Identification and validation of superior housekeeping gene(s) for qRT-PCR data normalization in Agave sisalana (a CAM-plant) under abiotic stresses

Abstract

The adaptive mechanisms in Agave species enable them to survive and exhibit remarkable tolerance to abiotic stresses. Quantitative real-time PCR is a highly reliable approach for validation of targeted differential gene expression. However, stable housekeeping gene(s) is prerequisite for accurate normalization of expression data by qRT-PCR. Till date, no systematic validation study for candidate housekeeping gene identification or evaluation has been carried-out in Agave species. A total of 17 candidate housekeeping genes were identified from the de novo assembled transcriptomic data of A. sisalana and rigorously analyzed for expression stability assessment under drought, heat, cold and NaCl stress. Different statistical algorithms like geNorm, BestKeeper, NormFinder, and RefFinder on expression data determined the superior housekeeping gene(s) for accurate normalization of the gene of interest (GOI). The comprehensive evaluation revealed the β-Tub 4, WIN-1 and CYC-A as the most stable, while EEF1α, GAPDH, and UBE2 were ranked as the least stable genes in pooled samples. Pairwise combination by geNorm showed that up to two housekeeping genes would be adequate to normalize the GOI expression data precisely. Validation of identified most and least stable housekeeping genes was carried-out by normalizing the expression data of AsHSP20 under abiotic stress conditions. Copy number of AsHSP20 gene supports the reliability of the genes used for normalization. This is the first report on the screening and validation of the housekeeping genes under abiotic stress condition in A. sisalana that would assist to understand the stress tolerance mechanisms by novel gene identification and accurate validation.

Electronic supplementary material

The online version of this article (10.1007/s12298-020-00760-y) contains supplementary material, which is available to authorized users.

Introduction

Abiotic stresses like drought, temperature (heat, cold, frost), salinity and nutrients deficiency are the key factors that drastically affect the growth, development, and productivity of crop plants worldwide (Sarwar et al. 2014). Plant transcriptomic behavior alters continuously with respect to time and place by these external environmental conditions. Advancements in molecular biology have enabled the researchers to study these changes at a single cell level via high-throughput techniques. Nowadays, RNA sequencing is the most extensively used approach to capture the comprehensive transcriptomic dynamics of the model and other plant species (Egan et al. 2012). For expression studies, Differential Expression of Gene (DEGs) requires further validation by using different procedures like Northern blotting, quantitative and semi-quantitative real-time PCR (q/qRT-PCR) (Nikalje et al. 2018). Among those, qRT-PCR is considered as a gold standard approach because of its robustness, reliability, sensitivity, and reproducibility. On the contrary, various factors like experimental design, nucleic acid purity, transcriptional efficiency and concentration of cDNA affect the reproducibility and efficiency of the qRT-PCR based validation experiments (Zhang et al. 2017). To fine-tune such unexpected variations, normalization of expression data with appropriate housekeeping genes (HKGs) becomes necessary (Barbierato et al. 2017). Generally, these genes are assumed to have an absolute stable response regardless of the external environmental conditions because of their involvement in the housekeeping metabolic functions (Gao et al. 2017). However, in a real-time situation, the expression of commonly used HKGs may vary considerably under different experimental and environmental conditions, developmental stages, different tissues, and even cell types, which affects the GOI (gene of interest) expression evaluation (Zhang et al. 2017).

To overcome this situation, the best possible approach is to carry out a comprehensive comparative qRT-PCR study to evaluate all possible combinations of HKGs for identification of the most suitable with respect to the specific condition prior to an experiment. Such screening and evaluation studies have been carried out for a number of plant species including Nicotiana tabacum (Schmidt and Delaney 2010), potato (Nicot et al. 2005), tomato (Wieczorek et al. 2013), Sapium sebiferum (Chen et al. 2017), Sesuvium portulacastrum (Nikalje et al. 2018) Sorghum bicolor (Reddy et al. 2016), Glycine max (Gao et al. 2017), Oryza sativa (Auler et al. 2017) etc. Several web sources and databases like Gene-quantification eportal www.Gene-Quantification.info, Minimum Information for Publication of Quantitative Real-time PCR Experiments (MIQE), a database of OMIC Tools, ERCC (External RNA controls consortium) etc. provides a suitable platform for discussion about the criteria for accurate normalization of experimental data. Aside, there are a number of other statistical tools designed to perform the stability analysis with their own advantages and pitfalls.

Agave, a member of the Agavaceae family, consists of about 166 cultivated species (Gil-Vega et al. 2006). It has been cultivated for food, fiber, medicine, beverages, biofuel, textile, other vital products etc. by indigenous tribes of the southwestern U.S., Mexico and Central America (Gross et al. 2013). Millions of years of selection pressure from evolution constructed the genomic structure of the agave in a way that enabled it to remain productive with least hydration and nutrition availability (Stewart 2015). “Sisal” is the sixth most important fiber, harvested from the Agave sisalana, representing 2% of the world’s natural fiber production (FAO 2018). A. sisalana displays exceptional abiotic stress tolerance as crassulacean acid metabolism (CAM) enables the plant to survive in extremely dry weather and extreme temperature (− 16.1 °C–77.4 °C) conditions (Gross et al. 2013; Sarwar et al. 2019). Remarkable tolerance to the abiotic stresses makes this species an ideal plant to explore the essential molecular and genetic information for abiotic stress tolerance traits.

As comprehensive whole-genome information of A. sisalana is not yet available and lack of transcriptomic information limits the ways for gene identification, evaluation characterization etc. Till now, there has been no published work aimed at identifying the effective housekeeping genes (HKGs) for qRT-PCR study in Agave species. Keeping this in view, we utilized a de novo assembled transcriptome data of A. sisalana published by Sarwar et al. (2019) against the locally developed database of reported housekeeping genes of model plant species. Based on homology, total seventeen genes were selected as candidate HKGs for abiotic stress tolerance in this study. Further stability analysis was performed by using algorithms and statistical analysis like BestKeeper (Pfaffl et al. 2004), NormFinder (Andersen et al. 2004), geNorm (Vandesompele et al. 2002), and RefFinder (Xie et al. 2012) under different abiotic stress treatments (drought, heat, cold, NaCl and rehydration). The expression of small heat shock protein gene AsHSP20 was normalized to test the topmost and least stable reference candidate genes in order to validate the data/results. This study will be effective for accurate gene expression analysis in A. sisalana as well as other crop species under abiotic stress conditions.

