https://orcid.org/0000-0002-0320-8865
This is an uncorrected proof.
Figures
Abstract
Abscisic acid (ABA) acts as a key signalling molecule that mediates plant responses to environmental cues as well as plant growth and development. Stress-induced and developmental changes in ABA content trigger a myriad of post-transcriptional and transcriptional events. Yet, ABA-dependent transcriptional responses are context dependent, and their temporal dynamics in roots under non-stress conditions remain poorly resolved. In this study, we characterised the hallmarks of ABA signalling and responses in the poplar root transcriptome (Populus nigra L.). We disturbed ABA homeostasis by exogenous ABA treatments, and we combined time-resolved transcriptomics with unsupervised gene network analysis to identify ABA-activated and ABA-inactivated gene co-expression modules. Considering the temporal dynamics of transcriptional events, we predicted the primary targets of the ABA signal, characterised early responding ABA-dependent processes, identified hub genes and revealed their putative functional links. We demonstrated that the properties of a master ABA-activated module induced by exogenous treatments were preserved in the transcriptome response to osmotic stress, revealing a core gene set of ABA-dependent stress responses. Our work sheds light on ABA repression of gene expression, the reprogramming of metabolism and the leaf-senescence pathway. Based on current functional knowledge and phylogenetic information, including poplar-specific features, we proposed a working model of ABA action on root transcriptome in poplar that integrates master genes, key responsive processes, and their putative regulatory architecture.
Author summary
Abscisic acid (ABA) is a plant hormone that dynamically governs plant development and stress responses. Core ABA signalling pathway conveys environmental and developmental cues to the nucleus, triggering extensive transcriptional reprogramming required to adjust plant physiology. Here, we quantified time-dependent changes in gene expression and in hormone concentrations, occurring within the root of poplar plants following exogenous ABA treatments. By combining time-resolved transcriptomics with unsupervised gene network analysis, we revealed the modular structure of ABA-dependant transcriptional responses and identified the ABA-activated and ABA-repressed processes. Through the integration of gene network architecture with in-depth literature analysis and phylogenetic information, we inferred regulatory mechanisms controlling primary ABA-dependent responses in roots, and identified lineage-specific programs. Interestingly, ABA signalling in roots engages regulatory pathways and functional responses previously associated with germination, stress responses, metabolic regulation, or leaf senescence. By integrating these multiple insights into a working model, we propose a unified framework for ABA-dependent regulatory mechanisms, providing new perspectives for research in both annual and perennial plant species.
Citation: Cohen D, De Freitas Pereira M, Pavlović I, Novák O, Bogeat-Triboulot M-B, Cookson SJ, et al. (2026) Gene network analysis predicts the primary regulators of ABA-dependent transcriptional activation and repression in Populus roots. PLoS Genet 22(8): e1012256. https://doi.org/10.1371/journal.pgen.1012256
Editor: Arun Sampathkumar, Max Planck Institute of Molecular Plant Physiology: Max-Planck-Institut fur molekulare Pflanzenphysiologie, GERMANY
Received: February 17, 2026; Accepted: July 15, 2026; Published: August 4, 2026
Copyright: © 2026 Cohen et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Reads were deposited in the Sequence Read Archive (SRA) under the Study Accession number GSE183780. The expression data are provided in the S3 Data. Identifier of genes mentioned in the main text are given in S2 Appendix.
Funding: The research has received funding from the European Community’s Seventh Framework Programme FP7/2007-2013 (grant agreement n°FP7-311929 to DC, MBBT and IH). This work was supported by Lorraine Université d’Excellence through the interdisciplinary program ARTEMIS (grant ANR-15-IDEX-04-LUE to DC, MBBT, IH) and by the Laboratory of Excellence ARBRE (grant ANR-11-LABX-0002 to DC, MFP, MBBT, IH). ON and IP were supported by the Ministry of Education, Youth and Sports of the Czech Republic (European Regional Development Fund-Project ‘Plants as a tool for sustainable global development’ No. CZ.02.1.01/0.0/0.0/16_019/0000827 to ON), and by the Palacky University Olomouc Young Researcher grant (No. JG_2020_002 to IP). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
The sesquiterpenoid phytohormone abscisic acid (ABA) is involved in the regulation of plant development, growth and functioning. ABA signalling plays key roles throughout the plant life cycle, from seed maturation to leaf senescence, also mediating both seed and bud dormancy [1–4]. As a stress signalling molecule, ABA orchestrates plant responses to challenging environments, including water deprivation and high salinity [5–8].
Local active ABA levels are finely tuned through the balance between de novo ABA biosynthesis and catabolism, compartmentation, transport and external uptake. Currently, the metabolic pathways underlying ABA homeostasis are mostly well known [9,10]. ABA biosynthesis begins in the plastids via the methylerythritol phosphate (MEP) pathway, which produces isoprenoid precursors for the biosynthesis of carotenoids, gibberellin, and cytokinin [11]. The plastid-localized cleavage of 9’-cis-neoxanthin or 9-cis-violaxanthin into xanthoxin by 9-CIS-EPOXYCAROTENOID DIOXYGENASE (NCED) initiates the entry into the ABA biosynthetic pathway, which is completed by two cytosolic enzymatic reactions converting xanthoxin into active ABA [12]. NCED is considered to be the rate-limiting step in de novo ABA biosynthesis [13,14]. ABA catabolism requires the ABA hydroxylation by the CYP707A subfamily of the cytochrome P450 monooxygenases [15,16]. CYP707As mainly produce 8’-hydroxy-ABA, which spontaneously isomerizes to phaseic acid (PA). Phaseic acid reductase (PAR) converts PA to dihydrophaseic acid (DPA). While DPA is an inactive form, PA retains residual activity and can bind to the ABA receptor and activate some of the ABA-responsive genes [17,18]. CYP707As also produce 9’-hydroxy-ABA, which is converted to neophasic acid (neoPA) by neoPA reductase, NeoPAR [19]. The enzymes involved in the production of 7’-hydroxy-ABA (7-OH-ABA) have yet to be characterized. Local ABA levels depend on the diffusion of uncharged ABA forms and the active translocation of ABA [2,20]. Several proteins have ABA transport activity, such as ATP-binding cassette (ABCG) transporters, multidrug and toxic compound extrusion (MATE)-type/DTX transporters, and members of the NRT1/PTR family (NPF) [21–26]. The AWPM-19 family protein, OsPM1, has been shown to mediate ABA influx in rice [27].
The schematic model of the ABA signalling pathway was established two decades ago using exogenous application of chemicals, forward and reverse genetics, protein-protein interactions, and modelling [28–31]. Core ABA signalling components include the regulatory component of ABA receptor (RCAR), clade A protein phosphatase (HAB), and SNF1-related protein kinase 2 (SNRK2) [3,28,32,33]. HAB phosphatases act as a negative regulator of ABA signalling by inhibiting the kinase activity of SNRK2s, thereby repressing the phosphorylation of downstream ABA effectors. RCAR receptors, when linked to ABA molecule, bind to and inhibit HABs, releasing SnRK2s from inhibition, which in turn mediates ABA signal to specific targets, such as ABA-responsive transcription factors or membrane channels [34–38]. SNRK2 phosphorylation targets the basic leucine zipper (bZIP) transcription factors, DPBF/ABI5 (dc3 promoter binding factor) and AREB/ABF members (ABA-responsive element binding), which in turn promote the transcription of themselves and their specific targets [39,40]. These transcription factors are identified as canonical ABA pathway regulators [41]. They recognize the cis ABA-responsive elements (ABREs) in the promoter of their targets and activate their expression [39]. Core ABA signalling components and canonical ABA pathway regulators are highly conserved across land plants [42,43]. The physical interactions between core ABA signalling components are functionally preserved in several species, including Populus species [38,44,45].
Activation of ABA signalling pathway elicits massive transcriptome remodelling, and this response is strongly context-dependent [10,46]. In Arabidopsis seedlings, thousands of genes have been identified as ABA responsive, displaying pronounced increases or decreases in expression following ABA treatment [47–50]. ABA orchestrates the expression of numerous drought-responsive genes, including those involved in detoxification, protection against cellular damage, water transport, and the biosynthesis of osmoprotectants [48,51,52]. Concurrently, ABA signalling triggers transcriptional feedback loops that modulate ABA action by targeting ABA metabolism, core signalling components, and canonical regulators of the ABA pathway [10,41,47]. ABA modulates gene regulatory networks by altering both the expression and binding patterns of multiple transcription factors, and thereby influencing the transcription of their downstream targets [41,50,53]. The complex, multi-layered, and dynamic architecture of ABA-dependent transcriptional responses is widely recognised, but remains largely overlooked in roots. This is an important lack of knowledge because ABA acts as a key regulator of root growth and development, and controls the whole-plant functioning, particularly under stress conditions.
Here, we used time-resolved transcriptomics and co-expression network analysis to construct a comprehensive map of the earliest ABA-responsive events occurring in the root transcriptome of the perennial species, Populus nigra. We hypothesized that time-resolved ABA treatments, followed by their release, would identify the cascade of ABA-dependent responses and their underlying regulatory networks. We took advantage of the adventitious rooting capacities of clonal poplar cuttings in hydroponics to analyse the root transcriptome in the absence of leaves and under complete darkness [54,55]. Specifically, we asked which genes are potentially primary ABA targets, what is the relative importance of transcriptional activation and repression, and how are these responses coordinated over time. We investigated the preservation of ABA-dependent co-expression modules in the context of osmotic stress. Following the principles of “guilt-by-profiling” and “guilt-by-association”, we deciphered ABA regulatory network in the poplar root tip. The architecture of ABA-dependent network and its integration with multiple stress-responsive pathways in a perennial species revealed ABA-specific and lineage-specific responses.
Results
Abscisic acid (ABA) exposure progressively disrupts ABA homeostasis in the root tip
ABA enters plant tissues either passively by diffusion or actively by transporter-mediated uptake [20]. Poplar adventitious roots were exposed to exogenous 2 µM ABA treatment (ABA+), and the changes in the contents of ABA and ABA catabolites were monitored in the root tip over the first three hours (Exp1, Fig 1, S1 Data). Under ABA-free conditions (CTL), ABA content per gram of root dry weight was in the picomole range. The content of ABA catabolites, namely 7-OH-ABA, PA, and neoPA, was also in the picomole range while DPA was found as the major ABA catabolite, reaching the nanomole range. Under ABA+, the ABA content in the root reached the nanomole range. A 5-min-long ABA exposure was sufficient to induce a 24-fold increase in ABA content in comparison with the CTL. And a 180-min-long exposure led to a 400-fold increase in ABA content relative to the CTL. Among ABA catabolites, PA exhibited the strongest response to ABA+, reaching the nanomole range 30 min after treatment began. DPA content oscillated and was significantly reduced by 5 and 60-min-long exposures relative to the CTL. NeoPA and 7-OH-ABA underwent delayed accumulation; their contents being 10-fold and 2-fold-higher 180 min after treatment began relative to the CTL respectively. In addition, pairwise correlation among samples revealed high positive correlations between ABA and its catabolites, except DPA (Fig 1B: ABA vs 7-OH-ABA r = 0.85***, ABA vs PA r = 0.91***, ABA vs neoPA r = 0.85***). These ABA catabolites showed a delayed accumulation relative to ABA.
Hormone contents were quantified in roots during Exp1, under ABA-free conditions (CTL) and after 2 μM ABA+ treatment for 5, 30, 60 or 180 min. (A) Hormone contents are presented as a function of treatments. Data are means ± SE, n = 4. Asterisks indicate significant differences, as determined by Student's t-test between ABA+ versus CTL (P-value: *** ≤ 0.001, ** ≤ 0.01, * ≤ 0.05). “<LOD” indicates value below the limit of detection. (B) The covariation of hormone contents is given as a correlation network (n = 20). Only the most significant Pearson coefficients are shown (P-value ≤0.01). Edge width denotes the statistical significance (thick line: P-value ≤ 0.001). Edge type denotes the direction of correlation (dash: negative coefficient, solid: positive coefficient). Grey shapes highlight the communities which were detected via optimizing modularity score (“i.graph” R package, “cluster_fast_greedy” function with default parameters).
In addition to ABA and ABA catabolites, 29 hormone-related compounds were quantified in the root tip under CTL and ABA+ (Fig 1, S1 Data). Half of them exhibited no significant changes in concentration under ABA+ compared with the CTL. Salicylic acid, jasmonic acid, and jasmonic acid isoleucine transiently accumulated 60 min after treatment began. The content of cis-12-oxo-phytodienoic acid, a precursor of jasmonate, approximately doubled under ABA+ while the content of 9,10-dihydrojasmonic acid transiently halved. Tryptophan content increased by 5-fold and the content of indole-3-acetic acid-aspartate decreased by 1.5-fold under ABA+. The contents of active cytokinin forms, isopentenyladenine and cis-zeatin, peaked 60 min after treatment began. The contents of isopentenyladenine riboside, trans-zeatin riboside, dihydrozeatin riboside and dihydrozeatin riboside monophosphate decreased under ABA+. Cis-zeatin riboside-O-glucoside also decreased. In contrast, trans-zeatin-7-glucoside accumulated under ABA+ relative to the CTL. Pairwise comparison among samples revealed some correlations between these compounds (Fig 1B). Yet, their differential accumulation did not reflect ABA+ exposure duration. Within the first three hours, ABA+ triggered the accumulation of ABA and ABA catabolites (PA, neoPA and 7-OH-ABA), which were clearly insulated from other hormone-related compounds (Fig 1B).
ABFs and ABI5 identified as potential orchestrators of early transcriptional events
ABA triggered transcriptional remodelling within the first hour of exposure (Exp1: CTL and ABA+ for 2 min to 60 min, Fig 2). iDREM, interactive Dynamic Regulatory Events Miner [56], reconstructed the temporal dynamics of early transcriptional events upon ABA+ and revealed the putative transcription factors most likely to control them (Fig 2, S2 Data). ABA exposure triggered six patterns of temporal expression (path A to F gathering 1,897 loci, Fig 2). The earliest event was the split between path A and path B, which occurred between 2 and 5 min after treatment began (Fig 2C). Path A corresponded to the loci with the strongest transcriptional activation (Fig 2B). Moderately responsive to ABA+, path C and path E bifurcated between 5 and 15 min after treatment began and then their expression increased in parallel with path B. The divergence between path D and path F occurred later, between 15 and 30 min after treatment began, but was more drastic (repression for path F vs activation for path D). The six transcriptional time-course profiles led to four distinct transcriptional states after a 60 min-long ABA+ exposure: 153 loci with highly up-regulated expression (path A), 876 with moderately up-regulated expression (path: B, C and D), 643 loci with weakly up-regulated expression (path E) and 225 loci with down-regulated expression (path F). Enrichments of the Gene Ontology (GO) terms and MapMan bins in the loci of the different paths were consistent with abiotic stress responses and short-term response to ABA (Fig 2D and S2 Data). All paths were enriched in the molecular function “DNA-binding transcription factor activity” and in the biological processes related to response to stimulus, including “GO:0009719”, “GO:0006950”, “GO:0009607” and “GO:0042221”. Path B and E were also enriched in “GO: secondary metabolic process”, “bin16: secondary metabolism”, “GO: kinase activity” and “bin30.2: receptor kinase”. Paths A, C, E and F were enriched in MapMan bins related to hormone metabolism, including “bin17: hormone metabolism”, “bin17.1: abscisic acid”, “bin17.4 cytokinin” and “bin17.5 ethylene”.
