstrated that the experimental results, formerly interpreted as Th1/Th2 differentiation, in effect comprise an observation of two consecutive processes. (i) An interaction between naive T cells and APC creates a mixture of mature cells irreversibly committed to Th1 or Th2 phenotype. (ii) Subsequent addition of regulatory cytokines, promotes expansion of one phenotype while suppressing the other. The consequent shift in the per culture production of marker cytokines mimics the appearance of a cellular phenotype switch. We present and analyse a mathematical model that extrapolates these experimental facts into systemic behavior during an immune response. Despite the fact that differentiation produces cells of Th1 and Th2 phenotypes with the same receptor specificity, our results indicate that competition for antigenic stimulation, mediated by the APCs, combines with cytokine mediated cross-suppression between phenotypes to yield a response that is eventually dominated by T helper cells that are uniform in both receptor specificity (clonotype) and in cytokine secretion phenotype.     

Spatial and temporal heterogeneity explain disease dynamics in a spatially explicit network model

There is an increasing recognition that individual-level spatial and temporal heterogeneity may play an important role in metapopulation dynamics and persistence. In particular, the patterns of contact within and between aggregates (e.g., demes) at different spatial and temporal scales may reveal important mechanisms governing metapopulation dynamics. Using 7 years of data on the interaction between the anther smut fungus (Microbotryum violaceum) and fire pink (Silene virginica), we show how the application of spatially explicit and implicit network models can be used to make accurate predictions of infection dynamics in spatially structured populations. Explicit consideration of both spatial and temporal organization reveals the role of each in spreading risk for both the host and the pathogen. This work suggests that the application of spatially explicit network models can yield important insights into how heterogeneous structure can promote the persistence of species in natural landscapes.     

Cutting edge: Th17 and regulatory T cell dynamics and the regulation by IL-2 in the tumor microenvironment

Th17 cells play an active role in inflammation and autoimmune diseases. However, the nature and regulation of Th17 in the context of tumor immunity remain unknown. In this study, we show that parallel to regulatory T (Treg) cells, IL-17(+) CD4(+) and CD8(+) T cells are kinetically induced in multiple tumor microenvironments in mice and humans. Treg cells play a crucial role in tumor immune pathogenesis and temper immune therapeutic efficacy. IL-2 is crucial for the production and function of Treg cells. We now show that IL-2 reduces IL-17(+) T cell differentiation in the tumor microenvironment accompanied with an enhanced Treg cell compartment in vitro and in vivo. Altogether, our work demonstrates a dynamic differentiation of IL-17(+) T cells in the tumor microenvironment, reveals a novel role for IL-2 in controlling the balance between IL-17(+) and Treg cells, and provides new insight of IL-17(+) T cells in tumor immune pathology and therapy.     

Balance of Th1 and Th17 effector and peripheral regulatory T cells

Transfer of antigen-specific T cells into antigen-expressing lymphopenic recipients leads to the sequential generation of Th1 and Th17 effector and protective CD25⁺FoxP3⁺ regulatory cells in the periphery with surprisingly different kinetics. Such an experimental model is potentially valuable for defining the stimuli that regulate lineage decision and plasticity of various T cell effectors and peripheral regulatory T cells. Our studies have shown that IL-17 production occurs rapidly and declines within the first week with the appearance of IFN-γ producing T cells. Regulatory T cells appear during the recovery phase of the disease. The factors that mediate this complex differentiation originating from a starting naïve T cell population remain to be defined.     

