Metabolic control of histone demethylase activity involved in plant response to high temperature

Abstract

Jumonji C (JmjC) domain proteins are histone lysine demethylases that require ferrous iron and alpha-ketoglutarate (or α-KG) as cofactors in the oxidative demethylation reaction. In plants, α-KG is produced by isocitrate dehydrogenases (ICDHs) in different metabolic pathways. It remains unclear whether fluctuation of α-KG levels affects JmjC demethylase activity and epigenetic regulation of plant gene expression. In this work, we studied the impact of loss of function of the cytosolic ICDH (cICDH) gene on the function of histone demethylases in Arabidopsis thaliana. Loss of cICDH resulted in increases of overall histone H3 lysine 4 trimethylation (H3K4me3) and enhanced mutation defects of the H3K4me3 demethylase gene JMJ14. Genetic analysis suggested that the cICDH mutation may affect the activity of other demethylases, including JMJ15 and JMJ18 that function redundantly with JMJ14 in the plant thermosensory response. Furthermore, we show that mutation of JMJ14 affected both the gene activation and repression programs of the plant thermosensory response and that JMJ14 and JMJ15 repressed a set of genes that are likely to play negative roles in the process. The results provide evidence that histone H3K4 demethylases are involved in the plant response to elevated ambient temperature.

Histone H3K4me3 demethylases JMJ14, JMJ15, and JMJ18 function redundantly in the plant thermosensory response, which is affected by mutation of the cytosolic isocitrate dehydrogenase gene.

Introduction

Histone modifications play essential roles in epigenetic regulation of gene expression and genome activity. Among the diverse histone modifications, lysine methylation is one of the most studied and characterized histone modification types (Kouzarides, 2007). In plants, a large number of genes are marked by methylation of histone H3 lysine 4 (H3K4; Zhang et al., 2009). H3K4 trimethylation (H3K4me3) is predominantly found at the promoter and 5′-end of genes and is strongly associated with gene induction by plant internal and external signals. In some cases, an H3K4me3 increase can lag behind gene activation, which is suggested to mark or to memorize gene expression states (Alvarez-Venegas et al., 2007; Kim et al., 2008, 2012; Hu et al., 2011; Jaskiewicz et al., 2011; Ding et al., 2012).

Histone methylation is reversed by histone demethylases. In mammals, lysine-specific demethylase 1 is the first identified histone demethylase to remove mono- and di-methyl groups from H3K4 (Shi et al., 2004). The second class of histone demethylases that contain the Jumonji C (JmjC) domain catalyzes demethylation of di- and tri-methylated histone lysine residues, based on mammalian proteins. Multiple JmjC domain-containing histone demethylases were identified and divided into distinct groups including JARID/KDM5, JMJD1/JHDM2/KDM3, JMJD2/KDM4, JMJD3/UTX/KDM6, JHDM1/FBX/KDM2, and the “JmjC domain-only” group. Each group targets specific methylated histone lysine residues including methylated H3K4, H3K9, H3K27, H3K36, and H4K20 (Klose et al., 2006; Black et al., 2012). For instance, the JARID/KDM5 group catalyzes H3K4me2/3 demethylation in mammalian cells. In Arabidopsis thaliana, ∼20 JmjC domain-containing protein genes were found (Lu et al., 2008; Sun and Zhou, 2008; Chen et al., 2011). Plant genomes contain a group of JmjC proteins homologous to the mammalian JARID/KDM5 group, which includes the Arabidopsis JMJ14, JMJ15, JMJ16, JMJ17, JMJ18, and JMJ19 proteins as well as the rice (Oryza sativa) JMJ703 protein (Lu et al., 2008; Sun and Zhou, 2008; Chen et al., 2011). JMJ14, JMJ15, JMJ18, and JMJ703 were reported to demethylate H3K4me3 and to regulate diverse aspects of chromatin function and plant development, as well as stress response (Deleris et al., 2010; Lu et al., 2010; Searle et al., 2010; Le Masson et al., 2012; Yang et al., 2012a, 2012b; Chen et al., 2013; Cui et al., 2013; Shen et al. 2014).

A recombinant JMJ14 demethylates H3K4m3 in vitro and in transiently transfected tobacco (Nicotiana benthamiana) cells (Jeong et al., 2009; Lu et al., 2010; Yang et al., 2010, 2018). The jmj14 knockout mutants showed early flowering in both long-day (LD) and short-day (SD) conditions (Lu et al., 2010; Yang et al., 2010; Ning et al., 2015). JMJ14 represses the flowering-promoting genes Flowering Locus T (FT) and Suppressor of overexpression of CO1 (SOC1) by removing H3K4me3 (Jeong et al., 2009; Yang et al., 2010; Lu et al., 2010, Ning et al., 2015). JMJ15 was found to demethylate H3K4me3 in vitro and in vivo (Yang et al., 2012c). The mutation of JMJ15 did not produce any obvious developmental phenotype (Yang et al., 2012c; Shen et al., 2014), but JMJ15 overexpression resulted in early flowering, which was associated with repression of Flowering Locus C (FLC) and reduction of H3K4me3 at the FLC locus, consequently leading to increased FT expression (Yang et al., 2012c). JMJ18 was shown to also demethylate histone H3K4me3 and H3K4me2 in vitro; mutations of JMJ18 resulted in a weak late-flowering phenotype, while JMJ18 overexpressors exhibited an obvious early-flowering phenotype. JMJ18 was shown to bind to the FLC locus and to demethylate H3K4me3 and H3K4me2 in chromatins of FLC clade genes (Yang et al., 2012d). Although overexpression of JMJ15 and JMJ18 led to early flowering, it remains unknown whether the gain-of-function phenotype reflected the exogenous function of the genes. The distinct effects of the three JmjC gene mutations on flowering time suggest that they target distinct downstream genes or chromatin regions.

JmjC domain-containing proteins catalyze demethylation of histones and other proteins from lysine residues via oxidative reactions requiring ferrous iron [Fe(II)] and alpha-ketoglutarate (α-KG, also known as 2-oxoglutarate) as cofactors (reviewed in Dimitrova et al., 2015). α-KG is produced by isocitrate dehydrogenases (ICDHs) in the mitochondria as an intermediate of the tricarboxylic acid (TCA) cycle and is also produced in the cytoplasm by cytosolic ICDH (cICDH). In addition to isocitrate, α-KG can be synthesized from amino acids such as arginine, glutamine, histidine, and proline. It was shown that cancer-associated mutations in ICDH (IDH1 and IDH2) enzymes converting α-KG into 2-hydroxyglutarate, as observed in gliomas, can alter KDM4C/JHDM3C-dependent demethylation of H3K9me3 in genes associated with differentiation of neural progenitor cells (Ward et al., 2010). Importantly, the activity of JmjC domain-containing histone demethylases can be inhibited by intermediates of the TCA cycle such as fumarate and succinate (when fumarate hydratase or succinate dehydrogenase is knocked down), which are downstream from α-KG (Xiao et al., 2012). This supports interconnection between the metabolic rate and histone demethylase activity, which is reflected by changes in chromatin modifications. However, the role of α-KG levels in regulating the activity of JmjC domain-containing enzymes is generally not yet clearly determined.

