Abstract
Age-dependent senescence is a multifaceted and highly coordinated developmental phase in the life of plants that is manifested with genetic, biochemical and phenotypic continuum. Thus, elucidating the dynamic network modeling and simulation of molecular events, in particular gene regulatory network during the onset of senescence is essential. Here, we constructed a computational pipeline that integrates senescence-related co-expression networks with transcription factor (TF)-promoter relationships and microRNA (miR)-target interactions. Network structural and functional analyses revealed important nodes within each module of these co-expression networks. Subsequently, we inferred significant dynamic transcriptional regulatory models in leaf senescence using time-course gene expression datasets. Dynamic simulations and predictive network perturbation analyses followed by experimental dataset illustrated the kinetic relationships among TFs and their downstream targets. In conclusion, our network science framework discovers cohorts of TFs and their paths with previously unrecognized roles in leaf senescence and provides a comprehensive landscape of dynamic transcriptional circuitry.
To elucidate the dynamic network modeling and simulation of the onset of senescence, we constructed a computational pipeline (Fig. 1a). It integrates co-expression networks, network topology analyses, TF-promoter associations and microRNA (miR)-target interactions. Subsequently, our platform models transcriptional regulatory dynamics using Scalable Models for the Analysis of Regulation from Time Series (SMARTS).1 Finally, it simulates the kinetic relationships among TFs and their downstream targets by employing Standardized Qualitative Dynamical systems (SQUAD)2 that combines Boolean and ordinary differential equation (ODE) models (Fig. 1a). Utilizing this platform, we first analyzed a high-resolution time-series senescence dataset spanning from four-day-old seedlings to 30-day-old leaves that are undergoing senescence.3 We compared the total number of differentially expressed genes (DEGs) and emergent genes, i.e. first significant change in the expression gradient, at each time point. While we found approximately 475 DEGs during each sampling time point, we observed a progressive increase in the number of emergent genes over time with maximum number of 1,317 emergent genes at day-30 (Supplementary Fig. S1a; Supplementary Table S1). This analysis allows us to identify phase-specific DEGs and a set of genes that play a role at the onset of leaf senescence.
To determine senescence-associated common gene signatures, we performed a weighted gene co-expression network analysis (WGCNA)4 in leaf senescence using two independent datasets encompassing 11,453 nuclear-encoded DEGs and 2,226 antisense DEGs. The resulting two co-expression networks are designated as Senescence CoExpression NeTwork (SCENT; 9,014 nodes and 993,699 edges; Fig. 1b; Supplementary Table S2) and Antisense Senescence CoExpression NeTwork (ASCENT; 1,158 nodes and 16,137 edges; Supplementary Fig. S5b), respectively. While the details on network analyses pertinent to ASCENT are described in Supplementary Information (Supplementary Fig. S5-S8), SCENT exhibits properties of a scale-free network (R2 > 0.6; a power-law degree distribution). Moreover, SCENT displays high mean connectivity as shown by topological overlap mapping metric (TOM) plot (Supplementary Fig. S1b). Based on cluster coefficient and connectivity of genes, we extracted 33 modules encompassing a minimum of 100 highly correlated genes that are differentially expressed during leaf senescence in SCENT (Supplementary Fig. S2). Turquoise and darkolivegreen are the largest and the smallest modules containing 1001 and 114 genes, respectively (Supplementary Table S2). A total of 366 genes were not assigned to any of the 33 modules, thus classified to a “gray” (or insignificant) 34th module (Supplementary Table S2). We observed that seven modules (darkolivegreen, greenyellow, violet, red, grey60, brown, and blue) exhibit heightened connectivity and average cluster coefficient compared to the other modules or SCENT as a whole (Supplementary Fig. S2b and c, Supplementary Table S2).
In network science, diverse centrality measures lead to the discovery of most influential nodes within a network.5,6,7,8 We analyzed degree (number of connections of a node), shortest paths (shortest distance between two nodes) and information centrality (propagation of information between vertices) for the nodes in SCENT and above mentioned seven modules. We discovered that information centrality and average shortest paths are significantly higher in seven modules compared to other modules or SCENT as a whole network (Fig. 1c; Supplementary Fig. S3a and b, Supplementary Table S2). Additionally, five modules (greenyellow, darkolivegreen, red, grey60, and blue) possess the majority of hub500 nodes in SCENT (Fig. 1b, Supplementary Table S2). Functional analyses in these seven modules revealed that most of genes in SCENT are involved in protein phosphorylation, response to salt stress, response to cold, oxidation-reduction process, protein folding, response to water deprivation and abscisic acid-activated signaling pathway followed by protein ubiquitination (Supplementary Fig. S4, Supplementary Table S3). Intriguingly, we discovered that hub500 nodes are significantly enriched in TFs, suggesting a central role of these key proteins in the co-regulation of their partners in SCENT (Hypergeometric test; p value = 7.198e−5).
Subsequently, we modeled our co-expression network, SCENT, to elucidate the intricate transcriptional relationships between the TFs and their target genes (TF-target association network; Fig. 2a, Supplementary Table S4). Intriguingly, we revealed the functions of 220 TFs in leaf senescence using our computational pipeline compared to previously reported 94 TFs in Woo et al.3 (Supplementary Fig. S9, Supplementary Table S4). Within SCENT, SMARTS highlighted major bifurcation events where the expression of a subset of genes diverged from the rest of the genes, and identified TFs potentially responsible for them. This analysis modeled 42 different paths that are governed by 18 different groups of TFs and several miRs (Fig. 1d, Supplementary Table S5). Taken together, this analysis inferred previously undiscovered dynamic and phase-specific transcriptional regulatory models that vastly expanded our horizons on the structural architecture and circuitry mechanisms in leaf senescence.
