Skip to main content
NCBI home page
The Plant Cell logoLink to The Plant Cell
. 2018 Jun 4;30(7):1387–1403. doi: 10.1105/tpc.18.00074

Disrupted Genome Methylation in Response to High Temperature Has Distinct Affects on Microspore Abortion and Anther Indehiscence[OPEN]

Yizan Ma a, Ling Min a,1, Maojun Wang a, Chaozhi Wang a, Yunlong Zhao a, Yaoyao Li a, Qidi Fang a, Yuanlong Wu a, Sai Xie a, Yuanhao Ding a, Xiaojun Su a, Qin Hu a, Qinghua Zhang a, Xueyuan Li b, Xianlong Zhang a,1
PMCID: PMC6096589  PMID: 29866646

Integrated multi-omics data analysis reveals the role of genome methylation in male fertility, shedding light on the mechanism underlying male sterility in response to high temperature.

Abstract

High-temperature (HT) stress induces male sterility, leading to yield reductions in crops. DNA methylation regulates a range of processes involved in plant development and stress responses, but its role in male sterility under HT remains unknown. Here, we investigated DNA methylation levels in cotton (Gossypium hirsutum) anthers under HT and normal temperature (NT) conditions by performing whole-genome bisulfite sequencing to investigate the regulatory roles of DNA methylation in male fertility under HT. Global disruption of DNA methylation, especially CHH methylation (where H = A, C, or T), was detected in an HT-sensitive line. Changes in the levels of 24-nucleotide small-interfering RNAs were significantly associated with DNA methylation levels. Experimental suppression of DNA methylation led to pollen sterility in the HT-sensitive line under NT conditions but did not affect the normal dehiscence of anther walls. Further transcriptome analysis showed that the expression of genes in sugar and reactive oxygen species (ROS) metabolic pathways were significantly modulated in anthers under HT, but auxin biosynthesis and signaling pathways were only slightly altered, indicating that HT disturbs sugar and ROS metabolism via disrupting DNA methylation, leading to microspore sterility. This study opens up a pathway for creating HT-tolerant cultivars using epigenetic techniques.

INTRODUCTION

DNA methylation represents an epigenetic mechanism for the regulation of gene expression (He et al., 2011; Zhang and Zhu, 2011; Jullien et al., 2012; Matzke and Mosher, 2014). DNA methylation of cytosines begins with the activation of DNA methyltransferases. The methylated cytosines are classified by sequence context as CG, CHG, and CHH (H represents A, T, or C) (Law and Jacobsen, 2010). Much research has shown that DNA methylation strongly influences many aspects of plant development, such as flower development, fruit ripening, and stress responses (Kim et al., 2009; Zhong et al., 2013; Yong-Villalobos et al., 2015).

In Arabidopsis thaliana, DOMAINS REARRANGED METHYLTRANSFERASE2 (DRM2) catalyzes de novo DNA methylation in all contexts (Cao and Jacobsen, 2002). After the establishment of DNA methylation, METHYLTRANSFERASE1 maintains methylation at CG sequences (Hu et al., 2014), CHROMOMETHYLASE3 (CMT3) maintains methylation at CHG (Cao and Jacobsen, 2002; Cao et al., 2003), and CMT2 maintains CHH methylation patterns (Shen et al., 2014). In plants, CG and CHG methylation patterns can be maintained by the recognition of hemimethylated signatures during DNA replication, but CHH methylation is not established by the same recognition process (Henderson and Jacobsen, 2007; Law and Jacobsen, 2010). An RNA-directed DNA methylation (RdDM) pathway guides de novo CHH methylation on strand-specific DNA sequences using a combination of 24-nucleotide small-interfering RNAs (siRNAs) (Matzke and Mosher, 2014). In the canonical RdDM pathway, specific transcripts are generated from plant-specific RNA polymerase (Pol IV) complexes. RNA-DEPENDENT RNA POLYMERASE2 (RDR2) then converts the transcripts into double-stranded RNAs. The 24-nt siRNAs are subsequently spliced by DICER-LIKE3 (DCL3). ARGONAUT4 (AGO4) binds to the 24-nt siRNAs to target the silencing complex to the transcripts generated by Pol V. Finally, DRM2 interacts with AGO4 to target the CHH sites (Chan et al., 2004; Jia et al., 2009; Havecker et al., 2010; Matzke and Mosher, 2014).

When plants encounter stress, genome-wide transcriptional regulation occurs, including the activation of stress defense genes and regulatory proteins (Jiang et al., 2014; Le et al., 2014; Secco et al., 2015). Plants also require a mechanism to terminate these responses after stress. The transcriptional regulation of genes is closely linked with their epigenetic status (Iwasaki and Paszkowski, 2014). CHG and CHH DNA methylation usually participate in the regulation of heterochromatin formation and transcriptional gene silencing, while the methylation sites in gene bodies are predominantly in the CG context (Sijen et al., 2001; He et al., 2011; Melnyk et al., 2011).

graphic file with name TPC_201800074DR1_fx1.jpg

Open in a new tab

There is increasing evidence that epigenetic regulation is essential for plant stress responses. In Arabidopsis, a large number of genes that respond to phosphate starvation are associated with hypomethylation in their upstream regions (Yong-Villalobos et al., 2015). Mutants in the RdDM pathway show a lower survival rate compared with the wild type under heat stress, indicating that the RdDM pathway is required for heat stress tolerance in plants (Popova et al., 2013).

Several studies have been performed on the role of DNA methylation in male reproductive development. In Arabidopsis, DNA demethylation occurs in vegetative cells and sperm cells and is associated with the reactivation of transposable elements (TEs) and transposition. However, reactivated TEs do not initiate transposition in fertilized zygotes (Slotkin et al., 2009). siRNAs generated from retrotransposons accumulate in pollen and sperm cells, suggesting that epigenetic reprogramming occurs during reproductive development (Slotkin et al., 2009). Other studies have revealed that CG and CHG DNA methylation sites remain stable in the plant germline during development, but CHH methylation levels are reduced in retrotransposons of sperm cells and microspores. The lost CHH methylation is subsequently restored by RdDM in both fertilized embryos and vegetative cells (Calarco et al., 2012). These findings indicate that the epigenetic reprogramming, especially via RdDM in germ cells, participates in the silencing of transposons and the regulation of development.

Previous DNA methylation studies have typically focused on model plants or seedlings under HT stress (Pecinka et al., 2010; Popova et al., 2013). Male reproductive organs are more sensitive to damage from environmental change than vegetative organs (Strømme et al., 2015). In a recent study, we detected changes in the levels of DNA methylation in cotton (Gossypium hirsutum) anthers under high temperature (HT) in both HT-tolerant and HT-sensitive cotton cultivars (Min et al., 2014). However, how DNA methylation is linked with male sterility under HT remains unclear. Here, we used the HT-tolerant cotton cultivar 84021 and HT-sensitive cotton cultivar H05 to uncover the role of DNA methylation during the HT response in anthers using whole genome bisulfite sequencing. Our results provide evidence for the mechanism by which HT disrupts DNA methylation to cause sterility in the anther, providing important insights into breeding using epigenetic techniques.

RESULTS

Single-Base Resolution Maps of DNA Methylation in Cotton Anthers under HT Stress

We previously identified two cotton lines that respond differentially to HT: 84021, which is HT tolerant and H05, which is HT sensitive. HT-induced male sterility in H05 is characterized by an indehiscent anther wall with abortive pollen grains, while 84021 shows normally developed anthers and pollen grains under HT (Figures 1A and 1B). HPLC analysis reveals different autologous genome methylation levels between 84021 and H05 under normal temperature (NT) conditions, and 84021 shows higher DNA methylation than H05 under HT (Min et al., 2014). To explore the roles of DNA methylation in male fertility under HT, we sampled anthers from 84021 and H05 at the tetrad stage, tapetum degradation stage, and anther dehiscence stage under NT and HT and performed whole-genome bisulfite sequencing (BS-seq).

Figure 1.

Figure 1.

Open in a new tab

Male Fertility Phenotypes and DNA Methylation Patterns during Anther Development under HT Stress.

(A) At the anther dehiscence stage, both 84021 and H05 released pollen grains normally under NT. Under HT, 84021 dehisced normally, while H05 showed indehiscence. Bar = 5 mm.

(B) 2,3,5-Triphenylterzolium staining of pollen in 84021 and H05 under NT and HT. Both 84021 and H05 produced normal pollen (red arrows) under NT. Under HT, 84021 produced normal pollen, while H05 produced sterile pollen (blue arrows). Bar = 50 μm.

(C) Distribution of CG, CHG, and CHH methylation sites on chromosome A01 in 84021 under NT (blue lines), H05 under NT (green lines), 84021 under HT (red lines), and H05 under HT (purple lines) at the tetrad stage. CG and CHG methylation sites were enriched in heterochromatic regions, while CHH methylation sites were relatively enriched in euchromatic regions. Significant changes in CHH methylation occurred, while CG and CHG remained stable under HT on a chromosome-wide scale. The approximate centromere regions are indicated by a gray rectangle. The y axis shows the relative methylation levels calculated as the number of mCs/number of totalCs (%) in each 1-Mb genomic region.

(D) Distribution of short (<0.5 kb, blue columns), medium (0.5–4 kb, green columns), and long (>4 kb, red columns) TEs on each chromosome. The x axis shows the G. hirsutum chromosome numbers, and the y axis shows the number of each type of TE in every 1-Mb region of each chromosome. Short TEs occupy a large proportion of total TEs. Long TEs are relatively rare and are spread uniformly on each chromosome.

(E) CHH methylation levels in 84021 under NT (blue lines), H05 under NT (green lines), 84021 under HT (red lines), and H05 under HT (purple lines) on short, medium, and long TEs at the tetrad stage. Short and long TEs showed hyper-CHH methylation in TE bodies compared with upstream (−2 kb) and downstream (+2 kb) regions, while medium TEs showed only slight changes in CHH methylation.