In plants, both NAD⁺ and NADP⁺-dependent ICDH isoforms are found. The nicotinamide adenine dinucleotide (NAD⁺) dependent isoform is restricted to mitochondria, where it takes part in the TCA cycle (Kruse et al., 1998; Hodges, 2002; Hodges et al., 2003). NADP⁺-dependent ICDH isoforms are found in the cytosol, chloroplasts, mitochondria, and peroxisomes. In Arabidopsis, a single isoform is found in each cell compartment. The cytosolic isoform (cICDH) is the most abundant form in leaves, as it is responsible for ˃80% of the extractible ICDH activity (Kruse et al., 1998; Hodges, 2002). However, experiments with knockout mutants for the single annotated Arabidopsis cICDH gene showed that cICDH is largely dispensable for Arabidopsis development, and that its loss of function does not greatly impact leaf metabolite profiles (Mhamdi et al., 2010). In contrast, icdh knockout plants showed accumulation of defense-related transcripts in the absence of pathogen attack, and an exacerbated phenotype and redox perturbation observed in the catalase2 oxidative stress background (Mhamdi et al., 2010). Whether fluctuation of cICDH-dependent α-KG levels regulates JmjC histone demethylases activity and gene expression is unknown.

Increases in ambient temperature affect plant growth and may have dramatic effects on plant architecture, biomass, and yield (reviewed in McClung et al., 2016). Plants have evolved sophisticated regulatory mechanisms of molecular and morphological changes to acclimate to elevated ambient temperatures for fitness and yield (Samach and Wigge, 2005; Liu et al., 2015). Recent results indicate that PHYTOCHROME-INTERACTING FACTOR4 (PIF4), a basic helix-loop-helix transcriptional regulator, plays a central role to regulate the gene expression network of plants in response to high ambient temperature (Koini et al., 2009; Leivar and Quail, 2011; Kumar et al., 2012; Proveniers and van Zanten, 2013; Leivar and Monte, 2014; Quint et al., 2016; Paik et al., 2017). One of the mechanisms by which PIF4 regulates the plant thermal response is that PIF4 stimulates the production of auxin by directly binding to the promoter regions of auxin biosynthesis genes, which subsequently triggers elongation growth at high ambient temperature (Franklin et al., 2011; Sun et al., 2012). Interestingly, low ambient light and high ambient temperature signals cross-communicate in determining plant adaptive growth responses (Delker et al., 2017; Legris et al., 2017). For instance, Arabidopsis phytochromes also function as thermosensors (Jung et al., 2016), and the well-described De-Etiolated1-Constitutive Photomorphogenesis1-Hypocotyl5 (HY5) photomorphogenic pathway controls PIF4-mediated high temperature-triggered elongation growth (Gangappa and Kumar, 2017).

Growing evidence indicates that epigenetic mechanisms are involved in the regulation of plant thermal response. For instance, the histone H2A variant, H2A.Z, is evicted from chromatin at transcriptonal start sites at elevated temperatures, thereby contributing to thermos-morphogenesis by allowing binding of the transcriptional machinery to temperature-regulated genes (Kumar and Wigge, 2010). Recently, PICKLE, an ATP-dependent chromatin remodeling factor (that alters nucleosome positioning), was shown to negatively control thermosensory HY growth in Arabidopsis by affecting the level of H3K27me3 in the loci of Indole-3-Acetic Acid Inducible19 (IAA19) and IAA29 and regulate their expression (Zha et al., 2017). Recent data indicated that the histone deacetylase HDA15 directly represses gene expression and plant responses to high ambient temperature (Shen et al., 2019). These studies highlight the importance of epigenetic regulation in the plant thermosensory response, but how the plant epigenetic system integrates metabolic activity to control the plant response to the changing environment remains largely unknown.

In this work, we studied the impact of a cICDH loss-of-function mutation on the function of histone demethylase genes in plant growth and gene expression in response to high ambient temperature. We found that the cICDH mutation led to an increase of overall levels of H3K4me3 and enhanced mutation defects of the histone demethylase gene JMJ14 in the plant response to high ambient temperature and flowering time. Genetic analysis revealed that JMJ14 functions redundantly with JMJ15 or JMJ18 in plant thermo-morphogenesis and suggested that the cICHD mutation may affect JMJ15 and JMJ18 function. Furthermore, our data indicate that JMJ14 is required to establish both the gene activation and repression programs of plant response to high temperature and that JMJ14 and JMJ15 directly represses a set of genes that are likely to play a negative role in the process.

Results

cICDH knockout mutants show slightly lower α-KG levels and enhanced H3K4me3

Previous analysis of cICDH transfer DNA (T-DNA) mutants (icdh-1, icdh-2, and icdh-3) indicated that mutations of the Arabidopsis cICDH gene did not produce very dramatic effects on leaf metabolites (Mhamdi et al., 2010). In all, 76 unique metabolites could be identified and relatively quantified by gas chromatography time-of-flight mass spectrometry, 8 of which were found to be significantly different (P < 0.05) in plants grown in ambient air, whereas 15 were different when plants were grown at high CO₂ (Mhamdi et al., 2010). In ambient air, notable icdh-linked changes were decreased fructose and increased citrate, whereas at high CO₂, the clearest effect was a decrease in sucrose and several respiratory intermediates, as well as three amino acids (Ala, Ser, and Thr; Mhamdi et al., 2010). Because of its low abundance, α-KG was not identified from the metabolic profiling. To evaluate α-KG levels in icdh mutants, we extracted 10-d-old seedlings and measured α-KG levels with a commercial kit (see Methods section). Significant decreases of α-KG levels were detected in two icdh mutant lines (Figure 1A).

Figure 1.

Effect of cICDH mutations on α-KG and histone methylation levels. A, α-KG levels in Col-0 and icdh mutant seedlings (10 DAG). Bars are means ± SD from three biological replicates; two-tailed Student’s t test, P < 0.05. B, Immunoblotting analysis of histone H3K4 methylation levels in two icdh knockout lines compared with the wild-type. Histones were extracted from 10-d-old seedlings under SD condition. Antibodies against H3K4me3, H3K4me2, and H3 were used. Five biological replicates of the Immunoblotting analyses are shown. The band intensities were measured and analyzed using the ImageJ software and the levels of each histone modification were normalized to that of H3, which was used as a loading control. Bars are means ± SD from four biological replicates. Significant differences were tested by two-sided Student’s t test (**P < 0.01).

To examine whether the relatively lower α-KG levels in icdh mutants affected overall histone methylation, we analyzed histone proteins in Col-0 and ichdh-1 and icdh-2 mutants by immunoblotting using specific antibodies against H3K4me3, H3K4me2, H3K27me3, and H3K36me3 marks and H3 as a control. The analysis with five biological replicates revealed that H3K4me3 levels were augmented in the mutants (Figure 1B). The levels of other histone methylation marks were not clearly changed (Figure 1B;Supplemental Figure S1). The data suggest that the mutations or lower α-KG levels may affect demethylase activities specific to H3K4me3.