To simulate the impact of transcriptional activities on their target genes, we performed SQUAD simulations on an integrated network that contains co-expression network and TF-target association data (Fig. 2a). For downstream detailed simulations and experimental validations, we selected two Hub500 TFs (ANAC046, and previously uncharacterized DREB; A-5 of ERF/AP2 factor; AT1G71520) that regulate gene expression at different phases of senescence. ANAC046 co-regulates with 137 genes of which 102 nodes exhibit hub500 property in SCENT (Fig. 2b), whereas AT1G71520 co-regulates with 22 genes (Fig. 2b). Among them, 10 genes have been previously reported to be involved in autophagy or senescence-associated processes.9 The activation of ANAC046 and AT1G71520 over time was used as an index to measure the behavior of its putative downstream target genes (Supplementary Fig. S10b and c; Supplementary Table S6). This analysis displayed varied levels of activation for these downstream target genes. To validate these simulations, we subjected wild-type Col-0, ANAC046-SRDX (anac046 mutant) and ANAC046-OX (ANAC046 overexpressor)10 plants to dark-induced senescence. We observed differential accumulation of SAG12 transcript and protein over time in different genotypes (Fig. 2c, e, Supplementary Fig. S11a). These data corroborate with Fv/Fm quantification (Fig. 2d). Finally, qRT-PCR and mathematical modeling verified the activation of selected genes via ANAC046 or AT1G71520 by Cγ0 method (Fig. 2e, f, Supplementary Fig. S11b-d). In summary, our network science framework identified most influential nodes within the senescence network (Supplementary Discussion). Network modeling and simulations predicted functional cohorts of co-regulated genes, revealed the complexity of transcriptional regulatory machinery in leaf senescence and discovered innovative pathways during the onset of senescence.
Data availability
All data generated or analyzed supporting the results reported in this published article (and its Supplementary Information files).
References
Wise, A. & Bar-Joseph, Z. SMARTS: reconstructing disease response networks from multiple individuals using time series gene expression data. Bioinformatics 31, 1250–1257 (2015).
Di Cara, A., Garg, A., De Micheli, G., Xenarios, I. & Mendoza, L. Dynamic simulation of regulatory networks using SQUAD. BMC Bioinf. 8, 462 (2007).
Woo, H. R. et al. Programming of plant leaf senescence with temporal and inter-organellar coordination of transcriptome in Arabidopsis. Plant Physiol. 171, 452–467 (2016).
Langfelder, P. & Horvath, S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinf. 9, 559 (2008).
Garbutt, C. C., Bangalore, P. V., Kannar, P. & Mukhtar, M. S. Getting to the edge: protein dynamical networks as a new frontier in plant-microbe interactions. Front. Plant Sci. 5, 312 (2014).
McCormack, M. E., Lopez, J. A., Crocker, T. H. & Mukhtar, M. S. Making the right connections: network biology and plant immune system dynamics. Curr. Plant Biol. 5, 2–12 (2016).
Mishra, B., Sun, Y., Ahmed, H., Liu, X. & Mukhtar, M. S. Global temporal dynamic landscape of pathogen-mediated subversion of Arabidopsis innate immunity. Sci. Rep. 7, 7849 (2017).
Tully, J. P. et al. Expression-based network biology identifies immune-related functional modules involved in plant defense. BMC Genom. 15, 421 (2014).
Reiser, L. et al. Sustainable funding for biocuration: The Arabidopsis Information Resource (TAIR) as a case study of a subscription-based funding model. Database 2016, https://doi.org/10.1093/database/baw018 (2016).
Oda-Yamamizo, C. et al. Corrigendum: the NAC transcription factor ANAC046 is a positive regulator of chlorophyll degradation and senescence in Arabidopsis leaves. Sci. Rep. 6, 35125 (2016).
Acknowledgements
We acknowledge Dr. Akemi Ohmiya from National Agriculture and Food Research Organization (NARO), Japan for providing ANAC046-OX and ANAC046-SRDX lines. Also, the authors wish to acknowledge Dr. Karolina Mukhtar and Mr. Thomas Detchemendy for critical reading of the manuscript. This work was supported by the National Science Foundation (IOS-1557796) to MSM and the Alabama State funded Graduate Research Scholars Program (GRSP) to T.C.H.
Author information
Authors and Affiliations
Contributions
M.S.M. designed the experiments. B. M., N.K., and M.S.M. performed network and computational analyses. Y.S. and T.C. H. provided experimental data on dark-induced senescence. B. M. and M.S.M. drafted the manuscript. All authors reviewed the manuscript and approved the final version.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Mishra, B., Sun, Y., Howton, T. et al. Dynamic modeling of transcriptional gene regulatory network uncovers distinct pathways during the onset of Arabidopsis leaf senescence. npj Syst Biol Appl 4, 35 (2018). https://doi.org/10.1038/s41540-018-0071-2
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41540-018-0071-2
This article is cited by
-
Transcriptome meta-analysis of abiotic stresses-responsive genes and identification of candidate transcription factors for broad stress tolerance in wheat
Protoplasma (2023)
-
Specificity of H2O2 signaling in leaf senescence: is the ratio of H2O2 contents in different cellular compartments sensed in Arabidopsis plants?
Cellular & Molecular Biology Letters (2022)
-
A genome-wide comparative evolutionary analysis of zinc finger-BED transcription factor genes in land plants
Scientific Reports (2022)
-
Transcriptional circuitry atlas of genetic diverse unstimulated murine and human macrophages define disparity in population-wide innate immunity
Scientific Reports (2021)