(F) CG and CHG showed little change on a genome-wide scale under HT, while CHH changed significantly under HT. 84021 showed hyper-CHH methylation at the tetrad stage (TS), tapetum degradation stage (TDS), and anther dehiscence stage (ADS) under HT. H05 showed hypo-CHH methylation at the tetrad stage (TS) and tapetum degradation stage (TDS) while displaying hyper-CHH methylation at the anther dehiscence stage (ADS) under HT. The comparison was performed by comparing the methylation levels of each 1-Mb genomic region between NT and HT (1947 comparison pairs). Significant difference from NT; statistical significances were determined by paired two-tailed Student’s t tests, *P < 0.05 and **P < 0.01.

After trimming adapters and filtering low-quality reads, ∼1.2 billion paired-end reads were generated. There were more than 100 million cytosines covered in each sample, a number sufficient for analysis (Supplemental Table 1). We used MethylKit (Akalin et al., 2012) software to evaluate the correlation between the replicates of each sample (Supplemental Figure 1). We used bisulfite-treated lambda DNA to evaluate the bisulfite conversion rate (Supplemental Table 2) and evaluated the false methylation rate by analyzing the methylation status of the mitochondrial genomes of each replicate per sample (Supplemental Table 3). The false methylation rates were relatively low compared with the results reported by Zhang et al. (2015a). We identified methylated cytosines using bismark_methylation_extractor software and binomial tests (Krueger and Andrews, 2011) (Supplemental Figure 2). Low methylated cytosines were all filtered out (Supplemental Figure 3), and ∼63 to 93 million methylated-cytosines (mCs) in each sample were identified. Of these mCs, ∼29 to 37 million mCs (∼40 to 52% of total mCs) were in the CG context, ∼23 to 30 million mCs (∼32 to 41% of total mCs) were in the CHG context, and ∼4 to 25 million mCs (∼6 to 27% of total mCs) were in the CHH context (Supplemental Figure 4 and Supplemental Table 4). Among the three contexts, a large fraction of cytosines in the CG and CHG contexts were strongly methylated, while few CHH sites were methylated (Supplemental Figure 4). From the chromosome-scale viewpoint, CG and CHG methylation sites were enriched in heterochromatic regions, and CHH sites were relatively enriched in chromosomal arms compared with the pericentromere (Figure 1C; Supplemental Figures 5A to 5C). In the figure, the approximate centromere regions (Wang et al., 2015) are indicated by gray rectangles (Figure 1C) and light-gray boxes (Supplemental Figures 5A to 5C). We constructed a website to show the profiles and details of DNA methylation of 84021 and H05 at the tetrad, tapetum degradation, and anther dehiscence stages under NT and HT (http://cgrd.hzau.edu.cn/cgi-bin/gb2/gbrowse/CottonReSequencing2016/).

The Hyper-CHH Methylation on Chromosomal Arms May Be Driven by Euchromatin-Preferential TEs in Cotton Anthers

DNA methylation usually occurs in heterochromatic regions, with the effect of preserving genome stability and silencing TEs. This has been confirmed in many species, including Arabidopsis, maize (Zea mays), rice (Oryza sativa), and cotton (Gossypium barbadense) (Li et al., 2015; Zhang et al., 2015a; Rigal et al., 2016; Wang et al., 2016). However, in this study, we found that CHH methylation sites were relatively enriched on chromosomal arms, i.e., in euchromatic regions.

Different CHH methylation patterns have been found in the different types of TEs (Song et al., 2013; Wang et al., 2016). To understand the basis of hyper-CHH methylation on chromosomal arms, we divided the TEs into short (<0.5 kb), medium (0.5–4 kb), and long (>4 kb) classes, as previously described (Wicker et al., 2007). We examined the distribution of these three types of TEs on the genome, finding that short and medium TEs represent a major proportion of TEs across the whole genome (Supplemental Data Set 1). There are more short TEs than medium and long TEs in every position of each chromosome, with long TEs distributed evenly on each chromosome in both the At and Dt subgenomes (Figure 1D).

We also analyzed the CHH methylation levels in the three types of TEs, including TE bodies and their upstream (−2 kb) and downstream (+2 kb) regions. No significant CHH methylation occurred in medium TE bodies, upstream or downstream regions. However, hyper-CHH methylation occurred in short and long TEs bodies, but not in their upstream or downstream regions (Figure 1E). These data demonstrate that short TEs are preferentially enriched on each chromosomal arm and that hyper-CHH methylation is initiated in the short TEs regions, which may lead to the hyper-CHH methylation observed on chromosomal arms.

HT-Induced Changes in CHH Methylation Patterns in 84021 and H05

To investigate the difference in DNA methylation patterns under HT, we examined the cytosines that were covered by sequencing reads in the two replicates of each sample and identified the methylated cytosines in both replicates for further analysis via binomial tests. We analyzed the changes in DNA methylation by comparing HT to NT treatments in the HT-tolerant 84021 and HT-sensitive H05 cultivars. No significant changes in the number of CG and CHG methylation sites were observed under HT conditions in 84021 or H05 at the three anther stages. However, significant changes in the number of methylation sites in the CHH context were observed (Figure 1F; Supplemental Figure 2 and Supplemental Table 4). At the tetrad and tapetum degradation stages, H05 had reduced CHH methylation under HT but increased CHH methylation under HT at the anther dehiscence stage (Figure 1F). 84021 showed increased CHH methylation under HT compared with NT conditions at the tetrad, tapetum degradation, and anther dehiscence stages.

To decipher the patterns of variation in DNA methylation across the whole genome, we divided the genome into 100-bp regions with no overlaps to identify differentially methylated regions (DMRs). All mCs were mapped to the corresponding genome (At and Dt). Putative DMRs were pooled via Fisher’s exact test (cutoff with P value < 0.05) (Calarco et al., 2012), followed by multiple testing correction (false discovery rate [FDR] <0.05), and were subsequently classified into the CG, CHG, and CHH contexts (Supplemental Table 5 and Supplemental Data Sets 2 to 4). We then calculated the ratio between the number of hyper-DMR and hypo-DMR sites in the whole genome to investigate the relationship between DMRs and changes in whole-genome methylation in 84021 and H05. At the tetrad and tapetum degradation stages, 84021 presented a higher ratio of hyper-DMRs than H05 in the CG, CHG, and CHH contexts, indicating the existence of genome-wide hyper-methylation levels in 84021 under HT. At the anther dehiscence stage, both 84021 and H05 showed more hyper-DMRs than hypo-DMRs under HT (Supplemental Table 5). To further explore the function of DMRs, we analyzed the distribution of DMRs along the genome. We mapped the DMRs to both the At and Dt subgenomes. DMRs in the CG and CHG methylation contexts were randomly distributed (Supplemental Figure 6). However, DMRs in the CHH context were distributed uniformly across the chromosomes (Figure 2B), which contrasts with the data for soybean (Glycine max), wheat (Triticum aestivum), and cotton fibers (Song et al., 2013; Gardiner et al., 2015; Wang et al., 2016). These results demonstrate that changes in DNA methylation across the genome occur widely when anthers suffer HT stress, with significant hyper-CHH methylation in HT-tolerant 84021 and hypo-CHH methylation in HT-sensitive H05 at the tetrad and tapetum degradation stages under HT.

Figure 2.

Figure 2.

Open in a new tab

RdDM Is Involved in Regulating CHH Methylation under HT.

(A) Identification of CHH methylation in regions containing 24-nt siRNA (siRNA+) and regions lacking 24-nt siRNA (siRNA−) in 1M genomic regions. Hyper-CHH methylation was identified in siRNA+ regions in 84021 and H05 under both NT and HT. The y axis represents CHH methylation levels. Significant difference between siRNA+ and siRNA− regions, Student’s t test, **P < 0.01.

(B) Circos plot showing the changes in CHH methylation and 24-nt siRNAs under HT in 84021 and H05 at the tetrad stage, tapetum degradation stage, and anther dehiscence stage. The outermost track represents the 26 chromosomes (A01–A13 for the At subgenome and D01–D13 for the Dt subgenome) of the G. hirsutum genome. The other tracks represent the following: 1, TE density; 2, number of protein-coding genes (PC genes); 3, CHH DMRs between 84021/NT/TS (84021 under NT at the tetrad stages) and 84021/HT/TS (84021 under HT at the tetrad stages); 4, CHH DMRs between H05/NT/TS (H05 under NT at the tetrad stage) and H05/HT/TS (H05 under HT at the tetrad stage); 5, CHH DMRs between 84021/NT/TDS (84021 under NT at the tapetum degradation stage) and 84021/HT/TDS (84021 under HT at the tapetum degradation stage); 6, CHH DMRs between H05/NT/TDS (H05 under NT at the tapetum degradation stage) and H05/HT/TDS (H05 under HT at the tapetum degradation stage); 7, CHH DMRs between 84021/NT/ADS (84021 under NT at the anther dehiscence stage) and 84021/HT/ADS (84021 under HT at the anther dehiscence stage); 8, CHH DMRs between H05/NT/ADS (H05 under NT at the anther dehiscence stage) and H05/HT/ADS (H05 under HT at the anther dehiscence stage); 9, DSRs between 84021/NT/TS and 84021/HT/TS; 10, DSRs between H05/NT/TS and H05/HT/TS; 11, DSRs between 84021/NT/TDS and 84021/HT/TDS; 12, DSRs between H05/NT/TDS and H05/HT/TDS; 13, DSRs between 84021/NT/ADS and 84021/HT/ADS; 14, DSRs between H05/NT/ADS and H05/HT/ADS. Data analysis for each chromosome was performed using 1-Mb sections. The approximate centromere regions are indicated by gray boxes. For tracks 3 to 8, each column represents the ratio between hypermethylated DMRs and hypomethylated DMRs, while for tracks 9 to 14, each column represents the ratio between hyper-DSRs and hypo-DSRs. The ratio between hypermethylated DMRs and hypomethylated DMRs is displayed as red columns versus green columns, and the ratio between hyper-DSRs and hypo-DSRs is presented as purple columns versus yellow columns.

(C) Identification of CHH methylation levels of DSRs in 84021 and H05 at the tetrad stage (TS), tapetum degradation stage (TDS), and anther dehiscence stage (ADS). Hyper-CHH methylation was identified in the hyper-DSRs, while hypo-CHH methylation was detected in the hypo-DSRs. The y axes in the upper panel show siRNA density normalized to 10*TPM, and the y axes in the lower panel show CHH methylation levels. The 10*TPM and CHH methylation levels under NT are indicated by blue boxes, and those under HT are indicated by red boxes. Significant difference, Student’s t test, **P < 0.01.

