- Research
- Open Access
A genetic algorithm-based boolean delay model of intracellular signal transduction in inflammation
- Chu Chun Kang†^{1},
- Yung Jen Chuang^{3},
- Kai Che Tung^{3},
- Chun Cheih Chao^{3},
- Chuan Yi Tang^{1, 4},
- Shih Chi Peng^{1}Email author and
- David Shan Hill Wong^{2}Email author
https://doi.org/10.1186/1471-2105-12-S1-S17
© Kang et al; licensee BioMed Central Ltd. 2011
- Published: 15 February 2011
Abstract
Background
Signal transduction is the major mechanism through which cells transmit external stimuli to evoke intracellular biochemical responses. Understanding relationship between external stimuli and corresponding cellular responses, as well as the subsequent effects on downstream genes, is a major challenge in systems biology. Thus, a systematic approach to integrate experimental data and qualitative knowledge to identify the physiological consequences of environmental stimuli is needed.
Results
In present study, we employed a genetic algorithm-based Boolean model to represent NF-κB signaling pathway. We were able to capture feedback and crosstalk characteristics to enhance our understanding on the acute and chronic inflammatory response. Key network components affecting the response dynamics were identified.
Conclusions
We designed an effective algorithm to elucidate the process of immune response using comprehensive knowledge about network structure and limited experimental data on dynamic responses. This approach can potentially be implemented for large-scale analysis on cellular processes and organism behaviors.
Keywords
- Tumor Necrosis Factor
- Boolean Network
- Boolean Model
- Chronic Inflammatory Response
- Qualitative Knowledge
Background
Resolving the complex cellular signal transduction is a grand challenge in systems biology. Signal transduction involves cascade of protein-protein interaction and complex feedback loops [1] across proteomic and genomic levels. Models of the dynamics of the combined regulatory networks provide in-depth analysis temporal characteristics of targeted biological process. Furthermore, in silico knockout experiments by these models could help biologists to prioritize target genes of interest and reduce time and cost of real experiments.
Types of dynamic network models include kinetic models [2, 3], hidden Markov models [4], and logic-based models [5, 6]. Kinetic models based on differential equations have been used to elaborate dynamics on numerous systems [7]. However, they need detailed information about network structure, reaction mechanism and the respective kinetic parameters; which, unfortunately, are not easily obtainable. Hidden Markov model (HMM) is a statistical model in acyclic pathway [8]. The state is hidden, but the outcome dependent on the state is visible. Hence, HMMs are usually used to model known results with unknown process mechanism. Boolean network is a qualitative logic-based model that was introduced in the 1960s [9]. In the past few decades, scientists have frequently used Boolean network to model gene regulatory networks (GRN), apoptosis, metabolic network, immune response and signaling pathways [5, 10–12]. Since logic-based or qualitative knowledge of interaction is abundant, the network structure can be easily established. Moreover, only minimal information is required to describe the dynamics of Boolean transfer function, they can be obtained using limited experimental data. Therefore Boolean model is an effective and extendable way of modeling the dynamics of signal transduction.
The transcription factor NF-κB controls various inflammation mediators to orchestra interwoven cellular responses to inflammatory stimuli such as TNF, IL-1 and TLR4 etc. In this study, Boolean model with time-delay was used to described the NF-κB signaling. The objective is to integrate qualitative information on network interactions from published datasets and dynamic response data in literatures to reveal the regulatory mechanism of infection and inflammation.
Results and discussion
Model
Boolean transfer functions were used for each edge. Each transfer function has two dynamic parameters. The delayed activation θ denotes the duration that the input to a node must turned on before the reception node is turned on. The sustained response r is the time that the output of a node can be sustained once it is turned on. These parameters were obtained by fitting a training data set published in [2, 17, 18] using GA.
Even with a simplified network and limited number of dynamic parameters, convergence to a set of reasonable parameters was not easy. In order to improve the modeling process, we trained model parameters in kernel pathway first with parameters of the rest of the edges set equal to 1, using wild type data containing measurements of both IKK and NFkB measurements. When kernel’s parameters were decided, the parameters of the remaining edges were determined by adding additional data involving A20 knockout, stimulus of various strengths and measurements of either IKK or NFkB only.