Materials and methods

Plant material, growth conditions, and abiotic stress treatment

The off-shoots of same age and size were harvested from the single mature A. sisalana plant and grown in the plastic pots (15.0 cm top diameter and 14.5 cm deep) having the peat moss and soil mixture as 1:1 under 16 h photoperiod at semi-controlled condition (32 ± 2 °C day/28 ± 2 °C night) in the glasshouse. Ninety days after sowing at 3 to 5 leaf stage, the plants were randomly shifted into five groups, containing six plants per group. These groups were subjected to abiotic stress treatments; drought, heat, cold, NaCl and rehydration. For low and high-temperature stress treatments, the plants were placed in a controlled cooling chamber (Ningbo Hinotek Technology Co., Ltd. BPC-250F) at 4 °C and 60 °C respectively. The leaves samples were harvested at 3, 6, 9 and 12 h time interval (Ma et al. 2016). For salt stress treatment, the NaCl solution (100 mM to 400 mM) was applied to plants as a substitute to normal water, while for drought induction, we stopped regular watering to the plants till the appearance of stress/wilting symptoms on leaves. For rehydration study, the drought-stressed plants were irrigated adequately with normal water and leaf sampling was carried out at 3, 6, 9 and 12 h interval after rehydration. We also maintained a control group of plants that were watered regularly and kept under normal temperature. A single juvenile leaf near to the middle of the rosette was harvested per plant of each group as reported by Sarwar et al. (2019). Sampling was done as three biological replicates from each treatment and harvested samples were immediately immersed into liquid nitrogen, ground well to a fine powder and stored at -80 °C for further analyses.

Total RNA isolation and cDNA synthesis

Total RNA was isolated from the leaf samples by using Trizol method followed by column-based purification as reported (Sarwar et al. 2019). The DNA contamination was removed by using RNase free amplification grade DNaseI (AMPD1-Sigma). The RNA quantity and quality parameters were ensured by Nanodrop ND-1000 (ThermoFisher Scientific) and Agilent 2100 Bioanalyzer. The first-strand cDNA synthesis was carried out with SuperScript™ IV First-Strand Synthesis Kit (Catalog number: 18091050-Invitrogen™) following the manufacturer protocol.

Candidate housekeeping gene identification and primer designing

De novo assembled database (TSA ID: GGRE00000000.1) was used to mine out the possible candidate HKGs from A. sisalana by considering the following parameters. Initially, the unigenes having the FPKM value less than 10 in the transcriptome count data were removed by assuming that sub-standard expressed genes would be least stable, difficult to detect and might be poor candidate as HKGs.

Average expression and Standard Deviation (Std) values of each transcript were used to determine the coefficient of variation (CV) via Microsoft Excel program. Means and Std were calculated over the leaf specific libraries (Sarwar et al. 2019), while CV was calculated by dividing the Std over mean expression (Std/mean) as mentioned by Czechowski et al. (2005). Screening parameters were as; a CV cutoff for each unigene below 0.3, with mean counts above 500 (Li et al. 2017). Further, a local database with proper annotation was developed as based on reported stably expressed HKGs from plant species including Arabidopsis thaliana (a dicot model) (Czechowski et al. 2005; Dekkers et al. 2012; Remans et al. 2008), Oryza sativa (monocot model) (Jain et al. 2006), Nicotiana benthamiana (Schmidt and Delaney 2010), Glycine max (Jian et al. 2008; Libault et al. 2008), Triticum aestivum (Paolacci et al. 2009), Sorghum bicolor (Reddy et al. 2016), Zea mays (Manoli et al. 2012), Pennisetum glaucum (L.) (Reddy et al. 2015) etc. and BLAST at e-value ≤ 10⁻⁹⁰ against the filtered TSA database by BLAST⁺ programme ftp.ncbi.nlm.nih.gov/blast/executables/blast+/LATEST/ (Additional File 1). Total 17 candidate HKGs were identified on the basis of top bit score and e-value for further validation (Additional File 2). Primers were designed by using Primer Premier 6.0 http://www.premierbiosoft.com/primredesign/ within the specific region of the sequence with T_m between 59 and 61 °C, primer length as 20–22 nucleotides, and amplicon length was within 120-150 base pair as per MIQE guidelines (Additional File 3).

Quantitative real-time PCR assay

MIQE guidelines were followed for conducting the qRT-PCR experiments. To verify the specificity, target genes were amplified by using the recombinant Taq DNA Polymerase (Invitrogen™ Cat # 10342020) and resolved on 2.5% agarose gel prepared in 1X TAE buffer. qRT-PCR experiments were carried out on the StepOnePlus™ Real-Time PCR System (Thermo Fisher Scientific) in 96 well microtiter plates using SYBR^® Green PCR Master Mix (cat # 4309155) with total final reaction volume of 25 µl. The primer amplification efficiency was individually calculated by using a 10-fold serial dilution (1, 0.1, 0.01, 0.001, 0.0001). The thermal profile includes the following conditions: Initial denaturation at 95 °C for 10 min followed by 40 amplification cycles at 95 °C for 30 s, 60 °C for 30 s and 72 °C for 30 s. A dissociation (melting) curve was also run after each qRT- PCR to determine the specific amplification. Relative expression software tool (REST™) software was used for expression analysis (Pfaffl et al. 2002). Three independent biological and technical replicates were used to carry out this gene expression study experiment.

Stability analysis of HKGs

Four algorithms, BestKeeper (Pfaffl et al. 2004), NormFinder (Andersen et al. 2004), geNorm-v:3.0; by Biogazelle, Belgium (Vandesompele et al. 2002) and the RefFinder tool (Zmienko et al. 2015) were considered for stability evaluation across all the experimental sets according to their corresponding manual instruction. The data obtained from the StepOnePlus™ Real-Time PCR System (Thermo Fisher Scientific) were directly exported to the MS excel datasheet and converted into required file formats for downstream analysis. BestKeeper algorithm requires raw C_T values to calculate the CV and Std for stability analysis (Pfaffl 2001; Pfaffl et al. 2004), while for geNorm and NormFinder, the raw C_T values were initially transformed to the relative quantities by using the formula 2^−ΔCT(ΔC_T = each corresponding C_T value − minimum C_T value) and then uploaded to the respective applet for stability measurement (M-value) (Wang et al. 2017). NormFinder employed an ANOVA based model to calculate the variation between candidate genes in all the analyzed samples as well as among inter- and intra groups (Andersen et al. 2004). The geNorm program identified the most stable reference genes based on the average pairwise variation of a reference gene and ranks them by their expression stability values (M) (Gao et al. 2017). In both cases, genes with the lowest M-values represents the highest stability and vice versa. Finally, RefFinder approach was employed to generate a comprehensive ranking of selected HKGs (Zhang et al. 2017).