The RdDM Pathway Plays a Central Role in Altering DNA Methylation in Cotton Anthers in Response to HT

Clear differences in CHH methylation were detected in 84021 and H05 anthers under HT, while CG and CHG showed only minor changes (Figure 1F). Given that RdDM can specifically initiate CHH methylation to regulate transcriptional gene silencing and contributes to chromatin remodeling during multiple stress responses, including responses to HT stress (Zhang and Zhu, 2011; Popova et al., 2013; Matzke and Mosher, 2014), we next focused on changes in the RdDM pathway in 84021 and H05 under HT conditions. First, we utilized the same samples used to perform BS-seq to perform small RNA sequencing in order to evaluate the effect of 24-nt siRNAs on DNA methylation (Supplemental Table 6). After adapter clipping and structural RNA filtering, the 24-nt siRNAs were mapped to the TM-1 cotton genome (Zhang et al., 2015b). We then selected uniquely mapped 24-nt siRNAs and analyzed their distribution on each chromosome. We found that the preferential enrichment of 24-nt siRNAs occurred on each chromosomal arm (Supplemental Figure 5D), which is similar to the distribution of hyper-CHH methylation sites. We performed correlation analysis between CHH methylation levels, the number of 24-nt siRNAs, protein-coding genes, and TEs in each 1-Mb chromosomal region. We detected a high correlation (R = 0.87) between CHH methylation levels and 24-nt siRNAs, as well as a high correlation (R = 0.82) between the number of 24-nt siRNAs and protein-coding genes (Supplemental Figure 7). These results point to a possible relationship between CHH methylation and the density of 24-nt siRNAs.

To further confirm that 24-nt siRNAs actively contribute to altering CHH methylation under HT, we studies the effects of 24-nt siRNAs in the 1-Mb genomic regions. We partitioned the genome into 1-Mb bins: regions harboring 24-nt siRNAs (siRNA-mapped regions) were regarded as siRNA+ regions, and regions lacking siRNA-mapped regions were regarded as siRNA− regions. CHH methylation levels in siRNA+ and siRNA− regions were also identified (Figure 2A). At all three stages of anther development, all siRNA+ regions showed a higher hypermethylation level than the siRNA− regions (Figure 2A). This result suggests that the genome-wide changes in CHH methylation under HT might be associated with 24-nt siRNAs. Therefore, we divided the genome into 100-bp regions to identify differentially siRNA-mapped regions (DSRs), as previously reported (Gent et al., 2014), to identify changes in CHH methylation levels in DSRs. The density of siRNAs in each bin, were normalized using the 10*transcripts per kilobase million (10*TPM) value for each sample. We compared HT to NT conditions and identified DSRs based on the criteria of 10*TPM of bins greater than 0 and changes of more than 2-fold. After identifying the DSRs, we mapped the hypo-DSRs and hyper-DSRs to 1-Mb genomic regions. The distribution of DSRs was similar to that of CHH DMRs (Figure 2B). We then calculated the CHH methylation levels of each DSR. The hyper-DSRs showed hyper-CHH methylation, while the hypo-DSRs showed hypo-CHH methylation (Figure 2C). Furthermore, we analyzed the expression of genes that participate in 24-nt siRNA generation and methylation initiation in the RdDM pathway, such as RDR2, DCL3, HUA ENHANCER1 (HEN1), AGO4, and DRM2. In H05, lower expression levels of these genes were detected at the tetrad and tapetum degradation stages under HT, while HT treatment increased the expression levels of these genes at the anther dehiscence stage. Meanwhile, these genes were all upregulated in 84021 at the tetrad, tapetum degradation, and anther dehiscence stages under HT. These results are consistent with the changes in CHH methylation levels detected in 84021 and H05 (Supplemental Figure 8). In conclusion, under HT, RdDM was disrupted in 84021 and H05 at the tetrad and tapetum degradation stages, which might be associated with reduced CHH methylation levels.

CHH Methylation Changes Significantly in the Promoter and Downstream Regions of Protein-Coding Genes in Anthers under HT Conditions

Our results reveal whole genome-wide changes in CHH methylation under HT conditions and relatively high levels of CHH methylation on chromosomal arms. Given the consensus that chromosomal arms are enriched in protein-coding genes, an interaction between DNA methylation and the expression of protein-coding genes would be expected. To test this hypothesis, we first explored the genome-wide DNA methylation density of protein-coding genes. All of the identified mCs were mapped to genic regions, including gene body, promoter (−2 kb) and downstream (+2 kb) regions. Minor changes were detected in the CG and CHG contexts in protein-coding gene regions in HT compared with NT conditions in anthers at the same stage (Supplemental Figure 9). However, CHH methylation levels were found to exhibit variation in the gene regions (Figure 3A), which is consistent with changes in the three methylation contexts (Figure 1F). We also found that CHH methylation mainly changed in the promoter and downstream regions, which is similar to the distribution of 24-nt siRNAs (Figures 3A and 3B). At the tetrad stage, 84021 showed lower CHH methylation levels than H05 under NT. However, 84021 showed little hyper-CHH methylation under HT, while H05 exhibited significantly reduced CHH methylation in the promoter and downstream regions under HT. At the tapetum degradation stage, compared with NT, there was little difference in CHH methylation in H05 under HT, but 84021 showed a small amount of hyper-CHH methylation under HT. Hyper-CHH methylation was found in both 84021 and H05 compared HT to NT at the anther dehiscence stage, while 84021 showed higher CHH methylation levels than H05 under HT (Figure 3A).

Figure 3.

Figure 3.

Open in a new tab

RdDM Helps Alter CHH Methylation in the Promoter and Downstream Regions of Protein-Coding Genes.

(A) Analysis of CHH methylation levels in gene regions including promoters (−2 kb), gene bodies, and downstream regions (+2 kb) at the tetrad stage (TS), tapetum degradation stage (TDS), and anther dehiscence stage (ADS) in 84021 under NT (blue lines), H05 under NT (green lines), 84021 under HT (red lines), and H05 under HT (purple lines). Significant differences were detected in the promoter and downstream regions. TSS, transcription start site; TTS, transcription termination site.

(B) Analysis of the density of 24-nt siRNAs in gene regions including promoters, gene bodies, and downstream regions in 84021 under NT (blue lines), H05 under NT (green lines), 84021 under HT (red lines), and H05 under HT (purple lines). The density of 24-nt siRNAs also differed in the promoters and downstream regions, as did CHH methylation pattern.

(C) Analysis of CHH methylation of promoters containing 24-nt siRNA sites (P_siRNA+, red boxes) and lacking 24-nt siRNA sites (P_siRNA−, shown in green boxes) in 84021 and H05 at the tetrad stage, tapetum degradation stage, and anther dehiscence stage under NT and HT. The promoters of P_siRNA+ genes showed hyper-CHH methylation compared with those of P_siRNA− genes. There was slight difference in CHH methylation levels among the promoters of P_siRNA− genes.

(D) Number of genes containing 24-nt siRNAs mapped in promoters in anthers at the tetrad stage, tapetum degradation stage, and anther dehiscence stage. In anthers at the same developmental stage, genes commonly detected in 84021 under NT, H05 under NT, 84021 under HT, and H05 under HT were defined as common genes (yellow columns), and the remaining genes were defined as 84021-specific genes (green columns) and H05-specific genes (blue columns). Enriched GO terms are shown on the right, with cutoff at P value < 0.05.

To visualize the relationship between DMR and protein-coding gene regions, we mapped DMRs in the CG, CHG, and CHH contexts to promoter (−2 kb), gene body, and downstream (+2 kb) regions. The results showed that more CG and CHG DMRs were mapped to gene bodies and that CHH DMRs mainly mapped to promoter and downstream regions (Supplemental Figure 10). Considering there were many more DMRs in the CHH context (Supplemental Table 5) and the higher enrichment of 24-nt siRNAs on the promoters (Figure 3B), we analyzed the effect of 24-nt siRNAs on the promoters of protein-coding genes. Genes with 24-nt siRNAs that were mapped to promoters were identified as P_siRNA+ genes, and other genes were classified as P_siRNA− genes. We then analyzed CHH methylation levels on the promoters of P_siRNA+ and P_siRNA− genes. At all stages of anther development and in both HT-tolerant and sensitive cotton under both HT and NT conditions, hyper-CHH methylation levels were higher on the promoter regions of P_siRNA+ genes compared with P_siRNA− genes (Figure 3C). Meanwhile, minor differences in CHH methylation levels were found between 84021 and H05 in anthers at the same stage, which further strengthens the conclusion that 24-nt siRNAs are responsible for altering CHH methylation levels in gene promoter regions and regulating gene expression under HT conditions in anthers.

Different changes in the RdDM pathway were found between 84021 and H05 (Supplemental Figure 8). Therefore, to explore the specific interactions between 24-nt siRNAs and protein-coding genes in 84021 and H05, we analyzed P_siRNA+ genes. In anthers at the same developmental stage, P_siRNA+ genes detectable in all four samples (84021 under NT and HT conditions; H05 under NT and HT conditions) were identified as common genes, and the remaining genes were classified as 84021-specific or H05-specific P_siRNA+ genes (Figure 3D). At the tetrad and tapetum degradation stages, there was little difference in the number of specific P_siRNA+ genes in 84021 versus H05 under NT conditions, but H05 had fewer specific P_siRNA+ genes than 84021 under HT, suggesting weaker RdDM regulation at these stages in H05 under HT (Figure 3D). At the anther dehiscence stage, both 84021 and H05 had more specific P_siRNA+ genes, indicating that RdDM was strengthened in these two samples under HT. We subjected the sample-specific P_siRNA+ genes to Gene Ontology (GO) analysis to identify any enriched pathways. Oxidoreductase activity and carbohydrate binding were found to be enriched (Figure 3D), suggesting that genes involved in energy metabolism and redox homeostasis might be regulated by RdDM under HT.

