A method for transcriptome profiling of cereals is presented. The microarray-based gene expression profiling starts with the isolation of high-quality total RNA from cereal grains and continues with the generation of cDNA. After cRNA labelling and microarray hybridization, recommendations are given for signal detection and quality control.
The characterization of gene expression is dependent on RNA quality. In germinating, developing and mature cereal seeds, the extraction of high-quality RNA is often hindered by high starch and sugar content. These compounds can reduce both the yield and the quality of the extracted total RNA. The deterioration in quantity and quality of total RNA can subsequently have a significant impact on the downstream transcriptomic analyses, which may not accurately reflect the spatial and/or temporal variation in the gene expression profile of the samples being tested. In this protocol, we describe an optimized method for extraction of total RNA with sufficient quantity and quality to be used for whole transcriptome analysis of cereal grains. The described method is suitable for several downstream applications used for transcriptomic profiling of developing, germinating, and mature cereal seeds. The method of transcriptome profiling using a microarray platform is shown. This method is specifically designed for gene expression profiling of cereals with described genome sequences. The detailed procedure from microarray handling to final quality control is described. This includes cDNA synthesis, cRNA labelling, microarray hybridization, slide scanning, feature extraction, and data quality validation. The data generated by this method can be used to characterize the transcriptome of cereals during germination, in various stages of grain development, or at different biotic or abiotic stress conditions. The results presented here exemplify high-quality transcriptome data amenable for downstream bioinformatics analyses, such as the determination of differentially expressed genes (DEGs), characterisation of gene regulatory networks, and conducting transcriptome-wide association study (TWAS).
The transcriptome represents the complete set of ribonucleic acid (RNA) transcripts expressed by the genome of an organism at a given time and in particular environmental and growth conditions. Each cell has its individual transcriptome, which reflects its current physiologic and metabolic state. A collection of cells derived from a similar tissue or organ is used in a typical transcriptome study, but single-cell and spatially-resolved transcriptomics are getting popular1. Transcriptomic analyses start with the extraction of the total RNA from a selected tissue at a specific time point, and in defined growth conditions. For this purpose, we are recommending the use of a newly developed method for the extraction of total RNA from samples high in starch or sugar content, such as cereal seeds2. The comparison of transcriptomes among different samples results in the identification of RNA molecules with different abundance. These RNA molecules are considered as differentially expressed genes (DEGs). The abundance of transcripts derived from specific marker genes can then be used to estimate the developmental status or determine the response of an organism to environmental fluctuations. Genes with no detectable changes in their transcript abundance across the developmental time points under study are often used as reference or housekeeping genes.
The RNA is typically detected and quantified by various methods, such as Northern blotting and quantitative reverse transcriptase polymerase chain reaction (qRT-PCR), but current high-throughput transcriptomics methods rely heavily on nucleic acid hybridization using microarray technology as well as RNA sequencing (RNA-Seq). RNA-Seq is very popular at present because it provides several advantages for high throughput transcriptomic applications as reviewed elsewhere3,4. Although an older technology, gene expression profiling using microarray chips is still widely used because it is a more established technology, which requires less of a background in bioinformatics. Compared to RNA-Seq, the data sets generated from microarray experiments are smaller and easier to analyze. In addition, it is more cost effective, especially if dealing with large sample numbers. In our laboratory, we routinely use transcriptomic analyses using the microarray technology to determine the role of central regulatory hubs that govern the molecular networks and pathways involved in the growth, development and metabolism of cereal grains5,6,7,8,9. We also routinely use it to conduct of genome-wide gene expression profiling studies to obtain a mechanistic understanding of the response of cereal grains to abiotic stresses10, as well as in the conduct of transcriptome-wide association (TWAS) and linkage mapping studies to identify genes responsible for cereal grain quality and nutrition11,12. Other groups have also used the microarray technology in providing development-specific gene expression atlas in barley13, rice14,15,16, sorghum17, and wheat18.