The Th1/Th2/Th17 cytokine profile of HIV-infected individuals: A multivariate cytokinomics approach

HIV infection causes the dysregulation of cytokine production. A cytokinomics approach employing cytometric bead array (CBA) technology, flow cytometry and multivariate analysis was applied to the investigation of HIV-induced T helper cell type 1 (Th1), Th2 and Th17 cytokine changes in the sera of treatment naive individuals. Stepwise linear discriminant analysis (LDA) and logistic regression identified interleukin (IL)-6 to be discriminatory for HIV infection with 74.6% and 71.2% of the cases correctly classified. Analysis of variance (ANOVA) confirmed IL-6 and IL-10 concentrations to be significantly (p = 0.001 and p = 0.025) different between the groups. A scatter plot of the log IL-6 and IL-10 concentrations for the groups largely overlapped, with improved differentiation where patients were advancing to the acquired immunodeficiency syndrome (AIDS). IL-17A levels were higher than other cytokines but did not significantly distinguish the groups suggesting that the HIV? and HIV+ individuals had similar immune profiles. This possibility was supported by other clinical indicators. Taken together, the measured cytokines (IL-6, 10 and 17) have potential prognostic value.     

A marine carotenoid, fucoxanthin, induces regulatory T cells and inhibits Th17 cell differentiation in vitro

Fucoxanthin is a non-provitamin A carotenoid contained in brown seaweeds. We found that it suppressed interleukin-17 secretion from CD4(+) T cells under IL-17-producing T (Th17) cell development conditions. By evaluating T cell differentiation in vitro, fucoxanthin and its metabolite fucoxanthinol inhibited T cell differentiation into Th17 cells. This suggests that fucoxanthin can improve inflammatory diseases due to Th17 cells.     

Restriction of de novo pyrimidine biosynthesis inhibits Th1 cell activation and promotes Th2 cell differentiation

Leflunomide, an inhibitor of de novo pyrimidine biosynthesis, has recently been introduced as a treatment for rheumatoid arthritis in an attempt to ameliorate inflammation by inhibiting lymphocyte activation. Although the immunosuppressive ability of leflunomide has been well described in several experimental animal models, the precise effects of a limited pyrimidine supply on T cell differentiation and effector functions have not been elucidated. We investigated the impact of restricted pyrimidine biosynthesis on the activation and differentiation of CD4 T cells in vivo and in vitro. Decreased activation of memory CD4 T cells in the presence of leflunomide resulted in impaired generation and outgrowth of Th1 effectors without an alteration of Th2 cell activation. Moreover, priming of naive T cells in the presence of leflunomide promoted Th2 differentiation from uncommitted precursors in vitro and enhanced Th2 effector functions in vivo, as indicated by an increase in Ag-specific Th2 cells and in the Th2-dependent Ag-specific Ig responses (IgG1) in immunized mice. The effects of leflunomide on T cell proliferation and differentiation could be antagonized by exogenous UTP, suggesting that they were related to a profound inhibition of de novo pyrimidine biosynthesis. These results indicate that leflunomide might exert its anti-inflammatory activities in the treatment of autoimmune diseases by preventing the generation of proinflammatory Th1 effectors and promoting Th2 cell differentiation. Moreover, the results further suggest that differentiation of CD4 T cells can be regulated at the level of nucleotide biosynthesis.     

白细胞介素23在呼吸道合胞病毒感染致Th1、Th2及Th17细胞分化中的作用及机制

本研究旨在探讨白细胞介素23(interleukin 23,IL-23)在呼吸道合胞病毒(respiratory syncytial virus,RSV)感染支气管上皮细胞BEAS-2B后对Th1、Th2和Th17细胞分化的影响及作用机制。将RSV感染BEAS-2B后的上清液与淋巴细胞共孵育,并分别阻断IL-23受体(IL-23 receptor,IL-23R)、IL-23p19亚基及p38丝裂原活化蛋白激酶(p38 mitogen-activated protein kinase,p38 MAPK)信号通路。应用酶联免疫吸附试验(enzyme-linked immunosorbent assay,ELISA)检测上清液中细胞因子γ干扰素(interferon γ,IFN-γ)、IL-4、IL-17的浓度。同时,应用实时聚合酶链反应(polymerase chain reaction,PCR)检测相关转录因子(t-bet、gata3、rorγt)和信号转导子(stat4、stat6、stat3)的表达。结果显示,RSV感染后IFN-γ、IL-4和IL-17蛋白表达上调,转录因子及信号转导子的表达也有所增加。阻断IL-23和p38 MAPK信号通路后,Th1、Th2和Th7细胞分泌的细胞因子及转录因子表达均明显下降。结果提示,阻断IL-23后可在基因转导层面抑制RSV感染上皮细胞后诱导的Th1、Th2和Th17细胞分化,此过程可能与p38 MAPK信号通路有关。     

