Selection and validation of reference genes for RT-qPCR analysis of different organs at various development stages in Caragana intermedia

Abstract Reverse transcription quantitative PCR (RT-qPCR) is a technique widely used to investigate the expression of genes. An appropriate reference gene (RG) is essential for RT-qPCR analysis to obtain accurate and reliable results. Caragana intermedia plays an important role in afforestation as a bush. However, due to the lack of appropriate RGs, the research on development-related genes is limited. In this study, the selection for suitable RGs of different organs at various development stages to normalize the results of RT-qPCR about development-related genes was performed. To test the expression stability across all samples, we used the software algorithms such as geNorm, NormFinder, BestKeeper, and RefFinder to evaluate all the candidate RGs. Our results showed that CiEF1α was the most stable RG with little fluctuation among all samples. In addition, CiGAPDH in roots, CiSKIP1 in stems and leaves, and CiEF1α in different organs were selected as the most stable RGs. To confirm the applicability of the most stable RGs, the relative expression of CiWRKY17 was normalized using different candidate RGs. Taken together, our research laid a foundation for the study of development-related genes in C. intermedia.


Introduction
Caragana intermedia, commonly known as a bush, belongs to the legume family and is widely distributed in north and northwest China along with semi-fixed or fixed sand dunes, barren land, and loess hills. C. intermedia has high ecological value via playing an important role in afforestation [1]. To make the bush perform its role, it is essential to study the growth and development of C. intermedia to achieve the goal of cultivating healthy seedlings quickly and transferring them efficiently for afforestation.
Today, with the wide recognition of the important role of growth and development of plants, there have been many studies about development [2,3], and the research on development-relevant genes in C. intermedia has also made some progress [4]. In order to better uncover the function of these genes, it is pivotal to analyze their spatio-temporal expression [5]. Reverse transcription quantitative PCR (RT-qPCR) was often used to study the expression of genes because of its high throughput, specificity, and sensitivity [6][7][8], but its accuracy is difficult to guarantee due to changes in mRNA quantity and quality, and other reasons. Using relatively stable reference genes (RGs) can be a good solution to this problem. Therefore, it is necessary to select appropriate RGs.
Under ideal conditions, the RGs should be relatively stable, but some reports indicated that the expression of these so-called RGs could fluctuate in different organs and at various stages in plants [9][10][11][12]. For example, the commonly used RGs, such as Actin (ACT) [13], Beta-tubulin (TUB) [14], Elongation factor 1-α (EF1α) [10,15], and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were often chosen for normalization of RT-qPCR in plants, but EF1α showed no expression stability in the root of Prunus spp. [16], and TUB was one of the least stable genes in the cells elicited with MeJA of Cichorium intybus [17]. So it is quite necessary to screen RGs for their suitability for different experimental designs. Owing to no perfect method to evaluate the stability of RGs, four different analytical software, such as geNorm [18], NormFinder [19], BestKeeper [20], and RefFinder [23], were usually used to identify the suitable RGs.
Previous studies showed that the RGs of C. intermedia [21] had been screened under various abiotic stresses. However, these RGs were not screened in different organs or at various development stages, so this study aims to select relatively stable RGs for the reference of the development-related genes of C. intermedia. First, we selected 11 commonly used RGs, such as CiACT, CiCAP (Cyclase-associated protein), CiEF1a, CiGAPDH, CiSKIP1/SKIP5-1/SKIP5-2 (F-box proteins), CiTUA (Alpha-tubulin), CiTUB/TUB3 (Betatubulins), and CiUBQ (Ubiquitin), which were proved to be relatively stable in its homologous species, C. korshinskii [22], and the expression level of these RGs were analyzed among samples from different organs or at various development stages by RT-qPCR. Next, all these selected RGs were ranked using geNorm, NormFinder, BestKeeper, and RefFinder software. Finally, CiWRKY17, which had been proved to express in different organs, might be involved in the growth and development of C. intermedia [4], was selected to validate these RGs. The above results will provide the most appropriate RGs for further study of development-related genes in C. intermedia.