(A) Experimental design for time-resolved transcriptomics. (B-D) iDREM revealed the transcriptional events occurring during the first hour of ABA treatment in Exp1. (B) iDREM reconstructed six distinct patterns of temporal expression (paths A to F). (C) iDREM regulatory map annotated with the putative transcription factors most likely to regulate the path transition events (Diamond, split cut-off score <0.0001). (D) Gene ontology (GO) and MapMan (Bin) enrichment analyses for the six iDREM paths are given as an heatmap with darkest colors being associated with the most significant enrichment.
iDREM predicted 16 P. trichocarpa putative transcription factors as the most likely to regulate the transition events (split cut-off score <0.0001, Fig 2C and S2 Data). They encompassed a Trihelix GT3 transcription factor (PtrGT29), a BES1/BZR1 transcription factor (Potri.003G106500), six basic helix–loop–helix (bHLH) transcription factors, and eight bZIP transcription factors (bZIP63, PtrGBF3, bZIP16/bZIP68 homolog Potri.003G097600, GBF1 homolog Potri.007G029400, PtrABI5, PtAREB7, PtrABF3, and PtrAREB1-2). These transcription factors were predicted to regulate the expression of 250 target genes on the regulatory map (Fig 2C and S2 Data). Among them, six transcription factors were transcriptionally activated by ABA, namely PtrGT29 in path B, PtrGBF3 and PtrABF3 in path C, and PtrbHLH165, PtrAREB1-2 and PtrABI5 in path E. The gene regulatory network (GRN) inferred for the 16 transcription factors comprised 965 predicted transcription factor-target interactions and 258 nodes in four ABA-activated paths (A, C, D and E) (S1A Fig and S2 Data). Most of these transcription factors are homologs of known master transcription factors of ABA signalling in Arabidopsis [41,53]. AREB/ABF proteins are activators of ABA response in the early steps of ABA signalling cascade [57]. In Arabidopsis, bZIP16 inhibited ABA action during seedling development [58]. In summary, the iDREM regulatory map and GRN revealed the rapid activation of the ABA-driven signalling cascade in poplar root through the action of poplar canonical ABA pathway regulators (ABFs and ABI5).
ABA shapes co-expression modules according to treatment duration
The transcriptome was characterized under CTL conditions, after ABA exposure (“ABA+” for 2–960 min) and after withdrawal of ABA (“ABA+/-”, ABA+ pretreated roots transferred to CTL conditions for 60–960 min) over two experiments (Exp1 and Exp2, Fig 2A). We analysed the transcriptome data jointly using weighted correlation network analysis with a signed network type (WGCNA [59]). This WGCNA organized 24,334 loci into 17 co-expression modules (M1-17) and provided gene expression as module eigengene (ME) (Fig 3A and S3 Data). WGCNA is based on unsupervised clustering; the modules are built without considering the experimental design. Heatmap visualization of ME as a function of treatments revealed that seven modules displayed clear patterns consistent with our experimental design (Fig 3A-B). These modules showed significant ME changes in response to ABA+ treatments relative to the CTL (pairwise Student’s t-tests, P-value ≤ 0.01, Fig 3B). ME1, ME2, ME10 and ME5 were shaped mostly by the 960-min-ABA+ treatment, whereas ME3 and ME4, and to a lesser extent ME16, changed progressively with the duration of ABA+ treatment. The WGCNA discriminated ABA-activated modules (M1, M10, M4 and M16) and ABA-inactivated modules (M2, M5 and M3) (Fig 3A-B). Among them, three modules were significantly enriched in loci previously associated with iDREM paths (Fig 2 and S2 Data). The ABA-inactivated M3 was significantly enriched in loci associated with ABA-inactivated iDREM path F. The ABA-activated M1 was significantly enriched in loci associated with iDREM paths A, B, C and E while the ABA-activated M4 was significantly enriched in loci associated with iDREM paths A, C, D and E. Consistently, the iDREM-derived GRN mostly included genes from M1 and M4, representing 24% and 55% of the nodes, respectively (S1B Fig and S2 Data). In addition, 11 out of the 16 transcription factors acting in the first layer of the GRN were associated with the ABA-activated M4 (S1B Fig and S2 Data).
WGCNA identified 17 co-expression modules within the reference expression matrix (27,379 loci x 46 libraries). (A) Heatmap of module eigengene (ME) for all samples and modules. (B) ME are given, as a function of treatments, for seven modules significantly structured according to experimental design; Panel B shows a focused view of Panel A. Treatments are ranked according to their duration. Asterisks indicate significant differences, as determined by Student's t-test between ABA treatments versus their respective control (P-value: *** ≤ 0.001, ** ≤ 0.01, * ≤ 0.05). (C) Gene ontology (GO) and MapMan (Bin) enrichment analyses for the seven modules are given as a heatmap with darkest colors being associated with the most significant enrichment. (D) Correlation network computed using the mean content of ABA responsive hormones and the mean ME of seven modules over five time points in Exp1 (Pearson correlation, n = 5, P-value ≤ 0.05). Edge color denotes significance level (black line: P-value ≤ 0.01), edge type denotes the direction of correlation (dash: negative coefficient, solid: positive coefficient), and grey shapes highlight the three communities.
The loci within the co-expression modules were enriched in various GO terms and MapMan bins (Figs 3C, S4 and S5 Data). The ABA-activated modules M1 and M10 shared enrichments in “bin29.5: protein degradation” and “GO: catabolic process”. The ABA-activated module M4 was enriched in “bin17.1: hormone metabolism.abscisic acid”, “bin3.1: minor CHO metabolism.raffinose”, “GO: response to chemical”, and “GO: DNA-binding transcription factor activity”, while the ABA-activated module M16 was enriched in “bin2: major CHO metabolism”. The ABA-inactivated modules M2 and M5 shared enrichments in “bin28: DNA”, “GO: DNA metabolic process”, “bin29.1: protein.aa activation”, and “GO: cellular component organization”, whereas the third ABA-inactivated module, M3, was enriched in transport, lipid metabolism, secondary metabolism and photosynthesis (“GO:0005215”, “bin34”, “GO:0006629”, “GO:0019748”, “bin16”, “GO:0015979” and “bin1”). Co-expression module analysis was completed by correlating MEs with hormone changes within the first three hours of ABA exposure (Fig 3D). In this time window, nine modules were correlated with at least one hormone profile. Yet only ME3, ME4, and ME16 were correlated with ABA and ABA catabolites. Early changes in ME4 and ABA content were highly correlated (P-values <0.01) while ME3 and ME16 captured the delayed accumulation of ABA catabolites (Fig 3D).
WGCNA computes the intramodular connectivity (kin) that is the connectivity of a given locus with respect to other loci from the same module. This kin metric allowed us to identify 678 intramodular hubs (scaled kin ≥ 0.75). Notably, the iDREM-derived GRN included 19 intramodular hubs, most of which were associated with M4 (S1B Fig and S2 Data). These 17 M4 hubs encompassed three transcription factors acting in the first layer of the iDREM-derived GRN (Potri.007G029400, PtrABF3 and PtrAREB1–2, S1B Fig). We investigated hub-hub co-expression relationships using a correlation-based approach and presented the most significant hub-hub correlation as a planar filtered network (Fig 4 and S6 Data). This network map highlighted the modular structure of gene co-expression. Intramodular hubs aggregated locally according to their module assignation. Moreover, intramodular hub edges exhibited a high value of topological overlap (TO). This TO metric shows the similarity of relationships between pairs of loci with respect to all other loci from the expression matrix [60]. The strongest intermodular relationships connected M2 and M5 hubs consistently with their ME covariation (Fig 3A). Intermodular hubs relationships showed the strong connection of ABA-activated and ABA-inactivated modules through highly significant and negative correlations (M3-M4, M16-M3, and M10-M5). Network layout revealed that M2, M5, M10 and M1 hubs formed a community that was distant from the community of M3, M4 and M16 hubs.
WGCNA hub network was summarized as a planar correlation network. Planar network was constructed on most significant pairwise correlation (FDR < 0.05) over the 678 hub genes (scaled intramodular connectivity, Kin ≥ 0.75). Network is drawn using Fruchterman and Reingold layout. Node color and size denote module assignation and scaled Kin, respectively. Edge color and edge size denote the direction of pairwise correlation and hub-hub topological overlap, respectively. Unigene identifiers are given for M3, M4 and M16 hub genes. They are ranked in descending order of Kin. Either P. trichocarpa V4.1 identifier or a common name is given for each unigene (detailed annotation and reference in S3 Data).
ABA-dependent transcriptional network is partially preserved in the transcriptome response to osmotic stress
WGCNA provides statistics for studying the properties of network modules in an independent dataset [61]. We tested whether the modular properties triggered by ABA treatments (Fig 3) were preserved during early responses to osmotic stress [62]. The multi-experiment dataset included 24,962 loci (Fig 5A and S3 Data). Strong evidence for a high preservation of modular properties in terms of connectivity and density was found for 13 out of the 17 modules (Zsummary >10 and Zmean.Adj > 10 respectively, Fig 5B-C). As Z-scores depend on module size, Zsummary values are expected to be the highest for largest modules. Within their respective size classes, the modular structure of ABA-activated modules M1 and M4 were best preserved (Fig 5A–C). Their mean adjacency scores were significantly higher than by chance with the highest significance (M1: Zmean = 104, Log P = -2328; M4: Zmean = 78, Log P = -1306). Among the small modules (<1,000 loci), M12 exhibited the strongest preservation of modular properties (M12: Zmean = 112, Log P = - 2728). The preservation of modular properties highlighted the high structural robustness and strong reproducibility of ABA-activated modules M1, M4 and M16, as well as ABA-inactivated modules M2, M3 and M5, in response to exogenous ABA treatment and osmotic stress (Fig 5B–C).
The modular structure revealed during ABA treatments (reference set) was tested for its preservation during early osmotic stress response (test set = CTL or 30 min and 180 min-long exposure to a polyethylene glycol-induced osmotic potential of −0.35 MPa). (A) Module size is given for reference and multi-experiment datasets. (B-C) WGCNA preservation statistics (nPerm = 200). Modules with Zsummary metrics (B) or Zmean.Adj metrics (C) > 10 (horizontal line) are highly preserved in reference and multi-experiment datasets. Light boxes denote weak to moderate preservation. (D-E) DGCA metrics. (D) Modular average differential connectivity, MeDC, is the change in correlation among module-embedded loci in reference vs test sets. Light boxes denote MeDC significantly different from 0 (nPerm = 75, empirical P-value for significance <0.01). (E) The distribution of differential correlation classes for pairwise relationships between intramodular hub loci is given.
Differential gene correlation analysis (DGCA) provides a bottom-up approach for assessing module preservation [63]. DGCA computes the modular average differential connectivity (MeDC) considering all module-embedded loci pairs. As shown in Fig 5D, ABA-activated modules M1, M4 and M16 exhibited no significant changes in modular connectivity between the two datasets (M1: MeDC = -0.7, P-value = 0.05; M4: MeDC = -0.6, P-value = 0.01; M16: MeDC = -1.1, P-value = 0.01). M12 also exhibited strong preservation of modular properties (M12: MeDC = -0.4, P-value = 0.5). In contrast, ABA-inactivated modules (M2, M3 and M5) and ABA-activated M10 exhibited a significant loss of average connectivity during the response to osmotic stress (P-value = 0, Fig 5D).
DGCA provides statistics for assessing the preservation of pairwise gene correlation between datasets [63]. We tested whether the hub-hub co-expression relationships triggered by ABA treatments were preserved under osmotic stress (S7 Data). DGCA retained 108,947 significant pairwise correlations between hubs in the multi-experiment dataset (P-value <0.01, S7 Data). DGCA computes correlation changes between the reference and test sets based on the mean difference in Z-score of the correlations and associated adjusted P-value (S7 Data). These results are summarized into nine DGCA classes in which positive, negative or non-significant correlations are coded as “+”, “-” or “0”, respectively, in both the reference and test sets (“reference/test” S7 Data). Intramodular hub-hub relationships were positive and highly significant under ABA treatments (30,226 “+/” correlations, P-value <0.01), and half of them were preserved under osmotic stress (DGCA class “+/+”: 15,066 correlations, P-value <0.01, Fig 5E). As shown in Fig 5E, the preservation of intramodular hub-hub relationships varied across modules. Among the ABA-activated and ABA-inactivated modules, M4 showed the best preservation of intramodular relationships (78% of DGCA class “+/+”). Within M1, M2 and M5, approximately half of intramodular relationships were preserved under osmotic stress. In contrast, more than 60% of intramodular relationships within M3, M10 and M16 were disrupted under osmotic stress. The divergent DGCA class “+/-” involved only 46 loci and occurred in ABA-responsive modules M3 and M10. Collectively, module-based statistics and hub-centred metrics showed that ABA-inactivated M3 and ABA-activated M10 were less preserved, and that ABA-activated M4 was the most preserved module in the context of osmotic stress.
Finally, our analysis also revealed that the expression of 17 hub genes acting in the iDREM-derived GRN was significantly up-regulated in response to both ABA treatments and osmotic stress (S1C Fig and S2 Data). Genes differentially expressed in response to osmotic stress represented 40% of the nodes, including four transcription factors acting in the first layer of the GRN. This analysis identified 85 genes whose expression rapidly increased in response to both ABA and moderate osmotic stress (S2 Data).
Identifying the hallmarks of ABA-dependent transcriptional network
In-depth interrogation of literature and public databases outlined the biologically relevant characteristics of the ABA network. We prioritised the most central loci in the ABA-activated module M4, the ABA-inactivated module M3 and those supported by iDREM analysis (Figs 2 and 4 and S1). While prioritising the functional information obtained in poplars, we also took into account homology-based information. The phylogenetic inference of orthologs was obtained by querying comparative genomics databases [64,65]. This functionally grounded analysis highlighted seven molecular processes and pathways that are the most likely to be coordinated by ABA. They are described in detail below.
Transcriptional feedback loops for ABA action.
Thirty-two M4 genes were associated with core ABA signalling pathway and its transcriptional regulators (Fig 6). Among these, all poplar genes encoding the canonical ABA pathway regulators (ABFs and ABI5) showed high intramodular connectivity in ABA-activated M4 (PtrAREB1-3, PtrAREB1-4, and PtrABF3, PtrAREB1-2, and PtrABI5). One-third of the genes harbouring predicted motifs for ABF and ABI5 binding sites were found in the ABA-activated M4 (18 hub genes and 411 genes, S3 Data). A feedback loop regulating core ABA signalling pathway involves the interaction of ABFs and ABI5 with ABI5-binding proteins (AFPs [66–68]). In poplar, PtrAREBs activate PagAFP2a expression and PtrAREBs-PagAFP2 interactions repress the expression of PtrAREB targets [68]. This feedback loop was ABA-activated in the root tip (Fig 6E). On the other hand, a negative transcriptional feedback acts on core ABA signalling components by modulating the expression of ABA receptor (RCARs) and the inhibitory protein phosphatase (HABs). As in Arabidopsis, ABA triggered the opposite regulation of RCAR and HAB expression [69–71]. Ten out of the14 RCARs clustered in the ABA-inactivated M3 while 11 out of the PtrHABs clustered in the ABA-activated M4. In Arabidopsis, ABA-dependent regulation of HABs and RCARs is mediated by homeodomain-leucine zipper transcription factors (ATHB7 and ATHB12 [72]). Their homologs, Pt-ATHB.3 and Pt-ATHB.5, were central in the ABA-activated M4 and are present in transcriptional networks underlying drought and stress memory [73,74]. ABA-INDuced Transcription Repressors (AITR) also control RCAR and HAB expression [75,76] and there was strong connectivity of the three AITR homologs in M4.
