Genome-wide gene expression profiling to investigate molecular phenotypes of Arabidopsis mutants deprived in distinct histone methyltransferases and demethylases

Abstract

Histone lysine (K) methylation is a type of epigenetic modification involved in regulation of DNA-based processes, including transcription, replication and repair. It can either activate or repress transcription depending on the histone K residue on which methylation occurs and on chromatin context of additional other modifications. In both animals and plants, methylation on one histone K residue can be deposited by several different histone methyltransferases and vice versa removed by different histone demethylases. It is of great interest to know which histone enzyme regulates which genes in the genome. Here we describe in details the contents and quality controls for the gene expression data of Arabidopsis mutants deprived in distinct histone methyltransferases (SDG26, SDG25, ATX1, CLF) and histone demethylases (LDL1, LDL2), in association with the study recently published by Berr and colleagues in The Plant Journal (Berr et al., 2015). The microarray dataset has been deposited in Gene Expression Omnibus with accession number GSE55167.

Specifications

Organism/cell line/tissue	Arabidopsis thaliana, 16-day-old seedlings.
Sex	NA
Sequencer or array type	Agilent's Whole Arabidopsis Gene Expression Microarray (4 × 44K).
Data format	Raw data: TAR of TXT files, normalized data: TXT and XLS of SOFT.
Experimental factors	Mutant vs. wild-type plants.
Experimental features	Seed germination and plant growth for all mutants and wild-type were performed under the same conditions; plant age used in analysis corresponds to 16 days counted after seed sawing.
Consent	NA
Sample source location	Strasbourg, France.

Direct link to deposited data

Deposited data can be found here: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE55167.

Experimental design, materials and methods

Plant materials and growth conditions

All Arabidopsis strains used in this work are in the Columbia (Col) ecotype background. Description of mutant strains can be found from previous publications: sdg26-1 [1], sdg25-1 [2], atx1-2 [3], clf-29 [4], ldl1-2 ldl2 [5], [6], and the combined mutants sdg26 sdg25, sdg26 atx1, sdg26 clf as well as sdg26 ldl1 ldl2 [7]. Seeds of wild-type (Col0) and the abovementioned mutant strains were produced from plants grown on soil in glasshouse. For seed surface sterilization, about 100 seeds were disposed in an open 2 ml Eppendorf tube and the tube was placed in a desiccator under a fume hood. A beaker containing 20 ml of bleach (FLOREAL Haagen GmbH, http://www.eau-de-javel.info/) was placed close to the seed tube, 5 ml of concentrated HCl (37%) was added to the bleach, and the lid of the desiccator was immediately closed to keep the produced chlorine gas in the desiccator compartment. After 3 to 6 h, sterilization was complete and the seeds could be used for in vitro plant culture.

Surface-sterilized seeds were plated on agar-solidified Murashige and Skoog (MS) medium M0255 (Duschefa, http://www.duchefa-biochemie.com/) supplemented with 0.9% sucrose in Petri Dishes. After stratification in the dark at 4 °C for 48 h that helps to synchronize seed germination time, the Petri Dishes were moved to the growth chamber and incubated for plant growth under Medium Day (MD; 12 h light and 12 h dark) photoperiods at 22 °C.

Sample collection and RNA preparation

Sample collection and RNA preparation were performed using plants 16 days after seed plating (16-day-old seedlings). For each sample set, about 30 to 40 plants were collected from three replicating Petri Dishes. For each plant genotype/strain, three independent sets were pooled. Thus, a total of 33 samples were analyzed in this work. Total RNA was isolated from plant samples using the Nucleospin RNA Plant kit (Macherey-Nagel, http://www.mn-net.com) followed by RNeasy mini kit (Qiagen, https://www.qiagen.com/) clean-up according to manufacturer's instructions. The quality of the RNA obtained from each sample was assessed based on the RNA profile generated by the Agilent Bioanalyzer 2100 (Agilent technologies, Santa Clara, CA). Our entire samples' RNAs are in high quality in line with the requirement for microarray analysis (Table 1).

Table 1.

RNA quality assessment of samples studied in this work.

