Identification of the most suitable reference gene for gene expression studies with development and abiotic stress response in Bromus sterilis

Bromus sterilis is an annual weedy grass, causing high yield losses in winter cereals. Frequent use of herbicides had led to the evolution of herbicide resistance in this species. Mechanisms underlying herbicide resistance in B. sterilis must be uncovered because this problem is becoming a global threat. qRT-PCR and the next-generation sequencing technologies can elucidate the resistance mechanisms. Although qRT-PCR can calculate precise fold changes, its preciseness depends on the expression of reference genes. Regardless of stable expression in any given condition, no gene can act as a universal reference gene. Hence, it is necessary to identify the suitable reference gene for each species. To our knowledge, there are no reports on the suitable reference gene in any brome species so far. Thus, in this paper, the stability of eight genes was evaluated using qRT-PCR experiments followed by expression stability ranking via five most commonly used software for reference gene selection. Our findings suggest using a combination of 18S rRNA and ACCase to normalise the qRT-PCR data in B. sterilis. Besides, reference genes are also recommended for different experimental conditions. The present study outcomes will facilitate future molecular work in B. sterilis and other related grass species.

www.nature.com/scientificreports/ of these methods is their inability to detect extremely low expression, resulting in replacing the pre-existing methods with microarrays and qRT-PCR 9,11 . Even though these modern techniques are highly sensitive and can calculate precise fold changes, their preciseness is highly dependent on the expression of a reference gene. Ideally, a reference gene refers to constitutive genes required to maintain the basic cellular functions of an organism. These genes are known to have stable gene expression in all cells under both normal and pathophysiological conditions [12][13][14] . However, the steps of qRT-PCR are reclined to variations; therefore, to overcome these variations, target gene transcription levels must be normalised to reference genes transcription levels. Any error in selecting a suitable reference gene may lead to misleading results. Hence, selecting a reliable reference gene is necessary for molecular biology-oriented studies 9, [14][15][16][17] . The most commonly used references genes for normalisation of plant gene expression studies are ubiquitin (UBQ), β-tubulin (β-TUB), ribosomal RNA genes (18S rRNA and 25S rRNA), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), eukaryotic elongation factor (eEF), eukaryotic translation initiation factor 1 (eIF1), actin (ACT ), acetyl-CoA carboxylase (ACCase) etc 9,18 . Although these genes are known to have a stable expression in any given condition, several studies documented variability in their expression level between species of plants or different stress conditions or developmental stages [19][20][21] . As no gene can act as a universal reference, it is necessary to systematically select and identify the suitable reference gene for each species 22,23 .
There are no reports of a suitable reference gene in B. sterilis or any other brome species. Our study aims to identify a suitable reference gene for gene expression studies in B. sterilis (or barren brome). Increasing the number of treatments might lead to more variations in results, which decreases the chance of identifying a suitable reference gene 9 . In this study, we had selected eight common candidate reference genes (UBQ, ACT , GAPDH, 18S rRNA, 25S rRNA, ACCase, β-TUB and eEF) identified in B. sterilis and evaluated the stability of their gene expression in three developmental stages (two-leaves, three-leaves and four-leaves), two different plant organs (shoots and leaves) and one abiotic stress (drought stress). Among the various severe issues with detrimental effects, climate change has remained a top priority. Global warming has resulted in an increase of air temperature and evapotranspiration, leading to agricultural droughts, affecting both crops and weeds.
Low soil moisture increase the competition for water and nutrients between weeds and the crop, thus making weed management complicated. Some (usually C4) weed species gain profit from this situation. Uptake and translocation of herbicides within the plant is reduced, thereby affecting the efficacy of the applied herbicides. Hence, interest for studies under drought is recently rising 24 , which might require expression studies with several genes of interest. Therefore, drought stress has been included in the present study, and our recommended reference genes will be helpful in future drought-related studies.
The most suitable candidate was selected based on the ranking provided by different widely used statistical software for reference gene analysis (comparative ΔCt, BestKeeper, NormFinder, geNorm and RefFinder). Additionally, the most suitable reference gene was used to validate a herbicide-stress experiment. Thus, our study provides a basis for identifying the suitable reference gene for future gene expression studies in B. sterilis and will aid in impending studies on the molecular basis underlying the herbicide resistance in barren brome.