Global Depression of DNA Methylation Leads to Pollen Sterility

The HT-sensitive line H05 showed reduced DNA methylation levels under HT conditions. We hypothesized that reduced DNA methylation has a negative effect on male fertility under HT. To further investigate the role of DNA methylation in male sterility caused by HT stress, we treated H05 plants with Zebularine (Zeb), a DNA methylation inhibitor. We applied four different treatments via spray application to buds: 150 μM Zeb solution to H05 under NT and HT conditions [HNZ (H05+NT+Zeb), HHZ (H05+HT+Zeb)], and water under NT and HT to H05 as controls [HNW (H05+NT+water), HHW (H05+HT+water)]. After treatment, we performed tissue sectioning of treated anthers to determine any developmental effects of the treatments.

Under control treatments, both HNW and HHW showed normal tetrad formation (Figures 4A and 4B) and normal callose staining with aniline blue (Figures 4A’ and 4B’) at the tetrad stage. No significant difference in tetrad or tapetum formation was found at the tetrad stage (Figures 4A and 4B). At the tapetum degradation stage, normally formed microspores were observed in HNW-treated anthers (Figure 4C), but HHW treatment led to the production of shriveled microspores with fewer inclusions (Figure 4D); this microspore phenotype is similar to a previously reported male sterility phenotype (Cecchetti et al., 2008, 2017). At the anther dehiscence stage, pollen was released normally from HNW anthers (Figure 4E), while plants treated with HHW had completely shriveled pollen grains and an indehiscent anther wall (Figure 4F). These results further confirmed the distinct male reproductive phenotype of H05 under NT and HT conditions.

Figure 4.

Figure 4.

Open in a new tab

The Suppression of DNA Methylation Induces Microspore Sterility but Does Not Affect Anther Dehiscence.

Toluidine Blue staining of anther sections of HNW (H05+NT+water), HHW (H05+HT+water), HNZ (H05+NT+Zeb), and HHZ (H05+HT+Zeb) at the tetrad stage (TS), tapetum degradation stage (TDS), and anther dehiscence stage (ADS). No obvious changes in tetrads or tapetum were detected under the four treatments at TS (A), (B), (a), and (b). At TDS, HT induced abnormal microspore formation in H05 following water treatment (D). H05 showed shrunken microspores following Zeb treatment under both NT (c) and HT (d). At ADS, H05 produced normal pollen only under NT with water treatment (E). Dehiscent anther walls and sterile pollen were detected in HNZ (e). Aniline blue staining of secondary wall thickening was performed under HNW, HHW, HNZ, and HHZ treatments at tetrad stage, tapetum degradation stage, and anther dehiscence stage. No obvious secondary wall thickening was identified at TS and TDS. At ADS, HT induced considerable secondary wall thickening in the endothecium (F’) and (f’), while Zeb treatment had little effect on secondary wall thickening in H05 under NT ([E’] and [e’]) and HT ([F’] and [f’]). Bars = 50 μm. Tds, tetrads; T, tapetum; MSP, microspore; PG, pollen grain; En, endothecium.

In response to Zeb treatment, both HNZ and HHZ showed normal tetrad formation at the tetrad stage (Figures 4a and 4b). Abnormal microspores were detected under both NT and HT in response to Zeb treatment at the tapetum degradation stage (Figures 4c and 4d). At the anther dehiscence stage, HHZ showed indehiscent anther walls and abnormal pollen grains (Figure 4f), which was similar to the phenotype observed under HHW treatment (Figure 4F). Unexpectedly, the anther endothecium of HNZ dehisced normally but contained barren pollen grains (Figure 4e).

Anther dehiscence is related to the thickness of secondary walls of the endothecium (Mitsuda et al., 2005; Zhao et al., 2010). We therefore performed aniline blue staining of anther tissue sections to examine secondary wall thickening in the endothecium. At the tetrad stage, no significant changes in tetrads were found in any of the treatment groups (Figures 4A’, 4B’, 4a’, and 4b’). At the tapetum degradation stage, only the HNW (H05+NT+water) treatment group had normally shaped microspores (Figures 4C’, 4D’, 4c’, and 4d’). At the anther dehiscence stage, secondary wall thickening in the endothecium was observed under HT (Figures 4F’ and 4f’). There was little difference in secondary wall thickening following Zeb treatment compared with the controls (Figures 4E’, 4e’, 4F’, and 4f’). Therefore, HT induced severe microspore sterility and secondary wall thickening of the endothecium in H05, but the suppression of DNA methylation disrupted microspore development, with minor effects on secondary wall thickening in the endothecium. These results suggest that HT stress disrupts DNA methylation, which affects microspore development and has minor effects on anther dehiscence.

Suppression of DNA Methylation Disrupts Gene and TE Transcription in Anthers

Disordered DNA methylation lead to abnormal development due to disrupted gene and TE expression (Hu et al., 2014; Zhang et al., 2015a). We hypothesized that the observed shrunken pollen grain observed in response to Zeb treatment might be due to disordered DNA methylation. First, we performed BS-seq on anthers at the tapetum degradation stage treated with HNZ and HHZ to evaluate the changes in DNA methylation under Zeb treatment (Supplemental Table 7). Reduced DNA methylation levels under Zeb treatment were observed based on the BS-seq data (Supplemental Figure 11). To further test this hypothesis, we subjected Zeb-treated and control samples to RNA-sequencing to identify transcriptional changes (Supplemental Table 8). We detected an increasing number of differentially expressed TEs in H05 under HT treatment, while Zeb treatment led to increased numbers of differentially expressed TEs under both NT and HT conditions (Supplemental Table 9). We also examined changes in gene expression in H05 following Zeb application and identified differentially expressed genes (DEGs) using TopHat2 and Cuffdiff software. There were more DEGs in H05 following Zeb treatment under both NT and HT treatment compared with the controls (Supplemental Figure 12 and Supplemental Table 10). Given the previous finding that the unexpected transcription of TEs could results in severe growth defects in plants (Hu et al., 2014; Zhang et al., 2015a), we propose that HT disrupts whole-genome methylation, removes DNA methylation on TEs and genes, and leads to their unregulated transcription in H05. Zeb treatment might mimic the HT-induced disruption of hypomethylation and induce unexpected transcription of TEs and genes in H05.

Changes in the Sugar Metabolism Pathway under HT and following Suppression of DNA Methylation

To understand how changes in DNA methylation under HT stress disrupt gene expression and lead to pollen abortion, we identified DEGs under control (water) versus Zeb treatment and subjected them to GO enrichment analysis. Under HT treatment, the DEGs from H05 sprayed with water were enriched in the GO categories “carbohydrate metabolic process,” “plant hormone response,” and especially “auxin signaling” (AUXIN RESPONSE FACTOR [ARF] signaling) and “oxidoreductase activity” (Supplemental Figure 13), which suggests that energy metabolism, auxin signaling, and redox homeostasis are altered by HT stress in anthers. Among DEGs under Zeb treatment, we found that the GO categories “carbohydrate metabolic process” and “response to oxidative stress” were further enriched, but no significant enrichment observed in the category “plant hormone response” (Supplemental Figure 13). These results suggest that greater energy consumption and changes in redox status occur in H05 under Zeb treatment than under HT stress.

To further investigate any changes in carbohydrate content under Zeb or HT treatment, we performed total soluble sugar and starch assays. At the tetrad and tapetum degradation stages, both HT and Zeb treatment induced the accumulation of soluble sugars, with combined HT and Zeb treatment affecting the soluble sugar content more strongly than single treatments (Figure 5A). At the anther dehiscence stage, significant accumulation of soluble sugar was detected following Zeb treatment under both NT and HT conditions compared with the respective controls (Figure 5A). Starch content showed an opposite trend to soluble sugar content in all samples (Figure 5B). We also found that DNA methylation levels on the promoters of several amylase genes were negatively correlated with the expression levels at the three developmental stages, suggesting that amylase genes are regulated by DNA methylation (Figure 5C). The results of starch (I2-KI) staining of pollen grains following HNW (H05+NT+water), HNZ (H05+NT+Zeb), HHW (H05+HT+water), and HHZ (H05+HT+Zeb) treatment strengthened our conclusion that HT induces DNA methylation that disrupts carbohydrate metabolism in pollen (Figure 5D). We therefore conclude that HT alters the DNA methylation status, leading to the excessive expression of amylase genes, thereby resulting in starch hydrolysis and a higher sugar concentration.

Figure 5.

Figure 5.

Open in a new tab

HT-Induced DNA Methylation Is Associated with Altered Sugar Metabolism.

(A) Measurement of total soluble sugar contents in HNW (H05+NT+water), HHW (H05+HT+water), HNZ (H05+NT+Zeb), and HHZ (H05+HT+Zeb) at the tetrad stage (TS), tapetum degradation stage (TDS), and anther dehiscence stage (ADS). Both HT and Zeb treatment induced sugar accumulation in H05. Values not sharing a common letter are considered significantly different (shortest significant range; P < 0.05). The values are means ± sd (n > 5).

(B) Measurement of starch contents in HNW, HHW, HNZ, and HHZ at TS, TDS, and ADS. Starch hydrolysis was induced by HT or Zeb, and increased starch hydrolysis was detected following combined HT and Zeb treatment. Values not sharing a common letter are considered significantly different (shortest significant range; P < 0.05).

(C) The left panel (heat map) shows the expression levels and DNA methylation levels in the promoters of amylase genes. The right panel (genome browser snapshot) shows DNA methylation levels (orange boxes) in different promoters and different expression levels (green boxes) of an amylase gene at TDS under NT and HT.

(D) I2-KI staining of starch in pollen of HNW, HHW, HNZ, and HHZ. H05 shows considerable starch hydrolysis under HT combined with Zeb. Bars = 50 μm.

HT-Induced DNA Methylation Is Associated with ROS Generation but Not with Auxin Accumulation in Anthers

The GO term “oxidoreductase activity” was further enriched following the DNA suppression assay, as described above. We then performed a H2O2 assay to investigate the redox status of pollen under HT or Zeb treatment. At the tetrad stage, HT or Zeb treatment induced a higher level of H2O2 in HT-sensitive H05 (Figure 6A). At both the tapetum degradation and anther dehiscence stages, H05 generated more H2O2 when under HT coupled with Zeb, similar to the results for soluble sugar (Figure 6A). Since it is generally considered that H2O2 is synthesized by respiratory burst oxidase homolog (RBOH) proteins (Mittler et al., 2004; Marino et al., 2012), we analyzed the expression and DNA methylation levels of all RBOH genes. Several RBOH genes were upregulated under HT, which was associated with hypomethylation of promoter regions (Figure 6B), as well as H2O2 concentrations (Figure 6A).

