Warming temperature leads to cessation at different floral developmental stages in broccoli
Green Harmony F1, a broccoli cultivar, is known to be highly sensitive to elevated temperature, especially in terms of broccoli head development (Duclos et al. 2008). We are interested in understanding the molecular mechanisms underlying the warming-induced floral development cessation in this broccoli cultivar. Plants were grown under three temperature regimes: 28 °C, 22 °C, and 16 °C, and the floral development of the plants were ceased at inflorescence meristem stage (IM; producing cauliflower-like curd at 28 °C), floral primordium stage (FP; intermediate curd at 22 °C) or developed into floral bud (FB; normal broccoli head that is harvested and consumed) and beyond at 16 °C (Fig. 1A). At the transition from vegetative growth to reproductive growth (VR) stage, the apical meristem became flat and wide, the leaf primordium ceased to divide, and the bract leaf began to differentiate (Fig. 1B, F). Subsequently the apical meristem will give rise to the secondary, tertiary and higher order of inflorescence meristems, resulting in a thick curd. The development of the apical meristem at 28 °C was ceased at the IM stage (cauliflower-like curd) (Fig. 1C, G). At 22 °C, there were inflorescences branching out of the periphery of inflorescence meristem, that soon differentiated into floral primordia, forming intermediate curd (Fig. 1D, H). In contrast, at 16 °C the floral primordia rapidly developed into floral buds and formed broccoli head after brief curd-thickening (Fig. 1E, I).
DNA methylation inhibitor eliminates warming temperature-induced floral development cessation
We hypothesized that DNA methylation might be involved in the cessation of the floral development described above. We thus treated the plants at the VR stage grown at 28 °C with 5-azaC-treated and ddH2O (control), respectively. Twenty-four days after the treatment, the apical meristems of both the 5-azaC plants and the control developed cauliflower-like curd (IM stage) (Fig. 2A, D). Thirty-one days after the treatment, the floral development of the control plants remained the cauliflower-like curd, that is, the development was ceased at the IM stage (Fig. 2B). In contrast, the IM of the 5-azaC-treated plants differentiated into the intermediate curd or FP (Fig. 2E). Fourty-one days after the treatment, the intermediate curd continued developing and formed the broccoli head (Fig. 2F) while the control remained ceased at the IM stage (Fig. 2C). The results strongly suggested that the floral development cessation caused by warming temperature is regulated by DNA methylation.
Global dynamics in DNA methylation under warming temperature regimes
The above results prompted us to investigate the global changes in DNA methylation during floral development at different temperature (28 °C, 22 °C, and 16 °C). Apexes at four development stages, VR, IM, FP, and FB, at different temperature as described in Fig. 1 were sampled and subjected to whole genome bisulfite sequencing, with two biological replicates. There were approximately 64.9 million clean reads for each sample, and more than 53% of the reads could be mapped to the Brassica oleracea reference genome (Supplementary Table S1). We identified 17,769,861 (28 °C), 16,564,204 (22 °C) and 16,384,000 (16 °C) methylated cytosines (mC) on average, the numbers of mC sites at 28 °C was significantly larger than those at 22 °C and 16 °C (Supplementary Table S2).
Analyses of the global DNA methylation levels at different stages under the three temperature regimes (Supplementary Table S3) revealed that the methylation levels were higher at 28 °C than at 22 °C and 16 °C in all contexts (CG, CHG, and CHH; H represents A, T or C). For examples, the average methylation levels in the CG context at the VR stage increased from 49.2% at 16 °C to 51.0% at 28 °C; at the IM stage it increased from 50.5% at 16 °C to 52.3% at 28 °C; and at the FP stage from 50.1% at 16 °C to 51.7% at 22 °C (Fig. 3A). Similarly, there was an increase in the global methylation levels in the CHG and CHH contexts at the high temperature regimes (Fig. 3A).
The DNA methylation profiles of coding genes (including gene body and their 2-kb flanking regions) at different temperature (Fig. 3E through G) also showed that higher temperature led to higher methylation levels in the upstream and downstream regions of genes in all contexts (CG, CHG and CHH, especially CHG). The DNA methylation levels in different genic regions were further analyzed. The average methylation levels in the promoter regions increased with the increasing temperature (from 16 °C, 22 °C to 28 °C) in all contexts (CG, CHG, and CHH) (Fig. 3B through D). For instances, in the CG context, the average methylation levels of promoter regions increased from 32.8% at 16 °C to 34% at 28 °C at the VR stage, from 34% at 16 °C to 35% at 28 °C at the IM stage, and from 33.6% at 16 °C to 34.9% at 22 °C at the FP stage. Similarly, the average methylation levels of the promoter regions in the CHG context at the VR stage increased from 16.7% at 16 °C to 18.5% at 28 °C, at the IM stage from 16.5% at 16 °C to 19.9% at 28 °C, and at the FP stage from 16.3% at 16 °C to 18% at 22 °C. In the CHH context, the average methylation levels of the promoter regions at the VR stage increased ~ 0.8% from 16 °C to 28 °C, at the IM stage it increased about 1% from 16 °C to 28 °C, and at the FP stage, it increased approximately 0.7% from 16 °C to 22 °C. These data indicated that warming temperatures led to global increases in DNA methylation levels, especially in the promoter regions.