Collection of plant materials
The samples were collected in the field located in Liangcheng County, Ulanchabu City, Inner Mongolia, China, with north latitude 41°23′ and an east longitude 111°41′. Totally 69 samples from seven different stages of development (Figure 1), and seven organs, including root, stem, leaf, bud, flower, young pods, and young seeds, were collected as indicated in Figure 1 and Table 1. Each sample was taken from an even mixture of the same tissues from three different plants, and three biological replicates were performed. All samples were immediately frozen in liquid nitrogen for RNA extraction in the follow-up steps.

RNA isolation and cDNA synthesis
RNA extraction and reverse transcription were performed using TaKaRa MiniBEST Plant RNA Extraction Kit and TaKaRa PrimeScriptTMRT reagent Kit with gDNA Eraser, respectively. RNA was measured using an ultraviolet spectrophotometer (Model DU800), and the ratio of A260 nm/ A280 nm and A260 nm/A230 nm was calculated to check the purity and the concentration (μg/mL) of the extracted RNA. RNA integrity was assessed using electrophoresis of the extracted RNA on a 1.0% (w/v) agarose gel. Any RNA sample with an A260/A280 ratio between 1.8 and 2.2 and an A260/A230 ratio greater than 2.0 was used for subsequent experiments. One microgram of total RNA from each sample was used to synthesize the first-strand cDNAs using the above-mentioned reverse transcription Kit according to the manufacturer's instructions. The cDNA acquired by reverse transcription was then diluted to 16-fold and used as the template for RT-qPCR.

Primer and RT-qPCR
The sequences for the primers of the selected 11 RGs were obtained from our previously published article [22] and are listed in Table S1. The primers' sequence, amplification efficiency, the regression coefficient, R 2 value, and the melting curve are also listed in Table S1 or in Figure S1. The cDNA was amplified using TB Green qPCR Master Mix (TaKaRa) with a Roche LightCycler 480 system. The thermal cycling program was 95°C for 60 s, followed by 40 cycles of 95°C for 5 s, 60°C for 30 s, and 72°C for 15 s. Each RT-qPCR reaction was performed with three technical replicates. The Ct values, which took the mean value of three technical replicates, were pooled for the RG evaluation.

Data analysis
Stability analysis of the RGs was assessed using the Excel-based geNorm [18], NormFinder [19], BestKeeper [20], and RefFinder [23], which were the most widely used software to screen RG by RT-qPCR. For geNorm and NormFinder algorithms, the raw Ct values from each sample were converted into relative quantity (RQ) values using the formula 2 −ΔCt (ΔCt = each Ct value − the minimum Ct value) [24]. The geNorm program first calculated an expression stability M-value for each gene, and M-values below 1.5 were supposed to be stably expressed, and a lower M-value indicated a more stable expression [18]. Moreover, the value of "n" was the optimal number of RGs when the pairwise value of variation (V n /V n+1 ) was below a cutoff value of 0.15 [18]. NormFinder was used to assess the stability of RGs based on the ANOVA model [19].
For BestKeeper and RefFinder programs, the raw Ct values were directly analyzed. BestKeeper examined the ranking of RGs based on the calculation of the variance and the standard deviation (SD) for each gene. Any gene with an SD value less than 1.0 was recommended as a gene with stable expression [20]. RefFinder, a user-friendly and webbased comprehensive tool, was developed for selecting RGs, and integrated the currently available computational programs to compare the rank of the tested RGs [23].

Validation of RGs
CiWRKY17 was selected to verify the most stable RG, the most stable RG combination, and the worst RG by RT-qPCR. Each qPCR reaction was performed with three technical replicates. The experiment was repeated three times (with three biological replicates), and the results were consistent. Finally, the data were calculated using the 2 −ΔΔCt method.