Figure 6.

Figure 6.

Open in a new tab

Suppression of DNA Methylation Induces Excessive ROS Generation in Anthers.

(A) Measurement of H2O2 contents in HNW (H05+NT+water), HHW (H05+HT+water), HNZ (H05+NT+Zeb), and HHZ (H05+HT+Zeb) at the tetrad stage (TS), tapetum degradation stage (TDS), and anther dehiscence stage (ADS). HT or Zeb induced the generation of H2O2, while the combination of HT and Zeb induced greater accumulation of H2O2. Values not sharing a common letter are considered significantly different (shortest significant range; P < 0.05). The values are means ± sd (n > 5).

(B) The left panel (heat map) shows the expression levels and DNA methylation levels in the promoters of RBOH genes. The right panel (genome browser snapshot) shows DNA methylation levels in different promoters (orange boxes) and different expression levels (green boxes) of a RBOH gene at TDS in H05 under NT and HT.

(C) 2’,7’-Dichlorodihydrofluorescein diacetate staining of ROS in pollen of HNW, HHW, HNZ, and HHZ. HT combined with Zeb treatment induced increased levels of H2O2 accumulation in pollen. Bars = 50 μm.

To further investigate reactive oxygen species (ROS) generation in pollen, we performed a ROS staining assay on the same four treatment groups used for starch staining [HNW (H05+NT+water), HNZ (H05+NT+Zeb), HHW (H05+HT+water), and HHZ (H05+HT+Zeb)]. The results show that Zeb or HT significantly induces ROS generation in H05 (Figure 6C), and HT combined with Zeb induced additional ROS generation in H05. We therefore speculate that HT induces hypomethylation to release the RBOH genes from silencing, thereby leading to the enhanced generation of H2O2 under HT.

Auxin contributes to the control of anther dehiscence by regulating endothecium lignification and the jasmonic acid pathway in Arabidopsis (Cecchetti et al., 2013). We previously showed that HT alters auxin metabolism and signaling and causes anther abortion in cotton (Min et al., 2014). The DEGs under HT treatment show significant enrichment for the GO term “hormone response pathway,” but the DEGs under Zeb treatment did not. Therefore, we investigated whether the changes in auxin metabolism and signaling pathways might be caused by disrupted DNA methylation under HT.

We investigated the expression of auxin biosynthesis genes at the anther dehiscence stage and found that several auxin biosynthesis genes (such as ALDEHYDE OXIDASE1 [AAO1], AAO2, NITRILASE4 (NIT4), YUCCA4, YUCCA5, and YUCCA6) were upregulated under HT in H05 following treatment with water (Figure 7A). Under Zeb treatment, fewer genes were found to be upregulated (Figure 7A). Given there were only minor changes in the expression levels of auxin biosynthesis genes, we investigated the expression of auxin signaling pathway genes. Several such genes (mostly ARFs) were upregulated in plants under HT conditions treated with water, but no obvious changes were detected under Zeb treatment (Figure 7A). These results suggest that the auxin biosynthesis and signaling pathways are slightly regulated by DNA methylation under HT.

Figure 7.

Figure 7.

Open in a new tab

Indehiscence of the Endothecium, Which Is Regulated by Auxin Biosynthesis and Signaling Pathways, Is Slightly Influenced by HT-Disrupted DNA Methylation.

(A) A heat map of the expression levels of auxin biosynthesis and signaling genes at the anther dehiscence stage in HNW (H05+NT+water), HHW (H05+HT+water), HNZ (H05+NT+Zeb), and HHZ (H05+HT+Zeb). Several auxin biosynthesis and signaling genes are upregulated in HHW, but not induced significantly in HNZ or HHZ. The genes with the same name (e.g., AAO1) represent different copies in the tetraploid cotton genome.

(B) Auxin concentration at anther dehiscence stage (ADS) in HNW (H05+NT+water), HHW (H05+HT+water), HNZ (H05+NT+Zeb), and HHZ (H05+HT+Zeb). HT induced significant accumulation of auxin in H05, but the auxin content was not affected by Zeb treatment under either NT or HT. Values not sharing a common letter are considered significantly different. The values are means ± sd (n > 5) (Student’s t test, P < 0.05).

(C) Immunohistochemical assay of auxin in the endothecium in HNW, HHW, HNZ, and HHZ.

(A) to (D) show the results of the negative control in HNW (A), HHW (B), HNZ (C), and HHZ (D), respectively. (a’) and (c’) show auxin accumulation in HNW (A) and HNZ (C). (b’) and (d’) show that HT induces auxin accumulation in the endothecium under HT in response to treatment with water (b’) and Zeb (d’). Zeb treatment alters the auxin content only slightly in the endothecium under both NT ([a’] and [c’]) and HT ([b’] and [d’]). Bar = 50 μm.

We therefore hypothesized that auxin concentrations would not change under Zeb treatment and performed auxin assays in anthers at the anther dehiscence stage under the following treatments: HNW (H05+NT+water), HNZ (H05+NT+Zeb), HHW (H05+HT+water), and HHZ (H05+HT+Zeb). We found that HT stress induced auxin accumulation in the anther, but no significant changes were detected between water and Zeb treatment under NT or HT (Figure 7B). To determine whether the auxin concentration in the endothecium differed under HT versus Zeb treatment, we investigated the distribution of auxin in the endothecium tissue of anthers at one day before anthesis via an immunohistochemical assay. As shown in Figure 7C, the controls showed no significant differences across the four treatments (Figure 7C, a to d). HT induced the accumulation of auxin in the endothecium (Figure 7C, b’ and d’), but Zeb treatment had no significant effect on auxin concentration (Figure 7C, a’ and c’). These results suggest that HT induces auxin accumulation in the endothecium and that this might cause anther indehiscence, but it is likely that HT-disrupted DNA methylation does not play a major role in this auxin-mediated process.

DISCUSSION

Global warming is increasing the mean temperature annually (Bita and Gerats, 2013), leading to HT stress to crops and resulting in male sterility and yield reductions in rice, wheat, maize, and cotton (Peng et al., 2004; Tang et al., 2006; Sakata et al., 2010; Min et al., 2014). We previously showed that two cotton lines, 84021 (HT-tolerant) and H05 (HT-sensitive), exhibit different male fertility phenotypes when subjected to 1 week of HT stress. Gene expression profile analysis revealed a significant change in the number of DEGs between 84021 and H05. Further analysis showed significant changes in the expression of genes involved in sugar metabolism and auxin signaling pathways, suggesting roles for energy metabolism and plant hormone pathways in HT stress responses in cotton anthers (Min et al., 2014). However, the mechanistic basis for the differential responses observed in 84021 and H05 had been unknown.

DNA methylation regulates gene expression through transcriptional gene silencing (Zhang and Zhu, 2011; Popova et al., 2013; Matzke and Mosher, 2014). By performing bisulfite sequencing of 84021 (HT-tolerant) and H05 (HT-sensitive) under NT and HT conditions at three different stages of anther development, we comprehensively analyzed the roles of DNA methylation during cotton anther development in response to HT. Our results revealed several intriguing DNA methylation patterns in anthers under HT. First, CG and CHG methylation sites were initiated in heterochromatic regions, but CHH methylation sites were relatively enriched on chromosomal arms. The hypermethylation in the CHH context occurred preferentially in the euchromatin-preferential TEs, which may have caused the unusual CHH methylation pattern detected in anthers. How this CHH methylation pattern is generated is still unknown. Second, few changes in CG and CHG methylation were identified in HT compared with NT, while CHH methylation sites changed significantly in both HT-tolerant 84021 and HT-sensitive H05 under HT. We conclude that hyper-CHH methylation may play a more important role than CG and CHG in the response to HT stress in anthers.

The methylation sites in gene bodies were predominantly in the CG context, while CHG and CHH showed hyperdensity at promoter regions (−2 kb). There were few changes in CG and CHG methylation in genic regions, including promoter (−2 kb), gene body, and downstream regions (+2 kb). Previous work indicated that disrupted DNA methylation disrupts gene expression, leading to seedling lethality in rice (Hu et al., 2014). Based on the consensus that CHH methylation participates in transcriptional gene silencing, the small changes in CG and CHG methylation under HT in anthers further support our hypothesis that disrupted CHH methylation under HT disrupts gene expression, leading to male sterility.

The analysis of hypermethylation levels in the siRNA+ regions indicated that 24-nt siRNAs participate in initiating DNA methylation across the genome. Further analysis of the DSRs confirmed the function of 24-nt siRNAs in altering genome-wide CHH methylation. The expression of genes that participate in the regulation of the RdDM pathway (RDR2, DCL3, AGO4, HEN1, and DRM2) was significantly altered under HT but showed contrary expression patterns in HT-tolerant 84021 versus HT-sensitive H05. The changes in CHH methylation and 24-nt siRNA density correspond to the expression patterns of the genes, perhaps implying that the RdDM pathway changes under HT conditions, although how the expression levels of genes involved in RdDM exhibited different changes in different samples remains unclear.

Disordered sugar metabolism is observed in anthers subjected to HT (Min et al., 2014). Both sugar and starch are vital during male reproductive development, as they serve as important energy sources (Yui et al., 2003; Zhang et al., 2010; Zhao et al., 2010; Zhu et al., 2015). Our transcriptome and experimental analysis also showed that starch hydrolysis was enhanced when DNA methylation was suppressed. These results suggest that HT-disrupted DNA methylation enhances the expression of amylase genes and resulted in the excess of consumption of starch, leading to male sterility in H05.

ROS-dependent cellular and metabolic processes occur during anther development (Hu et al., 2011; Xie et al., 2014), and unbalanced ROS metabolism results in male sterility. In rice, ROS levels are downregulated during late anther development to protect pollen grain maturation, as supported by our H2O2 measurements (Hu et al., 2011). Our results show that HT-induced hypo-DNA methylation levels on the promoters of RBOH genes led to their higher expression and the generation of H2O2 in microspores, which was detrimental to pollen development. These results show that hypo-DNA methylation in H05 under HT alleviates the silencing of amylase and RBOH genes, leading to excessive starch hydrolysis and ROS accumulation, thereby resulting in microspore abortion. This finding indicates that disrupted DNA methylation disrupts two different pathways to induce male sterility under HT.