The purpose of this publication is to provide a brief textual summary and a detailed visual description of the method we currently employ in our laboratory for the transcriptomic profiling of cereal grains using an Agilent microarray platform. Please note that other microarray platforms are available, but will not be covered in this method. We begin the protocol by presenting a detailed description of RNA extraction from developing or germinating cereal seeds. Based on our experience, obtaining a high quality and high quantity transcriptome amenable to downstream transcriptomic analyses is often the bottleneck when using cereal seed tissues. We have tried several commercially available RNA extraction kits, but none have provided satisfactory results. Hence, we developed a chemical extraction protocol to obtain a crude RNA extract that is then subjected to column purification using a commercially available kit. Using this method, we routinely and reproducibly obtain high quality RNA2 (Figure 1), which can be used for different downstream applications to generate a transcriptome profile.
1. Total RNA extraction and purification
NOTE: Always work under the fume hood as this method involves the use of harmful volatile organic solvents. Pre-warm a microfuge tube of nuclease-free water in 50 °C heat block before commencing the experiment. This nuclease-free water will be used to elute the total RNA from spin column in Step 1.8.
2. cDNA synthesis followed by cRNA transcription and labelling
NOTE: This step is suitable for processing 24 samples simultaneously. It is suggested that the entire step is conducted continuously in one day, but optional stop and pause points are identified. Be sure to pre-warm three microfuge tube heat blocks at 80 °C, 65 °C, and 37 °C before commencing the steps below. These temperature settings will be changed as noted in the steps below. For instance, the 37 °C heat block will be set to 40 °C after the Spike Mix is prepared (or if it has already been previously prepared). Prepare one-color Spike Mix, T7 Promoter Mix and cDNA master mix using the Low Input Quick Amp Gene Expression Labeling Kit (see Table of Materials) based on the manufacturer's instructions. This preparation is dependent on the amount of starting RNA extracts, which usually range from 10 to 200 ng for total RNA or 5 ng for PolyA RNA. We routinely use 50 ng total RNA as starting material for microarray hybridization2 and for the method that will be described below. In preparing a master mix for 24 samples, add 2 extra reactions to account for pipetting variations. Always use nuclease-free microfuge tubes and nuclease-free water in this step.
3. cRNA Purification
4. Microarray hybridization and scanning
NOTE: This step only takes 3-4 h and hence hybridization of 24 samples can be started after lunch. One operator can comfortably run up to 4 slides (32 samples). The morning of the following day is allocated for washing and scanning of microarray slides. Additional runs can then be performed in the afternoon of the second day. This step is repeated until all samples are hybridized and scanned. It is highly recommended to use color-free, powder-free latex gloves for handling and processing the slides to ensure that the samples are not contaminated with colored pigments that can interfere with the microarray analyses. Aside from commercially available microarrays, the following custom arrays are designed by our group and available for order from Agilent: Order Code 028827 for Barley (Hordeumvulgare), 054269 for Rice (Oryzasativa subs. Japonica), 054270 for Rice (Oryzasativa subs. Indica) and 048923 for Wheat (Triticumaestivum).
This method is optimized for extracting cereal seed samples containing significant amounts of contaminating starch or sugars. It is designed to extract total RNA from 24 seed samples per day. It should be conducted continuously in one day, but optional stop and pause points are identified throughout the protocol. Alternatively, the reader can use their preferred RNA extraction kits or manual chemical extraction method. However, based on our previous experience, commercially available plant RNA extraction kits are not appropriate for seeds due to significant amounts of starch, proteins, sugar and/or lipid contamination. In the described method, a chemical extraction of crude RNA is followed by column purification using a commercial RNA extraction kit. This typically provides RNA with higher quality and yield. Figure 1 shows the result of a BioAnalyzer run to test the quality of RNA extraction using the method described here. Results are presented for barley leaf (Samples 1 and 2 representing low starch content samples) and barley seeds (Samples 3 and 4 representing high starch content samples). The RNA integrity (RIN) value for all samples is 10.
Figure 1 shows a representative gel of the purified RNA extract, where the quality was tested using a Bioanalyzer. Samples 1 and 2 are typical results for low starch content samples such as barley leaves, where additional rRNA bands from chloroplasts are evident. Samples 3 and 4 are representative results for high starch content samples such as barley seeds, showing 18S and 28S rRNA. Please note that no automated RIN value can be calculated for green plant tissues such as leaf samples, due to chloroplast rRNA. However, the integrity and high quality can be visually ascertained by to the absence of any degradation products, which typically appears as low molecular smear. The RIN value for high-starch seeds samples such as barley seeds is typically 10 using the protocol described in this paper.