The icdh-1 mutation enhances the early flowering phenotype of the jmj14 mutant but attenuates the late-flowering phenotype of the jmj18 mutant

cICDH loss-of-function T-DNA mutants (icdh-1, icdh-2, and icdh-3) displayed only relatively minor effects on rosette growth (Mhamdi et al., 2010). To evaluate whether cICDH loss of function impacted the function of H3K4me3 demethylases in plant development, we produced double mutants by genetic crosses between icdh-1 (as the female) and jmj14-1, jmj15-3, or jmj18-1 mutants (as the male). All of them were in the Col-0 background. Reciprocal crosses between jmj14-1, jmj15-3, or jmj18-1 (as the female), and icdh-1 (as the male) were also obtained (Figure 2, A and B; Table 1). The double mutants showed no particular phenotype compared with the single mutants at seedling stages under normal conditions. When we examined the flowering time under both LD and SD conditions, we detected an early flowering phenotype of the jmj14-1 mutant (Table 1; Figure 2, C and D), as previously reported. Interestingly, we also detected an early flowering phenotype of icdh-1 under both LD and SD conditions (Table 1; Figure 2, C and D). The jmj14-1/icdh-1 double mutants displayed an even earlier flowering time than the single mutants under both LD and SD conditions (Table 1; Figure 2, C and D). Reverse transcription-quantitative polymerase chain reaction (RT-qPCR) analysis revealed lower expression of FLC and higher expression of FT in jmj14-1 plants grown in LD conditions (confirming previous results), and also in icdh-1 plants compared with Col-0 (Figure 2G). The increased FT expression in the icdh-1/jmj14-1 double mutant was significantly higher than in jmj14-1 (Figure 2G). The results indicated that the icdh-1 mutation enhanced the early flowering phenotype of jmj14-1. Possibly, the reduced α-KG level in icdh-1 plants was limiting for other demethylases that function redundantly with JMJ14. In contrast, in our conditions we only observed a slightly late flowering phenotype of jmj18-1 plants in LDs (Table 1), while the early flowering phenotype of icdh-1 was again observed in this set of experiments (Table 1; Figure 2, E and F). Interestingly, the icdh-1/jmj18-1 double mutants displayed an intermediate flowering time phenotype under both LD and SD conditions (Table 2; Figure 2, E and F). RT-qPCR analysis revealed higher expression of FLC in jmj18-1 and a lower expression of the gene in icdh-1 but similar levels of expression of the gene in the double mutants as in Col-0 under LD conditions (Figure 2H). FT expression was enhanced in icdh-1 but the increase was reduced in the icdh-1/jmj18-1 double mutant (Figure 2H). The data indicated that cICDH was not in the same pathway as JMJ18 in regulating flowering time. Rather they had opposite functions in the process. The Jmj15-1 single and the jmj15-1 icdh-1 double mutants showed normal flowering time in both LD and SD conditions (Figure 2, I and J). The analysis suggests that the three demethylase genes may have different functions in different pathways of flowering time.

Figure 2.

Characterization of icdh-1 and jmj14-1, jmj15-3, or jmj18-1 double mutants. A, Gene structures of cICDH, JMJ14, JMJ15, and JMJ18. T-DNA insertion sites are indicated by vertical arrows or open triangles. Exons are represented as black boxes, introns as lines, and untranslated regions as white boxes. Positions and directions of primers used are indicated by horizontal arrows. B, RT-PCR analysis of cICDH, JMJ14, JMJ15, and JMJ18 transcripts in the wild-type (Col-0) and single and double mutants. Actin2 was used as an internal control. C, D, jmj14/icdh-1 and icdh-1/jmj14-1 double mutants were early flowering compared with Col-0 or with the single mutants under both LD and SD conditions. E, F, flowering time of jmj18/icdh-1, and icdh-1/jmj18-1 double mutants compared with Col-0 and the single mutants. G, H, Relative expression levels of FLC and FT in icdh-1, jmj14-1, jmj14/icdh, icdh/jmj14, jmj18-1, jmj18/icdh, and icdh/jmj18 plants relative to Col-0. The transcript levels were normalized with that of ACTIN2. RNA was isolated from 14-d-old seedlings grown in LD conditions. The values are means ± sd from three or four independent biological replicates. The significant differences were tested by two-sided Student’s t test (*P < 0.05, **P < 0.01). I, J, Flowering time of jmj15-3/icdh-1 and icdh-1/jmj15-3 plants compared with Col-0 and the single mutants. Entire backgrounds of C–F were removed to extract rosettes digitally for comparison.

Table 1.

Flowering time icdh-1 and jmj14, jmj15 or jmj18 single and double mutants under LD and SD conditions^a

Day length	Genotype	Days to first flower open	Sig.	Rosette leaf no.	Sig.	Cauline leaf no.	Sig.	n
Long Day (LD)	Col-0	23.91 ± 0.23	A	14.83 ± 0.36	a	1.88 ± 0.12	a	40
	jmj14-1	23.03 ± 0.21	ab	13.05 ± 0.34	b	1.95 ± 0.10	a	21
	icdh-1	22.51 ± 0.18	B	12.38 ± 0.28	b	1.76 ± 0.10	a	45
	icdh/jmj14	21.35 ± 0.26	C	10.64 ± 0.19	c	2.00 ± 0.09	a	42
	jmj14/icdh	21.02 ± 0.25	C	9.84 ± 0.28	c	1.73 ± 0.011	a	37
Short Day (SD)	Col-0	59.44 ± 0.64	A	65.04 ± 2.88	a	2.58  ± 0.17	a	24
	jmj14-1	54.32 ± 1.52	C	60.75 ± 1.63	b	2.25 ± 0.24	a	16
	icdh-1	56.19 ± 0.66	bc	59.15 ± 2.40	bc	2.85 ± 0.22	a	22
	icdh/jmj14	50.78 ± 0.82	D	56.25 ± 1.72	b	2.55 ± 0.18	a	19
	jmj14/icdh	51.15 ± 0.89	D	55.60 ± 2.30	b	2.40 ± 0.20	a	22
	Genotype	Days to first flower open	Sig.	Rosette leaf no.	Sig.	Cauline leaf no.	Sig.	n
Long Day (LD)	Col-0	24.28 ± 0.25	A	14.73 ± 0.44	b	1.88 ± 0.11	a	40
	jmj18-1	24.66 ± 0.26	a	15.38 ± 0.49	b	2.23 ± 0.14	a	40
	icdh-1	22.46 ± 0.24	d	11.71 ± 0.32	c	1.62 ± 0.11	a	34
	jmj18/icdh	24.10 ± 0.25	b	15.98 ± 0.64	a	2.00 ± 0.15	a	42
	icdh/jmj18	23.09 ± 0.22	c	13.55 ± 0.34	b	1.81 ± 0.10	a	44
Short Day (SD)	Col-0	65.05 ± 0.74	a	61.91 ± 1.74	a	2.78 ± 0.22	a	24
	jmj18-1	63.18 ± 1.17	a	63.09 ± 1.96	a	3.09 ± 0.19	a	23
	icdh-1	56.47 ± 0.60	b	58.79 ± 0.92	b	2.45 ± 0.19	a	21
	jmj18/icdh	61.72 ± 0.70	bc	65.40 ± 2.16	a	2.68 ± 0.23	a	24
	icdh/jmj18	63.12 ± 0.89	Ba	64.96 ± 2.08	ab	2.33 ± 0.17	a	25
	Genotype	Days to first flower open	Sig.	Rosette leaf no.	Sig.	Cauline leaf no.	Sig.	n
Short Day (SD)	WT	64.00 ± 2.82	a	58.17 ± 3.76	a	2.91 ± 0.54	a	13
	icdh-1	58.08 ± 2.09	b	56.00 ± 2.63	b	2.58 ± 0.52	a	13
	jmj15-3	63.08 ± 2.14	a	58.42 ± 2.02	a	2.83 ± 0.62	a	12
	icdh/jmj15	62.92 ± 2.21	a	56.50 ± 0.17	a	2.67 ± 0.51	a	15
	jmj15/icdh	63.56 ± 2.70	a	57.42 ± 1.57	a	2.75 ± 0.65	a	14