Heat shock protein Gene20 from Agave sisalana (AsHSP20) cloning and plasmid copy number

The sequence of drought-responsive AsHSP20 gene was amplified by using sequence-specific primers: Forward Primer: 5′-TTCCGCCCTTTCTGCTCTCCTT-3′, Reverse Primer: 5′-CCGTATCTCGCCTGCTGTGTTG-3′ and cloned in the pJET1.2Vector (Additional File 4). The concentration of plasmid was measured by nanodrop ND-1000 at 260 nm. Following formula was used to determine the plasmid molecules/copies as described by (http://scienceprimer.c/copy-number-calculator-for-realtime-pcr).

$$\text{Number}\text{of}\text{copies}(\text{copies}u{\text{l}}^{-1})=\frac{6.0\times 1023\left(\frac{\text{copies}}{}\text{mol}\right)\times \text{plasmoid}\text{concentration}(\text{g/}u\text{l)}}{(\mathrm{number}\text{of}\text{base pair}\times 660\text{g}{\text{mol}}^{-1})}$$

Establishment of standard curve of AsHSP20 for absolute quantification

A standard curve was obtained by diluting the purified plasmid five times with ddH₂O from 3.848 × 10⁹ copies µl⁻¹ to 3.848 × 10⁵ copies µl⁻¹ (Additional File 5). To draw a standard curve based on the plasmid dilutions, the qPCR experiment was conducted by using 1 µl diluted plasmid as template per reaction with three replications per dilution. The standard curve was drawn in the excel program by using raw C_T values as detailed by (Karuppaiya et al. 2017).

Validation of HKGs by Relative and Absolute Quantification of AsHSP20

For data validation, the expression level of A. sisalana small heat shock protein gene (AsHSP20) was analyzed by using the most stable reference gene as recommended by RefFinder for data normalization under drought, heat, cold, NaCl stress and rehydration conditions. For comparison, the least stable reference genes were also used for the validation test. Relative Expression Software Tool-REST™ (http://www.gene-quantification.com/rest.html) was used to calculate the relative fold expression of the AsHSP20 (Livak and Schmittgen 2001), while the standard curve was used to determine the exact copy number of AsHSP20 gene under respective stress. One-way ANOVA was used to check the significance of data. The Statistical and graphical studies were conducted with GraphPad Prism7 software (https://www.graphpad.com/scientific-software/prism/).

Results

Pre-analytic assessment of candidate HKGs for determining primer specificity and amplification efficiency by qRT-PCR

In this study, a total of 17 HKGs were selected from the A. sisalana transcriptome data (Sarwar et al. 2019). These includes ADP-Ribosylation Factor 2 (ARF2), Cyclophilin A (CYC-A), Ribulose Bisphosphate Carboxylase Activase B (RcaB), RuBisCO activase (RCA), Actin 11 (ACT11), Beta-Tubulin 4 (β-Tub 4), Eukaryotic Elongation Factor 1 alpha(EEF1α), Eukaryotic Initiation Factor-4A (eIF-4A), Glyceraldehyde-3-Phosphate Dehydrogenase (GAPDH), Polyubiquitin (UB), RNA Polymerase II (RPII), RuBisCO Small Subunit (RBCs), Serine/Threonine-Protein Phosphatase Catalytic Subunit (PP2A-1), Cullin-1 (CUL-1), WIN-1, Ubiquitin 10 (UB10) and Ubiquitin-Conjugating Enzyme (UBE2) (Additional File 2). Table 1 indicates the exact annotation, cellular function, gene symbols, orthologous and NCBI accessions IDs. The specificity and amplification efficiency of primers for target genes were determined by gel electrophoresis and dissociation melt curve analysis. The genes with the specific and unique amplification at expected band size without any sign of the primer dimer formation were considered for further study. To ensure the single peak presence during amplification, a melt curve analysis of amplified products was carried out by qRT-PCR with 40 cycles of amplification (Fig. 1a, b). The presence of a single peak indicates the specific amplification of amplicons by specific primer. Despite the use of several alternate primers and variable thermal profiles for PCR based screening, non-specific amplification could not be avoided for RcaB, RCA, RBCS, and UB10 genes and later excluded from this study for further analysis. The amplification efficiency (E) by used primers displayed the good value in the range of 92 to 104% with linear Regression Coefficient (R²) value greater than the 0.92 (Additional File 3). All statistical results were obtained using three independent biological and technical replicates. No amplicons and signals were detected in (NTC) control samples. All these results indicate that these gene specific primers could be used for the next qRT-PCR analysis under abiotic stress conditions.

Table 1.

Agave sisalana candidate reference genes with gene symbols, accession numbers, descriptions, cellular functions and ortholog locus

Gene symbol	NCBI accession	Blast results		Cellular function
		Annotation	Ortholog Locus
ARF2	JZ979889.1	ADP-ribosylation factor 2	AT5G62000.1	Protein trafficking
CYC-A	JZ979890.1	Cyclophilin A	AT1G44110.1	Accelerate the folding of proteins
RcaB	JZ979892.1	Ribulose bisphosphate carboxylase activase B	AT2G39730.1	ATP binding
RCA	JZ979891.1	RuBisCO activase	AT2G39730.1	ATP binding
ACT11	JZ979893.1	Actin 11	AT3G12110.1	Cell shape determination, extension growth.
β-Tub 4	JZ979894	Beta-tubulin 4	AT5G44340.1	The major constituent of microtubules
EEF1α	JZ9798945	Eukaryotic elongation factor 1-alpha	AT5G60390.1	Promotes the GTP-dependent binding of aminoacyl-tRNA to the ribosomes during protein biosynthesis
eIF-4A		Eukaryotic initiation factor-4A	AT3G19760.1	ATP-dependent RNA helicase
GAPDH	JZ979896	Glyceraldehyde-3-phosphate dehydrogenase	AT3G04120.1	Encodes cytosolic GADPH (C subunit) involved in the glycolytic pathway b
UB	JZ979897	Polyubiquitin	AT4G05320.2	Covalently attached to substrate proteins targeting most for degradation
RPII	JZ979898	RNA polymerase II	AT2G15400.1	Catalyzes the transcription of DNA to synthesize precursors of mRNA
RBCS	JZ979899	RuBisCO small subunit	AT1G67090.1	Functions to yield sufficient Rubisco content for leaf photosynthetic capacity
PP2A-1	JZ97900.1	Serine/threonine-protein phosphatase catalytic subunit	AT1G59830.1	Phosphoprotein phosphatase activity
CUL1	JZ979901.1	Cullin-1	AT4G02570.1	Complexes involved in mediating responses to auxin and jasmonic acid
WIN-1		WIN-1	AT1G15360.1	Protein domain specific binding
UB10		Ubiquitin 10	AT4G05320.2	Membrane proteins, Genomic maintenance, Transcriptional regulation
UBE2	JZ9798902.1	Ubiquitin-conjugating enzyme	AT5G41700.4	Mediates the selective degradation of short-lived and abnormal proteins
Target genes
AsHSP20	Submitted	Small heat shock protein 20	AT5G59720.1	Response to heat, response to high light intensity, response to hydrogen peroxide, response to reactive oxygen species, response to salt stress

Fig. 1.