In addition, we also present here a representative data of a time course experiment of two elite barley inbred lines (Sofiara and Victoriana) used for analysis of malting quality19. During the malting process, starch is converted into sugar. Therefore, the samples represent tissues with varying proportions of starch and sugar contents. As the industrial malting process is similar to the germination process, the transcriptomes of two barley lines, differing in their malting quality, were analyzed. RNAs were extracted from germinating seeds at 2, 24, 48, 72, 120, 144 and 196 h after imbibition in biological triplicates. The RNA preparation and hybridization to the customized barley microarray chip were performed as described above. Figure 2 indicates the normalized grid read out from the microarray hybridization given in the quality control (QC) report (Figure 3) and the histogram plot for detected signals. The grid gives the example of derived signals from each corner of the chip, including background and spike-in read out dots used for calibration. The histogram indicates the deviation of detectable dots with respective signal intensities. A successful hybridization gives a broad Gaussian-shaped curve with only minor outliers as shown in the figure. Failed hybridizations can result in a strong shift towards one side ("green monsters").
Lastly, a representative result of the acceptable values is shown in column 4 of Figure 3. To indicate the reliability of the performed experiments, the results are further evaluated using the GeneSpring software. The collected data are presented as principal component analysis (PCA). The PCA integrates the values of selected dots (genes) as vector. The number of evaluated dots that are used can vary from several hundred to the entire chip and is dependent on the software used. Each chip (sample) results in one value (vector) that resulted from the integrated signal intensities for the analyzed dots. Therefore, the relative position in the graph (PCA) indicates the similarity of the samples to each other. The closer the samples are, the more similar they are. Technical replicates should be closer together than biological ones, and biological replicates of a sample should cluster closer together, than samples from different time points, tissues or conditions.
Figure 1: Electrophoresis file run summary obtained after checking the quality of RNA with Bioanalyzer. Samples 1 and 2 are barley leaf tissue as indicated by additional ribosomal bands from chloroplasts. Samples 3 and 4 are barley seed tissue with 18S and 28S rRNA bands shown. A RIN factor is not always calculated for green tissues such as leaf but according to the gel, the quality of RNA is very good. The RIN value for samples 3 and 4 is 10. Please click here to view a larger version of this figure.
Figure 2: QC Report from successful barley microarray hybridization. A + indicates detected signals on the grid from all corners of the chip. The histogram shows the number of signals categorized according to signal intensity (fluorescence) as logarithm after background subtraction. Please click here to view a larger version of this figure.
Figure 3: Summary of quality control (QC) after hybridization and scanning. The values for the hybridized slide are given in column 2 (value) and the range of acceptable values is shown in column 4. Please click here to view a larger version of this figure.
Wash Chamber Assembly | Content and Label | Purpose |
Dish 1 | Empty, leave on lab bench until next day | Fill with Wash Buffer 1 the next day, used to disassemble the microarray slides |
Dish 2 | Add one microarray slide rack and a small magnetic stir bar; label with "Wash Buffer 1", leave on lab bench until next day | Used to wash the microarray slides with Wash Buffer 1 the next day |
Dish 3 | Add one small magnetic stir bar, label with "Wash Buffer 2" and place in a 37 °C mini incubator. | Used to wash the microarray slides with Wash Buffer 2 the next day inside the 37 °C mini incubator |
Table 1: Preparation of the wash chamber assemblies.
Steps | Dish | Wash Buffer | Temperature | Time |
Disassembly | 1 | 1 | Ambient | As fast as possible |
(Step 4.13) | ||||
First wash | 2 | 1 | Ambient | 1 min |
(Step 4.14) | ||||
Second wash | 3 | 2 | 37 °C | 1 min |
(Step 4.15) |
Table 2: Incubation temperature and time for wash chamber assemblies.
The described method provides highly reproducible results for high-yielding and high-quality RNA extracts (Figure 1). Based on our experience, we recommend three biological replicates for one genotype, stage or condition for analysis. To be able to detect statistically meaningful differences, the general abundance of mRNA must be considered. Please note, however, that a starting amount of 200 mg tissue samples is required in triplicates for the RNA extraction step. Hence, for experiments that have a limited amount of samples such as genetically modified plant materials, this method may not be appropriate.