(A) Genes encoding REGULATORY COMPONENTS OF ABA RECEPTOR, RCAR. (B) Genes encoding CLADE A PROTEIN PHOSPHATASE, HAB. (C) Genes encoding SnrK2 PROTEIN KINASE. (D) Genes encoding canonical pathway regulators, ABFs and ABI5. (E) Genes encoding ABI5-BINDING PROTEINS, AFPs. (F) Genes encoding transcriptional regulators of ABA action. Heatmaps merge the main features from WGCNA (gene assignation to ABA-activated or ABA-inactivated modules, module assignation and scaled intramodular connectivity), from iDREM analysis (path), and from in silico prediction about ABI5 and/or ABFs gene interaction from plantregmap.gao-lab.org (Target). [number] indicates literature source and (h) indicates homology-based information by querying comparative genomics databases. Only unigenes expressed in roots are shown, identifiers refer to P. trichocarpa V4.1 genome.
Twelve M4 genes were associated with the post-transcriptional regulation of the core ABA signalling pathway (S2 Fig). In Arabidopsis, ABA sensing is subjected to a post-transcriptional feedback loop involving the WD40 protein ABT (ABA signalling Terminator), which disrupts the RCAR/HAB complex, and enhances HAB activity [40,77,78]. The ABT-centred feedback loop is likely to be present in poplar, as we found that the two poplar ABT homologs were transcriptionally activated by ABA (S2 Fig). Among the post-transcriptional regulators acting on SNRK2 phosphorylation [79], only one EGR phosphatase homolog exhibited a strong ABA activation (S2 Fig). Yet, a putative inhibitory loop involving SNRK2-Interacting Calcium Sensors, SCS [80], was transcriptionally activated by ABA in poplar root. Selective proteasomal degradation is known to occur for core ABA signalling components [3,81]. E3 ligases ATPUB22 and ATPUB23 are involved in the degradation of RCARs, while AIRP2 is involved in the activation of ABI5 [82–84]. The poplar homologs of ATPUB22 and ATPUB23 were transcriptionally activated by ABA, as well as the AIRP2 homolog, Potri.006G099600. In Arabidopsis, both ABA and drought activate the expression of E3 ligases CHYR1 and XERICO [85,86]. The expression of PtCHYR1 was not modified by ABA in poplar roots, while its paralogs (Potri.006G121801 and PtCHYR2) and PtXERICO were ABA-activated. The overexpression of PtXERICO, a M4 hub gene, triggers the expression of genes involved in the ABA homeostasis and enhances poplar drought tolerance [86]. ABA also activated the expression of PtrPUB72 and PtrPUB82 (S1 and S2 Figs). The expression of these genes is also up-regulated in response to drought, and PtrPUB72 positively regulates ABA-dependent drought response via the ubiquitination of PtrWRKY77, a repressor of the ABA response [87,88].
Cell protection against damage.
ABA increased the expression of genes whose products are putatively involved in cell protection against damage, and we further identified 15 M4 hub genes potentially involved in cell protection (S3 Fig). These genes are also transcriptionally activated in responses to water deficit in various organs and species [41,62,73,74,89–92]. The M4 hub gene, PtHsp100-01, is orthologous to DAA1, an ABA responsive gene that encodes a putative AAA-ATPase with a chaperone-like activity in Arabidopsis [93,94]. Other M4 hub genes or highly connected genes are related to redox homeostasis, another important component of cell protection [95–98]. They included nucleoredoxin, metallothionein, cytosolic glutathione peroxidase, and putative mitochondrial NDUFS4. Meanwhile, ABA activated gene expression for late embryogenesis abundant (LEA) proteins and for enzymes involved in raffinose family oligosaccharides (RFO) metabolism (S3 Fig). RFOs and LEA proteins have stabilizing functions, including membrane protection, and act as scavengers of reactive oxygen species or as osmoprotectants [98,99]. In poplar, PtrABF3, one of the IDREM predicted transcriptional regulators, has been shown to activate the expression of several genes in M4 with potential cell protectant ability, such as the LEAs PtrLEA55 and RD29b [8]. In Arabidopsis, RDs and ERDs encode several families of core drought-inducible proteins and are widely used as transcriptional drought reporters [89,100]. Here, ABA activated the expression of three ERD7 homologs, harbouring the senescence domain “IPR009686”. In Arabidopsis, ERD7 binds to negatively charged phospholipids and remodels membrane lipid composition [101]. In both Arabidopsis and poplar, ERD7 and LEAs are present in the proteome of lipid droplets [102,103]. These endoplasmic-reticulum-derived organelles, coated with lipid droplet associated proteins (LDAPs), OLEOSINs (OLEs) and SEIPINs, proliferate during stress responses [104–106]. The genes associated with lipid droplets and triacylglycerol synthesis were mostly found in the ABA-activated modules M4 and M1 (S3 Fig). PtLDAP1b, PtOLE5 and PtOLE6 were identified as putative targets of ABF/ABI5 (S3 Fig). Concordantly, LDAP expression is induced by ABA and drought in Arabidopsis [107]. The expression of AT2G25890, a PtOLE6 homolog, is activated by ABF1, ABF2 and ABI5, and repressed by AFP1 and AFP2 [67,108]. The ABA-activated M4 gene Potri.002G206100 is the ortholog of TSPO, an ABA-responsive gene encoding a multi-stress-inducible Translocator Protein that, in Arabidopsis, has been proposed to scavenge porphyrins and regulate lipid droplet formation [109–111].
Isoprenoid and hormone metabolism.
ABA reprogrammed the expression of genes regulating isoprenoid production via the MEP pathway and hormone metabolism (S4 Fig). The MEP pathway provides isoprenoid precursors for diverse pathways [11]. All enzymatic steps of the MEP pathway and its branch points were represented by at least one highly connected gene in the ABA-inactivated M3, including a hub gene encoding geranylgeranyl pyrophosphate synthase (S4A and S4B Fig). In a similar fashion to the MEP pathway, all enzymatic steps in the carotenoid biosynthesis were represented by at least one highly connected gene in the ABA-inactivated M3. For instance, in the case of PHYTOENE SYNTHASE (EC:2.5.1.32), although the PSY ortholog (Potri.017G138900) was present in ABA-activated M4, its paralog (Potri.001G007700) was a hub gene in the ABA-inactivated M3 and associated with the ABA-inactivated iDREM path F. In Arabidopsis, PSY is under the post-transcriptional control of ATOR-LIKE, which controls carotenoid biosynthesis [112,113]. Notably, the M3 hub gene Potri.006G210500 is orthologous to ATOR-LIKE. A key gene involved in ketocarotenoid biosynthesis was an ABA-inactivated M3 hub gene (Potri.004G197100) encoding capsanthin/capsorubin synthase (EC:5.3.99.8) [114]. Conversely, key genes involved in zeaxanthin and ABA biosynthesis were co-expressed in the ABA-activated M4, including β-carotene 3-hydroxylase, zea-epoxidase, NCED, and xanthoxin oxidase (EC:1.14.15.24, EC:1.14.15.21, EC:1.13.11.51 and EC:1.1.1.288). In addition, the ABA-activated M4 included four putative ABA-catabolizing CYP707s (EC:1.14.14.137), but did not include any of the nine poplar genes that we identified as putative PARs and NeoPARs based on their homology with CRL1, ABH2 and NeoPAR1 [18,19]. Carotenoid pathway feeds strigolactone synthesis, which was massively repressed by ABA (including the M3 hub genes: PtrMAX1b, MAX3, PtrMAX4b and PtD27c, S5 Fig). ABA rewired cytokinin pathway, with down-regulation of biosynthetic genes (PtIPT5a, Pt-CYP735.1 and Pt-CYP735.2), up-regulation of catabolic genes (seven PtCKXs) and differential ABA responses for other genes (S6 Fig). The ABA-inactivated M3 was enriched in “gibberellin metabolism” (S4 Data), indicating that gibberellin synthesis was down-regulated in response to ABA (S5 Fig). Three PtGA3oxs, encoding the enzyme that catalyses the synthesis of bioactive gibberellin, were present in M3, along with other biosynthetic genes. Six of these genes encodes 2-oxoglutarate-dependent dioxygenases and are associated with PtGA3ox1 in a gibberellin metabolic cluster located on the poplar chromosome 1 [115]. In support of the negative impact of ABA on the gibberellin pathway, three genes encoding gibberellin-inactivating enzymes were found in the ABA-activated M4, PtaGA2ox6 and PtaGA2ox1, as well as the ELA1/ELA2 homolog Potri.019G064200 [116].
Transporters and channels.
ABA remodelled the transporter landscape (S7 Fig). The ABA-inactivated M3 was enriched in “transport” with 206 annotated genes (Fig 3C; S4 and S5 Data). Notably, M3 was enriched in “nitrate transport” (S7A Fig, S5 Data and S1 Appendix). M3 was also enriched in “nitrogen metabolism” and “oxidative pentose phosphate (OPP) pathway” (S8 Fig and S5 Data). As the OPP pathway provides reductant for biosynthetic reactions, including nitrogen assimilation in heterotrophic tissues [117], ABA repressed, at least at the transcriptional level, key enzymatic steps in nitrogen assimilation. For the sake of brevity, more information about aquaporins (M3 and M4 hub genes), ion channels of OSCA family (highly-connected M4 genes) and transport of anthocyanin/flavonoid (ABA-inactivated M3) are given in S1 Appendix.
Anthocyanin/flavonoid pathway.
ABA antagonized the anthocyanin/flavonoid pathway, decreasing the expression of genes involved in transport, metabolism, and transcription factors (S7 and S9 Figs). The ABA-inactivated M3 was enriched in “secondary metabolic process” and more specifically in “flavonoids” (Fig 3, S4 and S5 Data). Most steps in anthocyanin/flavonoid synthesis were repressed at the transcriptional level. However, ABA activated the expression of genes encoding flavonol synthase, the key enzyme for flavonol production (M16, EC:1.14.11.9, S9 Fig). Such transcriptional regulations also occur in grapevine roots submitted to ABA [118]. R2R3-MYBs are key transcriptional regulators of the anthocyanin/flavonoid metabolism in poplar [119–125]. The ABA-inactivated M3 included most of MYBA1 activators, such as the hub gene PtrMYB153, as well as three MYB repressors, PtrMYB179 and the paralogs PtrMYB165/PtrMYB194 (S9 Fig). These paralogs belong to R2R3-MYB Subfamily S4, which also includes PtrMYB203, a M3 hub gene [126]. MYBs regulate anthocyanin/flavonoid pathway by forming an MBW complex with one bHLH and one WD40 repeat protein [127,128]. In poplar, the activators PtrMYB006 and PtrMYB115 interact with PtbHLH131, which also participates in the PtrMYB134-PtbHLH131-PtWD40 complex that activates the expression of PoptrBAN1, an anthocyanidin-reductase encoding gene [121,129]. These genes, except PtrMYB134, were well connected in the ABA-inactivated M3. In Arabidopsis and poplar, TTG1 and LTF15 are well-known MBW partners, respectively [129–132]. In Arabidopsis, MYC2 also acts as positive regulator of flavonoid biosynthesis [129,133]. The antagonistic effect of ABA on the anthocyanin/flavonoid pathway is strengthened by the co-repression of LTF15, TTG1 ortholog (Potri.015G002600), and MYC2 homologs (PtbHLH175 and PtbHLH178).
Tetrapyrrole and chlorophyll metabolism.
ABA decreased the expression of photosystem-related genes and tetrapyrrole biosynthetic genes (Figs 3C and S10). The ABA-inactivated M3 was enriched in GO term “Photosynthesis” and in bins “photosystem”” and “tetrapyrrole synthesis” (Fig 3 and S4, S5 Datas). Seventy-two genes contributed to these enrichments, but they showed low intramodular connectivity. Concerning tetrapyrrole synthesis, ABA decreased the expression of 15 M3 genes including the two genes encoding GUN4 (GENOMES UNCOUPLED 4), a regulatory subunit of Mg-chelatase, which is essential for chlorophyll accumulation [134]. In addition, we identified in the ABA-inactivated M3 two orthologs of key genes acting on chlorophyll homeostasis in Arabidopsis: Potri.001G143400 is orthologous to BCM1/BCM2 (BALANCE OF CHLOROPHYLL METABOLISM [135,136]) and Potri.010G211500 is orthologous to RPGE2/RPGE3 (REPRESSOR OF PHOTOSYNTHETIC GENES [137]). On the other hand, the ABA-activated M1 included PtSGR1 and Potri.001G112600, two homologs of SGR1, a Mg-dechelatase degrading chlorophyll and photosystem (STAY-GREEN1, EC: 4.99.1.10, [138]). The M4 hub gene, Potri.014G034700, is the ortholog of HEMD which encodes the enzyme responsible for the formation of uroporphyrinogen III (EC: 4.2.1.75), the precursor of all tetrapyrroles [139]. Potri.014G034700 was a predicted target of GBF1 homolog Potri.007G029400 on iDREM-derived GRN (S1 Fig). We identified two poplar homologs of senescence-associated genes, SAG113 and SAG201, involved in ABA-mediated and auxin-induced leaf senescence, respectively [140,141]. Their expression was ABA-activated in poplar roots, but increased also in leaf during autumn senescence [142].
Other transcription factors.
ABA modified the expression of numerous transcription factors (Figs 2 and 3, S4 and S5 Data, S1 Fig). We identified 41 genes in M4 and M3 putatively involved in the amplification and the diversification of the ABA signal (S11 Fig). In Arabidopsis, ABA regulatory network highlights GT3a and GBF3 as initial response regulators [53]. Their homologs, PtrGBF3 and PtrGT29, were found in the ABA-activated M4 and predicted to behave as master regulators by iDREM (Figs 2 and S1). Stress-responsive NACs, also called SNAC, and ANAC016 mediate stress responses and ABA-inducible leaf senescence [140,143–147]. Poplar SNACs and ANAC016 homologs were mostly associated with the ABA-activated M4 and were secondary transcription factors on the iDREM-derived GRN (S1 Fig). ABA-responsive SNACs are PtATAF1.1/PtATAF1.2, PtNAC049/PtNAC078 and PtRD26/PtNAC105, orthologs of ATAF1, ANAC032 and RD26/ANAC053/ANAC019, respectively, as well as NAC163/PtNAC118 and NAC095, orthologs of ANAC047 and ATNAP, respectively. Most poplar SNACs were predicted targets of ABF/ABI5, and PtrAREB1-2 binds to the ABRE motifs of three SNACs (PtATAF1.2, PtATAF1.1, and PtRD26 [148]). Meanwhile, our work confirmed the ABA-dependent transcriptional activation of R2R3-MYBs belonging to clades 16, 41 and 42, as previously reported in Arabidopsis [119,149–154]. The R2R3-MYBs of clade 16 comprise the ABA-activated Arabidopsis genes MYB49, MYB41, MYB102, and MYB74, and six poplar genes which were present in the ABA-activated modules M1 and M4. Among them, PtrMYB043 is known to be transcriptionally activated by PtrABF3 [8]. In a similar fashion to the R2R3-MYBs of clade 16, the R2R3-MYBs of clade 41 and 42 comprise three ABA-activated Arabidopsis genes (MYB121, MYB71, and MYB79) and five poplar genes which were present in the ABA-activated M4 (S11C Fig). In a variety of species, MYB121 homologs are transcriptionally activated in roots in response to ABA or abiotic stresses [118,155–158]. We also identified ABA-inactivated transcription factors. Six M3 hub genes encoded R2R3-MYB, bHLH, AP2/ERF, bZIP and GRAS transcription factors, that, to the best of our knowledge, have not been reported yet as ABA-responsive genes (S11 Fig). Among them, PtGRAS39 is orthologous to SCL23 in SCARECROW subfamily while PtSHR2B belongs to SHORTROOT subfamily [159]. Our work identified ABA-inactivated homologs of transcription factors acting on chloroplast maintenance and chlorophyll homeostasis in Arabidopsis. Potri.007G136901 and Potri.017G015800 are homologous to GLKs (GOLDEN2-LIKE [136,160]). Potri.006G075200 and Potri.018G142100 are homologous to CIA2/CIL (CHLOROPLAST IMPORT APPARATUS/CIA-LIKE [161,162]).