The primer specificity and amplicon size determination. a The melt curve analysis of the 12 candidate housekeeping genes after the quantitative real-time PCR showed single peak amplification. b Gel electrophoresis of the amplified product on 2% agarose gel indicated that the amplification of single PCR product of expected size for the 12 genes. M represents the 100 bp DNA marker

Quantification values of candidate HKGs

The qRT-PCR was performed to determine the stability of candidate HKGs at mRNA level under various abiotic stress conditions using StepOnePlus™ Real-Time PCR System (Thermo Fisher Scientific) and results were obtained as raw C_T values (Additional File 6). The raw C_T values directly indicate the abundance of GOI under a specific condition. Lower C_T values indicate higher expression level and vice versa. In all the samples used in this study, C_T values for the candidate HKGs were observed in the range of 14.5 to 30.7 (GAPDH to eIF-4A). This wide fluctuation in the expression of candidate genes suggests their diverse response towards different environmental conditions in A. sisalana (Fig. 2a–g). GAPDH was the most abundantly expressed HKG with the lowest C_T value (14.9-25.9 ± 0.92) followed by the ARF2 (17.9–25.5 ± 1.44), UBE-2 (18.8–25.9 ± 1.83), β-Tub 4 (19.2–25.9 ± 1.55), CYC-A (19.9–27.0 ± 1.61), RP-II (21.3–27.8 ± 1.21), eIF-4A (21.4–30.7 ± 2.52), EEF1-α (22.0–30.7 ± 2.75), WIN-1 (24.9–26.9 ± 0.83), PP2-A (25.2–26.7 ± 1.10), ACT11 (25.7–26.9 ± 1.18), and CUL-1 (26.3–28.1 ± 1.43). The primer sequences with their amplicon length, annealing temperature, average C_T value, Std, CV and amplification efficiency for individual HKGs are described in the Additional File 3. The genes with higher Std indicates more variable expression as compared to those with lower Std, while the CV indicates the stability of expressed genes. The CV values of these twelve genes were; CYC-A (8.2%), ARF2 (10.72%), ACT11 (4.052%), β-Tub 4 (8.74%), EEF1-α (15.08%), eIF-4A (9.68%), GAPDH (18.69%), RP-II (8.68%), PP2-A (5.17%), CUL-1 (4.76%), WIN-1 (3.09%) and UBE-2 (11.24%) respectively. The HKGs stability ranking as based on the CV values was followed as (from the least stable to the most stable) GAPDH, EEF1-α, UBE-2, ARF2, eIF-4A, β-Tub 4, RP-II, CYC-A, PP2-A, CUL-1, ACT11 and WIN-1 (Additional File 3 and 6). Briefly, these results indicate that the behavior of candidate reference genes varied across different experiments altogether.

Fig. 2.

The expression level of tested reference genes under various abiotic stress conditions. a the control condition, b drought stress, c rehydration condition, d high temperature 60 °C condition, e low temperature 4 °C condition, f NaCl stress condition (100 mM, 200 mM, 300 mM, 400 mM), g pooled samples

Stability analysis of candidate HKGs

geNorm, NormFinder, BestKeeper, and RefFinder statistical tools were used to screen out the suitable HKGs for specific stress condition in the leaves of the Agave sisalana. The stability ranking for each of the above-mentioned algorithms is detailed in the following sections.

geNorm indicators

All of the candidate HKGs were evaluated by using the geNorm software (Fig. 3). This determines a normalization value (M) based on geometric means and means pairwise variation of each HKG relative to others under a given set of condition(s). A default cutoff range to eliminate any candidate gene by the geNorm algorithm is < 1.5. Therefore, under a specific condition, genes with the lowest M value reflect the highest stability in terms of gene expression and vice versa. By following this criterion, RP II, β-Tub 4 and PP2A-1 were found to be the most stable HKGs under drought stress while GAPDH, RP II, and ARF2 gained the lowest M values respectively under rehydration condition (Fig. 3a, b). CUL-1, GAPDH, and CUL-1, WIN-1 were the most stable HKGs under high and low-temperature stress condition respectively (Fig. 3c, d), while ARF2 and β-Tub 4 were the most stable HKGs for NaCl-treated leaf samples Fig. 3e. Combined analysis of pooled stressed samples showed that a total of 6 candidate HKGs β-Tub 4, ARF2, CYC-A, eLF-4A, PP2-A, and CUL-1 justify the default limit of stability Fig. 3f. In addition to the stability value, geNorm also determines the minimum number of ideal gene(s) required for optimal normalization of expression data under a particular condition, based on the pairwise variation between ranked genes. Generally, a threshold of 0.15 (V_n value) is usually applied to calculate the best possible reference genes. When a paired value reaches to 0.15, additional reference genes (n + 1) would not be necessary. In this study, 2 solo set points of pairwise combination (V_n/V_n + 1) lower than the default limit were found for different conditions. First, (V_n/V_n+1:0.15) value was achieved at V_2/3 point for the drought (0.01), high temperature (0.09), low temperature (0.12) and NaCl stressed (0.07) sample. This indicates that only 2 HKGs would be sufficient for the accurate normalization of these samples under respective stress conditions, while for the samples under rehydration condition, the default set value was achieved at V_4/5 point of combinations, which specifies that involvement of up to four genes would be critical for the accurate normalization. Under pooled stressed samples, all pairwise variation (V_n/V_n+1) were above the 0.15 default limit (Additional File 7).

Fig. 3.