^aLD condition (16-h light, 20°C/8-h dark, 18°C). SD condition (8-h light, 20°C/16-h dark, 18°C). The values are the means ± standard deviations (sds). “n” indicates the number of plants scored for phenotype analysis. Different letters indicate the significant differences among multiple comparisons by LSD analysis based on P < 0.05.

Table 2.

RT-qPCR validated thermosensory genes that were reversely regulated in Col-0 and icdh/jmj14 double mutants

Locus tag	Gene name	Expression in Col at 20°C	Expression in Col at 27°C	Expression in icdh jmj14 at 20°C	Expression in icdh jmj14 at 27°C	Fold change of expression in icdh jmj14 relative to Col at 27°C
						Log2 (fold change)	P-value
AT3G55500	EXPA16 (expansin A16)	1.34086	94.48540	0.35088	1.00625	−6.60554	5.00E-05
AT5G13170	SAG29 (senescence-associated gene 29)	1.38274	57.60170	1.26165	1.72648	−5.11232	5.00E-05
AT5G52310	LTI78 or RD29A (low-temperature-responsive protein 78)	17.42290	127.40300	10.14990	8.29004	−3.99404	5.00E-05
AT3G45680	NPF2.3 (Major facilitator superfamily protein (AT3G45680)	1.04668	3.86706	0.80649	0.66422	−2.59161	0.00265
AT3G22550	DUF581 (NAD(P)H-quinone oxidoreductase subunit, putative)	1.88243	5.79453	5.52596	1.82809	−1.71496	0.01485
AT2G39700	EXPA4 (expansin A4)	24.68780	64.31000	35.81500	27.90490	−1.25497	0.0015
AT3G02480	ABR (Late embryogenesis abundant protein (LEA) family protein)	4.19259	506.27100	3.61883	2.32846	−7.81699	0.0004
AT1G03090	MCCA (methylcrotonyl-CoA carboxylase alpha chain)	43.91480	10.56820	16.28150	33.538600	1.74436	0.0001
AT1G11260	STP1 (sugar transporter 1)	179.56300	41.24790	63.27180	229.72900	2.04876	5.00E-05
AT1G28330	DYL1 (dormancy-associated protein-like1)	1029.08000	346.54100	505.03300	1020.74000	1.53575	0.0007
AT1G69600	ZFHD1 or ATHB29 (zinc finger homeodomain 1)	18.31910	7.27125	13.94390	16.41520	1.1287	0.02625
AT1G77210	STP14 (sugar transporter 14)	13.98330	5.06857	11.03530	15.27940	1.43669	0.0011
AT2G33830	DRM2 (Dormancy/auxin associated family protein)	1758.98000	477.40000	1113.18000	1693.41000	1.92551	5.00E-05
AT5G04190	PKS4 (phytochrome kinase substrate 4)	8.73486	2.83739	7.64864	6.13796	1.06795	0.03715
AT5G28770	BZO2H3 (bZIP transcription factor family protein)	34.39930	14.42540	44.35420	42.16350	1.31128	0.0008

The icdh/jmj14 double mutations lead to a synergistic effect on plant response to elevated temperature

At high ambient temperature (i.e. 27°C), Arabidopsis plants display a thermosensory growth with elongated HYs and petioles, increased leaf angles, and narrow and curled leaves. Under normal growth temperature (20°C), 10-d-old plants of icdh-1, jmj14-1, jmj15-3, and jmj18-1 single or double mutants did not show any visible phenotype (Figure 3). After shifting 7-d-old plants grown at 20–27°C for 3 d, the single and double mutants showed a similar thermosensory growth phenotype as the wild-type Col-0 plants (Figure 3), except for the icdh-1/jmj14-1 double mutants that displayed insensitivity to the elevated temperature, with shorter HYs and reduced leaf angles (which were measured after 4 d at 27°C) compared with the wild-type plants (Figure 3, A and B). The data indicated that the icdh-1/jmj14-1 double mutations had a synergistic effect on plant response to the elevated temperature.

Figure 3.

Effect of the cICDH mutation in jmj14, jmj15, or jmj18 mutants on plant growth at 20°C and 27°C. Phenotypes of single and double mutants of icdh-1/jmj14 (A, B), icdh/jmj15 (C), and icdh/jmj18 (D) grown at 20°C and 27°C. For HY length measurement, seedlings were grown on half MS-agar plates at 20°C under the sd condition for 7 d or first for 4 d then transferred to 27°C for 3 d before taking photographs (A, C, D). For leaf angle measurement (B), 7-d-old seedlings grown at 20°C then exposed to 27°C for 4 d were used. Bars are means ± sd from more than 100 plants. Significance of differences was tested using two-sided Student’s t test (P < 0.01).

JMJ14 is required to establish the high temperature-responsive gene expression program