Miscellaneous genes.
One-third of M4 hub genes and half of M3 hub genes, encoded miscellaneous enzymes, proteins of unknown function, or poorly documented proteins (S12 Fig). Based on our current knowledge, it is not yet possible to associate these genes with a specific process/pathway. Among them, Potri.008G194800, a DUF1338 protein encoded by a M4 hub, is orthologous to FLO7 in rice, AT1G07040 in Arabidopsis and DD15 in soybean, respectively involved in amyloplast development, lipid droplet organisation or nodule senescence [163–165]. AT1G07040 encodes a hydroxyglutarate synthase acting in the lysine catabolic pathway [164]. Supporting its potential role in ABA responses across species, DUF1338 encoding transcripts accumulated in response to drought in poplar, Arabidopsis, rice, barley, and brome [73,91]. Although these miscellaneous genes are not functionally linked to the previously mentioned genes, their high intramodular connectivity indicates that they are strongly co-expressed in roots with well-known ABA-dependent responses.
Discussion
ABA orchestrates a myriad of transcriptional and post-transcriptional events during environmental changes and plant development [3,4,41]. Here, we investigated transcriptome remodelling in poplar root tips during artificial modulation of ABA homeostasis. Our biological material and experimental set-up allowed to apply short-term exogenous ABA treatments with a fine temporal resolution while controlling for well-known confounding effects in root transcriptomics. Co-expression network analysis of time-resolved transcriptomics revealed that both up- and down-regulation of gene expression contributed to the ABA responses and followed similar temporal dynamics. Both transcriptional activation and inactivation shaped the early response and subsequent transcriptome remodelling. The properties of ABA-activated modules were more robustly preserved under osmotic stress than those of ABA-inactivated modules. This may explain the emphasis placed on ABA-activated transcriptional responses in the drought literature. By considering activation and repression with equal interest, we identified the hallmarks of primary ABA responses. In Fig 7, we propose a working model integrating the processes most likely responding to ABA, their putative molecular architecture, and their regulatory links.
ABA primary responses involve the coordinated activation and repression of multiple genes, organized into co-expressed modules. These responses are rapidly triggered, showing that the ABA-dependent transcriptional network is structured over time. Our working model provides a comprehensive view of early ABA-dependent processes, their functional links and their putative regulatory architecture. ABFs and ABI5 transcription factors mediate ABA signal, and act as central nodes in ABA regulatory network. In our model, ABF/ABI5 pathway recruits SNAC and GBF transcription factors, amplifying and diversifying transcriptional responses. We propose that ABA-dependent transcriptional repression of several transcription factors, including MYB, bHLH and GLK, is central to ABA regulatory network. They coordinate and reshape downstream ABA responses, with the GLK-BCM pathway explaining the overlap between leaf senescence signalling pathway and ABA responses in the root transcriptome. For clarity, only a subset of genes is shown. The model is supported by S2-S12 Figs.
Following exogenous ABA treatment, core ABA signalling components and downstream ABA-dependent phosphorylation cascade were activated, consistent with their previously demonstrated functions in poplar [38,44,45]. Similar to Arabidopsis, ABI5 and ABF pathways co-operated to mediate the ABA signal across the transcriptome, triggering the transcription of specific targets [50]. ABA-dependent transcriptional regulation involves feedback loops controlling ABA action [3,10,72], and we showed that these loops, including those involving ATHB7/ATHB12, AFPs, AITRs and ABT, are conserved and transcriptionally active in poplar. Lineage-specific responses were also observed, reflecting paralog gain, loss, and neo- or sub-functionalization following genome expansion after speciation [42,166,167]. For instance, while E3 ligase CHYR1 is a single copy gene and ABA-activated in Arabidopsis [85], its poplar paralogs, but not its ortholog, were transcriptionally activated by ABA, suggesting evolutionary divergence in promoter sequences. WGCNA identified a module of co-expressed genes rapidly induced by ABA, whose expression was sustained during continuous treatment and declined upon ABA withdrawal (M4). Notably, this ABA-activated module included genes encoding the core ABA signalling component (HAB) and regulatory feedback loops as well as genes involved in ABA metabolism, cell rescue, osmosensing, redox homeostasis, lipid droplet metabolism, water movement, and nitrate transport. These ABA responses functionally overlap with the transcriptional responses to water deficit in Arabidopsis and poplar [62,73,74,90,92], suggesting that evolutionary forces have conserved the coupling of ABA signal regulation and stress responses at the transcriptional level, even in the context of genome expansion.
In addition to ABA activation, ABA triggered widespread transcriptional inactivation. WGCNA identified a module of co-expressed genes whose expression was rapidly repressed and remained low as long as the ABA signal was active (M3). ABA inactivated a wide range of transporters, metabolic pathways, and transcription factors. To the best of our knowledge, the regulatory architecture underlying most of this ABA-mediated transcriptional repression has not yet been elucidated. Here, we showed that the ABA-inactivated module contained a suite of photosystem- and tetrapyrrole-related genes as well as the master regulators of leaf senescence, the GLK–BCM pathway (Fig 7). In Arabidopsis, GLK1 is essential for chloroplast maintenance, for the expression of photosystem- and tetrapyrrole-related genes, and for the activation of GUN4 and BCMs [135,136,160]. GLK1 overexpression induces chloroplast development in Arabidopsis roots [168]. BCM promotes chlorophyll synthesis by interacting with GUN4 and inhibits chlorophyll degradation by destabilizing the Mg-dechelatase SGR1 [135]. Reduced GLK1 activity initiates leaf senescence by down-regulating BCM expression, thereby promoting chlorophyll breakdown [136,160,169]. During leaf development, GLK1 is post-transcriptionally repressed by RPGE2 and transcriptionally activated or repressed by CIA2/CIL and ATAF1, respectively [137,146,162]. ATAF1 binds the NCED3 promoter, activates NCED3 expression, and increases ABA biosynthesis, forming a feedback loop that represses GLK1 expression [146,160,170]. In poplar, PtATAF1.1/PtATAF1.2, PtNCED3, and PtNCED3-1 were activated by ABA while GLK homologs were repressed, suggesting that these regulatory loops are conserved. SNAC transcription factors also transcriptionally up-regulate SGR1, which promotes chlorophyll degradation during senescence [171,172]. PtRD26 directly targets several senescence-associated genes, including PtSGR1 [142]. In Arabidopsis, SGR1 is directly activated by ABFs and ANAC016 [1,173]. Together, these findings show that ABA-dependent transcriptional regulation extends to key molecular regulators of leaf senescence in poplar roots. GLK expression is also strongly down-regulated under drought in poplar roots [74]. While the ABA signal is likely to account for these transcriptional patterns, it also raises intriguing questions about the specificity of processes typically associated with leaf senescence, but occurring in roots.
ABA decreased the expression of genes in several metabolic pathways, including flavonoid and anthocyanin biosynthesis. In Arabidopsis seedlings, GLK1 is a positive regulator of anthocyanin accumulation and the loss of glk1 function leads to reduced anthocyanin synthesis [174]. Consequently, ABA-dependent transcriptional repression of GLK homologs could inactivate the flavonoid/anthocyanin biosynthesis pathway in roots. Further functional studies are required to determine whether poplar GLKs directly regulate biosynthetic genes or act through MWB complex regulators, as previously proposed [174,175]. In poplar, ABA antagonised several regulators of the flavonoid/anthocyanin pathway, including well-known MYB repressors such as LTF15 and MYC2 homologs [40,119–124]. ABA-activated SNACs, PtNAC049/PtNAC078, may act as master repressors, consistent with the role of Arabidopsis ANAC032, which inhibits the anthocyanin/flavonoid pathway by down-regulating the expression of both biosynthetic and regulatory genes [176,177]. Furthermore, ABA selectively inactivated the MEP pathway and downstream hormone biosynthesis pathways, while promoting its own biosynthesis. These transcriptional patterns suggest a coordinated regulation of the isoprenoid pathway towards ABA synthesis, potentially reinforced by yet unexplored mechanisms diverting violaxanthin from the ketocarotenoid biosynthesis to ABA biosynthesis. In Arabidopsis leaves, GLKs act as transcriptional regulators that activate the transcription of GBF targets, including deoxy-D-xylulose 5-phosphate synthase, phytoene synthase, and β-carotene hydroxylase 2, as well as genes required for chloroplast function [178]. In poplar roots, ABA repressed GLKs, activated GBFs, and differentially regulated their putative targets. For instance, the multiple poplar PSY paralogs exhibited distinct ABA responses, suggesting that paralogy diversified trans-regulatory processes. However, these transcriptional patterns also support the long-standing hypothesis that transcriptional control orchestrates the secondary metabolites in an evolutionary conserved and lineage-specific manner [178–180]. Carotenoid and chlorophyll biosynthesis are also under the post-transcriptional control of OR family proteins in chloroplasts [181], suggesting that both transcriptional and post-transcriptional regulation co-occurred for secondary metabolism.
ABA extensively rewired the root transcriptional landscape and remodelled several molecular processes (Fig 7). A key finding is that ABA signalling rapidly engaged coordinated regulatory pathways and functional responses previously associated with processes such as germination, stress responses, metabolic regulation, and leaf senescence, although these processes have not previously been reported to share common regulatory features. ABA-activated responses are likely to contribute to cell protection against damage, while ABA-repressed responses are mainly involved transport and secondary metabolic pathways, whose importance in root functioning remain to be investigated. Among these, the transcriptional inactivation of the tetrapyrrole and chlorophyll metabolism in roots is functionally intriguing. In Arabidopsis cells, ABA-driven inhibition of the chlorophyll branch in the tetrapyrrole pathway has been proposed to redirect protoporphyrin IX towards the heme biosynthesis [110]. ABA-driven increases in heme levels may contribute to cell protection against oxidative stress. An antioxidant molecule (biliverdin) can be produced from the degradation of heme in plastid [110]. Additionally, heme may be exported from plastids and incorporated into heme-binding proteins, including enzymes involved in ROS detoxification (catalase or peroxidase) but also other hemoproteins such as ABA-8’-hydroxylases (CYP707A) involved in ABA catabolism and homeostasis [182]. Based on the proposed working model, we hypothesize that cost-benefit trade-offs could have shaped ABA responses in roots, repressing the expression of accessory, yet costly, metabolic processes and directing gene expression changes towards adaptive responses.
Many other genes were identified as early ABA-responsive, yet they were not included in our working model due to limited knowledge of their functions. ABA is known to modulate transcriptional networks that control root development. In Arabidopsis, the SHR–SCR–SCL23 interaction specifies cell fate in the root meristem [183]. However, gene duplication diversified poplar SHR functions: PtSHR1 retains canonical interaction with PtSCR, whereas PtSHR2B influences phellogen activity, reduces tree growth and alters stem anatomy [184,185]. Meanwhile, PtSHR2B was expressed at the root tip [185]. Here, the co-repression of PtSHR2B and PtGRAS39 in ABA-treated roots is particularly intriguing, and highlights an attractive target for future research. Overall, the comprehensive view of the early transcriptional landscape of ABA responses in poplar roots highlights primary targets of ABA, some lineage-specific innovations in the poplar genome, and core ABA-dependent responses encompassing both repression and activation.
Materials and methods
Woody cuttings (Populus nigra clone 6J29) were surface sterilized with hydrogen peroxide [186], and individually set in hydroponic tanks containing 18L of modified Hoagland solution (NS) [187]. To prevent light exposure, the culture system was opaque and placed in a dark room (mean air temperature: 23.7°C ± 0.9°C). The solution was aerated to provide sufficient oxygen to the roots and buds were removed to prevent leaf appearance. This experimental set-up therefore controlled for well-known confounding effects in time-resolved transcriptomic studies of roots, namely light and its impact on shoot-to-root communication. After one week, NS was renewed and rooted cuttings were randomly assigned to treatments. For CTL, fresh NS was supplemented with 238 µL ethanol 96%. A 150 mM stock solution of ABA was prepared in ethanol 96% (Sigma-Aldrich) and used to prepare 2 µM ABA-enriched NS (ABA+). Treatments were applied by carefully transferring the cuttings either into ABA+ for 2, 5, 15, 30, 60, 180, 480, or 960 min or into CTL. For ABA withdrawal modality (ABA+/-), cuttings were transferred twice as they were pretreated with ABA+ for 180 min prior being transferred into CTL for 60, 180, 240, or 960 min. Harvests were arranged according to a full randomized design. Two independent experiments (Exp1 and Exp2 with two overlapping treatment time points, Fig 2A) were performed under the same growth conditions in July and September 2018, respectively.
Hormone profiling
Hormone profiling was performed in Exp1 under ABA-free conditions (CTL) and under ABA+ for 5, 30, 60, or 180 min. 2-cm long root tips were collected, rinsed rapidly in distilled water, surface-dried, flash-frozen in liquid nitrogen and freeze-dried. Each of the four replicates consisted in 25 roots collected on 12 cuttings. Endogenous hormone contents and their metabolites were determined from 3 mg dry weight according to the method described by Šimura et al. [188] and previously used on poplar roots [189]. Analyte concentrations were calculated as ratios of non-labelled compounds to labelled internal standards or closely eluting stable isotope-labelled tracers using the standard isotope dilution method. The endogenous hormone contents were compared in ABA+-treated roots (for 5 min, 30 min, 60 min, and 180 min) versus CTL roots (P-value ≤ 0.05, Student's t-test). The covariation of hormone contents across samples was determined using Pearson correlation coefficients (n = 20 samples, P ≤ 0.01).
RNA isolation, cDNA library construction and RNA-Sequencing
In Exp1 and Exp2, 2-cm long root tips were collected and immediately fixed in RNAlater (Sigma-Aldrich). All samples were stored at -80°C and further processed together in a single batch for RNA extraction as well as for downstream RNA-seq library preparation and sequencing. The two to four replicates per time point were obtained by extracting RNA from a pool of eight root tips using a Spectrum Plant Total RNA kit (Sigma-Aldrich,) complemented by a DNase I treatment. RNA quality was checked using a Fragment Analyzer 5200 (Agilent Technologies). For details of cDNA library construction please refer to the S1A Protocol. The 48 cDNA libraries were normalized, multiplexed, and sequenced on an Illumina HiSeq3000 at the GeT-Plage core facility (http://www.get.genotoul.fr). RNA sequencing generated about 80 million paired-ends reads of 150 bp per library (S8 Data). RNA-seq data were pre-processed as described in Royer et al. [62] using P. trichocarpa genome v4.1 [166,190]. For details of RNA sequencing processing please refer to the S1B Protocol.
Gene annotation and Functional enrichments
Gene annotation was retrieved from P. trichocarpa V4.1 annotation files (S3 Data). Functional categorization was achieved using Gene Ontology (GO) and MapMan resources [191,192]. For details, please refer to the S1C Protocol. Significantly enriched GO terms and MapMan bins, as well as iDREM path assignation, were examined through a Fisher’s exact test combined with an FDR correction for multiple testing (corrected P-value threshold of 0.05).
Expression analysis
Normalized read counts, differential expression and significance statistics were computed with Deseq2 v1.24.0 [193]. Normalized count table was reduced to loci showing a robust expression for at least one treatment (i.e., no less than 10 aligned reads in all the libraries documenting one treatment). Hierarchical clustering on Log2 (normalized count+1) revealed two outlier libraries which were not further considered (S8 Data). No batch correction was applied prior to analysis. The expression matrix consisted in 27,379 loci over 46 libraries.