Dynamics implications
The model obtained by our approach
Component | Boolean transfer function |
---|---|
IRAK1(t) | = 5*IL1(t-1) OR 3*LPS(t-87) |
TAB1/TAB2/TAK1(t) | = IRAK1(t-13) OR LPS(t-125) |
NIK(t) | = TAB1/TAB2/TAK1(t-25) OR TNFR1(t-1) |
IKK(t) | = TAB1/TAB2/TAK1(t-100) OR NIK(t-1) |
IkBa(t) | = NOT 2* IKK(t-3) OR -62* NFkB(t-96) |
NFkB(t) | = NOT IkBa(t-1) |
A20(t) | = 7*NFkB(t-107) |
IL1(t) | = -62*NFkB(t-373) |
TNF(t) | = -62*NFkB(t-69) |
TNFR1(t) | = 4*TNF(t-1) AND NOT A20(t-10) |
Negative regulator
Zinc finger protein A20, also called TNFAIT3 (Tumor necrosis factor, alpha-induced protein 3), can also produce RIP- or TRAF2- mediated signal to indirectly block the NFkB activity. We have learned from the literatures that the negative regulator A20 blocked the NFkB activation while protecting the host cell from TNF-mediated apoptosis. To mimic an A20 knockout assay done in the wet-bench experiments, we set the output of the A20 nod to 0 and keep it in off state during simulation as the deletion. The resulting signaling pattern of wild type and A20 mutation in our model were shown in Figure 4. TNF induced IKK(left side of Figure 4.A) active from 5 min to 60 min in wild type and persist with A20 deletion. NF-κB(right side of Figure 4.A) caused secondary activation when A20 is knocked out. For LPS-induced IKK (left side of Figure 4.B) and NFkB (right side of Figure 4.B), there is no difference between active patterns obtained with wild-type and A20 deletion. This is because in a LPS induced response, the secondary TNF response will be triggered after the transcription of NFkB. Hence A20 is apparently a key component in TNF-induced pathway but with no significant influence in LPS-induced pathway.
Clinical implication
Conclusions
In this work, a dynamic Boolean model was generated by integrating and comprehensive qualitative knowledge about network structure and fitting a minimal amount of dynamic response data. The model is capable of capturing feedback and crosstalk dynamics between diverse signaling pathways. Using this model, mechanisms of and factors affecting periodic pro-inflammatory and anti-inflammatory responses can be elucidated.
The proposed approach integrated intracellular and intercellular process. Hence it is possible for us to use this approach to develop system models for host defense against the shock from environmental or pathogen stimuli and predict the inflammatory response. Such a model will potentially be able to provide insight to a feedback treatment scheme for clinical therapy.
Methods
Boolean transfer function
In other words, the activation of the i^{th} node by j^{th} node is on only if the j^{th} node has been on continuously for a period of θ_{ ij }. The effect of sustained and delayed response can be described by the following pseudocode:
when h({x_{1}(t)}, θ_{ i1 }) = 1
if r_{ ij } ≥ 1
τ_{ ij } = τ_{ ij } + r_{ ij } - 1
g_{ ij } = 1
else
τ_{ ij } = τ_{ ij } + 1
if τ_{ ij } ≥ |r_{ ij }|
g_{ ij } = 1
τ_{ ij } = τ_{ ij } - 1
end
end
Thus if r_{ ij } ≥ 1, the activation of x_{ i } by x_{ j } will be sustained; alternatively the activation of x_{ i } by x_{ j } will be delayed.
Data processing
Model fitting by genetic algorithm
Settings in genetic algorithm
Chromosome representation | θ_{ ij },r_{ ij } |
---|---|
Generation number | 800 |
Population size | 1000 |
Crossover rates | 1 |
Mutation rates | 0.02 |
Notes
Declarations
Acknowledgements
This study is supported by National Science Council, ROC, under grant NSC98-2627-B-007-017.
This article has been published as part of BMC Bioinformatics Volume 12 Supplement 1, 2011: Selected articles from the Ninth Asia Pacific Bioinformatics Conference (APBC 2011). The full contents of the supplement are available online at http://www.biomedcentral.com/1471-2105/12?issue=S1.
Authors’ Affiliations
References
- Neil A, Campbell JBR: Biology. 6th edition. 2002.Google Scholar
- Werner SL, Kearns JD, Zadorozhnaya V, Lynch C, O'Dea E, Boldin MP, Ma A, Baltimore D, Hoffmann A: Encoding NF-kappaB temporal control in response to TNF: distinct roles for the negative regulators IkappaBalpha and A20. Genes Dev 2008, 22: 2093–2101. 10.1101/gad.1680708PubMed CentralView ArticlePubMedGoogle Scholar
- Pettigrew MF, Resat H: Modeling signal transduction networks: a comparison of two stochastic kinetic simulation algorithms. J Chem Phys 2005, 123: 114707. 10.1063/1.2018641View ArticlePubMedGoogle Scholar
- Mark de Been CF, Roy Moezelaar, Tjakko Abee, Siezen aRJ: Comparative analysis of two-component signal transduction systems of Bacillus cereus, Bacillus thuringiensis and Bacillus anthracis. Microbiology 2006, 152: 3035–3048. 10.1099/mic.0.29137-0View ArticlePubMedGoogle Scholar