Experiment	Arabidopsis	Sample	A260/A280(a)	RIN(b)	28S/18S(b)
Repeat-1	Col	1	1.97	7.1	1.3
sdg25	2	2.04	7.0	1.3
sdg26	3	2.06	7.0	1.3
sdg25 sdg26	4	1.95	7.2	1.6
atx1	5	2.06	7.0	1.3
sdg26 atx1	6	2.03	7.0	1.4
clf	7	1.86	7.1	1.3
sdg26 clf	8	2.05	7.0	1.6
ldl1 ldl2	9	1.92	7.4	1.7
sdg25 ldl1 ldl2	10	2.04	7.0	1.3
sdg26 ldl1 ldl2	11	1.91	7.2	1.4
Repeat-2	Col	12	1.92	7.4	1.5
sdg25	13	1.88	7.0	1.3
sdg26	14	2.02	7.0	1.5
sdg25 sdg26	15	1.95	7.0	1.4
atx1	16	2.01	7.0	1.3
sdg26 atx1	17	2.03	7.0	1.3
clf	18	1.85	7.0	1.3
sdg26 clf	19	2.02	7.0	1.4
ldl1 ldl2	20	1.91	7.4	1.6
sdg25 ldl1 ldl2	21	1.96	7.6	1.4
sdg26 ldl1 ldl2	22	2.02	7.3	1.4
Repeat-3	Col	23	1.98	7.2	1.9
sdg25	24	1.93	7.2	1.8
sdg26	25	1.94	7.0	1.6
sdg25 sdg26	26	2.01	7.0	1.4
atx1	27	1.95	7.3	1.5
sdg26 atx1	28	1.97	7.1	1.4
clf	29	1.95	7.1	1.4
sdg26 clf	30	1.93	7.0	1.6
ldl1 ldl2	31	1.93	7.4	1.5
sdg25 ldl1 ldl2	32	1.95	7.1	1.7
sdg26 ldl1 ldl2	33	2.01	7.4	1.7

(a) A ratio of A260/A280 > 1.8 indicates little protein contamination.

(b) RNA integration number (RIN) ≥ 7.0 and 28S/18S ≥ 0.7 indicate good quality for microarray assay.

Microarray data

Microarray analyses were performed using Agilent's Whole Arabidopsis Gene Expression Microarray (G2519F, V4, 4 × 44K) via custom service of the Shanghai Huaguan Biochip Co. (http://www.bio-equip.cn/). Total RNA was amplified and Cy3-labeled using Low Input Quick Amp Labeling Kit, one-color by following the manufacturer's instructions (Agilent technologies, http://www.genomics.agilent.com/). Labeled cRNA was purified using RNeasy mini kit (Qiagen), slide hybridization at 65 °C for 17 h and image scanning (Scan resolution = 5μm, PMT 100%, 10%, 16 bits) were performed by following standard Agilent's protocols. Data extraction was performed using Agilent Feature Extraction software. Feature Extraction was used for background subtraction and normalization between the set of arrays was done by Quantile algorithm, a method that gives the same distribution of probe intensities to each array in the set [8]. Microarray and sample annotation data were deposited in the Gene Expression Omnibus (GEO) database under accession number GSE55167.

Quality control

In addition to good quality of RNAs (Table 1), our microarray hybridization and detection of expressed genes were also at high quality (Table 2). The percent coefficient of variation (%CV) values for our entire microarray sets were largely inferior to 15%, a value proposed as validation threshold by Agilent. A lower median %CV value indicates better reproducibility of signal across the microarray than a higher value. Moreover, independent RT-PCR analyses confirmed microarray data on several selected genes (see below).

Table 2.

Quality assessment of microarray hybridization reproducibility and detection rate for samples studied in this work.

Experiment	Arabidopsis	Sample	%CV(*)	Detection rate	GEO file
Repeat-1	Col	1	7.40	70.47	GSM1330673
sdg25	2	4.52	70.58	GSM1330674
sdg26	3	3.93	69.44	GSM1330675
sdg25 sdg26	4	7.17	65.57	GSM1330676
atx1	5	6.07	72.13	GSM1330677
sdg26 atx1	6	5.07	75.74	GSM1330678
clf	7	4.50	76.60	GSM1330679
sdg26 clf	8	5.61	72.02	GSM1330680
ldl1 ldl2	9	10.69	72.51	GSM1330681
sdg25 ldl1 ldl2	10	5.97	73.48	GSM1330682
sdg26 ldl1 ldl2	11	6.64	71.34	GSM1330683
Repeat-2	Col	12	5.52	70.18	GSM1330684
sdg25	13	3.22	75.06	GSM1330685
sdg26	14	3.80	78.41	GSM1330686
sdg25 sdg26	15	9.75	70.20	GSM1330687
atx1	16	2.83	76.88	GSM1330688
sdg26 atx1	17	3.09	76.88	GSM1330689
clf	18	4.47	76.22	GSM1330690
sdg26 clf	19	8.57	73.12	GSM1330691
ldl1 ldl2	20	3.17	75.81	GSM1330692
sdg25 ldl1 ldl2	21	2.33	76.61	GSM1330693
sdg26 ldl1 ldl2	22	9.97	72.81	GSM1330694
Repeat-3	Col	23	6.84	70.20	GSM1330695
sdg25	24	9.19	69.06	GSM1330696
sdg26	25	3.81	74.97	GSM1330697
sdg25 sdg26	26	3.39	75.79	GSM1330698
atx1	27	3.96	74.84	GSM1330699
sdg26 atx1	28	4.50	77.53	GSM1330700
clf	29	3.86	72.12	GSM1330701
sdg26 clf	30	3.90	76.70	GSM1330702
ldl1 ldl2	31	3.54	76.34	GSM1330703
sdg25 ldl1 ldl2	32	4.26	75.76	GSM1330704
sdg26 ldl1 ldl2	33	3.70	76.52	GSM1330705