The pattern of temporal expression was revealed using iDREM [56,194]. This software uses an input–output hidden Markov model to reconstruct regulatory networks while taking into account their dynamics [195]. iDREM model was inferred in the early part of the time series (Exp1: CTL and ABA+ for 2, 5, 15, 30, and 60 min), using default parameter, selecting Spot ID and allowing transformation of the time series to start at zero for CTL and Log-transformation of expression. iDREM was run using mean expression value per time points for the 1,914 loci differentially expressed in response to ABA at least at one of time point in comparison with CTL (absolute Log2-transformed fold-change expression ≥ 1 and a P-value ≤ 0.01, Wald test P-values corrected for multiple testing by applying FDR correction). iDREM filtered loci if the maximum absolute difference between transformed expression values of any two time points was less than one. The reconstructed temporal map described the groups of co-expressed loci as paths and highlighted the nodes where paths split. Additionally, we constructed an iDREM-derived gene regulatory network (GRN) for the putative transcription factors predicted to regulate transition events on the iDREM temporal map (split cut-off score < 0.0001). This GRN was inferred from the prediction of transcription factor-gene interaction, the temporal expression patterns of target genes, and the path transition events regulated by each transcription factor (S1 Fig and S2 Data). iDREM scored transcription factors with splits and paths based on both gene assignments and prediction about transcription factor-gene interaction in P. trichocarpa V3.0 genome (released on 2019-07, plantregmap.gao-lab.org).
Co-expression networks
Co-expression network analysis was performed using the R package WGCNA v1.70 [59,196]. The expression matrix (27,379 loci x 46 libraries, further referred to as the reference set) was Log2 (normalized count+1)-transformed. Scale-free topology was obtained by applying a soft power of 18 (S13 Fig). The network was constructed using the “blockwiseModules” function (parameter: signed network type, bicor, power = 18, maxBlockSize = 15,000, minModuleSize = 150, reassignThreshold = 1e-6, mergeCutHeight = 0.25, maxPOutliers = 0.05, and deepSplit = 3). WGCNA identifies modules of highly correlated loci and provides module eigengene (ME), which is the first principal component of each module. Module-based biological significance was computed as the value of the correlation between the mean ME values and the mean hormone levels over five time points in Exp1 (CTL and ABA+ for 5, 30, 60, and 180 min). Intramodular connectivity (Kin), which is the sum of a locus’ connection strengths with all other loci in the module, was computed using “intramodularConnectivity.fromExpr” function and scaled Kin by calling scaleByMax option. We selected as hubs any loci that exhibited a scaled Kin ≥ 0.75. The topological overlap (TO), which is the similarities of two loci relationships with all other loci in the network [60], was computed using “vectorTOM” function and hubs as input against the whole expression matrix.
Hub-hub correlation network was built using the hub expression matrix as input for R package MEGENA v1.4.1 [197]. Pearson correlation matrix was summarized into MEGENA-derived Planar Filtered Network, using default parameters. Networks were visualized using the R package igraph, v1.2.4.1 [198].
Module preservation in osmotically stressed samples
We re-analysed 32 libraries documenting CTL and osmotically-stressed root tips (−0.35 MPa, for 30 min or 180 min) in the division zone and the elongation zone of P. nigra roots (SRA: SRP067564 [62]). Taking advantage from paired-sampling procedure, the root transcriptome was built by concatenating the FastQ files from the two root zones. The resulting 16 libraries were processed as described above. Normalized and Log2 (normalized count+1)-transformed expression matrices from reference and test sets were embedded into a multi-experiment dataset (24,962 loci x 46 and 16 libraries).
WGCNA assesses the module robustness and replicability across datasets [61]. Preservation statistics were computed using “modulePreservation” function (networkType = “signed,” corFnc = “bicor,” maxModuleSize = 5000, nPermutations = 200). Zsummary, a composite statistic based on density and connectivity metrics, and Zmean.Adj, a module density-based metrics were extracted. A high Zsummary (>10) reveals a strong module structure preservation among the two sets while Zmean.Adj denotes how much module-embedded loci remain connected in the test set.
Modular structure preservation was assessed using DGCA, a bottom-up approach based on differential gene correlation analysis [63]. DGCA quantifies the modular average differential connectivity, MeDC, as the mean change in z-score difference among all module-embedded locus pairs between reference and test sets, and assesses its statistical significance using empirical P-values (moduleDC DGCA function, nperm = 75). Negative MeDC values indicated a loss of connectivity for the module in the test set with respect to the reference. Differential pairwise relationships were computed using the DGCA functions “getCors”, “pairwiseDCor”, and “dcTopPairs” with default parameters (except for: adjust = “fdr,” corSigThresh = 0.001).
Statistical analyses
Statistical analyses and plots were performed using R 3.6.1 (R Development Core Team, http://www.r-project.org/).
Data statement
Reads were deposited in the Sequence Read Archive (SRA) under the Study Accession number GSE183780. The expression data are provided in S3 Data. Identifier of genes mentioned in the main text are given in S2 Appendix.
Supporting information
S1 Fig. ABA-dependent gene regulatory network predicted in poplar roots (iDREM-derived GRN).
The first layer consists of transcription factors identified by the iDREM regulatory map as the most likely regulators of path transition events. The second layer consists of their target genes within each iDREM path. Node colors denote (A) iDREM paths, (B) WGCNA modules, (C) differentially expressed genes under osmotic stress in comparison with CTL (absolute Log2-transformed fold-change expression ≥ 1 and a P-value ≤ 0.01, Wald test P-values corrected for multiple testing by applying FDR correction).
https://doi.org/10.1371/journal.pgen.1012256.s001
(PDF)
S2 Fig. Genes encoding putative post-transcriptional regulators acting on the core ABA signalling components.
https://doi.org/10.1371/journal.pgen.1012256.s002
(PDF)
S3 Fig. M4 genes encoding proteins putatively involved in (A) cell protection and (B) raffinose family oligosaccharides metabolism.
(C) M4 genes encoding putative LATE EMBRYOGENESIS ABUNDANT (LEA) proteins. (D) Genes encoding putative lipid droplet associated proteins. (E) Further genes contributing to M4 enrichment in “Bin 3: minor CHO metabolism”.
https://doi.org/10.1371/journal.pgen.1012256.s003
(PDF)
S4 Fig. Genes encoding proteins involved in (A) the methylerythritol phosphate (MEP) pathway, in (B) the branch points of the MEP pathway, in (C) carotenoid metabolism, in (D) the conversion of β-carotene to zeaxanthin and ABA metabolism, and in (E) lutein and capsanthin synthesis.
The $ sign indicates that the AAO3 function linked to ABA has been questioned in poplar [42].
https://doi.org/10.1371/journal.pgen.1012256.s004
(PDF)
S5 Fig. Genes encoding proteins involved in (A) strigolactone synthesis and in (B) gibberellin metabolism.
https://doi.org/10.1371/journal.pgen.1012256.s005
(PDF)
S6 Fig. Genes encoding proteins involved in (A) cytokinin metabolism and (B) cytokinin signalling.
https://doi.org/10.1371/journal.pgen.1012256.s006
(PDF)
S7 Fig. Genes encoding putative transporters and channels.
(A) Genes encoding putative nitrate transporters. The $ sign indicates putative ABA transporter of the PtNPF family. (B) Genes encoding other putative ABA transporters of the AWPM-19, ABCG and DTX families. (C) Aquaporin and (D) OSCA family. (E) Genes related to anthocyanin/flavonoid transport. (F) Further M3 and M4 hub genes encoding miscellaneous transporters.
https://doi.org/10.1371/journal.pgen.1012256.s007
(PDF)
S8 Fig. Genes encoding proteins putatively involved in (A) nitrogen metabolism and in (B) oxidative pentose phosphate (OPP) pathway.
https://doi.org/10.1371/journal.pgen.1012256.s008
(PDF)
S9 Fig. Genes encoding proteins putatively involved in (A-B) flavonoid metabolism.
(B) Further genes contributing to M3 enrichment in “Bin16.8: secondary metabolism.flavonoids”. (C-F) Genes related to the regulation of anthocyanin/proanthocyanidin synthesis. R2R3-MYB transcription factors acting as (C) activators and (D) repressors. The $ sign indicates R2R3-MYB Subfamily S4 (Du et al. 2015). Genes encoding other putative members of the regulatory MBW complex belonging to (E) WD40 repeat family protein and (F) bHLH transcription factor family.
https://doi.org/10.1371/journal.pgen.1012256.s009
(PDF)
S10 Fig. Genes contributing to M3 enrichments in (A) “Bin1: photosystem” and (B) “Bin19: tetrapyrrole synthesis”.
(C) Poplar homologs and orthologs of key regulatory genes acting on chlorophyll homeostasis in Arabidopsis and SAG201 homologs.
https://doi.org/10.1371/journal.pgen.1012256.s010
(PDF)
S11 Fig. Genes encoding further transcription factors (A) predicted by iDREM or putatively involved in ABA signal amplification and diversification, including (B) Stress-responsive NAC (SNAC) and other NAC transcription factors, (C) R2R3-MYBs and (D) transcription factors from miscellaneous families.
https://doi.org/10.1371/journal.pgen.1012256.s011
(PDF)
S12 Fig. M4 and M3 hub genes not currently associated with a specific pathway or process, or encoding proteins of unknown function.
(A) Hub genes encoding miscellaneous enzymes and/or previously named. (B) Further hub genes with miscellaneous motif and domains.
https://doi.org/10.1371/journal.pgen.1012256.s012
(PDF)
S2 Data. IDREM analysis, including iDREM-derived GRN.
https://doi.org/10.1371/journal.pgen.1012256.s015
(XLSX)
S3 Data. Loci annotation, expression and differential expression and WGCNA information (module assignation, loci connectivity and ME x sample matrix).
https://doi.org/10.1371/journal.pgen.1012256.s016
(XLSX)
S4 Data. Gene Ontology enrichment for co-expressed unigenes per module.
https://doi.org/10.1371/journal.pgen.1012256.s017
(XLSX)
S5 Data. MapMan Bin enrichment for co-expressed unigenes per module.
https://doi.org/10.1371/journal.pgen.1012256.s018
(XLSX)
S1 Protocol. (A) cDNA library construction, (B) RNA sequencing processing, and (C) Gene Ontology and Mapman bin.
https://doi.org/10.1371/journal.pgen.1012256.s022
(DOCX)
S1 Appendix. Additional information about transporters and channels.
https://doi.org/10.1371/journal.pgen.1012256.s023
(DOCX)
S2 Appendix. Genes identifiers mentioned in the main text and Appendix.
https://doi.org/10.1371/journal.pgen.1012256.s024
(DOCX)
Acknowledgments
The authors thank Cyril Buré for the design of hydroponic devices and his excellent technical support. The authors acknowledge Emmanuelle Morin and Laure Fauchery for technical support in bioinformatics and laboratory analysis, respectively. The authors warmly thank Oliver Brendel and Didier Le Thiec for their helpful comments. The authors acknowledge Olivier Forestier from Guéméné-Penfao ONF Nursery for preparation of the cuttings.
References
- 1. Gao S, Gao J, Zhu X, Song Y, Li Z, Ren G, et al. ABF2, ABF3, and ABF4 Promote ABA-Mediated Chlorophyll Degradation and Leaf Senescence by Transcriptional Activation of Chlorophyll Catabolic Genes and Senescence-Associated Genes in Arabidopsis. Mol Plant. 2016;9(9):1272–85. pmid:27373216
- 2.
Seo M, Marion-Poll A. Abscisic acid metabolism and transport. Elsevier. 2019. https://doi.org/10.1016/bs.abr.2019.04.004
- 3. Yoshida T, Christmann A, Yamaguchi-Shinozaki K, Grill E, Fernie AR. Revisiting the Basal Role of ABA - Roles Outside of Stress. Trends Plant Sci. 2019;24(7):625–35. pmid:31153771
- 4. Sajeev N, Koornneef M, Bentsink L. A commitment for life: Decades of unraveling the molecular mechanisms behind seed dormancy and germination. Plant Cell. 2024;36(5):1358–76. pmid:38215009
- 5. Verslues PE, Zhu J-K. Before and beyond ABA: upstream sensing and internal signals that determine ABA accumulation and response under abiotic stress. Biochem Soc Trans. 2005;33(Pt 2):375–9. pmid:15787610
- 6. Yamaguchi-Shinozaki K, Shinozaki K. Transcriptional regulatory networks in cellular responses and tolerance to dehydration and cold stresses. Annu Rev Plant Biol. 2006;57:781–803. pmid:16669782
- 7. Dietrich D, Pang L, Kobayashi A, Fozard JA, Boudolf V, Bhosale R, et al. Root hydrotropism is controlled via a cortex-specific growth mechanism. Nat Plants. 2017;3:17057. pmid:28481327
- 8. Yang Y, Li H-G, Wang J, Wang H-L, He F, Su Y, et al. ABF3 enhances drought tolerance via promoting ABA-induced stomatal closure by directly regulating ADF5 in Populus euphratica. J Exp Bot. 2020;71(22):7270–85. pmid:32822499
- 9. Xiong L, Zhu J-K. Regulation of abscisic acid biosynthesis. Plant Physiol. 2003;133(1):29–36. pmid:12970472
- 10. Finkelstein R. Abscisic Acid synthesis and response. Arabidopsis Book. 2013;11:e0166. pmid:24273463
- 11. Vranová E, Coman D, Gruissem W. Network analysis of the MVA and MEP pathways for isoprenoid synthesis. Annu Rev Plant Biol. 2013;64:665–700. pmid:23451776
- 12. Seo M, Peeters AJ, Koiwai H, Oritani T, Marion-Poll A, Zeevaart JA, et al. The Arabidopsis aldehyde oxidase 3 (AAO3) gene product catalyzes the final step in abscisic acid biosynthesis in leaves. Proc Natl Acad Sci U S A. 2000;97(23):12908–13. pmid:11050171
- 13. Qin X, Zeevaart JA. The 9-cis-epoxycarotenoid cleavage reaction is the key regulatory step of abscisic acid biosynthesis in water-stressed bean. Proc Natl Acad Sci U S A. 1999;96(26):15354–61. pmid:10611388
- 14. Nambara E, Marion-Poll A. Abscisic acid biosynthesis and catabolism. Annu Rev Plant Biol. 2005;56:165–85. pmid:15862093
- 15. Krochko J, Abrams G, Loewen M, Abrams S, Cutler A. (+)-Abscisic acid 8’-hydroxylase is a cytochrome P450 monooxygenase. Plant Physiol. 1998;118(3):849–60. pmid:9808729
- 16. Saito S, Hirai N, Matsumoto C, Ohigashi H, Ohta D, Sakata K, et al. Arabidopsis CYP707As encode (+)-abscisic acid 8’-hydroxylase, a key enzyme in the oxidative catabolism of abscisic acid. Plant Physiol. 2004;134(4):1439–49. pmid:15064374
- 17. Weng J-K, Ye M, Li B, Noel JP. Co-evolution of Hormone Metabolism and Signaling Networks Expands Plant Adaptive Plasticity. Cell. 2016;166(4):881–93. pmid:27518563
- 18. Yin X, Bai Y-L, Ye T, Yu M, Wu Y, Feng Y-Q. Cinnamoyl coA: NADP oxidoreductase-like 1 regulates abscisic acid response by modulating phaseic acid homeostasis in Arabidopsis thaliana. J Exp Bot. 2022;73(3):860–72. pmid:34718526
- 19. Bai Y-L, Yin X, Xiong C-F, Cai B-D, Wu Y, Zhang X-Y, et al. Neophaseic acid catabolism in the 9’-hydroxylation pathway of abscisic acid in Arabidopsis thaliana. Plant Commun. 2022;3(5):100340. pmid:35585783
- 20. Kang J, Hwang J-U, Lee M, Kim Y-Y, Assmann SM, Martinoia E, et al. PDR-type ABC transporter mediates cellular uptake of the phytohormone abscisic acid. Proc Natl Acad Sci U S A. 2010;107(5):2355–60. pmid:20133880
- 21. Boursiac Y, Léran S, Corratgé-Faillie C, Gojon A, Krouk G, Lacombe B. ABA transport and transporters. Trends Plant Sci. 2013;18(6):325–33. pmid:23453706
- 22. Zhang H, Zhu H, Pan Y, Yu Y, Luan S, Li L. A DTX/MATE-type transporter facilitates abscisic acid efflux and modulates ABA sensitivity and drought tolerance in Arabidopsis. Mol Plant. 2014;7(10):1522–32. pmid:24851876
- 23. Kuromori T, Shinozaki K. ABA transport factors found in Arabidopsis ABC transporters. Plant Signal Behav. 2010;5(9):1124–6. pmid:20935463
- 24. Kanno Y, Hanada A, Chiba Y, Ichikawa T, Nakazawa M, Matsui M, et al. Identification of an abscisic acid transporter by functional screening using the receptor complex as a sensor. Proc Natl Acad Sci U S A. 2012;109(24):9653–8. pmid:22645333
- 25. Kang J, Yim S, Choi H, Kim A, Lee KP, Lopez-Molina L, et al. Abscisic acid transporters cooperate to control seed germination. Nat Commun. 2015;6:8113. pmid:26334616
- 26. Léran S, Noguero M, Corratgé-Faillie C, Boursiac Y, Brachet C, Lacombe B. Functional characterization of the arabidopsis abscisic acid transporters NPF4.5 and NPF4.6 in xenopus oocytes. Front Plant Sci. 2020;11:144.