- Saez-Rodriguez J, Alexopoulos LG, Epperlein J, Samaga R, Lauffenburger DA, Klamt S, Sorger PK: Discrete logic modelling as a means to link protein signalling networks with functional analysis of mammalian signal transduction. Mol Syst Biol 2009, 5: 331. 10.1038/msb.2009.87PubMed CentralView ArticlePubMedGoogle Scholar
- Julio Saez-Rodriguez LS, Jonathan LindquistA, Rebecca Hemenway, Ursula Bommhardt, Boerge Arndt, Utz-Uwe Haus, Robert Weismantel, Ernst GillesD, Steffen Klamt, Burkhart Schraven: A Logical Model Provides Insights into T Cell Receptor Signaling. PLoS Comput Biol 2007, 3: e163. 10.1371/journal.pcbi.0030163PubMed CentralView ArticlePubMedGoogle Scholar
- Arisi I, Cattaneo A, Rosato V: Parameter estimate of signal transduction pathways. BMC Neurosci 2006, 7(Suppl 1):S6. 10.1186/1471-2202-7-S1-S6PubMed CentralView ArticlePubMedGoogle Scholar
- Geiger D, Meek C, Wexler Y: Speeding up HMM algorithms for genetic linkage analysis via chain reductions of the state space. Bioinformatics 2009, 25: i196–203. 10.1093/bioinformatics/btp224PubMed CentralView ArticlePubMedGoogle Scholar
- Kauffman SA: Metabolic stability and epigenesis in randomly constructed genetic nets. Journal of Theoretical Biology 1969, 22: 437–467. 10.1016/0022-5193(69)90015-0View ArticlePubMedGoogle Scholar
- Kwon YK, Choi SS, Cho KH: Investigations into the relationship between feedback loops and functional importance of a signal transduction network based on Boolean network modeling. BMC Bioinformatics 2007, 8: 384. 10.1186/1471-2105-8-384PubMed CentralView ArticlePubMedGoogle Scholar
- Mai Z, Liu H: Boolean network-based analysis of the apoptosis network: Irreversible apoptosis and stable surviving. Journal of Theoretical Biology 2009, 259: 760–769. 10.1016/j.jtbi.2009.04.024View ArticlePubMedGoogle Scholar
- Thakar J, Pilione M, Kirimanjeswara G, Harvill ET, Albert R: Modeling Systems-Level Regulation of Host Immune Responses. PLoS Comput Biol 2007, 3: e109. 10.1371/journal.pcbi.0030109PubMed CentralView ArticlePubMedGoogle Scholar
- Holland JH: Genetic Algorithms and the Optimal Allocation of Trials. SIAM Journal on Computing 1973, 2: 88–105. 10.1137/0202009View ArticleGoogle Scholar
- Zheng Y, Yeh CW, Yang CD, Jang SS, Chu IM: On the local optimal solutions of metabolic regulatory networks using information guided genetic algorithm approach and clustering analysis. J Biotechnol 2007, 131: 159–167. 10.1016/j.jbiotec.2007.06.019View ArticlePubMedGoogle Scholar
- Wu FX, Poirier GG, Zhang WJ: Inferring gene regulatory networks with time delays using a genetic algorithm. Syst Biol (Stevenage) 2005, 152: 67–74.View ArticleGoogle Scholar
- Kanae Oda TK, Yukiko Matsuoka, Akira Funahashi, Masaaki Muramatsu, Hiroaki Kitano: Molecular Interaction Map of a Macrophage. Book Molecular Interaction Map of a Macrophage 2004., 2:Google Scholar
- Werner SL, Barken D, Hoffmann A: Stimulus Specificity of Gene Expression Programs Determined by Temporal Control of IKK Activity. Science 2005, 309: 1857–1861. 10.1126/science.1113319View ArticlePubMedGoogle Scholar
- Shih VF-S, Kearns JD, Basak S, Savinova OV, Ghosh G, Hoffmann A: Kinetic control of negative feedback regulators of NF-κB/RelA determines their pathogen- and cytokine-receptor signaling specificity. Proc Natl Acad Sci U S A 2009, 106: 9619–9624. 10.1073/pnas.0812367106PubMed CentralView ArticlePubMedGoogle Scholar
- Renner F, Schmitz ML: Autoregulatory feedback loops terminating the NF-kappaB response. Trends Biochem Sci 2009, 34: 128–135. 10.1016/j.tibs.2008.12.003View ArticlePubMedGoogle Scholar
- Bell E: Mediating endotoxin tolerance. Nat Rev Immunol 2004, 4: 750–750. 10.1038/nri1463View ArticleGoogle Scholar
- Akutsu T, Miyano S, Kuhara S: Identification of genetic networks from a small number of gene expression patterns under the Boolean network model. Pac Symp Biocomput 1999, 17–28.Google Scholar
- Fang-Xiang W, Kusalik AJ, Wen-Jun Z: A genetic algorithm for inferring time delays in gene regulatory networks. Computational Systems Bioinformatics Conference, 2004 CSB 2004 Proceedings 2004 IEEE 2004, 610–611.Google Scholar
- Assmann SM, Albert R: Discrete dynamic modeling with asynchronous update, or how to model complex systems in the absence of quantitative information. Methods Mol Biol 2009, 553: 207–225. full_textView ArticlePubMedGoogle Scholar
- Grefenstette J: Optimization of control parameters for genetic algorithms. IEEE Trans Syst Man Cybern 1986, 16(1):122–128. 10.1109/TSMC.1986.289288View ArticleGoogle Scholar
Copyright
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.