In Arabidopsis, auxin treatment reduces lignification of the endothecium to induce anther dehiscence (Cecchetti et al., 2013) and can rescue male sterility in wheat and Arabidopsis (Sakata et al., 2010). Our auxin analysis showed that H05 accumulates increased levels of auxin in the endothecium under HT, which is in contrast to the results in Arabidopsis, in which auxin exerts a positive effect on male sterility. The auxin assay also showed that the suppression of DNA methylation does not cause changes in auxin concentration in H05, suggesting that DNA methylation does not participate significantly in regulating auxin biosynthesis or signaling. These results suggest that auxin acts differently in different crops under HT stress.

METHODS

Plant Materials

The cotton (Gossypium hirsutum) lines 84021 (HT-tolerant) and H05 (HT-sensitive) used in this study were cultivated in the greenhouse under a 14-h-day/10-h-night photoperiod. All buds were sampled and divided into the tetrad stage (6–7 mm), tapetum degradation stage (9–14 mm), and anther dehiscence stage (>24 mm) by bud length (Ma et al., 2012; Min et al., 2014). The same stage of anthers from the same line were harvested, pooled in tubes, and stored in liquid nitrogen or at −70°C immediately for future use.

HT Treatment Procedures and in Vitro Application of the DNA Methylation Inhibitor Zeb

84021 and H05 were planted in the greenhouse for various treatments. Plants under NT (29–35°C daytime and 25–28°C at night) were used for the negative control. For HT treatment, plants were moved to a greenhouse with temperatures of 39 to 41°C in the daytime and 29 to 31°C at night.

The DNA methylation inhibitor Zeb (Selleck; catalog no. S7113) was dissolved in distilled water and sprayed onto buds to suppress DNA methylation. To evaluate the most suitable concentration of Zeb for use, a graded solution series (100, 150, 200, and 250 μM) was applied to H05 under either NT or HT conditions. Because all buds of H05 dropped after 200 and 250 μM Zeb treatment under NT, 150 μM Zeb was chosen for methylation suppression treatments under both NT and HT. Zeb solution (150 μM) was sprayed onto all buds under NT conditions. Five days later, half the plants were treated with HT, while the other half remained under NT. Plants cultivated under the same conditions and treated with distilled water were used as a control. The samples from different batches under HT or Zeb treatment were stored separately as different biological replicates.

DNA Extraction, Bisulfite Treatment, and Library Construction

The anthers from two different batches under HT or Zeb treatment were used for DNA extraction, which was regarded as two biological replicates. Total DNA was extracted using a Plant Genomic DNA Kit (Tiangen; catalog no. DP305). Approximately 3 μg of DNA was collected for bisulfite-conversion using an EZ DNA Methylation-Gold Kit (Zymo Research; catalog no. D5005). Illumina sequencing libraries were constructed using a TruSeq DNA methylation kit following the manufacturer’s instructions. Unmethylated lambda genomic DNA (Promega) was used as a control; the lambda DNA was treated each time anther libraries were constructed. The treated lambda DNA was sequenced together with the anther library to evaluate the conversion rate. The false methylation rate of each replicates per sample was evaluated by analyzing the methylation levels of mitochondrion (Liu et al., 2013). Sequencing was performed on the Illumina HiSeq 2000 platform.

DNA Methylation Data Analysis

Low-quality sequence data for 84021, H05, and lambda DNA were trimmed using Trimmomatic software (Bolger et al., 2014). Bisulfite nonconversion rates (0.003 of CG, 0.003 of CHG, and 0.003 of CHH) were evaluated by resequencing the BS-treated lambda DNA (Yong-Villalobos et al., 2015; Zhang et al., 2015a; Wang et al., 2016) (Supplemental Table 2).

Two biological replicates of clean reads of 84021 and H05 were mapped to the TM-1 genome using Bismark software (Krueger and Andrews, 2011) with the parameters –N 1 –L 30. The paired-end reads that uniquely mapped to the genome were reserved for further analysis. MethylKit software was used to evaluate the correlation between two replications of each sample (Akalin et al., 2012). The cytosines that were detected in two sets of BS-data were selected to identify putative mCs. Putative mCs were extracted using bismark_methylation_extractor software with the parameters –no_overlap–CX_context. All putative mCs were pooled into the classic binomial test with a cutoff P value < 1e-4. True mCs were determined based on P value under the binomial distribution Inline graphic, where mCs = number of mCs; totalCs = mCs + unmethylated Cs; error_rate is the error rate for the nonconversion rate of the lambda DNA.

For DMR calling, the whole genome was divided into 100-bp bins with no overlap, and all mCs identified by binomial test in two replicates were mapped to 100-bp bins. Fisher’s Exact test was then performed with the cutoff at 0.05 (Ausin et al., 2012; Calarco et al., 2012; Guo et al., 2014; Zhang et al., 2015a; Groth et al., 2016; Wang et al., 2016). Multiple testing correction (FDR < 0.05) was subsequently followed to test each window.

Small RNA Library Preparation and Sequencing

A total of 10 μg RNA from each sample was prepared using a modified Guanidine Thiocyanate method (Min et al., 2013), and 5S sections of RNA were separated by agarose gel electrophoresis. Small RNA libraries were constructed using TruSeq Small RNA Library Preparation Kits (Illumina) following the manufacturer’s protocol, with two biological replicates. Clean reads (18–26 bp) were obtained after adapter clipping and raw data trimming. Structural RNAs such as rRNA, snRNA, and tRNA were filtered via alignment to Rfam (http://rfam.xfam.org/) and the miRBase (http://www.mirbase.org/) database was used to predict putative microRNAs.

All remaining reads were mapped to the cotton genome using Bowtie (version 1.1.1) (Langmead et al., 2009), allowing no mismatches (-a -v 0 -m 200). Only 24-nt siRNAs that uniquely mapped to the genome, with no overlaps with each other, were selected for further analysis.

RNA-Seq and Data Analysis

Anthers from two different batches under HT treatment were sampled for RNA extraction. Total RNA was extracted using a modified Guanidine Thiocyanate method (Min et al., 2013). Approximately 3 μg RNA was used to construct libraries with a TruSeq Stranded Total RNA Kit with two biological replicates per sample (https://www.illumina.com/techniques/sequencing/rna-sequencing/total-rna-seq.html). Adapters and low-quality reads were clipped using Trimmomatic software (Bolger et al., 2014). All remaining reads were mapped to the cotton genome using TopHat2 (Ghosh and Chan, 2016). Further identification of DEGs was performed using Cuffdiff software with a cutoff P value < 0.05 (Ghosh and Chan, 2016).

Tissue Sectioning, Staining, and Imaging

Bracts and petals were removed from buds, which were subsequently immersed in 50% FAA (50 mL absolute ethanol, 10 mL 37% formaldehyde solution, and 5 mL acetic acid, diluted with water to 100 mL) and vacuum infiltrated three times for 15 min to fix the tissue. After infiltration, the solution was replaced with fresh FAA solution and postfixed at 4°C for at least 12 h. Fixed samples were dehydrated using a graded ethanol series (30, 50, 70, 95, and 100%) for 1 h at each concentration and embedded in paraffin. Embedded tissues were sectioned to 10 μm thickness. Toluidine blue solution (1%) and aniline blue solution (1%) were used to stain the anther sections. A Zeiss Axio Scope A1 microscope was used to image the samples under bright field for Toluidine Blue staining and at 395-nm excitation for aniline blue staining.

ROS and Starch Staining of Pollen Grains

Pollen from different treatments including H05+NT+water (HNW; H, H05; N, normal temperature; W, water), H05+HT+water (HHW; H, H05; H, high temperature; W, water), H05+NT+Zeb (HNZ; H, H05; N, normal temperature; Zeb, Zebularine), and H05+HT+Zeb (HHZ; H, H05; H, high temperature; Zeb, Zebularine) were stained for ROS and starch detection. Flowers on the day of blooming were carefully harvested, the petals were quickly removed, and the samples were immediately immersed in PBS (pH 7, prepared in 15-mL tubes and previously stored at room temperature) to avoid generating a ROS burst during the release of pollen grains. HT treatment was performed in an incubator for 2 h. Flowers under NT conditions were used as a negative control. After treatment, the pollen was incubated for 30 min in the dark in 10 μM 2′,7′-dichlorodihydrofluorescein diacetate dissolved in PBS for ROS staining. Samples were washed twice in PBS before imaging at excitation wavelength of 488 nm and emission wavelength of 522 nm (Tang et al., 2014).

I2-KI solution was used for starch staining in pollen. Samples were washed in PBS after staining in I2-KI solution for 3 min, and images were taken under bright field.

ROS Quantification

Anthers were collected in 2-mL tubes, ground, and extracted using 80% acetone for ∼30 min in the dark at 4°C. ROS quantification was performed using a H2O2 Quantitative Assay Kit (Sangon Biotech; C500069-0250), with at least five biological replicates (samples from different experiments) and two technological replicates (samples from the experiment) for each sample. Concentrations were calculated as μmol/g fresh weight (FW).

Plant Hormone Measurements

Approximately 50 mg anther tissue was sampled and stored in 2-mL tubes, ground using iron balls, and extracted using 80% methanol by shaking at 4°C overnight. Plant hormone measurements were performed on an ABI 4000 Q-Trap (Applied Biosystems) with indole-3-acetic-2,2-d2 acid (Sigma-Aldrich; catalog no. 24420-86-8) used as the internal standard.

Soluble Sugar and Starch Measurements

Soluble sugar was extracted using 80% acetone as described for ROS quantification, and the sediment was collected to measure starch content. Total soluble sugar was determined following the anthrone-sulfuric acid method (Min et al., 2014). Starch measurements were performed via perchloric acid hydrolysis of starch-anthrone sulfuric acid (Min et al., 2014). Both soluble sugar and starch contents were calculated as mg/g FW.