To further characterize the effects of jmj14 and icdh mutations on plant response to high temperature, we performed transcriptomic analysis of 10-d-old Col-0, jmj14-1, and icdh1-1/jmj14-1 plants grown at 20°C in SDs then shifted to 27°C or maintained at 20°C for 4 h. The plant samples were harvested at the same time at 3 p.m. (after 5 h light exposure) for RNA extraction. Two biological replicates were performed for RNA-sequencing. Before performing RNA-sequencing, we tested the expression of six known high-temperature-responsive marker genes (IAA9, IAA29, YUC8 [YUCCA8], SAUR22 [Small Auxin Up RNA22], ATHB2 [A. thaliana Homeobox2], and PIF4; Shen et al., 2019). They all showed increased expression in Col-0 plants after 4 h at 27°C versus 20°C (Supplemental Figure S2A), indicating that the high-temperature treatment was effective. The RNA-seq replicates were all highly correlated (r = 0.92–0.99; Supplemental Figure S2B). Analysis of differentially expressed (>2-fold, P < 0.05) genes (DEG) revealed 641 upregulated and 344 downregulated DEGs in Col-0 at 27°C versus 20°C (Supplemental Figure S2C; Supplemental Dataset S1). Gene ontology (GO) analysis revealed that the upregulated genes were enriched mainly in stress or hormone-responsive functions while the downregulated ones were enriched mainly for catabolic metabolism and plant response to low light (Supplemental Figure S3), similar to data previously reported in wild-type Arabidopsis plants (Cortijo et al., 2017; Shen et al., 2019). Compared with the wild-type, the jmj14-1 mutation resulted in 170 upregulated and 252 downregulated genes at 20°C and 209 upregulated, and 319 downregulated genes at 27°C (Supplemental Figure S2C; Supplemental Dataset S1). However, compared with the wild-type, the icdh-1/jmj14-1 double mutations led to much more upregulated (834) than downregulated (127) genes at 20°C. At 27°C, the double mutation resulted in 493 upregulated and 548 downregulated genes compared with the wild-type plants grown at the same temperature (Supplemental Figure S2C; Supplemental Dataset S1). The jmj14-1 DEGs detected at 20°C showed little overlap with those of icdh-1/jmj14-1 detected at 20°C. At 27°C, 52% (107/209) of the upregulated and 65% (207/319) of jmj14-1 downregulated genes overlapped with those detected in the double mutant at 27°C (Supplemental Figure S4). The analysis suggested that at 20°C the icdh-1/jmj14-1 double mutation affected a totally different gene expression program than the jmj14-1 single mutation, while at 27°C the double mutant displayed an overlapping and amplified effect on gene expression than the single mutant.

To investigate whether the jmj14-1 and icdh-1/jmj14-1 mutations impaired the high temperature-responsive gene expression program, we compared DEGs in Col-0 at 27°C versus 20°C with that of jmj14-1 or icdh-1/jmj14-1 (versus Col-0) at 27°C. We found that 58% (123/209) up- and 68% (220/319) downregulated DEGs in jmj14-1 overlapped, respectively, with the down- and upregulated DEGs in Col-0 at 27°C (Figure 4A;Supplemental Dataset S2). The icdh-1/jmj14-1 double mutations further increased the numbers of overlapping genes (154 upregulated and 264 downregulated; Figure 4A;Supplemental Dataset S2). The 264 downregulated genes in the double mutant showed similar enrichment of GO terms as those upregulated in Col-0 at 27°C. Similar to the genes downregulated in Col-0 at 27°C, the 154 upregulated genes in the double mutant were enriched for catabolic processes and responses to starvation and absence of light (Supplemental Figure S3). These data were consistent with the high-temperature-insensitive phenotype of the double mutant (Figure 2) and confirmed the synergistic effect of the double mutation on the plant response to the elevated temperature.

Figure 4.

Differentially expressed genes in jmj14-1 and jmj14-1/icdh-1 plants compared to Col-0 at 20°C and 27°C. A, Upper part: Overlaps between upregulated DEGs in Col-0 at 27°C versus 20°C and downregulated DEGs in jmj14-1 (left) and jmj14-1/icdh-1 (right) relative to Col-0 at 27°C. Lower part: overlaps between downregulated DEGs in Col-0 at 27°C versus 20°C and upregulated DEGs in jmj14-1 (left) and jmj14-1/icdh-1 (right) relative to Col-0 at 27°C. B, RT-qPCR validation of genes upregulated in Col-0 but downregulated in the double mutant at 27°C. C, RT-qPCR analysis of high-temperature-induced thermosensory marker genes in Col-0 at 20°C and 27°C and in the double mutant at 27°C. D, RT-qPCR validation of genes downregulated in Col-0 at 27°C versus 20°C but upregulated in the double mutant versus Col-0 at 27°C. RNA-seq data of the tested genes in B–D are listed in Table 2. Transcript levels in B–D are normalized with that of ACTIN2. Bars are means ± sd from three biological replicates using RNAs extracted from seedlings grown for 10 d in the sd condition at 20°C, then maintained at 20°C or transferred to 27°C for 4 h. Significance of differences between samples was tested by two-sided Student’s t tests (**P < 0.01, *P < 0.05).

To validate the RNA-seq data, we selected seven genes (Table 2) that were upregulated in Col-0 but downregulated in icdh-1/jmj14-1 (versus Col-0) at 27°C by RT-qPCR analysis. The analysis confirmed the high-temperature induction of the genes in Col-0, which was abolished in the double mutant at 27°C (Figure 4B). In addition, we confirmed the induction of known high-temperature marker genes in Col-0 at 27°C and their down-regulation in the double mutant at 27°C (Figure 4C). Similarly, we confirmed the expression levels of eight genes that were downregulated in Col-0 at 27°C versus 20°C but upregulated in the double mutant versus Col at 27° (Table 2; Figure 4D). Together, the analysis suggested that JMJ14 was required for both gene activation and repression of the high-temperature-responsive gene expression program, and that cICDH had a redundant effect in the process.

JMJ14 functions redundantly with JMJ15 and JMj18 in the plant thermosensory response

The high-temperature-insensitive phenotype of the icdh-1/jmj14-1 mutant and the more severe effect on gene expression than in the jmj14-1 mutant at 27°C suggested the icdh-1 mutation might affect the activity of additional demethylases that act redundantly with JMJ14 to regulate plant response to increases of ambient temperature. To study whether other members of the JARID/KDM5 group are redundant with JMJ14, we produced jmj14-1/jmj15-3, jmj15-3/jmj18-1, jmj14-1/jmj18-1, and icdh-1/jmj15-3 double and jmj14-1/jmj15-3/icdh-1 triple mutants (Figure 5A). The double and triple mutant plants showed no visible phenotype at 20°C (Figure 5, B–E). At 27°C, jmj14-1/jmj15-3 and jmj14-1/jmj18-1, but not jmj15-3/jmj18-1 or icdh-1/jmj15-3, displayed a high-temperature-insensitive phenotype at 27°C with reduced HY elongation (Figure 5, B–E). The jmj14-1/jmj15-3/icdh-1 triple mutant displayed a similar insensitive phenotype as the jmj14-1/jmj15-3 double mutant at 27°C (Figure 5E). As in jmj14-1/icdh-1, the expression of high-temperature-induced genes in Col-0 at 27°C was found to be repressed in the jmj14/jmj18 double and the jmj14-1/jmj15-3/icdh-1 triple mutants (Figure 6, A and B). Conversely, the expression of genes that were downregulated in Col-0 at 27°C versus 20°C but upregulated in the jmj14-1/icdh-1 mutant versus Col-0 at 27°C were found to be induced in jmj14/jmj15, jmj14/jmj18 double and jmj14-1/jmj15-3/icdh-1 triple mutants compared with Col-0 at 27°C (Figure 6, C–E). The data indicated that JMJ14 had a redundant function with JMJ15 and JMJ18 in the plant response to ambient temperature increases.

Figure 5.