(*) %CV indicates for percent coefficient of variation.

Basic and focused analysis

From previous studies we know that full-length transcripts of mutated genes are missing in the corresponding mutant strains used in this work. Therefore, it is predicted that our microarray data should confirm this information. Indeed, mutated genes are among the downregulated genes in the respective mutants (GSE55167; [7]), e.g. SDG26 is downregulated in sdg26, sdg26 sdg25, sdg26 atx1, sdg26 clf and sdg26 ldl1 ldl2 but not in the other mutants, SDG25 is downregulated in sdg25 and sdg26 sdg25 but not in the other mutants, etc. This confirmatory information on multiple mutated genes in various single, double and triple gene mutants in a same microarray analysis provides a good validation of our data.

Our focused analysis on flowering time regulatory genes revealed that FLOWERING LOCUS C (FLC) is upregulated or downregulated (−) to more than 2 folds in several of the mutant strains analyzed in microarray (GSE55167; Table 3). We validated the observed differential FLC expression in mutants by quantitative RT-PCR analysis. For qRT-PCR analyses, first-strand cDNA was synthesized from 2 μg of total RNA pretreated with 2 units of DNase I using the Impro-II Reverse Transcriptase system with oligo(dT)₂₀ primer by following the manufacturer's instruction (Promega, http://www.promega.com). The synthesized cDNA was analyzed by quantitative PCR in a 384-well optical plate on a BioRad i-cycler apparatus using 5 μl of PCR master mix (Roche) containing 480 SYBER® Green I fluorescent reporter with 2.5 μM forward and reverse FLC-specific primers. PCR reaction was performed by a step of 10 min at 95 °C, followed by 40 cycles of 15 s at 95 °C, 30 s at 60 °C and 15 s at 72 °C. Melting curves of PCR reactions were checked to insure the quality of PCR reaction and to avoid any DNA contamination. The threshold cycle value (CT) was set so that the fluorescent signal was above the baseline noise but as low as possible in the exponential amplification phase. The relative expression level of FLC was calculated using the Lightcycler 480 software and normalized using ACT2 and GAPDH as internal reference genes. Each sample was analyzed in triplicate, and mean ± SD was shown as result (Table 3). Our qRT-PCR data show an overall agreement with microarray results obtained on FLC expression changes in different mutants.

Table 3.

Validation of microarray data by quantitative RT-PCR analysis on FLC expression in different mutants as compared to wild-type (set as 1).

Mutant strain	Fold change in microarray	Relative level in RT-PCR
sdg25	− 5.5	0.15 ± 0.02
sdg26	2.4	3.24 ± 0.31
sdg25 sdg26	< 2.0	1.07 ± 0.15
atx1	− 6.0	0.24 ± 0.04
sdg26 atx1	< 2.0	1.08 ± 0.16
clf	3.8	3.80 ± 0.20
sdg26 clf	6.7	7.52 ± 0.11
ldl1 ldl2	< 2.0	1.92 ± 0.23
sdg25 ldl1 ldl2	< 2.0	1.17 ± 0.03
sdg26 ldl1 ldl2	5.1	4.57 ± 0.62

Discussion

We described here a unique dataset of microarray analyses on multiple different mutants deprived of one, two or three histone-methyltransferase(s)/histone-demethylase(s). This dataset has been recently used in a study focused on plant flowering time regulation [7]. We anticipate that the dataset may also be useful for comparative study of other mutants as well as for investigation of roles of histone methylations in processes beyond plant flowering time control.