Expression profiles of the RGs
The expression level of the 11 RGs was analyzed across all samples via determination of the Ct value, and lower the Ct value the higher the gene expression level was. The Ct value of these genes ranged from 18 to 34 in all studied samples, with most of them between 20 and 26, and the smaller the range of the Ct value the more stable the RG was ( Figure 2).
To evaluate the relative stability value of RGs, we took the average Ct value and Ct value range. On the one hand, CiACT (with an average Ct of 21.15) showed the highest expression level, whereas CiSKIP5-1 (with an average Ct of 24.9), UBQ (with an average Ct of 24.9), and CiCAP (with an average Ct of 25.2) showed relatively low expression level ( Figure 2). On the other hand, the six genes with the minimum Ct value range were CiSKIP1 (4.61), CiGAPDH (5.67), CiSKIP5-2 (5.83), CiEF1α (6.14), CiTUA (8.12), and CiSKIP5-1 (8.57) ( Figure 2). In brief, based on both the higher expression level and the lower Ct value range, CiSKIP1, CiGAPDH, CiSKIP5-2, and CiEF1α were more appropriate as RGs.

Stability analysis by geNorm
According to geNorm analysis, the M-values of all tested RGs were below 1.5, indicating that they were relatively stable, and the lowest M-value indicated the highest stability [18]. Among all of the tested samples (Figure 3e), CiSKIP1 was found to be the stable RGs successively, while CiSKIP5-2 and CiSKIP5-1 represented the most stable RG combination. On the contrary, CiTUB3 was found to be the most inappropriate RG based on their fluctuating expression levels. In addition, these screening results of RGs varied in different organ samples (Figure 3a-d).
The optimal number of RGs for normalizing RT-qPCR data was determined by calculating the pairwise variations V n /V n+1 in the geNorm program, and when the value of V n /V n+1 is less than 0.15, the n-value is the optimal number of RGs [18]. As shown in Figure 3f, since the values of V2/3 and V3/4 were greater than 0.15 from left to right, and the value of V4/5 was below the cutoff value of 0.15, these four RGs were ideal for normalizing RT-qPCR data in all samples.

Stability analysis by NormFinder
The NormFinder was similar to geNorm in algorithm to calculate different M-values of RGs. The difference was that NormFinder selected only one optimal gene, while geNorm selected two or more genes. The results calculated with NormFinder ( Figure 4) showed that CiEF1α was the most stable gene in all the tested samples, while CiTUB3 was considered to be a weakly stable gene, and there were some differences in different organ samples.

Stability analysis by BestKeeper
BestKeeper examined the ranking of RGs based on the calculation of the SD value for each gene. Any gene with an SD value less than 1.0 was recommended as a gene with stable expression, and the lowest SD value was the highest stability of a gene. According to the above principles, the result showed CiSKIP1 was the most stable gene, and CiACT was the least stable gene in all tested samples, and the results differed in different organ samples (Figure 5a-c).

Comprehensive stability analysis of RGs by RefFinder
In order to comprehensively evaluate the stability of RGs, we made a comprehensive ranking by RefFinder to get the most stable RGs. As shown in Figure 6, CiEF1α was the most stable RGs with little fluctuation; moreover, CiACT was the least stable RGs due to large expression of fluctuations in all samples. Similarly, CiGAPDH in roots, CiSKIP1 in stems and leaves, and CiEF1α in different organs were found to be the most stable RGs; CiTUA in roots and CiTUB3 in stems, leaves, and different organs were not suitable to be used as RGs.

Verification of the selected RGs
In order to validate the identified RGs and to demonstrate the application of the stable RGs selected by geNorm, NormFinder, BestKeeper, and RefFinder under the studied conditions, the transcript profile of CiWRKY17, which had been confirmed to express in leaf, root, and stem previously [4], was assayed again using the RGs (including the least and most stable RGs and their combinations). As shown in Figure 7, using the 2 −ΔΔCt method, the expression level of CiWRKY17 was similar when we used the most stable RG and RG combination to normalize the RT-qPCR data. However, the results varied using the least stable RG.