Thermosensory phenotypes of jmj14/jmj15, jmj15/jmj18, and jmj14/jmj18 double and jmj14/jmj15/icdh triple mutants. A, Validation of the double mutants by RT-qPCR analysis of JMJ14, JMJ15, JMJ18, and cICDH expression in the single and double mutants. Actin2 was used as an internal control. B, HY lengths of jmj14-1/jmj15-3 double mutants compared with Col-0 and the single mutants at 20°C and 27°C. C, HY lengths of jmj14-1/jmj18-1 double mutants compared to Col-0 and the single mutants at 20°C and 27°C. D, HY lengths of jmj15-3/jmj18-1 double mutants compared to Col-0 and the single mutants at 20°C and 27°C. E, HY lengths of jmj14-1/jmj15-3/icdh-1 triple mutants compared to Col-0 and the single mutants at 20 and 27°C. Bars are means ± standard errors of at least 30 plants. Significance of differences between samples indicated by different letters was tested by two-sided Student’s t tests (P < 0.05).

Figure 6.

Expression levels of temperature-sensitive mark genes in the jmj14-1/jmj15-3, jmj14-1/jmj18-1, and jmj14-1/jmj15-3/icdh-1 plants were compared with Col-0 at 20°C and 27°C. A, B, Expression levels of genes upregulated in Col-0 but downregulated in the jmj14/icdh mutant at 27°C were tested in jmj14/jmj18 and jmj14/jmj15/icdh plants. C–E, Expression levels of genes downregulated in Col-0 but upregulated in the jmj14/icdh mutant at 27°C were tested in jmj14/jmj15 and jmj14/jmj18 double and jmj14/jmj15/icdh triple mutants. Transcript levels in A–E are normalized with that of ACTIN2. Bars are means ± sd from three biological replicates using RNAs extracted from 10-d-old seedlings grown under the sd condition at 20°C, then maintained at 20°C or transferred to 27°C for 4 h. Significance of differences between samples was tested by two-sided Student’s t tests (^**P < 0.01, ^*P < 0.05).

JMJ14 and JMJ15 directly associate with a set of genes repressed during the plant response to elevated temperature

JMJ14, as an H3K4me3 demethylase, may directly target a set of high temperature-downregulated genes to undertake H3K4me3 removal and transcriptional repression. To test this hypothesis, using chromatin immunoprecipitation (ChIP)-qPCR, we analyzed H3K4me3 levels at two regions (one in the promoter and one in the 5′-exon near the ATG codon) of the seven tested genes (Figure 7, A and B) that were downregulated in Col-0 but upregulated in jmj14-1 and/or icdh-1/jmj14-1 mutants at 27°C (Table 2; Figure 4). Except for Dormancy-Associated Protein-Like1 (DYL1), the tested genes showed significantly higher H3K4me3 levels at the promoter and/or the exon regions in icdh-1/jmj14-1 compared with Col-0 plants. In order to study whether JMJ14 was directly associated with the genes, using ChIP-qPCR with anti-cMYC antibody, we analyzed the jmj14-1 plants complemented with 9 X MYC-tagged JMJ14 driven by the endogenous JMJ14 promoter (Myc-JMJ14; Deleris et al., 2010), grown at 20°C, then kept at 20°C or transferred to 27°C for 4 h. Col-0 plants (without the construct for the tagged protein) grown at 20°C were used as controls. The analysis revealed that MYC-JMJ14 was enriched in all of the tested genes at 20°C and the enrichments were significantly higher at 27°C (Figure 7C), suggesting that the JMJ14 binding to target genes was enhanced by high temperature. To test whether JMJ15 could also directly associate with the same genes, we analyzed the 35S:JMJ15-HA plants by ChIP-qPCR using anti-HA antibody (Shen et al., 2014). The analysis revealed that at 20°C JMJ15-HA was enriched in Phytochrome Kinase Substrate4 (PKS4; promoter and exon), Sugar Transport Protein14 (STP14; promoter), DYL1 5 (exon), and ATHB29 (promoter) relative to the negative controls (Col-0; Figure 7D). Similarly, at 27°C JMJ15-HA binding levels were significantly higher at the promoter and/or exon regions of the tested genes with the exception of STP1 (Figure 7D). The results indicated that the recruitment of both JMJ14 and JMJ15 to the genes was induced at high ambient temperature. The genes were also detected as among the JMJ14 genome-wide targets (Supplemental Figure S5; Zhang et al., 2015).

Figure 7.

JMJ14 and JMJ15 bind to and demethylate H3K4me3 of genes downregulated at high temperature. A, Structure and the ChIP tested promoter (P) and exon (E) regions of the eight genes that were downregulated in Col-0 but upregulated in jmj14 and/or icdh/jmj14 double mutants at 27°C (Figure 4D;Table 2). B, ChIP analysis of H3K4me3 levels of the genes in Col-0 and the double mutants grown at 20°C in the sd condition for 10 d then transferred to 27°C for 4 h. Bars are means ± sd of data from three biological replicates. Ribulose Bisphosphate Carboxylase Small Subunit and Chlorophyll A/B-Binding Protein2 were tested as positive controls. Significance of differences between 27°C and 20°C was tested using two-sided Student’s t tests (*P < 0.05 and **P < 0.01). C, D, ChIP analysis of binding of JMJ14 and JMJ15 to the genes. The jmj14 35S:MYC-JMJ14 (JMJ14) (C) jmj15 35S:JMJ15-HA-FLAG (JMJ15) (D) as well as Col-0 seedlings were grown at 20°C under the sd condition for 10 d then transferred to 27°C for 4 h before harvest for chromatin isolation. Anti-cMYC (C) and anti HA (D) antibodies were used for the ChIP-qPCR analysis. Bars are means ± sd of data from three biological replicates (*P < 0.05 and **P < 0.01, Student’s t test).

Discussion

Metabolic regulation of histone demethylase activity

Higher H3K4me3 levels in icdh mutants suggest that H3K4me3 JmjC demethylases might be particularly sensitive to the reduced α-KG levels found in these mutants to regulate genome-wide H3K4me3 homeostasis. Nevertheless, our data do not exclude the possibility that reduced α-KG levels could also affect activities of other oxidoreductases including JmjC proteins demethylating other histone methylation marks at specific chromatin regions. The impaired response to elevated temperature of icdh-1/jmj14 -1 indicates that the cICDH mutation may affect proteins that function redundantly with JMJ14 in the plant response to high temperature. The similar insensitive phenotype of jmj14-1/jmj15-3 and jmj14-1/jmj18-1 as icdh-1/jmj14-1 at the elevated temperature suggests that the icdh-1 mutation may affect JMJ15 and JMJ18 function in the process. Within the plant JARID/KDM5 family, JMJ15 and JMJ18 are mostly closely related. The reduced α-KG level in icdh mutants could possibly affect the activity of JMJ15 and JMJ18, while an indirect effect of the icdh mutation on their function is not excluded.