Immunohistochemical Assay of IAA

An immunohistochemical assay of IAA was performed as described previously (Hou and Huang, 2005). IAA in anthers was fixed with carbodiimide hydrochloride (EDAC; Sangon; catalog no. C600433) under a vacuum for 1 h, followed by 50% FAA. Before immunochemistry, anthers were incubated in blocking solution (10 mM PBS, pH 7.2, 0.1% Tween 20, 1.5% glycine, and 5% BSA [Biosharp; catalog no. BS043E]) for 45 min, and IAA antibody (Sigma-Aldrich; catalog no. A0855-200UL) diluted in PBS/BSA (10 mM PBS and 0.8% BSA) as the primary antibody was incubated with the tissue sections. Tissue sections were incubated with an alkaline phosphatase secondary antibody, covered with Parafilm, and incubated with Western Blue Stabilized Substrate for Alkaline Phosphatase (Promega; catalog no. S3841). Incubation was stopped by washing in water or PBS, and the sections were mounted on a cover glass before imagining. Sections incubated with anti-mouse IgG diluted (Promega; catalog no. S3721) in PBS/BSA acted as a negative control.

qRT-PCR

For qRT-PCR, 3 µg RNA was reverse-transcribed using M-MLV (Promega) following the manufacturer’s protocol. qRT-PCR was performed using an ABI 7500 real-time PCR system. Relative gene expression levels were calculated using the 2−ΔCt method as previously described (Min et al., 2014). Expression levels were normalized to GhUBIQUITIN7 as an internal control to standardize RNA content.

Accession Numbers

Sequence data from this article can be found in the GenBank/EMBL libraries under the following accession numbers: GhUBIQUITIN7 (DQ116441), GhRDR2 (Gh_A12G2496), GhRDR2 (Gh_A13G0247), GhDCL3 (Gh_D13G2027), GhHEN1 (Gh_A06G1061), GhAGO4 (Gh_D07G1699), GhAGO4 (Gh_A07G1540), GhAGO4 (Gh_A08G1752), and GhDRM2 (Gh_A09G0264). The whole-genome bisulfite sequencing reads, small RNA sequencing reads, and RNA-sequencing reads have been deposited with the National Center for Biotechnology Information under Sequence Read Archive under accession number PRJNA393079. The profiles of DNA methylation and gene expression levels of 84021 and H05 under NT and HT are presented at the website: http://cgrd.hzau.edu.cn/cgi-bin/gb2/gbrowse/CottonReSequencing2016/.

Supplemental Data

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

Acknowledgments

This work was supported by the China Agriculture Research System (CARS-18-09) and the National Key Research and Development Program of China (2016YFD0101402). We thank Hongbo Liu (Huazhong Agricultural University, China) for assistance with liquid chromatography/mass spectrometry analysis.

AUTHOR CONTRIBUTIONS

X.Z. and L.M. conceived and designed the experiments. Y.M., L.M., Y.L., Y.W., Y.D., X.S., and Q.H. performed HT treatment. Y.M. and Y.W. performed tissue section, pollen staining, and imaging. Y.M., C.W., Y.Z., and Q.F. measured soluble sugar, starch, and ROS contents. Q.Z. constructed the Illumina sequencing libraries. Y.M. and M.W. contributed to data analysis. Y.M. wrote the article, and L.M and X.Z. revised it.

Footnotes

[OPEN]

Articles can be viewed without a subscription.