Ranking of 12 candidate housekeeping genes based on average stability values (M) determined by geNorm software. a Drought stress condition; b Rehydration condition; c High-temperature stress; d Cold stress condition; e samples submitted to NaCl stress and f pooled dataset

NormFinder ranking

NormFinder is an excel based mathematical algorithm that analyzes each sample set individually and estimates the intra- and inter-group variations in terms of expression across different sample sets. Like the geNorm, NormFinder also ranked the HKGs based on their stability values (SV), where lower values represent the higher stability and vice versa (Table 2). Under drought-treated samples β-Tub 4, PP2A-1 and CYC-A emerged as most stable HKGs while under rehydration condition, ARF2 and β-Tub 4 showed the highest stability. Under high and low-temperature stresses, GAPDH and CUL-1 revealed the highest stability respectively while under NaCl stress condition, β-Tub 4 give stable expression. Combined analysis of pooled samples by NormFinder identified the β-Tub 4, WIN-1 and CYC-A genes as the three most stably expressed with stability values of 0.381, 0.441 and 0.491 respectively while UBE2, RP II, and EEF1-α genes exhibited the utmost variation with SV of 1.254, 0.961 and 0.839 respectively. Both the geNorm and NormFinder ranked the β-Tub 4 and CYC-A as the most stable genes respectively in all the pooled samples. However, comparatively the ranking of candidate HKGs generated by the NormFinder was slightly different from those generated by geNorm in most of the samples.

Table 2.

Expression stability (M value) of candidate reference genes calculated by NormFinder under different abiotic stress treatment of Agave sisalana

Rank	Drought		Rehydration		High temperature (± 60 °C)		Cold (± 4 °C)		NaCl (100, 200, 300, 400) mM		Total
	Gene	Stability value	Gene	Stability value	Gene	Stability value	Gene	Stability value	Gene	Stability value	Gene	Stability value
1	β-Tub 4	0.389	ARF2 β-Tub 4	0.236	GAPDH	0.152	CUL-1	0.190	β-Tub 4	0.166	β-Tub 4	0.381
2	PP2A-1 CYC-A	0.464			CYC-A	0.193	WIN-1	0.232	RP II	0.171	WIN-1	0.441
3			GAPDH	0.320	ACT11	0.205	ARF2	0.353	CYC-A	0.188	CYC-A	0.494
4	CUL-1	0.504	WIN-1 ACT11	0.457	WIN-1	0.274	GAPDH	0.630	ARF2	0.282	ACT11	0.585
5	ARF2	0.524			eIF-4A	0.303	CYC-A	0.720	GAPDH	0.294	PP2A-1	0.619
6	RP II	0.599	PP2A-1	0.509	ARF2	0.315	PP2A-1	0.739	WIN-1	0.316	CUL-1	0.666
7	ACT11	0.610	RP II	0.548	CUL-1	0.335	RP II	0.786	ACT11	0.339	ARF2	0.671
8	GAPDH	0.622	EEF1α	0.592	β-Tub 4	0.343	β-Tub 4	0.810	CUL-1	0.343	eIF-4A	0.787
9	eIF-4A	0.687	CYC-A	0.700	EEF1α	0.525	eIF-4A	0.820	UBE2	0.360	GAPDH	0.822
10	WIN-1	0.893	eIF-4A	0.744	PP2A-1	0.819	UBE2	0.845	eIF-4A	0.381	RP II	0.839
11	EEF1α	0.924	CUL-1	0.770	UBE2	1.148	ACT11	0.953	EEF1α	0.485	EEF1α	0.961
12	UBE2	1.825	UBE2	1.802	RP II	1.199	EEF1α	1.433	PP2A-1	0.501	UBE2	1.254

BestKeeper ranking

The BestKeeper is a Microsoft Excel-based statistical tool with an altogether different algorithm than geNorm and NormFinder. It ranked the candidate HKGs in term of stable expression on the following three indicators, (a) the coefficient of covariance (CV), (b) standard deviation (Std), and (c) coefficient of correlation (r). Genes with > 1 Std were considered to have an acceptable range of variation. Lower Std values represent a higher level of stability and vice versa. The stability indicators calculated by the BestKeeper for the candidate HKGs are shown in Table 3. β-Tub 4, EEF1α and ACT 11 were ranked as top three HKGs under drought treatment, while GAPDH was ranked as the top first stable reference gene followed by the PP2A-1 and RP II under rehydration condition. CYC-A, eIF-4A, GAPDH and β-Tub 4 were on the top of the list of ranking under the high-temperature while CUL-1 and RP II were the most stable genes in the cold and NaCl-stressed leaf samples. Under the pooled condition, only three genes (WIN-1, ACT11, and CUL-1) got the Std values less than 1 and ranked on top of the list. Detailed comparison of results generated by the geNorm, NormFinder and BestKeepr respectively (Fig. 3, Table 2 and 3) ranked the candidate HKGs in comparable order under the stress conditions are summarized in Additional File 8.

Table 3.

Expression analysis of 12 candidate reference genes by BestKeeper under abiotic stress condition in Agave sisalana

Rank	Drought			Rehydration			Temperature (± 60 °C)			Cold (± 4 °C)			NaCl (100, 200, 300, 400) mM			Total
	GENE	Std.	CV	GENE	Std.	CV	GENE	Std.	CV	GENE	Std.	CV	GENE	Std.	CV	GENE	Std.	CV
1	β-Tub 4	± 0.25	1.15	GAPDH	± 0.12	0.87	CYC-A	± 0.12	0.49	CUL-1	0.06	0.23	RP II	0.12	0.49	WIN-1	0.79	3.02
2	EEF1α	± 0.32	1.12	PP2A-1	± 0.36	1.47	eIF-4A GAPDH β-Tub 4	± 0.30	1.18	ACT11 WIN-1	0.23	0.84	β-Tub 4 CYC-A	0.19	0.78	ACT11	0.89	3.31
3	ACT11		1.10	RP II ARF2	± 0.42	1.98			1.84			1.19			0.73	CUL-1	0.95	3.45
4	ARF2	± 0.36	1.51			2.28			1.56	CYC-A	0.46	1.73	ARF2	0.22	1.02	PP2A-1	1.13	4.32
5	GAPDH	± 0.37	1.67	β-Tub 4	± 0.54	2.68	CUL-1	± 0.34	1.24	β-Tub 4	0.58	2.30	CUL-1	0.26	0.96	CYC-A	1.54	6.40
6	PP2A-1	± 0.42	2.53	CUL-1	± 0.61	2.39	ACT11	± 0.40	1.57	RP II	0.73	2.71	GAPDH	0.33	1.82	RP II ARF2	1.70	7.09
7	CYC-A	± 0.58	2.17	eIF-4A	± 0.73	3.29	WIN-1	± 0.46	1.72	PP2A-1 UBE2	0.82 0.84	3.0	UBE2	0.39	1.71
8	CUL-1	± 0.72	3.01	ACT11	± 0.78	2.96	EEF1α ARF2	± 0.60	2.51				WIN-1	0.39	1.46	β-Tub 4	1.76	7.76
9	WIN-1	± 0.72	2.51	EEF1α	± 0.95	4.26			3.01	eIF-4A	0.93	3.22	eIF-4A	0.44	1.61	UBE2	1.82	8.03
10	eIF-4A	± 0.72	2.81	CYC-A	± 1.11	5.21	PP2A-1	± 1.16	4.61	ARF2	1.07	4.21	ACT11	0.53	1.98	eIF-4A	2.10	7.96
11	RP II	± 0.79	2.83	WIN-1	± 1.51	6.05	RP II	± 1.55	6.17	GAPDH	1.99	8.25	EEF1α	0.60	1.99	GAPDH	2.38	13.3
12	UBE2	± 1.18	5.08	UBE2	± 2.85	13.8	UBE2	± 1.88	8.11	EEF1α	3.06	9.27	PP2A-1	0.71	2.65	EEF1α	3.45	12.5