JMJ4 and JMJ15 redundantly regulate key regulatory genes in the plant response to high temperature

The observations that the high temperature-related DEGs in the icdh-1/jmj14-1 double mutants overlapped with about two-thirds of those in jmj14-1 and that the double mutation resulted in many additional high-temperature-related DEGs (Supplemental Dataset S3) indicated that JMJ14 and cICDH (likely via JMJ15 or JMJ18) synergistically regulate additional key genes of the plant response to elevated temperature, the deregulation of which was sufficient to produce the insensitive phenotype to the elevated temperature. Among these genes were DUF581 (encoding NAD(P)-H-quinone Oxidoreductase Subunit), PKS4, and Zinc Finger Homeodomain1 (ZFHD1) /ATHB29 (Figure 4; Table 2). PKS4 was shown to promote plant response to low light (for phototropic growth) but also limit plant response to high light, which depends on the blue light receptor phot1-mediated phosphorylation switch of the protein (Schumacher et al., 2018). As low light and high-temperature signals cross-communicate in determining plant adaptive growth responses (Delker et al., 2017; Legris et al., 2017) and phytochromes act as thermosensors in Arabidopsis (Jung et al., 2016), downregulation of PKS4 in Col-0 at 27°C may be linked to the plant response to high temperature. Although the function of the Zinc Finger Homeodomain (ZFHD) family members is presently unclear, recent results indicated that ZFHD10 regulates the expression of transcriptional regulators, light-responsive and growth-promoting genes required for HY elongation (Perrella et al., 2018). It remains to be determined whether ZFHD1/ATHB29 also plays a role in HY elongation during the plant response to elevated temperature.

JMJ14 and JMJ15 regulate both the gene activation and repression programs of the plant response to high temperature

Our data indicate that JMJ14 together with JMJ15 or JMJ18 redundantly promote the plant response to high temperature. Although the high-temperature-induced gene activation program involving the PIF4-regulated network has been well documented, detail on the gene repression program is less clear. The present results identified JMJ14 and JMJ15 as direct regulators of the gene repression program of the plant response to high temperature. The observations that JMJ14 and JMJ15 binding to downstream targets was enhanced at 27°C suggest an active recruitment of the demethylases during the response. The factors involved in JMJ14 and JMJ15 recruitment remain to be identified. In contrast, being H3K4me3 demethylases, the JMJ14 and JMJ15 function in promoting gene activation should be of indirect effects. One possibility would be that the demethylases inhibit transcriptional repressors of the high-temperature-responsive gene activation program. Previous results identified Arabidopsis histone deacetylase HDA15 as a direct repressor of gene activation induced by high temperature (Shen et al., 2019). The mutation of HDA15 resulted in induction of many high-temperature-induced genes at both 20°C and 27°C and hda15 mutants show a constitutive high-temperature-response phenotype (Shen et al., 2019). Although HDA15 was not among the DEGs (>2-fold, P < 0.05) in jmj14-1 or icdh-1/jmj14-1 mutants detected by RNA-seq, RT-qPCR analysis revealed that the gene was significantly induced in the icdh-1 and jmj14-1 single and double mutants at 27°C (Supplemental Figure S6). Conversely, two other histone deacetylases, namely HDA9 and HDA19, which were shown to play an opposite role to that of HDA15 in the plant response to high temperature (Shen et al., 2019) were repressed in the icdh-1/jmj14-1 double mutants at 27°C (Supplemental Figure S6). Like icdh1/jmj14-1 plants, both hda9 and hda19 mutants show a reduced response to high temperature. HDA9 appears to act downstream of HDA15, as the hda15/hda9 double mutants displayed the same phenotype as hda9 (Shen et al., 2019). Together, the data allow hypothesizing that JMJ14 and JMJ15 function upstream of the histone deacetylases to regulate the high-temperature-responsive gene expression program.

The present data showing that genes involved in catabolic activity and in response to starvation are repressed in Col-0 but activated in jmj14-1 or icdh-1/jmj14-1double mutant at 27°C (Supplemental Figure S3) are consistent with previous results showing that altered primary metabolite accumulation in hda15 mutants are related to the hypersensitive response to high temperature (Shen et al., 2019). Repression of catabolic activity and in response to starvation (e.g. sugar mobilization) during plant response to high temperature would lead to fluctuation of a-KG levels, which may eventually affect JmjC demethylases activity. Collectively, the present results reveal an important role played by JmjC demethylases in the plant response to high ambient temperature, and pinpoint a functional relationship between members of the JARDI/KDM5 family of JmjC proteins.

Materials and methods

Plant materials and growth conditions

The A. thaliana jmj mutants (jmj14-1, jmj15-3, and jmj18-1) and icdh mutants (icdh-1 and icdh-2) used in this study have been previously characterized (Lu et al., 2010; Mhamdi et al., 2010; Yang et al., 2012d; Shen et al., 2014). The tagged transgenic lines: MYC-JMJ14, JMJ15-HA-FLAG, and JMJ18-GFP were provided by Dr A. Deleris, Dr Y. Shen, and Dr H-C Yang (Deleris et al., 2010; Yang et al., 2012d; Shen et al., 2014). The jmj14 icdh, jmj15 icdh, and jmj18 icdh reciprocal double mutants were obtained by genetic crosses between jmj mutants and icdh-1 and confirmed by genotyping and RT-PCR analysis. The jmj14 jmj15 and jmj15 jmj18 double mutants were obtained by using the same method.

The Arabidopsis seeds were surface-sterilized by 5% (w/v)sodium hypochlorite for 7 min and washed with 95% (v/v)ethanol twice, then sown on half Murashige and Skoog (MS) basal salt medium (Duchefa, Haarlem, The Netherlands). After stratification for 2 d at 4°C, seeds were transferred into a growth chamber or cabinet under various conditions. Seeds were planted in soil directly for phenotype observation.

Flowering time measurement

All plants were grown in soil side by side at 20°C under both LD (16/8 light-dark) and SD (8/16 light-dark) conditions. Flowering time was measured as the total number of rosette and cauline leaves when the plants flowered. At least 16 plants were counted for each line.

Warm temperature treatment

Plants were grown on half MS-agar plates at 20°C for 4 d under SD conditions and incubated at either 20°C or 27°C for 3 or 4 additional days for phenotype observation. At least 50 seedlings were photographed; HY length and leaf angle were measured using ImageJ software. Three independent experiments were performed. Data were analyzed using two-sided Student’s t test. For gene expression and chromatin analysis, 10-d-old seedlings grown under the SD condition were harvested after being maintained at 20°C or incubated at 27°C for 4 h.

RNA sequencing and data analysis

Ten-day-old seedlings of wild-type (Col-0), single mutant (jmj14-1) and icdh-1/jmj14-1 double mutant incubated at 20°C or 27°C for 4 h were pooled for RNA extraction and transcriptomic analysis. Plants harvested from two independent cultures were used as the biological replicates. Total RNA was extracted with TRIzol Reagent (Molecular Research Center, Inc; TR118), and treated with DNase I (Promega, Madison, WI, USA; M6101) to remove the genomic DNA. Then the mRNA was enriched by using oligo(dT) magnetic beads and broken into short fragments (200 bp) using the fragmentation buffer. The first-strand cDNA was synthesized by using a random primer. RNase H, DNA polymerase I, and dNTPs were used to synthesize the second strand. The double-strand cDNA was purified with magnetic beads. cDNA ends were repaired and a nucleotide A (adenine) was added at the 3′-end. Finally, PCR amplification was performed with fragments ligated by sequencing adaptors. The Agilent 2100 Bioanalyzer and the ABI Step One Plus real-time PCR system were used to qualify and quantify the QC library. The library products were sequenced with the Illumina HiSeq 2000 platform, and the library construction and sequencing were completed at BGI (Shenzhen, China). The sequencing reads were mapped to the TAIR10 version of the A. thaliana genome. The RNA-seq data are deposited to NCBI-SRA databases under the accession PRJNA565103.

