Global profiling of rice and poplar transcriptomes highlights key conserved circadian-controlled pathways and cis-regulatory modules

Abstract

Background: Circadian clocks provide an adaptive advantage through anticipation of daily and seasonal environmental changes. In plants, the central clock oscillator is regulated by several interlocking feedback loops. It was shown that a substantial proportion of the Arabidopsis genome cycles with phases of peak expression covering the entire day. Synchronized transcriptome cycling is driven through an extensive network of diurnal and clock-regulated transcription factors and their target cis-regulatory elements. Study of the cycling transcriptome in other plant species could thus help elucidate the similarities and differences and identify hubs of regulation common to monocot and dicot plants.

Methodology/principal findings: Using a combination of oligonucleotide microarrays and data mining pipelines, we examined daily rhythms in gene expression in one monocotyledonous and one dicotyledonous plant, rice and poplar, respectively. Cycling transcriptomes were interrogated under different diurnal (driven) and circadian (free running) light and temperature conditions. Collectively, photocycles and thermocycles regulated about 60% of the expressed nuclear genes in rice and poplar. Depending on the condition tested, up to one third of oscillating Arabidopsis-poplar-rice orthologs were phased within three hours of each other suggesting a high degree of conservation in terms of rhythmic gene expression. We identified clusters of rhythmically co-expressed genes and searched their promoter sequences to identify phase-specific cis-elements, including elements that were conserved in the promoters of Arabidopsis, poplar, and rice.

Conclusions/significance: Our results show that the cycling patterns of many circadian clock genes are highly conserved across poplar, rice, and Arabidopsis. The expression of many orthologous genes in key metabolic and regulatory pathways is diurnal and/or circadian regulated and phased to similar times of day. Our results confirm previous findings in Arabidopsis of three major classes of cis-regulatory modules within the plant circadian network: the morning (ME, GBOX), evening (EE, GATA), and midnight (PBX/TBX/SBX) modules. Identification of identical overrepresented motifs in the promoters of cycling genes from different species suggests that the core diurnal/circadian cis-regulatory network is deeply conserved between mono- and dicotyledonous species.

Figure 1. Sizable proportions of poplar and rice transcriptomes display daily oscillations in RNA abundance with peak expression encompassing all phases of the day.

Proportions of poplar and rice genes rhythmically (red sector) expressed under diurnal and free running (circadian) conditions. Transcripts were considered cyclically expressed if the Pearson correlation coefficient r between the data and respective HAYSTACK pattern model (, ; http://haystack.cgrb.oregonstate.edu/) was 0.75 or greater. The proportions were calculated as ratios of the number of the unique cycling genes to the total number of unique gene models represented on array. Diurnal and circadian segments of each time course were separated by a spacer period of forty eight hours of continuous light as described in the Materials and Methods.

Figure 2. Overlap of oscillating transcripts under various driven conditions.

Venn diagrams show distribution of oscillating transcripts in poplar (left) and rice (right) under photo-, thermo- and photo/thermocycles. Collectively, 20,619 and 21,364 (for poplar and rice, respectively) unique gene models were rhythmically expressed under all diurnal conditions. Of these, 1,942 poplar (11.7%) and 5,176 rice (24.6%) transcripts oscillated under all three tested diurnal conditions. To compile the lists of genes cycling under each condition, the array probe sets displaying cycling patterns of expression were matched to their respective gene models as described in the Materials and Methods.

Figure 3. Rhythmic rice and poplar transcripts encompass all phases of the day peaking a few hours before light/dark transitions.

Histograms of phase call frequency distributions among cycling poplar and rice genes according to phase of day. The distribution of the phase frequencies is broadly consistent with that of Arabidopsis plants grown under similar conditions . Each phase bin corresponds to a one-hour increment.

Figure 4. Under similar diurnal conditions Arabidopsis-rice-poplar orthologs are phased to similar times of day.

28–34% of rhythmically expressed Arabidopsis-rice-poplar orthologs were phased within three hours of each other under photo- (LDHH), thermo-(LLHC) and photo/thermocycles (LDHC). The plots depict the number of orthologs between Arabidopsis and poplar or Arabidopsis and rice for every phase of day under LDHH, LDHC and LLHC. The number of genes is represented by a color code (below). The red line represents positions of identical phases between both species being compared. Only orthologs that were present in all three species were selected in order to allow comparisons between all three species. The sum of every phase call combination between orthologs was compiled and plotted using a Microsoft Excel surface contour plot. The color scale was adjusted depending on the number of orthologs found under each condition.

Figure 5. The diurnal expression profiles of the core circadian clock and several clock-associated genes are conserved among rice, poplar and Arabidopsis.