References

  1. Akalin A., Kormaksson M., Li S., Garrett-Bakelman F.E., Figueroa M.E., Melnick A., Mason C.E. (2012). methylKit: a comprehensive R package for the analysis of genome-wide DNA methylation profiles. Genome Biol. 13: R87. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Ausin I., et al. (2012). INVOLVED IN DE NOVO 2-containing complex involved in RNA-directed DNA methylation in Arabidopsis. Proc. Natl. Acad. Sci. USA 109: 8374–8381. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bita C.E., Gerats T. (2013). Plant tolerance to high temperature in a changing environment: scientific fundamentals and production of heat stress-tolerant crops. Front. Plant Sci. 4: 273. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bolger A.M., Lohse M., Usadel B. (2014). Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30: 2114–2120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Calarco J.P., Borges F., Donoghue M.T., Van Ex F., Jullien P.E., Lopes T., Gardner R., Berger F., Feijó J.A., Becker J.D., Martienssen R.A. (2012). Reprogramming of DNA methylation in pollen guides epigenetic inheritance via small RNA. Cell 151: 194–205. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cao X., Jacobsen S.E. (2002). Locus-specific control of asymmetric and CpNpG methylation by the DRM and CMT3 methyltransferase genes. Proc. Natl. Acad. Sci. USA 99 (suppl. 4): 16491–16498. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Cao X., Aufsatz W., Zilberman D., Mette M.F., Huang M.S., Matzke M., Jacobsen S.E. (2003). Role of the DRM and CMT3 methyltransferases in RNA-directed DNA methylation. Curr. Biol. 13: 2212–2217. [DOI] [PubMed] [Google Scholar]
  8. Cecchetti V., Altamura M.M., Falasca G., Costantino P., Cardarelli M. (2008). Auxin regulates Arabidopsis anther dehiscence, pollen maturation, and filament elongation. Plant Cell 20: 1760–1774. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Cecchetti V., Altamura M.M., Brunetti P., Petrocelli V., Falasca G., Ljung K., Costantino P., Cardarelli M. (2013). Auxin controls Arabidopsis anther dehiscence by regulating endothecium lignification and jasmonic acid biosynthesis. Plant J. 74: 411–422. [DOI] [PubMed] [Google Scholar]
  10. Cecchetti V., Celebrin D., Napoli N., Ghelli R., Brunetti P., Costantino P., Cardarelli M. (2017). An auxin maximum in the middle layer controls stamen development and pollen maturation in Arabidopsis. New Phytol. 213: 1194–1207. [DOI] [PubMed] [Google Scholar]
  11. Chan S.W., Zilberman D., Xie Z., Johansen L.K., Carrington J.C., Jacobsen S.E. (2004). RNA silencing genes control de novo DNA methylation. Science 303: 1336. [DOI] [PubMed] [Google Scholar]
  12. Gardiner L.J., Quinton-Tulloch M., Olohan L., Price J., Hall N., Hall A. (2015). A genome-wide survey of DNA methylation in hexaploid wheat. Genome Biol. 16: 273. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Gent J.I., Madzima T.F., Bader R., Kent M.R., Zhang X., Stam M., McGinnis K.M., Dawe R.K. (2014). Accessible DNA and relative depletion of H3K9me2 at maize loci undergoing RNA-directed DNA methylation. Plant Cell 26: 4903–4917. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Ghosh S., Chan C.K. (2016). Analysis of RNA-seq data using TopHat and Cufflinks. Methods Mol. Biol. 1374: 339–361. [DOI] [PubMed] [Google Scholar]
  15. Groth M., et al. (2016). MTHFD1 controls DNA methylation in Arabidopsis. Nat. Commun. 7: 11640. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Guo H., et al. (2014). The DNA methylation landscape of human early embryos. Nature 511: 606–610. [DOI] [PubMed] [Google Scholar]
  17. Havecker E.R., Wallbridge L.M., Hardcastle T.J., Bush M.S., Kelly K.A., Dunn R.M., Schwach F., Doonan J.H., Baulcombe D.C. (2010). The Arabidopsis RNA-directed DNA methylation argonautes functionally diverge based on their expression and interaction with target loci. Plant Cell 22: 321–334. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. He X.J., Chen T., Zhu J.K. (2011). Regulation and function of DNA methylation in plants and animals. Cell Res. 21: 442–465. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Henderson I.R., Jacobsen S.E. (2007). Epigenetic inheritance in plants. Nature 447: 418–424. [DOI] [PubMed] [Google Scholar]
  20. Hou Z.X., Huang W.D. (2005). Immunohistochemical localization of IAA and ABP1 in strawberry shoot apexes during floral induction. Planta 222: 678–687. [DOI] [PubMed] [Google Scholar]
  21. Hu L., et al. (2014). Mutation of a major CG methylase in rice causes genome-wide hypomethylation, dysregulated genome expression, and seedling lethality. Proc. Natl. Acad. Sci. USA 111: 10642–10647. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hu L., Liang W., Yin C., Cui X., Zong J., Wang X., Hu J., Zhang D. (2011). Rice MADS3 regulates ROS homeostasis during late anther development. Plant Cell 23: 515–533. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Iwasaki M., Paszkowski J. (2014). Identification of genes preventing transgenerational transmission of stress-induced epigenetic states. Proc. Natl. Acad. Sci. USA 111: 8547–8552. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Jia Y., Lisch D.R., Ohtsu K., Scanlon M.J., Nettleton D., Schnable P.S. (2009). Loss of RNA-dependent RNA polymerase 2 (RDR2) function causes widespread and unexpected changes in the expression of transposons, genes, and 24-nt small RNAs. PLoS Genet. 5: e1000737. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Jiang C., Mithani A., Belfield E.J., Mott R., Hurst L.D., Harberd N.P. (2014). Environmentally responsive genome-wide accumulation of de novo Arabidopsis thaliana mutations and epimutations. Genome Res. 24: 1821–1829. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Jullien P.E., Susaki D., Yelagandula R., Higashiyama T., Berger F. (2012). DNA methylation dynamics during sexual reproduction in Arabidopsis thaliana. Curr. Biol. 22: 1825–1830. [DOI] [PubMed] [Google Scholar]
  27. Kim D.H., Doyle M.R., Sung S., Amasino R.M. (2009). Vernalization: winter and the timing of flowering in plants. Annu. Rev. Cell Dev. Biol. 25: 277–299. [DOI] [PubMed] [Google Scholar]
  28. Krueger F., Andrews S.R. (2011). Bismark: a flexible aligner and methylation caller for Bisulfite-Seq applications. Bioinformatics 27: 1571–1572. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Langmead B., Trapnell C., Pop M., Salzberg S.L. (2009). Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol. 10: R25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Law J.A., Jacobsen S.E. (2010). Establishing, maintaining and modifying DNA methylation patterns in plants and animals. Nat. Rev. Genet. 11: 204–220. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Le T.N., Schumann U., Smith N.A., Tiwari S., Au P.C., Zhu Q.H., Taylor J.M., Kazan K., Llewellyn D.J., Zhang R., Dennis E.S., Wang M.B. (2014). DNA demethylases target promoter transposable elements to positively regulate stress responsive genes in Arabidopsis. Genome Biol. 15: 458. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Li Q., et al. (2015). RNA-directed DNA methylation enforces boundaries between heterochromatin and euchromatin in the maize genome. Proc. Natl. Acad. Sci. USA 112: 14728–14733. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Liu G., Cao D., Li S., Su A., Geng J., Grover C.E., Hu S., Hua J. (2013). The complete mitochondrial genome of Gossypium hirsutum and evolutionary analysis of higher plant mitochondrial genomes. PLoS One 8: e69476. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Ma J., Wei H., Song M., Pang C., Liu J., Wang L., Zhang J., Fan S., Yu S. (2012). Transcriptome profiling analysis reveals that flavonoid and ascorbate-glutathione cycle are important during anther development in upland cotton. PLoS One 7: e49244. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Marino D., Dunand C., Puppo A., Pauly N. (2012). A burst of plant NADPH oxidases. Trends Plant Sci. 17: 9–15. [DOI] [PubMed] [Google Scholar]
  36. Matzke M.A., Mosher R.A. (2014). RNA-directed DNA methylation: an epigenetic pathway of increasing complexity. Nat. Rev. Genet. 15: 394–408. [DOI] [PubMed] [Google Scholar]
  37. Melnyk C.W., Molnar A., Bassett A., Baulcombe D.C. (2011). Mobile 24 nt small RNAs direct transcriptional gene silencing in the root meristems of Arabidopsis thaliana. Curr. Biol. 21: 1678–1683. [DOI] [PubMed] [Google Scholar]
  38. Min L., Zhu L., Tu L., Deng F., Yuan D., Zhang X. (2013). Cotton GhCKI disrupts normal male reproduction by delaying tapetum programmed cell death via inactivating starch synthase. Plant J. 75: 823–835. [DOI] [PubMed] [Google Scholar]
  39. Min L., Li Y., Hu Q., Zhu L., Gao W., Wu Y., Ding Y., Liu S., Yang X., Zhang X. (2014). Sugar and auxin signaling pathways respond to high-temperature stress during anther development as revealed by transcript profiling analysis in cotton. Plant Physiol. 164: 1293–1308. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Mitsuda N., Seki M., Shinozaki K., Ohme-Takagi M. (2005). The NAC transcription factors NST1 and NST2 of Arabidopsis regulate secondary wall thickenings and are required for anther dehiscence. Plant Cell 17: 2993–3006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Mittler R., Vanderauwera S., Gollery M., Van Breusegem F. (2004). Reactive oxygen gene network of plants. Trends Plant Sci. 9: 490–498. [DOI] [PubMed] [Google Scholar]
  42. Pecinka A., Dinh H.Q., Baubec T., Rosa M., Lettner N., Mittelsten Scheid O. (2010). Epigenetic regulation of repetitive elements is attenuated by prolonged heat stress in Arabidopsis. Plant Cell 22: 3118–3129. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Peng S., Huang J., Sheehy J.E., Laza R.C., Visperas R.M., Zhong X., Centeno G.S., Khush G.S., Cassman K.G. (2004). Rice yields decline with higher night temperature from global warming. Proc. Natl. Acad. Sci. USA 101: 9971–9975. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Popova O.V., Dinh H.Q., Aufsatz W., Jonak C. (2013). The RdDM pathway is required for basal heat tolerance in Arabidopsis. Mol. Plant 6: 396–410. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Rigal M., Becker C., Pélissier T., Pogorelcnik R., Devos J., Ikeda Y., Weigel D., Mathieu O. (2016). Epigenome confrontation triggers immediate reprogramming of DNA methylation and transposon silencing in Arabidopsis thaliana F1 epihybrids. Proc. Natl. Acad. Sci. USA 113: E2083–E2092. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Sakata T., Oshino T., Miura S., Tomabechi M., Tsunaga Y., Higashitani N., Miyazawa Y., Takahashi H., Watanabe M., Higashitani A. (2010). Auxins reverse plant male sterility caused by high temperatures. Proc. Natl. Acad. Sci. USA 107: 8569–8574. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Secco D., Wang C., Shou H., Schultz M.D., Chiarenza S., Nussaume L., Ecker J.R., Whelan J., Lister R. (2015). Stress induced gene expression drives transient DNA methylation changes at adjacent repetitive elements. eLife 4: 4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Shen X., De Jonge J., Forsberg S.K., Pettersson M.E., Sheng Z., Hennig L., Carlborg Ö. (2014). Natural CMT2 variation is associated with genome-wide methylation changes and temperature seasonality. PLoS Genet. 10: e1004842. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Sijen T., Vijn I., Rebocho A., van Blokland R., Roelofs D., Mol J.N., Kooter J.M. (2001). Transcriptional and posttranscriptional gene silencing are mechanistically related. Curr. Biol. 11: 436–440. [DOI] [PubMed] [Google Scholar]
  50. Slotkin R.K., Vaughn M., Borges F., Tanurdzić M., Becker J.D., Feijó J.A., Martienssen R.A. (2009). Epigenetic reprogramming and small RNA silencing of transposable elements in pollen. Cell 136: 461–472. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Song Q.X., Lu X., Li Q.T., Chen H., Hu X.Y., Ma B., Zhang W.K., Chen S.Y., Zhang J.S. (2013). Genome-wide analysis of DNA methylation in soybean. Mol. Plant 6: 1961–1974. [DOI] [PubMed] [Google Scholar]
  52. Strømme C.B., Julkunen-Tiitto R., Krishna U., Lavola A., Olsen J.E., Nybakken L. (2015). UV-B and temperature enhancement affect spring and autumn phenology in Populus tremula. Plant Cell Environ. 38: 867–877. [DOI] [PubMed] [Google Scholar]
  53. Tang J.H., Fu Z.Y., Hu Y.M., Li J.S., Sun L.L., Ji H.Q. (2006). Genetic analyses and mapping of a new thermo-sensitive genic male sterile gene in maize. Theor. Appl. Genet. 113: 11–15. [DOI] [PubMed] [Google Scholar]
  54. Tang W., Tu L., Yang X., Tan J., Deng F., Hao J., Guo K., Lindsey K., Zhang X. (2014). The calcium sensor GhCaM7 promotes cotton fiber elongation by modulating reactive oxygen species (ROS) production. New Phytol. 202: 509–520. [DOI] [PubMed] [Google Scholar]
  55. Wang S., et al. (2015). Sequence-based ultra-dense genetic and physical maps reveal structural variations of allopolyploid cotton genomes. Genome Biol. 16: 108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Wang M., Wang P., Tu L., Zhu S., Zhang L., Li Z., Zhang Q., Yuan D., Zhang X. (2016). Multi-omics maps of cotton fibre reveal epigenetic basis for staged single-cell differentiation. Nucleic Acids Res. 44: 4067–4079. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Wicker T., et al. (2007). A unified classification system for eukaryotic transposable elements. Nat. Rev. Genet. 8: 973–982. [DOI] [PubMed] [Google Scholar]
  58. Xie H.T., Wan Z.Y., Li S., Zhang Y. (2014). Spatiotemporal production of reactive oxygen species by NADPH oxidase is critical for tapetal programmed cell death and pollen development in Arabidopsis. Plant Cell 26: 2007–2023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Yong-Villalobos L., González-Morales S.I., Wrobel K., Gutiérrez-Alanis D., Cervantes-Peréz S.A., Hayano-Kanashiro C., Oropeza-Aburto A., Cruz-Ramírez A., Martínez O., Herrera-Estrella L. (2015). Methylome analysis reveals an important role for epigenetic changes in the regulation of the Arabidopsis response to phosphate starvation. Proc. Natl. Acad. Sci. USA 112: E7293–E7302. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Yui R., Iketani S., Mikami T., Kubo T. (2003). Antisense inhibition of mitochondrial pyruvate dehydrogenase E1alpha subunit in anther tapetum causes male sterility. Plant J. 34: 57–66. [DOI] [PubMed] [Google Scholar]
  61. Zhang T., et al. (2015b). Sequencing of allotetraploid cotton (Gossypium hirsutum L. acc. TM-1) provides a resource for fiber improvement. Nat. Biotechnol. 33: 531–537. [DOI] [PubMed] [Google Scholar]
  62. Zhang H., Zhu J.K. (2011). RNA-directed DNA methylation. Curr. Opin. Plant Biol. 14: 142–147. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Zhang H., Liang W., Yang X., Luo X., Jiang N., Ma H., Zhang D. (2010). Carbon starved anther encodes a MYB domain protein that regulates sugar partitioning required for rice pollen development. Plant Cell 22: 672–689. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Zhang J., Liu Y., Xia E.H., Yao Q.Y., Liu X.D., Gao L.Z. (2015a). Autotetraploid rice methylome analysis reveals methylation variation of transposable elements and their effects on gene expression. Proc. Natl. Acad. Sci. USA 112: E7022–E7029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Zhao Q., Gallego-Giraldo L., Wang H., Zeng Y., Ding S.Y., Chen F., Dixon R.A. (2010). An NAC transcription factor orchestrates multiple features of cell wall development in Medicago truncatula. Plant J. 63: 100–114. [DOI] [PubMed] [Google Scholar]
  66. Zhong S., Fei Z., Chen Y.R., Zheng Y., Huang M., Vrebalov J., McQuinn R., Gapper N., Liu B., Xiang J., Shao Y., Giovannoni J.J. (2013). Single-base resolution methylomes of tomato fruit development reveal epigenome modifications associated with ripening. Nat. Biotechnol. 31: 154–159. [DOI] [PubMed] [Google Scholar]
  67. Zhu X., Liang W., Cui X., Chen M., Yin C., Luo Z., Zhu J., Lucas W.J., Wang Z., Zhang D. (2015). Brassinosteroids promote development of rice pollen grains and seeds by triggering expression of Carbon Starved Anther, a MYB domain protein. Plant J. 82: 570–581. [DOI] [PubMed] [Google Scholar]

Articles from The Plant Cell are provided here courtesy of Oxford University Press

RESOURCES