Sugar modulation of anaerobic-response networks in maize root tips

Abstract

Sugar supply is a key component of hypoxia tolerance and acclimation in plants. However, a striking gap remains in our understanding of mechanisms governing sugar impacts on low-oxygen responses. Here, we used a maize (Zea mays) root-tip system for precise control of sugar and oxygen levels. We compared responses to oxygen (21 and 0.2%) in the presence of abundant versus limited glucose supplies (2.0 and 0.2%). Low-oxygen reconfigured the transcriptome with glucose deprivation enhancing the speed and magnitude of gene induction for core anaerobic proteins (ANPs). Sugar supply also altered profiles of hypoxia-responsive genes carrying G4 motifs (sources of regulatory quadruplex structures), revealing a fast, sugar-independent class followed more slowly by feast-or-famine-regulated G4 genes. Metabolite analysis showed that endogenous sugar levels were maintained by exogenous glucose under aerobic conditions and demonstrated a prominent capacity for sucrose re-synthesis that was undetectable under hypoxia. Glucose abundance had distinctive impacts on co-expression networks associated with ANPs, altering network partners and aiding persistence of interacting networks under prolonged hypoxia. Among the ANP networks, two highly interconnected clusters of genes formed around Pyruvate decarboxylase 3 and Glyceraldehyde-3-phosphate dehydrogenase 4. Genes in these clusters shared a small set of cis-regulatory elements, two of which typified glucose induction. Collective results demonstrate specific, previously unrecognized roles of sugars in low-oxygen responses, extending from accelerated onset of initial adaptive phases by starvation stress to maintenance and modulation of co-expression relationships by carbohydrate availability.

Figure 1

Root transcriptome responses to different levels of oxygen and sugar. (A) Hierarchical clustering analysis of normalized samples using average linkage based on Euclidean distance revealed significant differences (two-fold change, P < 0.05) between oxygen and sugar levels. (B) Number of sugar-responsive genes that were differentially expressed (P < 0.05) in ambient air and low oxygen (low O₂). (C) Number of hypoxia-responsive genes differentially expressed at each glucose level (LG or AG). Treatments included two oxygen levels: ambient air (21% O₂ v/v) or low oxygen (0.2% O₂ v/v) in combination with two glucose concentrations: LG (0.2% w/v) or AG (2% w/v).

Figure 2

GO-term enrichment among sugar- and oxygen-responsive genes over time. (A) functional categories of transcripts over- (red) or under-represented (blue) in root-tips exposed to low oxygen (Low O₂; 0.2% v/v) with either LG (0.2% w/v) or AG (2% w/v). (B) GO categories over- (red) or under-represented (blue) among glucose-responsive genes under ambient air (21% O₂ v/v) or low oxygen (Low O₂; 0.2% O₂ v/v). Data were analyzed using the Wilcoxon test in Pageman (P < 0.05, Usadel et al., 2006) and resulting heat-maps indicate log2 changes during treatments.

Figure 3

Changes in relative gene expression during responses to oxygen and sugar treatments over time. The heatmap represents the top 30 genes with the greatest statistical variance and shows treatment effects on expression of each gene relative to its overall standard deviation. Treatments included two levels of glucose: LG (0.2% w/v) or AG (2% w/v) in combination with two oxygen treatments: ambient air (21% O₂ v/v) and low oxygen (Low O₂; 0.2% v/v). Red text indicates genes that encode the core “ANPs” initially identified using maize by Sachs et al. (1980). Transcripts are ranked based on significant differences in their relative expression (P < 0.05; Supplemental Datasets S1 and S2).

Figure 4

Effect of glucose levels on transcript profiles associated with primary and anaerobic metabolism over time. (A) Schematic representation of reactions associated with primary and hypoxic metabolism. (B) Heat-map of mRNA abundance (rlog-transformed, normalized count values) for transcripts associated with pathways in A. (C) Schematic representation of SPS and SPP reactions. (D) Heat-map of relative RNA abundance as number of normalized sequence reads (rlog-transformed) genes encoding SPS and SPP. Treatments included two levels of glucose: LG (0.2% w/v) or AG (2% w/v) in combination with two oxygen treatments: ambient air (21% O₂ v/v) and low oxygen (Low O₂; 0.2% v/v). Significance of sugar effects on gene expression was evaluated at each time point using the DESEq2 pipeline (Supplemental Dataset S1).