- 27. Yao L, Cheng X, Gu Z, Huang W, Li S, Wang L, et al. The AWPM-19 Family Protein OsPM1 Mediates Abscisic Acid Influx and Drought Response in Rice. Plant Cell. 2018;30(6):1258–76. pmid:29716991
- 28. Cutler SR, Rodriguez PL, Finkelstein RR, Abrams SR. Abscisic acid: emergence of a core signaling network. Annu Rev Plant Biol. 2010;61:651–79. pmid:20192755
- 29. Umezawa T, Nakashima K, Miyakawa T, Kuromori T, Tanokura M, Shinozaki K, et al. Molecular basis of the core regulatory network in ABA responses: sensing, signaling and transport. Plant Cell Physiol. 2010;51(11):1821–39. pmid:20980270
- 30. Yoshida T, Mogami J, Yamaguchi-Shinozaki K. Omics Approaches Toward Defining the Comprehensive Abscisic Acid Signaling Network in Plants. Plant Cell Physiol. 2015;56(6):1043–52. pmid:25917608
- 31. Hsu P-K, Dubeaux G, Takahashi Y, Schroeder JI. Signaling mechanisms in abscisic acid-mediated stomatal closure. Plant J. 2021;105(2):307–21. pmid:33145840
- 32. Park S-Y, Fung P, Nishimura N, Jensen DR, Fujii H, Zhao Y, et al. Abscisic acid inhibits type 2C protein phosphatases via the PYR/PYL family of START proteins. Science. 2009;324(5930):1068–71. pmid:19407142
- 33. Soon F-F, Ng L-M, Zhou XE, West GM, Kovach A, Tan MHE, et al. Molecular mimicry regulates ABA signaling by SnRK2 kinases and PP2C phosphatases. Science. 2012;335(6064):85–8. pmid:22116026
- 34. Yoshida R, Hobo T, Ichimura K, Mizoguchi T, Takahashi F, Aronso J, et al. ABA-activated SnRK2 protein kinase is required for dehydration stress signaling in Arabidopsis. Plant Cell Physiol. 2002;43(12):1473–83. pmid:12514244
- 35. Hubbard KE, Nishimura N, Hitomi K, Getzoff ED, Schroeder JI. Early abscisic acid signal transduction mechanisms: newly discovered components and newly emerging questions. Genes Dev. 2010;24(16):1695–708. pmid:20713515
- 36. Vlad F, Droillard M-J, Valot B, Khafif M, Rodrigues A, Brault M, et al. Phospho-site mapping, genetic and in planta activation studies reveal key aspects of the different phosphorylation mechanisms involved in activation of SnRK2s. Plant J. 2010;63(5):778–90. pmid:20561261
- 37. Wang P, Xue L, Batelli G, Lee S, Hou Y-J, Van Oosten MJ, et al. Quantitative phosphoproteomics identifies SnRK2 protein kinase substrates and reveals the effectors of abscisic acid action. Proc Natl Acad Sci U S A. 2013;110(27):11205–10. pmid:23776212
- 38. Song X, Ohtani M, Hori C, Takebayasi A, Hiroyama R, Rejab NA, et al. Physical interaction between SnRK2 and PP2C is conserved in <i>Populus trichocarpa</i>. Plant Biotechnology. 2015;32(4):337–41.
- 39. Fujita Y, Yoshida T, Yamaguchi-Shinozaki K. Pivotal role of the AREB/ABF-SnRK2 pathway in ABRE-mediated transcription in response to osmotic stress in plants. Physiol Plant. 2013;147(1):15–27. pmid:22519646
- 40. Wang X, Guo C, Peng J, Li C, Wan F, Zhang S, et al. ABRE-BINDING FACTORS play a role in the feedback regulation of ABA signaling by mediating rapid ABA induction of ABA co-receptor genes. New Phytol. 2019;221(1):341–55. pmid:30019753
- 41. Song L, Huang S-SC, Wise A, Castanon R, Nery JR, Chen H, et al. A transcription factor hierarchy defines an environmental stress response network. Science. 2016;354(6312):aag1550. pmid:27811239
- 42. Hanada K, Hase T, Toyoda T, Shinozaki K, Okamoto M. Origin and evolution of genes related to ABA metabolism and its signaling pathways. J Plant Res. 2011;124(4):455–65. pmid:21626211
- 43. Komatsu K, Takezawa D, Sakata Y. Decoding ABA and osmostress signalling in plants from an evolutionary point of view. Plant Cell Environ. 2020;43(12):2894–911. pmid:33459424
- 44. Papacek M, Christmann A, Grill E. Interaction network of ABA receptors in grey poplar. Plant J. 2017;92(2):199–210. pmid:28746755
- 45. Rigoulot SB, Petzold HE, Williams SP, Brunner AM, Beers EP. Populus trichocarpa clade A PP2C protein phosphatases: their stress-induced expression patterns, interactions in core abscisic acid signaling, and potential for regulation of growth and development. Plant Mol Biol. 2019;100(3):303–17. pmid:30945147
- 46. Himmelbach A, Yang Y, Grill E. Relay and control of abscisic acid signaling. Curr Opin Plant Biol. 2003;6(5):470–9. pmid:12972048
- 47. Hoth S, Morgante M, Sanchez J-P, Hanafey MK, Tingey SV, Chua N-H. Genome-wide gene expression profiling in Arabidopsis thaliana reveals new targets of abscisic acid and largely impaired gene regulation in the abi1-1 mutant. J Cell Sci. 2002;115(Pt 24):4891–900. pmid:12432076
- 48. Seki M, Ishida J, Narusaka M, Fujita M, Nanjo T, Umezawa T, et al. Monitoring the expression pattern of around 7,000 Arabidopsis genes under ABA treatments using a full-length cDNA microarray. Funct Integr Genomics. 2002;2(6):282–91. pmid:12444421
- 49. Choudhury A, Lahiri A. Comparative analysis of abscisic acid-regulated transcriptomes in Arabidopsis. Plant Biol (Stuttg). 2011;13(1):28–35. pmid:21143722
- 50. Lumba S, Toh S, Handfield L-F, Swan M, Liu R, Youn J-Y, et al. A mesoscale abscisic acid hormone interactome reveals a dynamic signaling landscape in Arabidopsis. Dev Cell. 2014;29(3):360–72. pmid:24823379
- 51. Shinozaki K, Yamaguchi-Shinozaki K. Gene Expression and Signal Transduction in Water-Stress Response. Plant Physiol. 1997;115(2):327–34. pmid:12223810
- 52. Cuming AC, Cho SH, Kamisugi Y, Graham H, Quatrano RS. Microarray analysis of transcriptional responses to abscisic acid and osmotic, salt, and drought stress in the moss, Physcomitrella patens. New Phytol. 2007;176(2):275–87. pmid:17696978
- 53. Aerts N, Hickman R, Van Dijken AJH, Kaufmann M, Snoek BL, Pieterse CMJ, et al. Architecture and dynamics of the abscisic acid gene regulatory network. Plant J. 2024;119(5):2538–63. pmid:38949092
- 54. Youssef C, Bizet F, Bastien R, Legland D, Bogeat-Triboulot M-B, Hummel I. Quantitative dissection of variations in root growth rate: a matter of cell proliferation or of cell expansion?. J Exp Bot. 2018;69(21):5157–68. pmid:30053124
- 55. Bannoud F, Bellini C. Adventitious Rooting in Populus Species: Update and Perspectives. Front Plant Sci. 2021;12:668837. pmid:34093625
- 56. Ding J, Ahangari F, Espinoza CR, Chhabra D, Nicola T, Yan X, et al. Integrating multiomics longitudinal data to reconstruct networks underlying lung development. Am J Physiol Lung Cell Mol Physiol. 2019;317(5):L556–68. pmid:31432713
- 57. Yoshida T, Fujita Y, Sayama H, Kidokoro S, Maruyama K, Mizoi J, et al. AREB1, AREB2, and ABF3 are master transcription factors that cooperatively regulate ABRE-dependent ABA signaling involved in drought stress tolerance and require ABA for full activation. Plant J. 2010;61(4):672–85. pmid:19947981
- 58. Hsieh W-P, Hsieh H-L, Wu S-H. Arabidopsis bZIP16 transcription factor integrates light and hormone signaling pathways to regulate early seedling development. Plant Cell. 2012;24(10):3997–4011. pmid:23104829
- 59. Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008;9:559. pmid:19114008
- 60. Miller JA, Horvath S, Geschwind DH. Divergence of human and mouse brain transcriptome highlights Alzheimer disease pathways. Proc Natl Acad Sci U S A. 2010;107(28):12698–703. pmid:20616000
- 61. Langfelder P, Luo R, Oldham MC, Horvath S. Is my network module preserved and reproducible?. PLoS Comput Biol. 2011;7(1):e1001057. pmid:21283776
- 62. Royer M, Cohen D, Aubry N, Vendramin V, Scalabrin S, Cattonaro F, et al. The build-up of osmotic stress responses within the growing root apex using kinematics and RNA-sequencing. J Exp Bot. 2016;67(21):5961–73. pmid:27702994
- 63. McKenzie AT, Katsyv I, Song W-M, Wang M, Zhang B. DGCA: A comprehensive R package for Differential Gene Correlation Analysis. BMC Syst Biol. 2016;10(1):106. pmid:27846853
- 64. Tello-Ruiz MK, Naithani S, Gupta P, Olson A, Wei S, Preece J, et al. Gramene 2021: harnessing the power of comparative genomics and pathways for plant research. Nucleic Acids Res. 2021;49(D1):D1452–63. pmid:33170273
- 65. Kolberg L, Raudvere U, Kuzmin I, Adler P, Vilo J, Peterson H. g:Profiler-interoperable web service for functional enrichment analysis and gene identifier mapping (2023 update). Nucleic Acids Res. 2023;51(W1):W207–12. pmid:37144459
- 66. Garcia ME, Lynch T, Peeters J, Snowden C, Finkelstein R. A small plant-specific protein family of ABI five binding proteins (AFPs) regulates stress response in germinating Arabidopsis seeds and seedlings. Plant Mol Biol. 2008;67(6):643–58. pmid:18484180
- 67. Lynch TJ, Erickson BJ, Miller DR, Finkelstein RR. ABI5-binding proteins (AFPs) alter transcription of ABA-induced genes via a variety of interactions with chromatin modifiers. Plant Mol Biol. 2017;93(4–5):403–18. pmid:27942958
- 68. Zhou B, Wang L, Ji Z, Chen X, Sun X, Xu N, et al. The PagAFP2a-PagAREB1 Module Form a Negative Feedback Loop to Regulate Salt Tolerance in Populus. Plant Cell Environ. 2025;48(7):5198–216. pmid:40165385
- 69. Merlot S, Gosti F, Guerrier D, Vavasseur A, Giraudat J. The ABI1 and ABI2 protein phosphatases 2C act in a negative feedback regulatory loop of the abscisic acid signalling pathway: ABI2 is a negative regulator of ABA signalling. Plant J. 2001;25:295–303.
- 70. Santiago J, Rodrigues A, Saez A, Rubio S, Antoni R, Dupeux F, et al. Modulation of drought resistance by the abscisic acid receptor PYL5 through inhibition of clade A PP2Cs. Plant J. 2009;60(4):575–88. pmid:19624469
- 71. Vlad F, Rubio S, Rodrigues A, Sirichandra C, Belin C, Robert N, et al. Protein phosphatases 2C regulate the activation of the Snf1-related kinase OST1 by abscisic acid in Arabidopsis. Plant Cell. 2009;21(10):3170–84. pmid:19855047
- 72. Valdés AE, Overnäs E, Johansson H, Rada-Iglesias A, Engström P. The homeodomain-leucine zipper (HD-Zip) class I transcription factors ATHB7 and ATHB12 modulate abscisic acid signalling by regulating protein phosphatase 2C and abscisic acid receptor gene activities. Plant Mol Biol. 2012;80(4–5):405–18. pmid:22968620
- 73. Cohen D, Bogeat-Triboulot M-B, Tisserant E, Balzergue S, Martin-Magniette M-L, Lelandais G, et al. Comparative transcriptomics of drought responses in Populus: a meta-analysis of genome-wide expression profiling in mature leaves and root apices across two genotypes. BMC Genomics. 2010;11:630. pmid:21073700
- 74. Georgii E, Kugler K, Pfeifer M, Vanzo E, Block K, Domagalska MA. The systems architecture of molecular memory in poplar after abiotic stress. Plant Cell. 2019;31:346–67.