To investigate the role of DNA methylation in regulating warming temperature-induced floral development cessation, we further analyzed the DNA methylomes in two pairwise comparison groups: 28 °C vs. 16 °C and 22 °C vs. 16 °C. Seven thousand, five hundred thirty-two differentially methylated regions (DMRs) in group of 28 °C vs. 16 °C were identified with 7011 hypermethylated (hyper-DMRs) and 521 hypomethylated (hypo-DMRs) (Fig. 4A). The numbers of hyper-DMRs in CHG and CHH contexts were also larger than those of hypo-DMRs in the group of 22 °C vs. 16 °C (Fig. 4B). These results collectively indicated that warming temperature caused the increases in genome-wide DNA methylation levels in general, and the numbers of hyper-DMRs in particular (although there were also hypo-DMRs).
Warming temperature alters gene expression patterns during floral development
To determine whether the DNA methylation contribute to changes in gene expression during floral development, transcriptomes at different stages of floral development under various temperature regimes were generated, with three biological replications. After data filtering, approximately 87% of reads were uniquely mapped to the B. oleracea reference genome (Supplementary Table S4). We identified differentially expressed genes (DEGs) by pairwise analyses among three temperature regimes (28 °C, 22 °C and 16 °C) with emphases on two groups of them (28 °C vs. 16 °C; 22 °C vs. 16 °C). Comparing the transcriptomes at 28 °C vs. 16 °C led to the identification of 2238 up- and 2500 down-regulated DEGs at the VR stage, and 3322 up- and 3126 down-regulated DEGs at the IM stage (Fig. 5A). Similarly, comparison of the transcriptomes at 22 °C vs. 16 °C resulted in the identification of 1386 up- and 1944 down-regulated DEGs at the IM stage, and 8736 up- and 7045 down-regulated DEGs at the FP stage. These results clearly showed that the warmer temperatures remarkably altered the gene expression dynamics during floral development.
There were overlapping DEGs between “28°C vs. 16°C” and “22°C vs. 16°C” at all floral development stages as revealed by Venn analysis (Fig. 5B through D). It is reasonable to hypothesize that the non-overlapping DEGs found at and/or before a specific floral cessation stage may most likely be involved in the warming temperature-induced floral cessations, we thus focused our further analyses on the non-overlapping DEGs. For instance, the floral development was ceased at the IM stage at 28 °C, but not at 22 °C (Fig. 1), and the non-overlapping 2355 up- and 1982 down-regulated DEGs identified at the IM stage from 28 °C vs. 16 °C were most likely responsible for the floral development cessation at the IM stage (Fig. 5C). The non-overlapping 1591 up- and 1939 down-regulated DEGs identified at the VR stage from 28 °C vs. 16 °C might also be responsible for the cessation at the IM stage (Fig. 5B). Similarly, the non-overlapping 1631 up- and 1567 down-regulated DEGs identified at the FP stage from 22 °C vs. 16 °C were most likely responsible for the floral development cessation at the FP stage (Fig. 5D), and the non-overlapping 419 up- and 800 down-regulated DEGs identified at the IM stage from 22 °C vs. 16 °C might be responsible for the cessation at the FP stage as well (Fig. 5C).
Warming temperature-induced hypermethylation regulates apex-highly-expressed genes such as those ribosome biogenesis-related genes
We hypothesized that a subset of the above identified DEGs potentially responsible for the floral development cessation at the IM stage or FP stage might be regulated by DNA methylation. We thus performed association analyses between DMRs and DEGs, and identified a number of differentially-methylated DEGs (methDEGs) that may be involved in the floral development cessations. For the floral cessation at the IM stage at 28 °C, 318 up- and 232 down-regulated genes were hypermethylated, and 18 up- and 20 down-regulated genes hypomethylated (Fig. 6A). For the floral cessation at the FP stage (22 °C), 193 up- and 168 down-regulated genes were hypermethylated, and 44 up- and 39 down-regulated genes hypomethylated (Fig. 6B). For both cessations (IM and FP) hypermethylated genes, up- or down-regulated, were predominant.