Atorvastatin ameliorates experimental autoimmune neuritis by decreased Th1/Th17 cytokines and up-regulated T regulatory cells

Statins have anti-inflammatory and immune-regulating properties. To investigate the effects of atorvastatin on experimental autoimmune neuritis (EAN), an animal model of Guillain–Barré syndrome (GBS), atorvastatin was administered to Lewis rats immunized with bovine peripheral myelin in complete Freund’s adjuvant. We found that atorvastatin ameliorated the clinical symptoms of EAN, decreased the numbers of inflammatory cells as well as IFN-γ⁺ and IL-17⁺ cells in sciatic nerves, decreased the CD80 expression and increased the number of CD25⁺Foxp3⁺ cells in mononuclear cells (MNC), and decreased the levels of IFN-γ in MNC culture supernatants. These data provide strong evidence that atorvastatin can act as an inhibitor in EAN by inhibiting the immune response of Th1 and Th17, decreasing the expression of co-stimulatory molecule, and up-regulating the number of T regulatory cells. These data demonstrated that statins could be used as a therapeutic strategy in human GBS in future.     

Identification of Th1/Th2 regulatory switch to promote healing response during leishmaniasis: a computational approach

Leishmania devices its survival strategy by suppressing the host’s immune functions. The antigen molecules produced by Leishmania interferes with the host’s cell signaling cascades and consequently changes the protein expression pattern of the antigen-presenting cell (APC). This creates an environment suitable for the switching of the T-cell responses from a healing Th1 response to a non-healing Th2 response that is favorable for the continued survival of the parasite inside the host APC. Using a reconstructed signaling network of the intracellular and intercellular reactions between a Leishmania infected APC and T-cell, we propose a computational model to predict the inhibitory effect of the Leishmania infected APC on the T-cell and to identify the regulators of this Th1-/Th2-switching behavior as observed during Leishmania infection. In this work, we hypothesize that a complete removal of the parasite could only be achieved with a simultaneous up-regulation of the healing Th1 response and stimulation of nitric oxide (NO) production from the APCs, and downregulation of the non-healing Th2 response and thereby propose several unique combinations of protein molecules that could elicit this anti-Leishmania immune response. Our results indicate that TLR3 may play a positive role in eliciting NO synthesis, while TLR2 may be responsible for inhibiting an anti-Leishmania immune response. Also, TLR3 overexpression (in the APC), when combined with SHP2 inhibition (in the T cell), produces an anti-Leishmania response that is better than the conventional IFN-gamma or IL12 treatment. A similar anti-Leishmania response is also obtained in another combination where TLR3 (in APC) is overexpressed, and SHC and MKP (of T cell) are inhibited and activated, respectively. Through our study, we also observe that Leishmania infection may induce an upregulation of IFN-beta production from the APC that may lead to an upregulation of the RAP1 and SOCS3 proteins inside the T cell, the potential inhibitors of MAPK and JAK-STAT signaling pathways, respectively, via the TYK2-mediated pathway. This study not only enhances our knowledge in understanding the Th1/Th2 regulatory switch to promote healing response during leishmaniasis but also helps to identify novel combinations of proteins as potential immunomodulators.     

