Figure 1.
Experimental design of diurnal and circadian time series.
Whole rosettes were harvested at the indicated ZT (zeitgeber time, in hours) and used for transcript and metabolite profiling. Temperature and light conditions are indicated. White, black and grey bars indicate the corresponding day, night and subjective night periods.
Figure 2.
Contribution of low temperature and diurnal regulation to the variation in transcripts and metabolites in diurnal and circadian time courses.
PCA (Principal Component Analysis) was applied to transcript (A) and metabolite (B) profiling datasets from diurnal and circadian time courses. The color indicates the different light and temperature conditions of the studied time courses: light/dark cycles either at 20°C (red) or 4°C (dark blue) and continuous light at 20°C (orange) or 4°C (light blue). Circadian time courses are denoted by c in lowercase. Sampling time is indicated by ZT (zeitgeber, in hours). PC1 and PC2 correspond to principal component 1 and 2, respectively. Each point in (A) represents a single transcript profile, whilst in (B) it represents the mean metabolite profile of five biological replicates.
Figure 3.
Effect of cold on cyclic gene expression and metabolite content.
(A) Venn diagram showing the number and overlaps between of transcripts classified into each of the following three groups: 1) diurnally regulated at 20°C, 2) circadian regulated at 20°C or 3) cold-responsive. Circadian and diurnal regulated transcripts were identified using the HAYSTACK tool or by autocorrelation, respectively. Cold responsive genes were identified using limma with a sliding window of three sequential time points (FDR<0.05) (see Methods for details). (B) Venn diagram showing the number and overlaps between metabolites classified into each of the following four groups: 1) diurnally regulated at 20°C, 2) circadian regulated at 20°C, 3) diurnally regulated at 4°C, 4) circadian regulated at 4°C. Circadian and diurnal regulated transcripts were identified using the HAYSTACK tool or by autocorrelation, respectively (see Methods for details).
Table 1.
Summary of the changes in primary metabolism during cold acclimation in diurnal and circadian time courses.
Figure 4.
Global overview of the dynamic changes in primary metabolism during cold acclimation in diurnal and circadian time courses.
Heatmaps showing the changes occurring across time in the pool sizes of different metabolites during the four performed time courses. From left to right, these represent 16 h/8 h light/dark at 20°C, 16 h/8 h light/dark at 4°C, continuous light at 20°C and continuous light at 4°C. Along the top axis white bars indicate light, black bars indicate dark and grey bars indicate the subjective dark periods (i.e. corresponding to the entrainment conditions). Metabolite levels are the mean log2 values from five biological replicates depicted in a false color scale where red indicates low and blue indicates high values in the range of −5 to +5 (log2). *Due to the high levels of glucose, fructose and starch content at low temperature in comparison to other metabolites, the false color scale ranges from −15 to +15 (log2). As not all cyclical regulation is visible in this figure (especially where a metabolite is strongly cold responsive), it should be interpreted in conjunction with the detailed classification of circadian and diurnal regulation of the different metabolites shown in Table 1.
Table 2.
Selected network parameters of metabolite correlation networks.
Figure 5.
Core network of metabolite correlations observed in diurnal and circadian conditions both at 20°C and 4°C.
In the depicted network, nodes (spots) represent metabolites and edges (lines) indicate highly significant positive (blue) and negative (red) pairwise correlations between metabolites. The core network represents thirteen correlations between seventeen metabolites which were stably in diurnal and circadian conditions both at 20°C and 4°C (i.e. in all four studied time series). Node color codes indicate compound classes as described in the figure. The significance threshold of the Spearman correlations was set at <0.001 for the Bonferroni corrected p-values (see Methods).
Figure 6.
Coordinated transcriptional regulation of conventional cold induced metabolites.
Summary of the metabolic pathway, transcript and metabolite profiles of GABA (A) and proline (B). For transcripts, relative expression (log2) from a pool of five biological replicates is indicated. Metabolite content (log2) corresponds to the normalized peak apex intensities from five biological replicates with bars indicating the largest standard deviation. GAD, glutamate decarboxylase (At1g65960); GABA-T, GABA transaminase; SSADH, succinic semialdehyde dehydrogenase (At1g79440); SSA, succinate semialdehyde; Glu, glutamate; GABA, 4-aminobutyrate; P5CS, Δ1-pyrroline-5-carboxylate synthase (At2g39800); δ-OAT, ornithine-δ-aminotransferase; P5C, pyrroline-5-carboxylate reductase; PD, proline dehydrogenase (At3g30775); P5CDH, 1-pyrroline-5-carboxylate dehydrogenase.
Figure 7.
Integration of gene expression and metabolite accumulation for the aspartic acid biosynthetic pathway.
For transcripts, relative expression (log2) from a pool of five biological replicates is indicated. Metabolite content (log2) corresponds to the normalized peak apex intensities from five biological replicates with bars indicating the largest standard deviation. AS, asparaginase; ASN, asparagine synthase (ASN1: At3g47340); AKII-HSDHII, aspartate-kinase-homoserine-dehydrogenase.
Figure 8.
Integration of gene expression and metabolite accumulation for OAS.
For transcripts, relative expression (log2) from a pool of five biological replicates is indicated. Metabolite content (log2) corresponds to the normalized peak apex intensities from five biological replicates with bars indicating the largest standard deviation. SAT, serine O-acetyltransferase (SAT1: At1g55920; SAT3: At3g13110); oas, cysteine synthase; OAS, O-acetylserine.
Figure 9.
Integration of gene expression and metabolite accumulation for starch degradation pathway.
For transcripts, relative expression (log2) from a pool of five biological replicates is indicated. Metabolite content (log2) corresponds to the normalized peak apex intensities from five biological replicates. GWD, glucan water dikinase; PWD, phosphoglucan, water dikinase; ISA3, isoamylase3 (At4g09020); DPE1, disproportionating enzyme; DPE2, cytosolic transglucosidase (At2g40840). HK, hexokinase (At2g19860); α-amylase (At1g69830); BAM2/BMY9, β-amylase (At4g00490); BAM3/BMY8, β-amylase (At4g17090).
Figure 10.
Freezing tolerance of circadian mutants cca1-11, lhy-21 and cca1-11/lhy-21 compared to the wild-type Ws.
A, Electrolyte leakage was measured in whole seedlings frozen at different temperatures, either before (NA) or after cold acclimation (ACC). A total of four replicates were measured. B, The survival of the circadian mutants after exposure to freezing temperatures was estimated as the percentage of NA and ACC seedlings surviving the different temperatures after 7 d of recovery under control conditions.