RNA isolation and RT-qPCR analysis

Total RNA was extracted from ∼0.1 g Arabidopsis rosette leaves, flowers or seedlings using TRIzol Reagent (TR118; Molecular Research Center, Inc., Cincinnati, OH, USA). RNA concentration was measured by using a NanoDrop. For RT-qPCR analysis, 4 µg of total RNA treated with DNase (RQ1; Promega, M6101) were used to synthesize first-strand cDNA with Oligo(dT)15 primers using ImPromII reverse transcriptase (M3104A; Promega). RT-qPCR was performed with LightCycler 480 SYBR Green I Master mix (04707516001; Roche, Basel, Germany) on the LightCycler 480 (Roche). The reactions were performed in triplicate for each run and at least three biological replicates were carried out for each reaction. Relative transcript levels were standardized to the housekeeping gene β-actin. Primer sequences used in this study are summarized in Supplemental Table S1.

α-KG content measurement

α-KG was measured using the Abcam α-KG Assay Kit (Ab 83431) according to the manufacturer’s instructions. Briefly, 100 mg 10-d-old seedlings were ground and resuspended in assay buffer included in the kit, and then incubated with substrate for 30 min at 37°C. The α-KG substrate was probed to fluorescence (Ex/Em = 535/587 nm). The α-KG content was calculated according to the instructions in the assay kit.

Histone extraction from Arabidopsis and immunoblotting

Ten-day-old Arabidopsis seedlings were harvested in liquid nitrogen and ground in cold extraction buffer containing 10 mM Tris–HCl pH 7.5, 2 mM ethylenediaminetetraacetic acid (EDTA), 0.4 M HCl, 5 mM beta-mercaptoethanol, 0.2 mM phenylmethylsulfonyl fluoride, and phenylmethylsulfonyl fluoride (PMSF). The extract was centrifuged at 12,000g for 10 min at 4°C and after centrifugation protein was precipitated from the supernatant using 15–20% (v/v) trichloroacetic acid. Then the samples were centrifuged at 17,000g for 30 min at 4°C, the pellet was washed with 100% cold acetone once or twice, air-dried, and dissolved in ddH₂O with protease inhibitors. For immunoblots, ∼20 µg of protein extract was loaded on 4–12% (w/v) sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to Immobilon PVDF membranes (Eurogentec, Seraing, Belgium) by using iBlot Gel Transfer Stacks (Invitrogen, Carlsbad, CA, USA). Histone modifications were detected using anti-histone H3 antibody (ab1791), anti-dimethyl-histone H3 (Lys4) antibody ( 07030; Millipore, Burlington, MA, USA) anti-trimethyl-histone H3 (Lys4) antibody (07473; Millipore), anti-trimethylation H3 (Lys27) antibody ( ABE44; Sigma, St Louis, MO, USA), and anti-trimethylation H3 (Lys36) antibody (ab9050) and visualized using a goat peroxidase-conjugated anti-rabbit IgG secondary antibody and ECL detection system (WBKLS0100; Millipore). The signals were captured on a ChemiDoc MP imaging system (Bio-Rad, Hercules, CA, USA). Quantification of the western blotting signal was performed with ImageJ software. Three independent experiments were performed.

ChIP and ChIP-qPCR

ChIP assays were performed as described previously with minor modifications (Bowler et al., 2004). Ten-day-old plants were maintained at 20°C or incubated at 27°C for 4 h before harvest. About 2 g seedlings were harvested and fixed in 1% formaldehyde (v/v) under vacuum. The samples were ground in extraction buffer containing 0.4 M sucrose, 10 mM Tris–HCl, pH 8.5, 5 mM β-mercaptoethanol, 1 mM PMSF, and protease inhibitors and fragmented to 200–500 bp by sonication, and ChIP was performed using the following antibodies: anti-H3K4me2, anti-H3K4me3, anti-HA, anti-GFP, and anti-MYC. The precipitated and input DNA fragments were analyzed by qPCR with gene-specific primers listed in Supplemental Table S1. ChIP-qPCR was performed with three biological replicates, and the results were calculated as a percentage of input DNA.

Statistical analysis

The statistical analysis was performed using Student’s t test. Calculations were from a minimum of three independent data sets, assuming two sample equal variance and a two-tailed distribution. Significant difference is expressed with P <0.05 or <0.01.

Accession numbers

The genes in this articles can be found in the Arabidopsis Genome Initiative with the following accession numbers: JMJ14 (AT4G20400), JMJ15 (AT2G34880), JMJ18 (AT1G30810), FLC (AT5G10140), FT (AT1G65480), PIF4 (AT2G43010), YUC8 (AT4G28720), IAA19 (AT3G15540), IAA29 (AT4G32280), SAUR22 (AT5G18050), ATHB2 (AT4G16780), PKS4 (AT5G04190), STP14 (AT1G77210), EXPA4 (AT2G39700), MCCA (AT1G03090), BZO2H3 (AT5G28770), DRM2 (AT2G33830), and DRM1 (AT1G28330).

Supplemental data

The following materials are available in the online version of this article.

Supplemental Figure S1 Histone methylation levels in icdh mutant plants.

Supplemental Figure S2 Transcriptomic analysis of Col-0, jmj14-1, and icdh-1/jmj14-1 plants at 20°C and 27°C.

Supplemental Figure S3 GO term enrichment of Col-0 DEGs at 27°C and the overlapped DEGs between Col-0 and the icdh-1/jmj14-1 double mutant.

Supplemental Figure S4 Overlap of icdh-1/jmj14-1 DEGs with jmj14-1 DEGs

Supplemental Figure S5 JMJ14 binding genes shown in Figure 4 are found in the JMJ14 genome-wide ChIP-seq data.

Supplemental Figure S6 Transcript levels of histone deacetylase HDA9, HDA15, and HDA19 genes in Col-0, icdh-1, jmj14-1, and icdh-1/jmj14-1mutants.

Supplemental Table S1 Primers used in this study

Supplemental Dataset S1. Lists of DEGs in Col-0 at 27°C versus 20°C and in jmj14-1 and icdh-1/jmj14-1 mutants versus Col-0 at 20°C and 27°C.

Supplemental Dataset S2. List of overlapping DEGs in Col-0 with those in jmj14-1 and icdh-1/jmj14-1.

Supplemental Dataset S3. List of non-overlapping DEGs in the icdh-1/jmj14-1 double mutant compared with jmj14-1.

 

X.C. performed most of the experiments; Y.Z. participated in genetic analysis; Y.L. analyzed the RNA-seq data; E.I.B. participated in and supervised the experiments; D.X.Z. designed the project, analyzed the data, and wrote the paper. All authors read and approved the final manuscript.