During microarray hybridization, the following steps are most critical: (1) loading the samples to the gasket slide (Step 4.7) and (2) washing the arrays after hybridization (Steps 4.13 and 4.14). Sample loading needs to be done very carefully to avoid bubble formation and liquid spilling. Caution is necessary when putting the microarray slide on the gasket slide containing the loaded samples: the DNA-side (with spotted probes) should be facing down to allow hybridization. For washing the microarrays, the second wash step is especially critical: here, removing the slides from wash buffer 2 needs to be performed very slowly (10 s) to avoid buffer artefacts on the slides, which can interfere with data acquisition and analyses.
In order to receive results from the hybridization experiment that are eligible for down-stream bioinformatics analysis, one should follow a few important criteria. The QC report, which is generated after the performance of the protocol above, gives a good idea of what should be improved, and which data are in a usable condition (Figure 3). The first information received is the visualized grid (Figure 2). Here, one can directly see if all areas for the fluorescence read-out are properly aligned. If there is a visually detectable shift, this will influence all other values (Background, signal intensity, etc.) and the read out of this chip cannot be used. On the overview (Spatial Distribution of all Outliers), one can easily spot any contamination (e.g., dust or hair), which affected the outcome. Dirt can be removed by softly blowing and rescanning the chip. The histogram of signal plot as mentioned above should result in a broad Gauss-shaped curve, with only minor outliers (Figure 2). Depending on the tissue analyzed (Figure 1), this curve can look different and can be appearing to have two peaks, or one predominant peak with a shoulder. This will not influence the quality of data. Only a strong shifted peak, or high signals on the edges indicate problems, such as "green monsters" (too high fluorescence intensities), which cannot be removed by washing and should be reported to the chip manufacturer. Also, the "Spatial Distribution graph for Median Signals" should vary around a common value. This should be in the range between 40 and 100, depending on the general intensity readout of the chip analyzed. Another important quality control is the evaluation of the spike in data: The log-graph for the spike in signals should result in a linear and regular line. Finally, the table of "evaluation metrics" gives a good and reliable view of the received results. Here the manufacturer provides a range of values which can be classified as Excellent, Good and Evaluate (Figure 3). According to our experience, some of the values cannot always be applied for all chips. For rice custom chips, the "nonControl value" was frequently out of range but does not significantly affect further data processing. In addition, the range for "NegControl" might be extended, based on tissue, organism and general output of the Chip to <80. The range for evaluation for value E1 med CV might be extended to <9, instead of <8 according to our experience. Following this, a proper evaluation of all chip read out data is possible. In general, one should always keep in mind that independent biological replicates always give the most reliable result.
The advantage of this technique compared to other methods lies in the fact that it represents a cost-effective, high-throughput method for transcriptional profiling. Moreover, the downstream data analysis pipeline is easier and more established. Despite advantages, there are certain limitations of this technique, such as the number of genes that can be analyzed. The microarray can only facilitate the analysis of genes previously spotted probes on the array. Therefore, this technique is only applicable to plant species with sequenced genomes and fully annotated genes. We envision that the microarray-based transcriptomics will continue to be very useful and relevant not only for gene expression profiling but also for other downstream bioinformatics pipelines such as gene regulatory network analyses and transcriptome-wide association study.
The authors have nothing to disclose.
This work has been supported under the CGIAR thematic area Global Rice Agri-Food System CRP, RICE, Stress-Tolerant Rice for Africa and South Asia (STRASA) Phase III, and IZN (Interdisciplinary Centre for Crop Plant Research, Halle (Saale), Germany. We thank Mandy Püffeld for her excellent technical assistance and Dr. Isabel Maria Mora-Ramirez for sharing information and experience with the wheat chip. We thank Dr. Rhonda Meyer (RG Heterosis, IPK Gatersleben, Germany) for comments and critically reading the manuscript.