Results
Primer efficiency and candidate genes expression. 1.2% agarose gel electrophoresis was used to check the integrity of the RNA. In addition, the quantity and quality of RNA were evaluated by a nanodrop spectrophotometer (Thermo Scientific™, US). The A260/A280 values ranged from 1.90 to 2.05. These samples were further used to synthesise cDNA, which was used for the qRT-PCR experiments. In all the qRT-PCR amplification, a single peak was obtained (supplementary Fig. 1). The selected primers for this study showed a single band in the 1.5% agarose gel (supplementary Fig. 2) and had efficiency values ranged between 92.32 and 106.79%, which falls under the acceptable range. The correlation coefficient values ranged from 0.980 to 0.999 ( Table 1). The expression profile of the 8 candidate genes under different experimental conditions is shown in the Fig. 1. 18s rRNA showed the lowest cycle threshold value (Ct), indicating high expression of the gene, whereas ACT showed the highest Ct value indicating low expression.
Gene expression stability analysis. Developmental stages-related experiments. 18S rRNA was identified as the stable reference gene by comparative ΔCt and RefFinder. BestKeeper software identified ACCase as the most stable reference gene ( Table 2). NormFinder analysis revealed 18S rRNA and eEF as the most stable genes, whereas geNorm analysis ranked 18S rRNA and ACCase as the best reference gene for developmental stages-related experiments in B. sterilis (Table 2, Fig. 2). Except, comparative ΔCt, all the used software identified GAPDH as the least stable gene. According to the comparative ΔCt analysis, eEF is the least stable gene.
Plant organs related studies. In gene expression studies with the plant organs, 18S rRNA has been ranked as the most stable gene by comparative ΔCt, BestKeeper and RefFinder (Table 2). NormFinder analysis identified ACCase and eEF as the most stable genes ( Table 2). Based on the geNorm analysis, 18S rRNA and ACCase might be the best reference gene for plant organs-related studies in B. sterilis (Fig. 2). β-TUB was identified as the least stable gene by comparative ΔCt, BestKeeper and RefFinder, whereas NormFinder and geNorm analysis identified GAPDH as the least stable gene.

Under drought stress.
For studies under drought stress, comparative ΔCt and RefFinder identified 18S rRNA as the most suitable reference gene ( Table 2). BestKeeper software identified ACCase as the most stable reference gene ( Table 2). NormFinder analysis revealed GAPDH and 18S rRNA as the most stable genes, whereas geNorm analysis ranked 18S rRNA and β-TUB as the best reference gene ( Combined conditions. When all the conditions were taken together, 18S rRNA was identified as the most stable reference gene, irrespective of the method ( Table 2, Fig. 2). NormFinder analysis results for combined conditions revealed that UBQ and ACCase could be considered the best reference gene, and the geNorm algorithm ranked 18S rRNA and ACCase, as the best reference gene ( Table 2, Fig. 2). β-TUB was identified as the least stable gene by comparative ΔCt, BestKeeper and RefFinder, whereas NormFinder and geNorm analysis identified GAPDH as the least stable gene.
Pairwise variation analysis. The pairwise variation (Vn/Vn + 1) was calculated based on the geNorm algorithm. The optimal number of the reference genes required for the normalisation were determined from the pairwise variation results, based on the average expression stability (M) values (cutoff: M < 1.5). The optimal number of the reference genes required for the normalisation for experiments related to the developmental stages and plant organs are 1 and 2, respectively. However, to avoid any biases in the normalization, we recommend using 2 reference genes for developmental stages. Hence, we recommend using 18S rRNA and ACCase, as housekeeping genes for developmental stages and plant organ-related studies in B. sterilis. Under drought stress, 4 candidate genes (18S rRNA, β-TUB, 25S rRNA and ACCase) were considered suitable for normalisation. When all the conditions were considered together, the pairwise variation result suggested that 2 reference genes will be required for the normalisation (Fig. 3). Therefore, 18S rRNA and ACCase were identified as the most suitable gene when all the conditions were considered together.

Relative expression of the acetolactate synthase (ALS) gene under herbicide stress.
Based on the analysis of the commonly used software for reference gene analysis, 18S rRNA and a combination of 18S rRNA and ACCase were identified as the most suitable candidate genes for gene expression studies in Bromus sterilis, whereas β-TUB as the most unstable gene. To validate the reliability of the candidate genes, relative expression of the acetolactate synthase under herbicide stress was evaluated using the best and the least stable candidate genes. When normalised with 18S rRNA and a combination of 18S rRNA and ACCase, B. sterilis biotype showed twofold ALS gene overexpression after herbicide treatment compared to the control, whereas with β-TUB, the result is almost eight times (Fig. 4).