Gene Ontology (GO) analyses of these methDEGs revealed that for the 28 °C-induced floral cessation at the IM stage, genes involved in expression, ribosome biogenesis and transmembrane transport were abundant among the down-regulated methDEGs, while 16 of the up-regulated methDEGs were responsible for tissue development (Fig. 6C, D). Further analyses of individual genes with their counterpart genes in Arabidopsis revealed that some methDEGs are highly expressed in inflorescence meristem; examples of these methDEGs are LOC106315696 (counterpart in Arabidopsis AT1G60080, a 3–5-exoribonuclease family protein), LOC106294921 (AT2G34480, RPL18AB, a nuclear localized member of the ribosomal L18ae/LX protein family), LOC106300910 (AT1G48920, NUC1, a nucleolin protein involved in rRNA processing, ribosome biosynthesis, and vascular pattern formation), LOC106331732 (AT4G33250, EIF3K, the initiation factor 3 k), LOC106298457 (AT5G41520, RPS10B, the eukaryote-specific protein S10e of the small cytoplasmic ribosomal subunit), and LOC106336129 (AT3G09630, SAC56, a ribosomal protein L4/L1 family) (Supplementary Fig. S1, Supplementary Table S5). The subset of the methDEGs identified here are highly expressed in the shoot apex and most likely responsible for the floral development cessation at the IM stage under 28 °C, which are designated as floral development cessation-associated genes (FCGs). I.e., these FCGs are required for the floral development from IM to FP (FCGs for IM to FP).
For the 22 °C-induced floral cessation at the FP stage, the up-regulated methDEGs were rich in genes involved in RNA metabolic process, transcription, and response to hormones while the down-regulated methDEGs were well represented by genes involved in response to stress, ribonucleoprotein complex, and RNA binding (Fig. 6E, F). In inferring from the counterpart genes in Arabidopsis, a number of methDEGs were found to be highly expressed in the shoot apex inflorescence: LOC106336011 (AT5G02960, a ribosomal protein S12/S23 family protein), LOC106316876 (AT5G48870, SAD1, a polypeptide similar to multifunctional Sm-like snRNP proteins that are required for mRNA splicing, export, and degradation), LOC106315720 (AT5G20920, the protein synthesis initiation factor eIF2 beta), LOC106313474 (AT5G19510, the translation elongation factor EF1B), LOC106337101 (AT3G53870, a ribosomal protein S3 family protein), and LOC106342054 (AT1G56450, PBG1, 20S proteasome β subunit) (Supplementary Fig. S1, Supplementary Table S5). These methDEGs or FCGs are required for the floral development from FP to FB (FCGs for FP to FB).
The tripartite relation among temperature, methylation levels and transcript abundance of individual FCGs
The above co-profiling of methylomes and transcriptomes at warming temperature enabled the identification of FCGs that might be responsible for the floral cessations. We further examined the methylation levels and expression levels of individual FCGs at different temperatures, which revealed a positive relation between methylation levels and temperature, and a negative relation between the methylation levels and transcript abundances of FCGs (Fig. 7A, B). That is, the increasing temperature increased the methylation levels of FCG’s promoters, and reduced the FCG’s transcript levels, supporting the hypothesis that warming temperature induced hypermethylation, which in turn suppressed FCGs’ expression, resulting in the floral cessation at IM or FP.
The warming temperature-induced hypermethylation is the cause of the suppression of FCGs’ expression
To investigate that the warming temperature-induced hypermethylation was responsible for the suppression of FCGs’ expression revealed above, we performed real-time quantitative PCR (qRT-PCR) analysis of randomly selected FCGs from samples treated with or without the DNA methylation inhibitor 5-azaC at warming temperature. The expression levels of these FCGs in samples treated with the methylation inhibitor were significantly higher than those in controls at 28 °C or 22 °C (Fig. 7C, D), confirming that the warming temperature-induced DNA hypermethylation is the cause of the suppression of FCGs.
The DMRs on the FCG promoters are transcription factor (TF) binding sites
The findings that DNA hypermethylation suppressed gene expression as described above prompted us to investigate whether the DMRs on the FCG promoters are TF binding sites; the hypermethylation may prevent TF bindings. We used DMR sequences of the promoters of the FCGs in Fig. 7 to search against JASPAR2022 database; the JASPAR CORE database contains a curated, non-redundant set of profiles, derived from published collections of experimentally defined transcription factor binding sites for eukaryotes (https://jaspar.genereg.net). All the DMRs were predicted to be cis elements to which various TF binds (Fig. 8A, B). Among the TFs binding to the promoters of the FCGs responsible for 28 °C-induced floral cessation are ARR10 (a type-B ARR TF involved in the cytokinin signaling), PHL2 (a MYB-CC protein involved in regulation of response to phosphate starvation), TRB1 (telomere repeat binding factor 1, functioning together with MYB domain), MYB55 and DOFs (DNA-binding one finger), and the DOF family TFs are predominant (Fig. 8A). In contrast, ARR2 (also a TF involved in the cytokinin signaling), GATA15 (a Zinc finger TF involved in cytokinin response), AGL55 (AGAMOUS-like 55 TF highly expressed in shoot apical meristem) and DOFs are among the TFs binding to the promoters of the FCGs responsible for 22 °C-induced floral cessation (Fig. 8B).