Figure 5

Responses by hypoxia-regulated genes over time, which carry motifs with potential to form four-stranded, DNA-quadruplex structures (G4s). (A) Proportion of G4 motifs among the total number of DEGs under low oxygen in LG (0.2% w/v) or AG (2% w/v). The probability of the observed frequency for G4-motifs in a DEG set (induced or repressed) occurring by random chance was determined using a binomial distribution test where n=total number of DEG at each time point, k=number of DEG’s that contain G4s at each time point, p=proportion of G4-containing genes in the whole maize genome (28%, blue line), and q = 1–0.28. Asterisks indicate where the probability was significant (P < 0.05) for DEG in low (black) or abundant (red) glucose. (B) Number of G4-containing genes that were newly upregulated at each time point.

Figure 6

Profiles of sugars and their primary metabolites in roots exposed to different levels of oxygen and sugars over time. Metabolite levels throughout the treatments were quantified by GC-MS (A–D) or LC–MS–MS (E and F). Treatments included two oxygen levels: ambient air (+O₂; 21% O₂ v/v) or low oxygen (−O₂; 0.2% O₂ v/v) in combination with two glucose concentrations: LG (0.2% w/v) or AG (2% w/v). Data represent mean values ± sem; n = 3. Asterisks indicate a significant difference in responses to oxygen concentrations (black) or between glucose levels (red) according to two-way ANOVA (P ≤ 0.05).

Figure 7

Co-expression network of genes for core ANPs. The co-expression network was constructed using 15 genes coding for ANPs (designated by Sachs et al. (1980); Supplemental Table S4) as baits. Co-expressed targets were identified in the transcriptome data from 60 different RNA-Seq libraries, each representing responses of 0.6-cm maize root tips under either aerobic (21% O₂ v/v) or low-oxygen conditions (0.2% O₂ v/v) using CoExpNetViz (Tzfadia et al., 2016). Data were analyzed using the Pearson correlation co-efficient, with the 5 lowest and 95 uppermost percentile rankings as thresholds for co-expression. Dots with the same color represent genes co-expressed with a given bait, based on positive correlations where R² > 0.8. For gene identifiers, see Supplemental Table S1. Wsl1, Receptor-like protein kinase-related.

Figure 8

Changes in the co-expression network of genes for core ANPs throughout oxygen and sugar treatments in maize. The overall network is as shown in Figure 7, modified here with altered color to show regulation of component genes and changes in their co-expression relationships in each glucose concentration: LG or physiologically AG. Visualized data represent target genes that were positively correlated (R² > 0.8) and differentially expressed (P < 0.05). Downregulated genes are shown in blue, with strongly downregulated genes (more than two-fold) in black. Upregulated genes are shown in yellow, with strongly upregulated genes (more than two-fold) in orange. The Table shows topological characteristics of co-expression networks among low-oxygen-responsive genes under each glucose condition.

Figure 9

Modules of the most highly interconnected co-expression networks for genes expressed under low-oxygen with either AG (2% w/v) or LG (LG; 0.2% w/v) levels. (A) Modules were identified using the Cytoscape MCODE algorithm plug-in. Nodes represent DEG (P < 0.05) and edges represent significant co-expression between DEG. Color indicates expression levels: blue (downregulated); yellow (upregulated); orange (more than two-fold upregulated). (B) Cis-regulatory elements identified in co-expressed members of the Gcp4–Pdc3 modules. Promoter sequences were analyzed using the PlantPAN2.0 (Chow et al., 2015). Partners in these modules shared four of the motifs noted when glucose was limiting (LG, green oval), but only two where distinct to module genes when glucose was abundant (AG, blue oval).