Regulation of Th1/Th2 and Th17/Treg by pDC/mDC imbalance in primary immune thrombocytopenia

This study investigates the regulatory effect of plasmacytoid dendritic cells (pDC)/myeloid dendritic cells (mDC) imbalance on balance of Th1/Th2 and Th17/Treg in primary immune thrombocytopenia (ITP). A total of 30 untreated ITP patients and 20 healthy controls were recruited. Compared with healthy control, the pDC proportion of ITP patients was significantly reduced (P = 0.004), while the mDC proportion was not significantly changed (P = 0.681), resulting in a decrease in the pDC/mDC ratio (P = 0.001). Additionally, compared with controls, serum levels of interleukin (IL)-6, IL-12, and IL-23 were increased in ITP patients (P < 0.001), and mRNA levels of IL-12p40, IL-12p35, and IL-23p19 were also increased (P =0.014, P = 0.043, P < 0.001). Compared with the healthy control, the proportion of Th1 and Th17 cells in ITP patients increased (P = 0.001, P = 0.031). Serum levels of interferon gamma (IFN-γ) and IL-17 in ITP patients also increased (P = 0.025, P = 0.005). Furthermore, T-bet and RORγt mRNA levels were increased in peripheral blood of ITP patients (P = 0.018, P < 0.001). Correspondingly, the proportion of Th2 and Treg cells decreased (P = 0.007, P < 0.001), along with a decrease in serum IL-4 and transforming growth factor beta (TGF-β) (P = 0.028, P = 0.042), and an increase in GATA-3 mRNA (P < 0.001). However, there was no significant difference in Foxp3 mRNA levels (P = 0.587). Pearson correlation analysis showed that the proportion of total dendritic cells (DCs) was positively correlated with IL-12 (r = 0.526, P = 0.003) and IL-23 (r = 0.501, P = 0.005) in ITP patients. Th1/Th2 ratio, IFN-γ, and IL-12 levels were negatively correlated with platelet counts (r = −0.494, P = 0.009; r = –0.415, P = 0.028; r = –0.492, P = 0.032). However, IL-23 was positively correlated with IL-17 (r = 0.489, P = 0.006) and negatively correlated with platelet count (r = –0.564, P = 0.001). The ratio of IL-6 and Th17 cells was negatively correlated with platelet count (r = –0.443, P = 0.014; r = –0.471, P = 0.011). The imbalance of pDC/mDC and the increase of IL-6, IL-12, and IL-23 lead to the increased differentiation of CD4+ T cells into Th1 and Th17 cells, which might be the important mechanisms underlying the imbalance of Th1/Th2 and Th17/Treg in ITP patients.     

CD2 facilitates differentiation of CD4 Th cells without affecting Th1/Th2 polarization.

The role of CD2 in murine CD4 helper T cell differentiation and polarization was examined using TCR-Cyt-5CC7-I transgenic recombination activating gene-2-/- H-2(a) mice on CD2+/+ or CD2-/- backgrounds. In the absence of CD2, thymic development was abnormal as judged by reduction in the steady state number of total, double-positive, and CD4 single-positive (SP) thymocytes, as well as a defect in their restorative dynamics after peptide-induced negative selection in vivo. In addition, in CD2-/- animals, lymph node CD4 SP T cells manifest a 10- to 100-fold attenuated activation response to cytochrome c (CytC) agonist peptides as judged by induction of CD25 and CD69 cell surface expression or [(3)H]TdR incorporation; differences in the magnitude of responsiveness and requisite molar peptide concentrations were even greater for altered peptide ligands. Although the presence or absence of CD2 did not impact the final Th1 or Th2 polarization outcome, CD2 expression reduced the CytC peptide concentration threshold necessary to facilitate both Th1 and Th2 differentiation. In vivo administration of CytC peptide to CD2-/- animals yielded an impaired CD4 SP T cell effector/memory phenotype compared with similarly treated CD2+/+ mice. Analysis of TCR-Cyt-5CC7-I human CD2 double-transgenic mice similarly failed to reveal a preferential Th1 vs Th2 polarization. Collectively, these results indicate that CD2 is important for the efficient development of CD4 SP thymocytes and TCR-dependent activation of mature CD4 lymph node T cells, but does not direct a particular helper T cell subset polarity.