- 75. Tian H, Chen S, Yang W, Wang T, Zheng K, Wang Y, et al. A novel family of transcription factors conserved in angiosperms is required for ABA signalling. Plant Cell Environ. 2017;40(12):2958–71. pmid:28857190
- 76. Chen S, Zhang N, Zhou G, Hussain S, Ahmed S, Tian H, et al. Knockout of the entire family of AITR genes in Arabidopsis leads to enhanced drought and salinity tolerance without fitness costs. BMC Plant Biol. 2021;21(1):137. pmid:33726681
- 77. Wang Z, Ren Z, Cheng C, Wang T, Ji H, Zhao Y, et al. Counteraction of ABA-Mediated Inhibition of Seed Germination and Seedling Establishment by ABA Signaling Terminator in Arabidopsis. Mol Plant. 2020;13(9):1284–97. pmid:32619606
- 78. Ali A, Pardo JM, Yun D-J. ABAting the Response: A Novel ABA Signal Terminator that Disrupts the Hormone Co-receptor Complex. Mol Plant. 2020;13(9):1241–3. pmid:32717348
- 79. Li C, Li X, Deng Z, Song Y, Liu X, Tang XA, et al. EGR1 and EGR2 positively regulate plant ABA signaling by modulating the phosphorylation of SnRK2.2. New Phytol. 2024;241(4):1492–509. pmid:38095247
- 80. Tarnowski K, Klimecka M, Ciesielski A, Goch G, Kulik A, Fedak H, et al. Two SnRK2-Interacting Calcium Sensor Isoforms Negatively Regulate SnRK2 Activity by Different Mechanisms. Plant Physiol. 2020;182(2):1142–60. pmid:31699848
- 81. Sirko A, Wawrzyńska A, Brzywczy J, Sieńko M. Control of ABA Signaling and Crosstalk with Other Hormones by the Selective Degradation of Pathway Components. Int J Mol Sci. 2021;22(9):4638. pmid:33924944
- 82. Cho SK, Ryu MY, Seo DH, Kang BG, Kim WT. The Arabidopsis RING E3 ubiquitin ligase AtAIRP2 plays combinatory roles with AtAIRP1 in abscisic acid-mediated drought stress responses. Plant Physiol. 2011;157(4):2240–57. pmid:21969385
- 83. Zhao J, Zhao L, Zhang M, Zafar SA, Fang J, Li M, et al. Arabidopsis E3 Ubiquitin Ligases PUB22 and PUB23 Negatively Regulate Drought Tolerance by Targeting ABA Receptor PYL9 for Degradation. Int J Mol Sci. 2017;18(9):1841. pmid:28837065
- 84. Xia F-N, Zeng B, Liu H-S, Qi H, Xie L-J, Yu L-J, et al. SINAT E3 Ubiquitin Ligases Mediate FREE1 and VPS23A Degradation to Modulate Abscisic Acid Signaling. Plant Cell. 2020;32(10):3290–310. pmid:32753431
- 85. Ding S, Zhang B, Qin F. Arabidopsis RZFP34/CHYR1, a Ubiquitin E3 Ligase, Regulates Stomatal Movement and Drought Tolerance via SnRK2.6-Mediated Phosphorylation. Plant Cell. 2015;27(11):3228–44. pmid:26508764
- 86. Kim M-H, Cho J-S, Park E-J, Lee H, Choi Y-I, Bae E-K, et al. Overexpression of a Poplar RING-H2 Zinc Finger, Ptxerico, Confers Enhanced Drought Tolerance via Reduced Water Loss and Ion Leakage in Populus. Int J Mol Sci. 2020;21(24):9454. pmid:33322558
- 87. Tong S, Chen N, Wang D, Ai F, Liu B, Ren L, et al. The U-box E3 ubiquitin ligase PalPUB79 positively regulates ABA-dependent drought tolerance via ubiquitination of PalWRKY77 in Populus. Plant Biotechnol J. 2021;19(12):2561–75. pmid:34382303
- 88. Jiang Y, Tong S, Chen N, Liu B, Bai Q, Chen Y, et al. The PalWRKY77 transcription factor negatively regulates salt tolerance and abscisic acid signaling in Populus. Plant J. 2021;105(5):1258–73. pmid:33264467
- 89. Taji T, Seki M, Yamaguchi-Shinozaki K, Kamada H, Giraudat J, Shinozaki K. Mapping of 25 drought-inducible genes, RD and ERD, in Arabidopsis thaliana. Plant Cell Physiol. 1999;40(1):119–23. pmid:10189710
- 90. Bray EA. Classification of genes differentially expressed during water-deficit stress in Arabidopsis thaliana: an analysis using microarray and differential expression data. Ann Bot. 2002;89 Spec No(7):803–11. pmid:12102506
- 91. Shaar-Moshe L, Hübner S, Peleg Z. Identification of conserved drought-adaptive genes using a cross-species meta-analysis approach. BMC Plant Biol. 2015;15:111. pmid:25935420
- 92. Grünhofer P, Herzig L, Sent S, Zeisler-Diehl VV, Schreiber L. Increased cuticular wax deposition does not change residual foliar transpiration. Plant Cell Environ. 2022;45(4):1157–71. pmid:35102563
- 93. Neuwald AF, Aravind L, Spouge JL, Koonin EV. AAA+: A class of chaperone-like ATPases associated with the assembly, operation, and disassembly of protein complexes. Genome Res. 1999;9(1):27–43. pmid:9927482
- 94. Fujita Y, Fujita M, Satoh R, Maruyama K, Parvez MM, Seki M, et al. AREB1 is a transcription activator of novel ABRE-dependent ABA signaling that enhances drought stress tolerance in Arabidopsis. Plant Cell. 2005;17(12):3470–88. pmid:16284313
- 95. Kohler A, Blaudez D, Chalot M, Martin F. Cloning and expression of multiple metallothioneins from hybrid poplar. New Phytol. 2004;164(1):83–93. pmid:33873478
- 96. Navrot N, Collin V, Gualberto J, Gelhaye E, Hirasawa M, Rey P, et al. Plant glutathione peroxidases are functional peroxiredoxins distributed in several subcellular compartments and regulated during biotic and abiotic stresses. Plant Physiol. 2006;142(4):1364–79. pmid:17071643
- 97. Chibani K, Wingsle G, Jacquot J-P, Gelhaye E, Rouhier N. Comparative genomic study of the thioredoxin family in photosynthetic organisms with emphasis on Populus trichocarpa. Mol Plant. 2009;2(2):308–22. pmid:19825616
- 98. ElSayed AI, Rafudeen MS, Golldack D. Physiological aspects of raffinose family oligosaccharides in plants: protection against abiotic stress. Plant Biol (Stuttg). 2014;16(1):1–8. pmid:23937337
- 99. Hernández-Sánchez IE, Maruri-López I, Martinez-Martinez C, Janis B, Jiménez-Bremont JF, Covarrubias AA, et al. LEAfing through literature: late embryogenesis abundant proteins coming of age-achievements and perspectives. J Exp Bot. 2022;73(19):6525–46. pmid:35793147
- 100. Wu G, Tian N, She F, Cao A, Wu W, Zheng S, et al. Characteristics analysis of Early Responsive to Dehydration genes in Arabidopsis thaliana (AtERD). Plant Signal Behav. 2023;18(1):2105021. pmid:35916255
- 101. Barajas-Lopez J de D, Tiwari A, Zarza X, Shaw MW, Pascual JS, Punkkinen M, et al. EARLY RESPONSE TO DEHYDRATION 7 Remodels Cell Membrane Lipid Composition during Cold Stress in Arabidopsis. Plant Cell Physiol. 2021;62(1):80–91. pmid:33165601
- 102. Brocard L, Immel F, Coulon D, Esnay N, Tuphile K, Pascal S, et al. Proteomic Analysis of Lipid Droplets from Arabidopsis Aging Leaves Brings New Insight into Their Biogenesis and Functions. Front Plant Sci. 2017;8:894. pmid:28611809
- 103. Veerabagu M, Rinne PLH, Skaugen M, Paul LK, van der Schoot C. Lipid Body Dynamics in Shoot Meristems: Production, Enlargement, and Putative Organellar Interactions and Plasmodesmal Targeting. Front Plant Sci. 2021;12:674031. pmid:34367200
- 104. Gidda SK, Park S, Pyc M, Yurchenko O, Cai Y, Wu P, et al. Lipid Droplet-Associated Proteins (LDAPs) Are Required for the Dynamic Regulation of Neutral Lipid Compartmentation in Plant Cells. Plant Physiol. 2016;170(4):2052–71. pmid:26896396
- 105. Fernández-Santos R, Izquierdo Y, López A, Muñiz L, Martínez M, Cascón T, et al. Protein Profiles of Lipid Droplets during the Hypersensitive Defense Response of Arabidopsis against Pseudomonas Infection. Plant Cell Physiol. 2020;61(6):1144–57. pmid:32219438
- 106. Kang B-H, Anderson CT, Arimura S-I, Bayer E, Bezanilla M, Botella MA, et al. A glossary of plant cell structures: Current insights and future questions. Plant Cell. 2022;34(1):10–52. pmid:34633455
- 107. Kim EY, Park KY, Seo YS, Kim WT. Arabidopsis Small Rubber Particle Protein Homolog SRPs Play Dual Roles as Positive Factors for Tissue Growth and Development and in Drought Stress Responses. Plant Physiol. 2016;170(4):2494–510. pmid:26903535
- 108. Fang Y, Zhu R-L, Mishler BD. Evolution of oleosin in land plants. PLoS One. 2014;9(8):e103806. pmid:25105766
- 109. Guillaumot D, Guillon S, Déplanque T, Vanhee C, Gumy C, Masquelier D, et al. The Arabidopsis TSPO‐related protein is a stress and abscisic acid‐regulated, endoplasmic reticulum–Golgi‐localized membrane protein. The Plant Journal. 2009;60(2):242–56.
- 110. Vanhee C, Zapotoczny G, Masquelier D, Ghislain M, Batoko H. The Arabidopsis multistress regulator TSPO is a heme binding membrane protein and a potential scavenger of porphyrins via an autophagy-dependent degradation mechanism. Plant Cell. 2011;23(2):785–805. pmid:21317376
- 111. Jurkiewicz P, Melser S, Maucourt M, Ayeb H, Veljanovski V, Maneta-Peyret L, et al. The multistress-induced Translocator protein (TSPO) differentially modulates storage lipids metabolism in seeds and seedlings. Plant J. 2018;96(2):274–86. pmid:30003614
- 112. Cazzonelli CI, Pogson BJ. Source to sink: regulation of carotenoid biosynthesis in plants. Trends Plant Sci. 2010;15(5):266–74. pmid:20303820
- 113. Zhou X, Welsch R, Yang Y, Álvarez D, Riediger M, Yuan H, et al. Arabidopsis OR proteins are the major posttranscriptional regulators of phytoene synthase in controlling carotenoid biosynthesis. Proc Natl Acad Sci U S A. 2015;112(11):3558–63. pmid:25675505
- 114. Bouvier F, Hugueney P, d’Harlingue A, Kuntz M, Camara B. Xanthophyll biosynthesis in chromoplasts: isolation and molecular cloning of an enzyme catalyzing the conversion of 5,6-epoxycarotenoid into ketocarotenoid. Plant J. 1994;6(1):45–54. pmid:7920703
- 115. Schläpfer P, Zhang P, Wang C, Kim T, Banf M, Chae L, et al. Genome-Wide Prediction of Metabolic Enzymes, Pathways, and Gene Clusters in Plants. Plant Physiol. 2017;173(4):2041–59. pmid:28228535
- 116. Zhang Y, Zhang B, Yan D, Dong W, Yang W, Li Q, et al. Two Arabidopsis cytochrome P450 monooxygenases, CYP714A1 and CYP714A2, function redundantly in plant development through gibberellin deactivation. Plant J. 2011;67(2):342–53. pmid:21457373
- 117. Jiang Z, Wang M, Nicolas M, Ogé L, Pérez-Garcia M-D, Crespel L, et al. Glucose-6-Phosphate Dehydrogenases: The Hidden Players of Plant Physiology. Int J Mol Sci. 2022;23(24):16128. pmid:36555768
- 118. Rattanakon S, Ghan R, Gambetta GA, Deluc LG, Schlauch KA, Cramer GR. Abscisic acid transcriptomic signaling varies with grapevine organ. BMC Plant Biol. 2016;16:72. pmid:27001301
- 119. Chai G, Wang Z, Tang X, Yu L, Qi G, Wang D, et al. R2R3-MYB gene pairs in Populus: evolution and contribution to secondary wall formation and flowering time. J Exp Bot. 2014;65(15):4255–69. pmid:24821954
- 120. Yoshida K, Ma D, Constabel CP. The MYB182 protein down-regulates proanthocyanidin and anthocyanin biosynthesis in poplar by repressing both structural and regulatory flavonoid genes. Plant Physiol. 2015;167(3):693–710. pmid:25624398
- 121. James AM, Ma D, Mellway R, Gesell A, Yoshida K, Walker V, et al. Poplar MYB115 and MYB134 Transcription Factors Regulate Proanthocyanidin Synthesis and Structure. Plant Physiol. 2017;174(1):154–71. pmid:28348066
- 122. Wan S, Li C, Ma X, Luo K. PtrMYB57 contributes to the negative regulation of anthocyanin and proanthocyanidin biosynthesis in poplar. Plant Cell Rep. 2017;36(8):1263–76. pmid:28523445
- 123. Ma D, Reichelt M, Yoshida K, Gershenzon J, Constabel CP. Two R2R3-MYB proteins are broad repressors of flavonoid and phenylpropanoid metabolism in poplar. Plant J. 2018;96(5):949–65. pmid:30176084
- 124. Ma D, Tang H, Reichelt M, Piirtola E-M, Salminen J-P, Gershenzon J, et al. Poplar MYB117 promotes anthocyanin synthesis and enhances flavonoid B-ring hydroxylation by up-regulating the flavonoid 3’,5’-hydroxylase gene. J Exp Bot. 2021;72(10):3864–80. pmid:33711094
- 125. Wang L, Lu W, Ran L, Dou L, Yao S, Hu J, et al. R2R3-MYB transcription factor MYB6 promotes anthocyanin and proanthocyanidin biosynthesis but inhibits secondary cell wall formation in Populus tomentosa. Plant J. 2019;99(4):733–51. pmid:31021017
- 126. Du H, Liang Z, Zhao S, Nan M-G, Tran L-SP, Lu K, et al. The Evolutionary History of R2R3-MYB Proteins Across 50 Eukaryotes: New Insights Into Subfamily Classification and Expansion. Sci Rep. 2015;5:11037. pmid:26047035
- 127. Koes R, Verweij W, Quattrocchio F. Flavonoids: a colorful model for the regulation and evolution of biochemical pathways. Trends Plant Sci. 2005;10(5):236–42. pmid:15882656
- 128. Ramsay NA, Glover BJ. MYB-bHLH-WD40 protein complex and the evolution of cellular diversity. Trends Plant Sci. 2005;10(2):63–70. pmid:15708343
- 129. Gesell A, Yoshida K, Tran LT, Constabel CP. Characterization of an apple TT2-type R2R3 MYB transcription factor functionally similar to the poplar proanthocyanidin regulator PtMYB134. Planta. 2014;240(3):497–511. pmid:24923676
- 130. Debeaujon I, Nesi N, Perez P, Devic M, Grandjean O, Caboche M, et al. Proanthocyanidin-accumulating cells in Arabidopsis testa: regulation of differentiation and role in seed development. Plant Cell. 2003;15(11):2514–31. pmid:14555692
- 131. Baudry A, Heim MA, Dubreucq B, Caboche M, Weisshaar B, Lepiniec L. TT2, TT8, and TTG1 synergistically specify the expression of BANYULS and proanthocyanidin biosynthesis in Arabidopsis thaliana. Plant J. 2004;39(3):366–80. pmid:15255866
- 132. Lepiniec L, Debeaujon I, Routaboul J-M, Baudry A, Pourcel L, Nesi N, et al. Genetics and biochemistry of seed flavonoids. Annu Rev Plant Biol. 2006;57:405–30. pmid:16669768
- 133. Dombrecht B, Xue GP, Sprague SJ, Kirkegaard JA, Ross JJ, Reid JB, et al. MYC2 differentially modulates diverse jasmonate-dependent functions in Arabidopsis. Plant Cell. 2007;19(7):2225–45. pmid:17616737
- 134. Larkin RM, Alonso JM, Ecker JR, Chory J. GUN4, a regulator of chlorophyll synthesis and intracellular signaling. Science. 2003;299(5608):902–6. pmid:12574634
- 135. Wang P, Richter AS, Kleeberg JRW, Geimer S, Grimm B. Post-translational coordination of chlorophyll biosynthesis and breakdown by BCMs maintains chlorophyll homeostasis during leaf development. Nat Commun. 2020;11(1):1254. pmid:32198392
- 136. Yamatani H, Ito T, Nishimura K, Yamada T, Sakamoto W, Kusaba M. Genetic analysis of chlorophyll synthesis and degradation regulated by BALANCE of CHLOROPHYLL METABOLISM. Plant Physiol. 2022;189(1):419–32. pmid:35348770
- 137. Tachibana R, Abe S, Marugami M, Yamagami A, Akema R, Ohashi T, et al. BPG4 regulates chloroplast development and homeostasis by suppressing GLK transcription factors and involving light and brassinosteroid signaling. Nat Commun. 2024;15(1):370. pmid:38191552
- 138. Shimoda Y, Ito H, Tanaka A. Arabidopsis STAY-GREEN, Mendel’s green cotyledon gene, encodes magnesium-dechelatase. Plant Cell. 2016;28:2147–60.