Comprehensive assessments by RefFinder

geNorm, NormFinder, and BestKeeper software generated a heterogeneous ranking for the candidate HKGs based on stability values. To address this heterogeneity issue and get a clear picture, RefFinder approach was used to establish a comprehensive ranking of the candidate HKGs (Fig. 4a–f). This comprehensive highest to a least stable ranking by RefFinder was based on the individual signed-rank and geometric mean (GeoMean) of the weight of the respective genes by the used algorithms (Additional File 9). These stability indicators by RefFinder were ranged from 2 to 11.7 for the 12 candidates HKGs in combined analysis across all experimental conditions. β-Tub 4, WIN-1, and CYC-A were the most stable genes while EEF1α, GAPDH, UBE2, and RP II genes were observed as least stable HKGs by RefFinder approach in pooled samples (Fig. 4f). Individually β-Tub 4 HKG has also been ranked as the most stable gene, with lowest geomean ≤ 2 value for drought, rehydration, and NaCl stress condition (Fig. 4 and Additional File 9). CYC-A with GAPDH and CUL 1 with WIN-1 were emerged as best candidate genes for the GOI calibration under high and low-temperature treatments respectively. UBE2 and EEF1 α showed the least reliability under the majority of the treatment conditions.

Fig. 4.

The comprehensive ranking of 12 candidate reference genes based on their expression stability by the RefFinder. The Y-axis indicates the stability values of the genes based on the normalization stability, while x-axis indicates the sequence of the candidate’s reference genes from most stable to least stable in each group. a Drought; b Rehydration; c High temperature; d Low-temperature stress; e NaCl group and f pooled samples

In short, while selecting the best reference gene under altered abiotic stress conditions in A. sisalana, the above mentioned four algorithms yielded the flexible combinations (Fig. 5A a, b). β-Tub4 and CYC-A exhibited relatively stable expression (Fig. 5A-a) while UBE2 and eIF-4A showed the least stability under abiotic stress conditions (Fig. 5A-b). CYC-A should be used carefully for normalization of GOI under rehydration condition as its stability was least. RP II showed stability under drought and NaCl stress condition while contrary results were observed under high-temperature stress condition. Similarly, WIN-1 also behaved oppositely under cold and drought stress condition. A detailed comparison of the top four most and least stable genes ranked by the statistical tools individually and under different abiotic stresses can be observed in Fig. 5A, B.

Fig. 5.

A Venn diagram showing the (a) most stable and (b) least stable candidate reference genes in common up to four as determined by the NormFinder, geNorm, RefFinder, and BestKeeper statistical algorithms. The map data for this Venn diagram was derived from the Additional File 8; section (pooled samples). B Venn diagram of (a) most stable and (b) least stable candidates reference genes in common based on the recommended comprehensive ranking (RefFinder) under abiotic stress condition. The Mapping data were derived from Tables 2, 3 and Fig. 3

Validation of the identified HKGs by target gene normalization

It has been proven that the use of inappropriate HKGs for normalization of GOI can lead to significantly wrong interpretation of expression data under specific stress condition. To validate the usage efficiency of the identified superior HKGs, expression of a target gene ‘AsHSP20’ was normalized in response to drought, heat (high temperature), cold, NaCl and rehydration condition (Fig. 6). Small heat shock proteins (sHSPs) gene is a member of an abiotic stress-inducible gene family, which plays an important and critical role in plant stability under abiotic stress condition especially under high temperature and drought stresses. Based on comprehensive ranking, the most stable β-Tub 4, PP2A-1 and least stable WIN-1, eIF-4A HKGs were selected for normalization of AsHSP20 target gene data under drought stress condition. The relative fold expression of the target gene was found to be upregulated as compared to the control under drought stress at 3 h interval and experienced a slight down-regulation at 6 h when normalized by the β-Tub 4, PP2A-1 stable HKGs. This trend of GOI expression significantly differed when normalization was carried out by the least stable WIN-1, eIF-4A reference genes (p < 0.05 Tukey multiple range tests) (Fig. 6a). The peak expression pattern of AsHSP20 was extremely down-regulated when WIN-1 was used as a reference gene. Similarly, under heat stress CYC-A, GAPDH was used as the most stable reference genes while UBE2, RP II were used as the least stable calibrator for GOI data normalization. The results showed that the expression of AsHSP20 was increased upon the heat induction when stable reference genes were used while it showed significant overestimation under least stable genes’ normalization (Fig. 6b). Under NaCl stress, the expression level of GOI was up-regulated up to twofold after the normalization by β-Tub 4, RP II stable genes which were significantly different from PP2A-1, EEF1α as least stable genes (p < 0.05 Tukey multiple range tests). Uniformity in expression at each point of NaCl stress (100 mM–400 mM) was noted when normalization was performed by β-Tub 4 and RP II HKGs (Fig. 6c). Under Cold stress condition, CUL-1, WIN-1 were identified as the most stable while EEIFα and eIF-4A were the least stable reference genes in A. sisalana. The target genes exhibited uniform expression pattern when normalized with CUL-1 and WIN-1 as reference genes and that showed significant variation when normalized with EEIFα and eIF-4A genes (Fig. 6d). A similar trend was observed when data normalization was done with the least and most stable reference genes under rehydration condition. Relative expression pattern of the target AsHSP20 gene was found to be different when normalization was carried out by the most stable reference genes β-Tub 4, ARF 2 as compared to the least stable reference genes eIF-4A and CUL-1 (Fig. 6e).

Fig. 6.