Discussion
The powerful technique RT-qPCR has been widely used for the detection and quantification of gene expression in plants. In order to interpret RT-qPCR data accurately and reliably, appropriate RGs are essential. Reports on several plant species, such as Undaria pinnatifida [25], Raphanus sativus L. [26], Lycoris aurea [27], Lagerstroemia indica and L. speciosa [28], Eucommia ulmoides Oliver [29], Pyrus L. [30], and Davidia involucrata Baill. [31], had shown the importance of validating appropriate RGs for normalizing RT-qPCR data. In addition, some studies have shown that the best RG was different for different samples of organs or experimental conditions [26,29,32]. RGs of C. intermedia [21] had been screened under various abiotic stresses but had not been screened in different organs collected from different developmental stages, which led to difficulties in the normalization of the development-related gene. Therefore, it is particularly important to screen RGs in different organs at various development stages. Current studies showed no perfect analysis software to evaluate the stability of RGs. In order to obtain the stable RG, we selected four kinds of analysis software to evaluate. Among them, geNorm [18], NormFinder [19], and BestKeeper [20] obtained the best RGs, respectively: CiEF1α, CiGAPDH, and CiSKIP1 in roots; CiUBQ, CiSKIP1, and CiGAPDH in stems; CiSKIP5-2, CiEF1α, and CiCAP in leaves; CiUBQ, CiTUA, and CiSKIP1 in different organs; and CiSKIP1, CiEF1α, and CiSKIP1 in all samples. The above results indicated that different analysis software indeed screened different RGs, which might be caused by the different algorithms of this software, and this might also be the case in other reports [29,33,34]. Since different software came to a diverse conclusions, in order to obtain the best RG, we chose RefFinder [22] to integrate these conclusions and finally concluded that the best RG was CiGAPDH in roots, CiSKIP1 in stems, CiSKIP1 in leaves, CiTUB in different organs, and CiEF1α in all samples. In addition, geNorm could obtain the most suitable number of RGs according to V n /V n+1 [18]. For example, V4/V5 was less than 0.15 in all samples, indicating that four RGs were the most suitable. However, considering that multiple RGs would make the experiment more complicated and timeconsuming, 1-2 RG was suitable to normalize the target genes for convenient operation. Therefore, according to the analysis results of geNorm, the optimal combination of RGs was obtained: SKIP5-1/SKIP5-2 in roots, SKIP5-1/ SKIP5-2 in stems, SKIP5-1/SKIP1 in leaves, TUA/EF1α in different organs, and TUB/EF1α in all samples.
To sum up, we found that the optimal RGs varied in different organ samples, and the optimal combination of RGs was different except for root and stem. In addition, previous studies on the selection of plant RGs mainly focused on different hormones [11,35], stress treatments [29,36,37], and different tissues/organs [9,12,38], but there were no studies on the selection of plant RGs in different organs collected from various development stages. This study aimed to screen the RGs of different organs at different developmental periods. We also used the most and least stable RGs, and their combination to normalize the transcript of a known gene, CiWRKY17, to evaluate the practicality of the selected RGs, and the results indicated that the selected RGs were reliable.

Conclusion
In conclusion, the 11 RGs were systematically selected and evaluated using RT-qPCR in seven organ types and at seven different developmental stages, using geNorm, NormFinder, BestKeeper, and RefFinder software. Then the selected RGs were further validated by analysis of the CiWRKY17 expression in different organs. And the best RGs and the best combination of RGs were obtained. This study will improve the accuracy of the RT-qPCR results and lay the foundation for future studies on development-related genes of C. intermedia.    (d) different organs (buds collected from the stage of S3, flowers collected from S4, pods collected from S5, and roots, stems, leaves, and seeds collected from S6); (a-d) green, blue, and red represent the most stable RG, the most stable RG combination, and the worst RG, respectively; (a-d) represent samples containing three biological replicates (data were presented as means ± SE of three independent biological replicates; * p < 0.05, * * p < 0.01, and * * * p < 0.001, by Student's t-test).