Discussion
Recent reports from the United Kingdom and the Czech Republic on B. sterilis, developing resistance against commonly used herbicides, indicate that if they remained uncontrolled, these species might become a concern worldwide 25,26 . Herbicide resistance mechanisms can be target-site based (TSR) and/or non-target site-based (NTSR). Target-site based mechanisms involve nucleotide polymorphisms 27 , gene amplification 6 and gene overexpression 28 , whereas increased detoxification by enhanced metabolism 29,30 and/or reduced herbicide uptake and translocation 29 fall under non-target site-based herbicide resistance. Irrespective of the mechanism/s of resistance, qRT-PCR and the next-generation sequencing technologies have been used recently as a common technique to investigate the resistance mechanism in different weed species 5 . qRT-PCR experiments require an appropriate reference gene to normalise the target transcript levels. Any misapprehension in selecting a stable reference gene might lead to ambiguous results. Hence, the selection of a reliable reference gene is obligatory. Even though suitable candidate genes under different experimental condition were identified in many weedy species, like Alopecurus sp. 20 , Eleusine sp. 8 , Avena sp. 31,32,33 , Descurainia sp. 34 etc., but to date, there are no reports on the systematic selection of stable reference genes under any conditions for barren brome or any other related brome species. This study used qRT-PCR to evaluate the expression stability of eight candidate reference genes in barren brome under different experimental conditions. The most stable reference genes for each experimental condition were identified exclusively. geNorm software identified the ideal pair of genes with the minor variation in their expression ratios for each experimental condition. For studies related to life stages, geNorm identified that combining two reference genes would be suitable for normalising the qRT-PCR based gene expression values. 18S rRNA and ACCase was chosen as the best reference gene for the studies with life stages of B. sterilis. For studies related to plant organs and under drought stress, pairwise variation analysis recommended using two and four genes, respectively. 18S rRNA and ACCase were chosen as the most suitable candidates for plant organs-related studies, whereas, for studies under drought stress, we recommend using 18S rRNA, β-TUB, 25S rRNA and ACCase. When all the conditions were considered together, 18S rRNA and ACCase were identified as the most suitable gene. Validation under herbicide stress indicated that both 18S rRNA and the combination of 18S rRNA and ACCase could be suitable. 18S rRNA, a component of the 40S ribosomal small subunit in eukaryotes, has been recognised to have a steady expression in grasses under different stresses in earlier studies 35,36 . 18S rRNA is a primary constituent of all eukaryotic cells. Hence, 18S rRNA is known to have extremely high expression in most cell types, so it can be challenging to use it as an endogenous normaliser gene. Moreover, synchronized www.nature.com/scientificreports/ use of multiple reference genes will also decrease the chance of biased normalisations. Finally, from our study results, 18S rRNA and ACCase appeared to be the most suitable reference genes to normalise the qRT-PCR data in B. sterilis. Rapid advances in molecular biology techniques in plant biotechnology have increased the demand for identification of reference genes, which will be more stable than the traditional reference genes. The reference genes identified and validated in our study will assist the studies related to the elucidation of abiotic stress and its regulatory mechanisms. Comparative RNA-seq transcriptome analysis between the control and experimental plants www.nature.com/scientificreports/ can be regarded as the most straightforward way to identify the genes involved in abiotic stresses like herbicide stress 20 . Recent studies on herbicide resistance mechanisms of B. sterilis suggest that both TSR and NTSR can be linked with the herbicide resistance in these species 25,26 . Nevertheless, detailed follow-up studies are essential to delineate further the regulatory mechanisms underlying the observed herbicide resistance mechanism 24 . However, among the herbicide resistance mechanisms, NTSR mechanisms are considered more complex to elucidate than the TSRs 20 . Comparative RNA-seq studies between the herbicide-resistant and susceptible plants will facilitate unravelling plausible resistance mechanisms in barren brome. Nevertheless, the RNA-seq data  www.nature.com/scientificreports/ should be further cross-checked via qPCR, whose reliability depends on selecting the reference genes. This is the first study to evaluate and validate experiment-condition specific reference genes in brome species to the best of our knowledge. We had identified and validated internal reference gene suitable for normalising qRT-PCR experiments. Thus, our reference genes can be used during any RNA-seq based transcriptome or gene expression studies on B. sterilis. Our findings provide a basis for future molecular work on B. sterilis and can also be used during gene expression studies in other related species after preliminary validation.  8 . Besides, comparative ΔCt 37 and RefFinder 8 were also used. Before and after herbicide treatment, the relative ALS gene expression was calculated using the 2 −ΔΔCt method 38,39 . NormFinder software estimates the intra-and intergroup variation. These variations are then combined into a stability value. The gene with minimal variation is ranked as the best by the software. geNorm program estimates an expression stability value (M) for each gene. Genes with the lowest M values have the most stable expression. BestKeeper ranks the candidate genes based on standard deviation values of cycle threshold (Ct) or crossing point values (CP) and coefficient of correlation (r) values. A gene with a standard deviation value below 1 and a coefficient of correlation value close to 1 is considered to have more stable gene expression than others. RefFinder integrates the available well-known programs for reference gene screening (geNorm, NormFinder, BestKeeper, and the comparative Delta-Ct method) and calculates the geometric mean of ranking values to give the overall ranking. The genes with a minimal geometric mean of ranking values are categorized as the best 37 .

Methods
Ethical approval. No permissions were necessary to collect plant samples. All experiments conducted in this study, including the collection of plant material, are in compliance with relevant institutional, national, and international guidelines and legislation.
Ethics statement. This article does not contain any studies with human or animal subjects. www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.