Normalization of AsHSP20 gene expression for validation of selected reference genes under a Drought, b High Temperature, c NaCl, d, Cold and e Rehydration condition. The Error bar represents the standard deviation among the biological replicates. The solid blue and red lines represent the topmost stable reference while doted green and indigo lines represent the least stable reference genes under abiotic stress treatments respectively. f the copy number of the AsHSP20 genes in the A. sisalana leaves. The copy number of AsHSP20 gene was determined by relative absolute quantification (Additional File 5) in response to abiotic stress condition (colour figure online)

Absolute quantification of AsHSP20 by qRT-PCR for copy number determination

To validate the actual expression of AsHSP20 in leaves of A. sisalana plants under abiotic stress condition, the copy number of AsHSP20 gene was determined by drawing a standard curve (see Materials and Methods) (Additional File 5). High copy number of the AsHSP20 was observed under the heat and drought stress respectively, while least expression was observed under cold and rehydration treatment. These results were in complement to the trends generated by the stable reference genes in particular (Fig. 6a–e). Figure 6f showed the actual copy number of AsHSP20 gene under drought, cold, NaCl, heat and rehydration conditions. This relative absolute quantification of AsHSP20 further proved the accuracy of identified reference genes. Figure 7 represents the list of final validated most and least stable HKGs under respective abiotic stress condition in Agave sisalana.

Fig. 7.

List of recommended top most and least stable HKGs for different abiotic stress conditions. Green leaf mark indicates the HKGs that endorsed for expression data normalization, while red leaf mark represents the HKGs ensure to use them under their respective abiotic stress condition (colour figure online)

Discussion

The qRT-PCR is one of the most recognized technique to accurately monitor the response of GOI towards a specific condition in biological research (Pfaffl 2001). This accuracy is compromised by numerous factors like RNA quality, non-specificity of primers, enzyme efficiencies, housekeeping genes etc. used for normalization and selection of statistical methods for normalization. These factors could lead to misinterpretation of results (Nikalje et al. 2018). Thus, the use of appropriate HKGs for the GOI quantification becomes critical to correct the sample-to-sample variation and errors arising due to said limitation. Commonly used HKGs include Polyubiquitin (UBQ) and tubulin (TUB4), translation elongation factor (TEF), ribosomal RNA (18S rRNA), 25S ribosomal RNA (25S rRNA), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), cyclophilin (CYP2), β-actin (ACT11), eukaryotic initiation factor (EIF4α), protein phosphates 2 (PP2A) etc. (Dheda et al. 2004). These genes should behave like invariant, irrespective to tissue, organ, developmental stage, environmental factors etc. (Zhang et al. 2017). In general, they are primarily associated with primary metabolism and other cellular processes, which are important for plant survival. The harsh environmental condition disrupts plant metabolism and may alter their expression altogether. Many studies have reported the variable behaviors of HKGs under the specific condition that lead to a false interpretation of GOI (Andersen et al. 2004; Nikalje et al. 2018; Zhang et al. 2017). For instance, the normalized expression of stress-responsive factors HSF5 and HSF15 changed altogether when tested in sorghum with Glyceraldehyde-3-phosphate Dehydrogenase (GAPDH), and ubiquitin-conjugating enzyme E2 (UBC1) HKGs, while no relative change in expression was observed when normalization was carried out with stable HKG like eukaryotic initiation factor (EIF4α) and protein phosphates 2 (PP2A) (Reddy et al. 2016).

The same state happened in time-dependent expression study of SsCHS (Chalcone synthase) gene that showed variable expression after normalization with different calibrator genes in Sapium sebiferum (Chen et al. 2017). Thus, it becomes clear that appropriate selection of HKGs become a bottleneck for an accurate estimation of gene expression and inappropriate selection may lead to biased profiling of GOI and false conclusion about their actual biological action (Barbierato et al. 2017). The stability of the used reference genes determines the response of the studied GOI towards different treatments in invitro.

No single reference gene has been reported yet with verified stable expression across samples and tissues in Agave species. High throughput sequencing provides an opportunity for the screening of genes from the non-model organism if whole-genome information is lacking. The current study is the first report for identification of appropriate HKGs in A. sisalana leaf tissue under abiotic stress condition by using the transcriptome database for accurate normalization of GOI (Sarwar et al. 2019). We identified 17 HKGs from the leaf specific transcriptome sequencing data and further accessed their expression towards different abiotic stress conditions by qRT-PCR (Additional File 2). A good range of amplification efficiencies was observed for the primer pairs of 12 candidate HKGs including GAPDH, ARF2, UBE-2, β-Tub 4, CYC-A, RP-II, eIF-4A, EEF1- α, WIN-1, PP2-A, ACT11, and CUL-1. The raw C_T value for these each gene was in the range of 14.5 to 30.7 (Fig. 2; Additional File 6). This variability in C_T range indicates the suitability of genes for further analysis by using the geNorm, NormFinder, and BestKeeper algorithms to verify the stability of candidate genes.

Different software ranked the candidate HKGs inconsistently in variable order because of different approaches of working algorithms. Under drought condition β-Tub 4 was ranked first by NormFinder and BestKeeper while it stood at the second place by the geNorm. CUL-1 gene showed stability under cold stress and was ranked in the first position by all the algorithms. Under heat stress condition, CUL-1, GAPDH and CUL-1 were ranked as the most stable candidate HKGs by geNorm, NormFinder, and BestKeeper respectively, while the similar conflicting trend was observed under rehydration condition and NaCl stress treatment where β-Tub 4, ARF2, GAPDH and ARF2, β-Tub 4, RPII were ranked 1^st respectively (Additional File 8). Overall, the order of least stable HKGs was more-or-less consistent across the abiotic stress treatments. For stability, no definitive decision could be achieved as based on such conflicting ranking by the software’s used for analysis. Therefore, as a final point, RefFinder approach made a comprehensive ranking of candidate HKGs (Nikalje et al. 2018; Xiao et al. 2014). It determined the stability indicators of genes based on respective rankings by the above-mentioned algorithms. Lower value by the RefFinder indicates high stability of the genes while higher value indicates the opposite effect. Hence, final selection for ranking of the candidate HKGs was made as based on the output from the RefFinder (Fig. 4; Additional File 9). Lastly β-Tub 4 was ranked as the most stable HKG under drought, rehydration and NaCl treated samples with the lowest RefFinder value, while CYC-A and CUL-1 showed the lowest value under high and low-temperature stress respectively. No consensus was achieved for the second-best gene in all the samples under stress conditions. This might be due to differential plant response towards different stress conditions triggering a different set of metabolomics and proteomic factors (Nikalje et al. 2018). When we conducted the combined analysis of pooled samples (drought, rehydration, heat, cold and NaCl stress) altogether, again β-Tub 4 emerged as the most stable gene followed by the WIN-1, CYC-A, and ACT11. Under drought stress condition, β-Tub 4 and PP2A-1 genes achieved a stable expression. Under rehydration, β-Tub 4 and ARF2 performed best as the most stable reference genes. For heat stress, CYC-A and GAPDH were found to be the optimal HKGs, while CUL-1 and WIN-1 showed the highest stability under cold stress treatment. Under NaCl stress, a consensus was achieved for the β-Tub 4 along with RPII as reference genes.