Putative O. sativa (ssp. japonica), poplar, and Brachypodium clock gene orthologs were identified using criterion of the best mutual BLAST hit against the respective Arabidopsis counterparts. Time points 0 and 12 correspond to subjective dawn and dusk, respectively. Putative poplar/rice/Brachypodium orthologs correspond to Arabidopsis CCA1 (AT1G01060)/LHY (AT2G46830)/estExt_Genewise1_v1.C_LG_XIV1950/LOC_Os08g06110/Bradi3g16510, GI (AT1G22770)/estExt_fgenesh4_pg.C_LG_V1131/LOC_Os01g08700/Bradi2g05230, TOC1 (AT5G61380)/fgenesh4_pg.C_scaffold_129000038/LOC_Os02g40510/Bradi3g48880, RVE1 (At5g17300)/gw1.IV.3973/LOC_Os02g46030/Bradi3g51960, LUX (AT3G46640)/gw1.IX.4105/LOC_Os01g74020/Bradi2g62070, FKF1 (At1g68050)/estExt_fgenesh4_pg.C_LG_X0958/LOC_Os11g34460/Bradi4g16630, PRR5 (AT5G24470)/gw1.XII.1231/LOC_Os09g36220/Bradi4g36080, ELF3 (AT2G25930)/estExt_fgenesh4_pm.C_LG_VI0700/LOC_Os06g05060/Bradi2g14290), and ERP1 (At1g18330)/fgenesh4_pg.C_scaffold_122000043/LOC_Os06g51260/Bradi1g29680, respectively.

Figure 6. Several Arabidopsis, rice, poplar and Brachypodium orthologs are phased to similar times of day.

A. An example of the cycling expression profile of a gene involved in cellulose biosynthesis (Arabidopsis gene AT2G32530, CELLULOSE SYNTHASE LIKE B3) under multiple diurnal/circadian conditions. B, C, and D. Expression of orthologs involved in cell wall biosynthesis in Arabidopsis, rice, poplar, and Brachypodium. B. Beta-fructofuranosidase/invertase, involved in sucrose catabolic process (Arabidopsis/poplar/rice, ssp. japonica/Brachypodium loci: - AT1G12240/estExt_fgenesh4_pg.C_LG_III0902/rice/LOC_Os04g45290/Bradi5g16900). C. Trehalose-6-phosphate synthase, involved in trehalose metabolism (AT1G78580/estExt_fgenesh4_pg.C_1680018/LOC_Os05g44210/Bradi2g19640). D. Cinnamoyl CoA reductase, involved in lignin biosynthesis (AT1G15950/estExt_fgenesh4_kg.C_LG_III0056/LOC_Os08g34280/Bradi3g36890).

Figure 7. All major steps of auxin biosynthesis and response pathways in rice, poplar, and Arabidopsis include at least one cycling ortholog.

Only array probe sets identified by HAYSTACK as cycling under the tested diurnal conditions are shown. Time points 0 and 12 correspond to subjective dawn and dusk, respectively. A.th - Arabidopsis thaliana, O. sativa - Oryza sativa, ssp. japonica, P.t – Populus trichocarpa, clone Nisqally 1.

Figure 8. Gene ontology (GO) categories overrepresented among rice (ssp. japonica) cycling genes under photoperiods (LDHH).

The shaded circles indicate overrepresented GO categories (categories with FDR≤0.05 are shown in yellow) based on significance level. The radius of each circle denotes the number of genes in each category. The list of cycling genes was generated using HAYSTACK with best fitting model cutoff value r≥0.9. GO overrepresentation graph was generated using the network visualization tool Cytoscape (http://www.cytoscape.org/).

Figure 9. Rhythmic expression of poplar and rice transcription factor transcripts.

A. A large portion of the P. trichocarpa and O. sativa TFs represented on arrays are rhythmically expressed under photo-, thermo-, and photo/thermocycles. B. Rhythmically expressed TF transcripts encompass all phases of the day peaking a few hours before light/dark transitions. Poplar and rice TFs cycling in driven condition are listed in Table S1. B. Peak expression of cycling TFs occurs at all phases of the day. Expression heat maps of Populus trichocarpa and O. sativa (ssp. japonica) TFs oscillating under photocyles (LDHH). Mean centered expression levels are depicted in yellow (high expression) and blue (low expression). The heat map was generated using HAYSTACK output filtered using a Pearson correlation coefficient cutoff value r≥0.9.

Figure 10. Identification of conserved cis-elements in promoters of Arabidopsis, rice and poplar genes cycling in the LDHH condition.

ELEMENT-based enumerative promoter analysis and Z-score profile comparisons between rice, poplar and Arabidopsis. Z-score profiles were summarized for diurnal and circadian associated elements exhibiting conserved time-of-day overrepresentation across rice, poplar and Arabidopsis. A Z-score cutoff threshold value 2.33 corresponding to a p-value of 0.01 was selected arbitrarily. Time points 0 and 12 correspond to subjective dawn and dusk, respectively.