- 139. Tan F-C, Cheng Q, Saha K, Heinemann IU, Jahn M, Jahn D, et al. Identification and characterization of the Arabidopsis gene encoding the tetrapyrrole biosynthesis enzyme uroporphyrinogen III synthase. Biochem J. 2008;410(2):291–9. pmid:18042043
- 140. Zhang K, Gan S-S. An abscisic acid-AtNAP transcription factor-SAG113 protein phosphatase 2C regulatory chain for controlling dehydration in senescing Arabidopsis leaves. Plant Physiol. 2012;158(2):961–9. pmid:22184656
- 141. Hou K, Wu W, Gan S-S. SAUR36, a small auxin up RNA gene, is involved in the promotion of leaf senescence in Arabidopsis. Plant Physiol. 2013;161(2):1002–9. pmid:23250625
- 142. Wang H-L, Zhang Y, Wang T, Yang Q, Yang Y, Li Z, et al. An alternative splicing variant of PtRD26 delays leaf senescence by regulating multiple NAC transcription factors in Populus. Plant Cell. 2021;33(5):1594–614. pmid:33793897
- 143. Fujita M, Fujita Y, Maruyama K, Seki M, Hiratsu K, Ohme-Takagi M, et al. A dehydration-induced NAC protein, RD26, is involved in a novel ABA-dependent stress-signaling pathway. Plant J. 2004;39(6):863–76. pmid:15341629
- 144. Kou X, Watkins CB, Gan S-S. Arabidopsis AtNAP regulates fruit senescence. J Exp Bot. 2012;63(17):6139–47. pmid:23066145
- 145. Kim Y-S, Sakuraba Y, Han S-H, Yoo S-C, Paek N-C. Mutation of the Arabidopsis NAC016 transcription factor delays leaf senescence. Plant Cell Physiol. 2013;54(10):1660–72. pmid:23926065
- 146. Garapati P, Xue G-P, Munné-Bosch S, Balazadeh S. Transcription Factor ATAF1 in Arabidopsis Promotes Senescence by Direct Regulation of Key Chloroplast Maintenance and Senescence Transcriptional Cascades. Plant Physiol. 2015;168(3):1122–39. pmid:25953103
- 147. Takasaki H, Maruyama K, Takahashi F, Fujita M, Yoshida T, Nakashima K, et al. SNAC-As, stress-responsive NAC transcription factors, mediate ABA-inducible leaf senescence. Plant J. 2015;84(6):1114–23. pmid:26518251
- 148. Li S, Lin Y-CJ, Wang P, Zhang B, Li M, Chen S, et al. The AREB1 Transcription Factor Influences Histone Acetylation to Regulate Drought Responses and Tolerance in Populus trichocarpa. Plant Cell. 2019;31(3):663–86. pmid:30538157
- 149. Stracke R, Werber M, Weisshaar B. The R2R3-MYB gene family in Arabidopsis thaliana. Curr Opin Plant Biol. 2001;4(5):447–56. pmid:11597504
- 150. Denekamp M, Smeekens SC. Integration of wounding and osmotic stress signals determines the expression of the AtMYB102 transcription factor gene. Plant Physiol. 2003;132(3):1415–23. pmid:12857823
- 151. Lippold F, Sanchez DH, Musialak M, Schlereth A, Scheible W-R, Hincha DK, et al. AtMyb41 regulates transcriptional and metabolic responses to osmotic stress in Arabidopsis. Plant Physiol. 2009;149(4):1761–72. pmid:19211694
- 152. Wilkins O, Nahal H, Foong J, Provart NJ, Campbell MM. Expansion and diversification of the Populus R2R3-MYB family of transcription factors. Plant Physiol. 2009;149(2):981–93. pmid:19091872
- 153. Zhang P, Wang R, Yang X, Ju Q, Li W, Lü S, et al. The R2R3-MYB transcription factor AtMYB49 modulates salt tolerance in Arabidopsis by modulating the cuticle formation and antioxidant defence. Plant Cell Environ. 2020;43(8):1925–43. pmid:32406163
- 154. Cheng Y, Ma Y, Zhang N, Lin R, Yuan Y, Tian H, et al. The R2R3 MYB Transcription Factor MYB71 Regulates Abscisic Acid Response in Arabidopsis. Plants (Basel). 2022;11(10):1369. pmid:35631794
- 155. Song L, Prince S, Valliyodan B, Joshi T, Maldonado dos Santos JV, Wang J, et al. Genome-wide transcriptome analysis of soybean primary root under varying water-deficit conditions. BMC Genomics. 2016;17:57. pmid:26769043
- 156. Heydarian Z, Yu M, Gruber M, Coutu C, Robinson SJ, Hegedus DD. Changes in gene expression in Camelina sativa roots and vegetative tissues in response to salinity stress. Sci Rep. 2018;8(1):9804. pmid:29955098
- 157. Wang A, Liang K, Yang S, Cao Y, Wang L, Zhang M, et al. Genome-wide analysis of MYB transcription factors of Vaccinium corymbosum and their positive responses to drought stress. BMC Genomics. 2021;22(1):565. pmid:34294027
- 158. Avico EH, Acevedo RM, Duarte MJ, Rodrigues Salvador A, Nunes-Nesi A, Ruiz OA, et al. Integrating Transcriptional, Metabolic, and Physiological Responses to Drought Stress in Ilex paraguariensis Roots. Plants (Basel). 2023;12(13):2404. pmid:37446965
- 159. Liu X, Widmer A. Genome-wide Comparative Analysis of the GRAS Gene Family in Populus, Arabidopsis and Rice. Plant Mol Biol Rep. 2014;32(6):1129–45.
- 160. Waters MT, Wang P, Korkaric M, Capper RG, Saunders NJ, Langdale JA. GLK transcription factors coordinate expression of the photosynthetic apparatus in Arabidopsis. Plant Cell. 2009;21(4):1109–28. pmid:19376934
- 161. Sun CW, Chen LJ, Lin LC, Li H. Leaf-specific upregulation of chloroplast translocon genes by a CCT motif–containing protein, CIA2. Plant Cell. 2001;13:2053–61.
- 162. Yang C-Y, Yan W-Y, Chang H-Y, Sun C-W. Arabidopsis CIA2 and CIL have distinct and overlapping functions in regulating chloroplast and flower development. Plant Direct. 2022;6(1):e380. pmid:35106435
- 163. Zhang L, Ren Y, Lu B, Yang C, Feng Z, Liu Z, et al. FLOURY ENDOSPERM7 encodes a regulator of starch synthesis and amyloplast development essential for peripheral endosperm development in rice. J Exp Bot. 2016;67(3):633–47. pmid:26608643
- 164. Thompson MG, Blake-Hedges JM, Pereira JH, Hangasky JA, Belcher MS, Moore WM, et al. An iron (II) dependent oxygenase performs the last missing step of plant lysine catabolism. Nat Commun. 2020;11(1):2931. pmid:32523014
- 165. Yu H, Xiao A, Wu J, Li H, Duan Y, Chen Q, et al. GmNAC039 and GmNAC018 activate the expression of cysteine protease genes to promote soybean nodule senescence. Plant Cell. 2023;35(8):2929–51. pmid:37177994
- 166. Tuskan GA, Difazio S, Jansson S, Bohlmann J, Grigoriev I, Hellsten U, et al. The genome of black cottonwood, Populus trichocarpa (Torr. & Gray). Science. 2006;313(5793):1596–604. pmid:16973872
- 167. Cohen D, Bogeat-Triboulot MB, Vialet-Chabrand S, Merret R, Courty PE, Moretti S. Developmental and environmental regulation of aquaporin gene expression across Populus species: divergence or redundancy?. PLoS ONE. 2013;8:e55506.
- 168. Kobayashi K, Sasaki D, Noguchi K, Fujinuma D, Komatsu H, Kobayashi M, et al. Photosynthesis of root chloroplasts developed in Arabidopsis lines overexpressing GOLDEN2-LIKE transcription factors. Plant Cell Physiol. 2013;54(8):1365–77. pmid:23749810
- 169. Rauf M, Arif M, Dortay H, Matallana-Ramírez LP, Waters MT, Gil Nam H, et al. ORE1 balances leaf senescence against maintenance by antagonizing G2-like-mediated transcription. EMBO Rep. 2013;14(4):382–8. pmid:23459204
- 170. Jensen MK, Lindemose S, de Masi F, Reimer JJ, Nielsen M, Perera V, et al. ATAF1 transcription factor directly regulates abscisic acid biosynthetic gene NCED3 in Arabidopsis thaliana. FEBS Open Bio. 2013;3:321–7. pmid:23951554
- 171. Li S, Gao J, Yao L, Ren G, Zhu X, Gao S, et al. The role of ANAC072 in the regulation of chlorophyll degradation during age- and dark-induced leaf senescence. Plant Cell Rep. 2016;35(8):1729–41. pmid:27154758
- 172. Mahmood K, El-Kereamy A, Kim S-H, Nambara E, Rothstein SJ. ANAC032 Positively Regulates Age-Dependent and Stress-Induced Senescence in Arabidopsis thaliana. Plant Cell Physiol. 2016;57(10):2029–46. pmid:27388337
- 173. Sakuraba Y, Han S-H, Lee S-H, Hörtensteiner S, Paek N-C. Arabidopsis NAC016 promotes chlorophyll breakdown by directly upregulating STAYGREEN1 transcription. Plant Cell Rep. 2016;35(1):155–66. pmid:26441053
- 174. Zhao D, Zheng Y, Yang L, Yao Z, Cheng J, Zhang F, et al. The transcription factor AtGLK1 acts upstream of MYBL2 to genetically regulate sucrose-induced anthocyanin biosynthesis in Arabidopsis. BMC Plant Biol. 2021;21(1):242. pmid:34049482
- 175. Zeng X, Ye L, Zhang R, Wang P. GLK2, a GOLDEN2-LIKE Transcription Factor, Directly Regulates Anthocyanin Accumulation by Binding With Promoters of Key Anthocyanin Biosynthetic Genes in Arabidopsis. Plant Cell Environ. 2025;48(9):7055–71. pmid:40518742
- 176. Mahmood K, Xu Z, El-Kereamy A, Casaretto JA, Rothstein SJ. The Arabidopsis Transcription Factor ANAC032 Represses Anthocyanin Biosynthesis in Response to High Sucrose and Oxidative and Abiotic Stresses. Front Plant Sci. 2016;7:1548. pmid:27790239
- 177. Maki H, Sakaoka S, Itaya T, Suzuki T, Mabuchi K, Amabe T, et al. ANAC032 regulates root growth through the MYB30 gene regulatory network. Sci Rep. 2019;9(1):11358. pmid:31388054
- 178. Sun T, Hazra A, Lui A, Zeng S, Wang X, Rao S, et al. GLKs directly regulate carotenoid biosynthesis via interacting with GBFs in plants. New Phytol. 2025;246(2):645–65. pmid:39953697
- 179. Cordoba E, Salmi M, León P. Unravelling the regulatory mechanisms that modulate the MEP pathway in higher plants. J Exp Bot. 2009;60(10):2933–43. pmid:19584121
- 180. Meier S, Tzfadia O, Vallabhaneni R, Gehring C, Wurtzel ET. A transcriptional analysis of carotenoid, chlorophyll and plastidial isoprenoid biosynthesis genes during development and osmotic stress responses in Arabidopsis thaliana. BMC Syst Biol. 2011;5:77. pmid:21595952
- 181. Sun T, Wang P, Rao S, Zhou X, Wrightstone E, Lu S, et al. Co-chaperoning of chlorophyll and carotenoid biosynthesis by ORANGE family proteins in plants. Mol Plant. 2023;16(6):1048–65. pmid:37202926
- 182. Vanhee C, Batoko H. Arabidopsis TSPO and porphyrins metabolism: A transient signaling connection?. Plant Signal Behav. 2011;6:1383–5.
- 183. Long Y, Goedhart J, Schneijderberg M, Terpstra I, Shimotohno A, Bouchet BP, et al. SCARECROW-LIKE23 and SCARECROW jointly specify endodermal cell fate but distinctly control SHORT-ROOT movement. Plant J. 2015;84(4):773–84. pmid:26415082
- 184. Xuan L, Xu M, Chen C, Yang C, Xu L, Huang M. Identification and characterization of three PeSHRs and one PeSCR involved in adventitious root development of Populus. Plant Cell Tiss Organ Cult. 2014;117(2):253–64.
- 185. Miguel A, Milhinhos A, Novák O, Jones B, Miguel CM. The SHORT-ROOT-like gene PtSHR2B is involved in Populus phellogen activity. J Exp Bot. 2016;67(5):1545–55. pmid:26709311
- 186. Bonito G, Hameed K, Ventura R, Krishnan J, Schadt CW, Vilgalys R. Isolating a functionally relevant guild of fungi from the root microbiome of Populus. Fungal Ecology. 2016;22:35–42.
- 187. Bizet F, Hummel I, Bogeat-Triboulot M-B. Length and activity of the root apical meristem revealed in vivo by infrared imaging. J Exp Bot. 2015;66(5):1387–95. pmid:25540436
- 188. Šimura J, Antoniadi I, Široká J, Tarkowská D, Strnad M, Ljung K, et al. Plant Hormonomics: Multiple Phytohormone Profiling by Targeted Metabolomics. Plant Physiol. 2018;177(2):476–89. pmid:29703867
- 189. de Freitas Pereira M, Cohen D, Auer L, Aubry N, Bogeat-Triboulot M-B, Buré C, et al. Ectomycorrhizal symbiosis prepares its host locally and systemically for abiotic cue signaling. Plant J. 2023;116(6):1784–803. pmid:37715981
- 190. Goodstein DM, Shu S, Howson R, Neupane R, Hayes RD, Fazo J, et al. Phytozome: a comparative platform for green plant genomics. Nucleic Acids Res. 2012;40(Database issue):D1178-86. pmid:22110026
- 191. Thimm O, Bläsing O, Gibon Y, Nagel A, Meyer S, Krüger P, et al. MAPMAN: a user-driven tool to display genomics data sets onto diagrams of metabolic pathways and other biological processes. Plant J. 2004;37(6):914–39. pmid:14996223
- 192. Conesa A, Götz S, García-Gómez JM, Terol J, Talón M, Robles M. Blast2GO: a universal tool for annotation, visualization and analysis in functional genomics research. Bioinformatics. 2005;21(18):3674–6. pmid:16081474
- 193. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550. pmid:25516281
- 194. Ding J, Bar-Joseph Z. Analysis of time-series regulatory networks. Curr Opin Syst Biol. 2020;21:16–24.
- 195. Ernst J, Vainas O, Harbison CT, Simon I, Bar-Joseph Z. Reconstructing dynamic regulatory maps. Mol Syst Biol. 2007;3:74. pmid:17224918
- 196. Langfelder P, Horvath S. Fast R Functions for Robust Correlations and Hierarchical Clustering. J Stat Softw. 2012;46(11):i11. pmid:23050260
- 197. Song WM, Zhang B. Multiscale embedded gene co-expression network analysis. PLOS Comput Biol. 2015;11:e1004574.
- 198. Csardi G, Nepusz T. The igraph software package for complex network research. InterJ Complex Syst. 2005;1695.