Several recent studies have reported that the target gene expression would be more appropriate, steady and precise when normalization was carried out by two or more internal reference genes under a specific experimental setup (Expósito-Rodríguez et al. 2008; Gutierrez et al. 2008; Karuppaiya et al. 2017; Reid et al. 2006). Therefore, we also figured out the maximum number of the required reference genes for optimal normalization of GOI by geNorm software. This applet initially calculates the successive normalization factor NF (NF_n and NF_n + 1) for two highly stable reference genes and then moves onwards by adding one by one in an order to determine the pairwise variation (V_n/V_n + 1) (Kumar et al. 2013). Usually, a threshold of 0.15 value (V_n/V_n+1) is applied to calculate the best combinations of reference genes and then no additional reference genes (V_n + 1) would be required for GOI normalization (Vandesompele et al. 2002). In this study, all pairwise variation was below the 0.15 limit (V_n/V_n+1) which suggests that only two top best identified internal reference genes could be appropriate for GOI normalization in A. sisalana under abiotic stress condition. Few reports regarding the cut-off value also pointed that the 0.15 cutoff value should not be considered as a rigorous standard but rather as an ideal value, and could be found higher (V_n/V_n+1) for optimal normalization of data (Chen et al. 2015; Karuppaiya et al. 2017; Wan et al. 2010; Yang et al. 2015). Under the combined analysis of stressed samples, all the pairwise variation was out of the default limit of 0.15.

The reliability of identified most and least stable HKGs was validated by studying the relative profiling of stress-responsive AsHSP20 as target gene under abiotic stress conditions. Small heat shock proteins belong to an abiotic stress-inducible family, which play important and extensive roles to maintain plant stability under abiotic stress conditions especially under drought and high temperature. The relative expression of AsHSP20 exhibited a significant uniformity in expression pattern when single or combination of stable genes (β-Tub 4, PP2A-1); (β-Tub 4, ARF2); (CYC-A, GAPDH); (CUL-1, WIN-1) and (β-Tub 4, RP II) were used, while significant disparities in expression pattern revealed when least stable reference genes/combinations (WIN-1, eIF-4A); (CUL-1, eIF-4A); (UBE2, RP II); (EEF1α, eIF-4A) and (PP2A-1, EEF1α) were used for normalization of qPCR generated data under abiotic stress conditions. These findings signify the importance of stable HKGs selection for accurate normalization. Additionally, absolute quantification of AsHSP20 was also carried out to determine the copy number of GOI under the respective stress condition. Highest copy number of the AsHSP20 were observed under the heat, drought and NaCl treatments respectively, while the lowest copy number were detected under the rehydration and cold treatment respectively. Hence, absolute quantification of AsHSP20 gene confirms the reliability of stable reference genes used to normalize the target gene expression under abiotic stress conditions. To the best of our knowledge, this is the first report on the identification and validation of the suitable candidate’s HKGs for accurate qRT-PCR data normalization in A. sisalana under abiotic stress conditions.

Conclusion

We identified and validated the stability of seventeen HKGs in A. sisalana under abiotic stress conditions. Their comprehensive stability was determined by using different algorithms. We identified the β-Tub 4, PP2A-1, RP II as reference genes under drought stress; β-Tub 4, ARF2, and GAPDH under rehydration condition. CYC-A, GAPDH, and CUL-1 were found to be the most stable genes under heat stress; CUL-1, WIN-1, and ACT11under cold stress; while β-Tub 4, RP II and ARF2 were identified as the top three stable genes under NaCl stress condition. Keeping in view this disparity among the abiotic stresses, the selection of reference genes and their usage should be based on the stress being investigated. Furthermore, the relative quantification and copy number determination of AsHSP20 target gene signify the importance of the stable internal gene(s) for normalization of the qRT-PCR data under specific conditions. The use of stable reference genes would be useful to improve the precision of qRT-PCR analysis in A. sisalana and related bioenergy crops for functional genomic studies.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

ACT

Actin

CT

Thershold cycle

CUL-1

Cullin

CV

Coefficient of covariance

CYP

Cyclophilin

DEG

Differentially expressed genes

DAS

Days after sowing

E

Efficiency

EIF

Eukaryotic initiation factor

eIF-4A

Eukaryotic initiation factor-4A

FPKM

Fragments per kilobase of transcript per million mapped reads

GAPDH

Glyceraldehyde-3-phosphate dehydrogenase

GOI

Gene of interest

HKGs

Housekeeping genes

MIQE

Minimum information for publication of quantitative real-time PCR experiments

NF

Normalization factor

NTC

No template control

PP2A

Protein phosphates 2

r

Coefficient of correlation

R²

Regression coefficient

RCA

Rubisco activase

RPII

RNA polymerase II

rRNA

Ribosomal RNA

Std

Standard deviation

sq/qRT-PCR

Semi quantitative/quantitative real time polymerase chain reaction

SV

Stability values

TEF

Translation elongation factor

TSA

Transcriptome Shotgun assembly

TUB

Tubulin

UBQ

Polyubiquitin

Authors contribution

MBS and BR conceived and designed the project. AB, MA and MS conducted the lab experiments. MBS, ZA and SH performed the statistical analysis and conducted the validation studies. MBS drafted the manuscript. TH and BR read and edited the final manuscript. All authors read and approved the final manuscript.

Availability of data and materials

All data generated or analyzed during this study are included in this published article and its additional files. The sequence data of this study are available in the NCBI Transcriptome Shotgun Assembly (TSA) under Accession Number GGRE00000000.1 of BioProject PRJNA359581 https://www.ncbi.nlm.nih.gov/sra/PRJNA359581.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.

Ethics approval and consent to participate

Not applicable. Human and animal samples were not used and no field permissions were necessary to collect the plant samples for this study. The authors declare that experimental research works on the plants described in this paper comply with institutional, national and international guidelines.

Consent for publication

Not applicable.