ß-mercaptoethanol | Roth | 4227.3 | Add 300 ul to 20 mL RNA Extraction Buffer immediately prior to use |
Bioanalyzer 2100 | Agilent Technologies | To determine quality and quantity of RNA extract | |
Crushed ice maker | Various brands | To keep samples on ice during sample processing | |
Dewar flask | Various brands | Used as liquid nitrogen container | |
Ethanol, absolute | Roth | 9065.1 | |
Gene Expression Hybridization Kit | Agilent Technologies | 5188-5242 | Kit components: 2X Hi-RPM Hybridization Buffer 25X Fragmentation Buffer 10X Gene Expression Blocking Agent |
Gene Expression Wash buffer Kit | Agilent Technologies | 5188-5325 | Kit components: Gene Expression Wash Buffer 1 Gene Expression Wash Buffer 2 Triton X-102 |
Heat block | Various brands | It is recommended to have at least one heat block for 1.5 ml tubes and another one for 2.0 ml tubes | |
Hybridization oven | Agilent | G2545A | Stainless-steel oven designed for microarray hybridization From Sheldon Manufacturing, used to hybridize the sample in microarray overnight. |
Hybridization gasket slide kit | Agilent | G2534-60014 | |
iQAir air cleaner | To ensure that the experiment is conducted in low-ozone area | ||
Isopropanol | Roth | 9866.5 | |
Lint-free paper, Kimwipes | Kimberly-Clark Professional | KC34155 | |
Low Input Quick Amp Labeling Kit | Agilent Technologies | 5190-2305 | Kit components: T7 Primer 5x First Strand Buffer 0.1M DTT 10 mM dNTP Mix AffinityScript RNAse Block Mix 5x Transcription Buffer NTP Mix T7 RNA Polymerase Blend Nuclease-free water Cyanine 3-CTP |
Metal spatula, small | Ensure that the small metal spatula can fit in the microfuge tubes to ensure easy scooping of samples. | ||
Microarray scanner | Agilent Technologies | Agilent SureScan or Agilent C microarray scanner are recommended | |
Microfuge tubes, nuclease-free | Various brands | 2.0 mL and 1.5 mL volume | |
Microcentrifuge | Eppendorf | 5810 R | It is recommended to have at least one ambient and cold temperature microfuge with rotors that can hold 24 each of 1.5 ml and 2.0 ml microfuge tubes |
Mini incubator | Labnet | I5110-230V | Any small incubator will do. |
Mortar and pestle | Any small ceramic or marble mortar and pestle that can withstand cryogenic grinding using liquid nitrogen. | ||
NaCl | Sigma-Aldrich | S7653-1KG | |
Nanodrop 1000 | Peqlab / Thermofisher | ||
Nuclease-free water | Biozym | 351900302 | |
One-Color RNA Spike-In Kit | Agilent | 5188-5282 | Kit components: One color RNA Spike-Mix Dilution Buffer |
Ozone barrier slide cover kit | Agilent | G2505-60550 | Optional but highly recommended |
PCR tube, 0.2 mL, RNase-free | Stratagene | Z376426 | |
Phenol:chloroform:isoamyl mixture (25:24:1) | Roth | A156.2 | |
Pipette tips, nuclease free, filter tips | Various brands | To accommodate 2, 20, 20, 100, 200 and 1000 ul volumes | |
Pipettor set | Various brands | For 2, 20, 20, 100, 200 and 1000 ul volumes | |
RNA Extraction Buffer | 10 mM Tri-HCl pH 8.0 150 mM LiCl 50 mM EDTA 1.5% EDTA 15 ul/mL β-mercaptoethanol |
We routinely use Roth or Sigma chemicals; Add β-mercaptoethanol fresh daily | |
RNase-free DNase set | Qiagen | 79254 | |
RNeasy Mini Kit | Qiagen | 74104 | Kit components: RNeasy mini spin column (pink) 1.5 ml collection tubes 2 ml collection tubes Buffer RLT Buffer RW Buffer RPE Nuclease-free water |
RNeasy Plant Mini Kit | Qiagen | 74904 | Kit components: RNeasy mini spin column (pink) QIAshredder spin column (purple) 1.5 ml collection tubes 2 ml collection tubes Buffer RLT Buffer RLC Buffer RW1 Buffer RPE Nuclease-free water |
Slide holders for DNA microarray scanner | Agilent | G2505-60525 | |
Slide staining dish with removable rack | DWK Life Sciences 900200 | Fisher Scientific 08-812 | We recommend DWK Life Sciences Wheaton™ Glass 20-Slide Staining Dish with Removable Rack (Complete with dish, cover, and glass slide rack) |
Sodium chloride | Roth | P029.1 | |
Ultralow temperature freezer | Various brand | Capable of storing sample at -80 °C | |
Vortex mixer | Various brands | At least one for single tube vortex-mixer and another one for that can vortex-mix multiple tubes |