Selection of reference genes for tissue/organ samples of adults of Eucryptorrhynchus scrobiculatus

Eucryptorrhynchus scrobiculatus is an important wood-boring pest of Ailanthus altissima in China, where it causes a large number of these trees to weaken or even die. To identify genes related to economic traits or specific cellular processes in E. scrobiculatus, gene expression in multiple tissue/organ samples is commonly surveyed, and reference genes are required in this process as a control for normalization. In the present study, 18 candidate reference genes from E. scrobiculatus were identified, and the expression levels of these reference genes were estimated through quantitative real-time PCR. Differences in expression levels were analyzed with four algorithms (geNorm, NormFinder, BestKeeper and delta Ct) and comprehensively with RefFinder. With the most stable levels of expression in different tissues, RPL13, RPS3 and RPL36 were determined to be suitable for use as candidate reference genes. Moreover, the expression profile of one target gene (glycoside hydrolase family 45, GH45) confirmed the reliability of the selected candidate reference genes. This study provides the first set of suitable candidate reference genes for gene expression studies in E. scrobiculatus, and the findings will facilitate subsequent transcriptomics and functional gene research on this pest.


Introduction
Eucryptorrhynchus scrobiculatus Motschulsky (Coleoptera: Curculionidae), an important wood-boring pest of the tree-of-heaven Ailanthus altissima Swingl [1][2][3] that is widely distributed in 21 provinces in China [4], is one of the most damaging invasive insects in this country. E. scrobiculatus larvae feed on A. altissima roots, and adults feed on A. altissima twigs, which usually kills the trees within 3-5 years [5]. Considered an invasive species, A. altissima is a serious threat to ecosystems in North America; as E. scrobiculatus specifically feeds on this tree, it has great promise as a potential biological control agent [1,6]. To study the biology of E. scrobiculatus, it is important to identify plant cell wall-degrading enzymes and functional genes involved in host-parasite interactions; however, reference genes must first be selected. Fortunately, E. scrobiculatus transcriptomes are available and provide comprehensive information to facilitate such studies. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 Gene expression analysis can be used to identify genes related to biological processes and physiological conditions [2]. Real-time quantitative RT-PCR (RT-qPCR) is the most commonly used technology for accurately detecting gene expression, and it has been widely applied in molecular biology [7,8]. Compared with classical molecular techniques, RT-qPCR has major advantages of higher sensitivity, better reproducibility and specificity; moreover, with its high-throughput nature, it has become the "gold standard" for gene expression quantification [9]. Nonetheless, the results of RT-qPCR vary due to differences in initial sample size, template RNA integrity, mRNA recovery, reverse transcription efficiency, and primer design. Thus, to obtain accurate and reliable gene expression results, RT-qPCR data must be normalized with appropriate reference genes to make the results more accurate and standardized, and the expression of which should not fluctuate during treatment [2,[10][11][12]. β-Actin (ACTB), elongation factor 1 (EF1), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), β-tubulin (TUB), and 18S ribosomal RNA (18S rRNA) have been widely used as reference genes in different organisms because they are considered to have uniform expression [2,[12][13][14][15]. In addition, although ACTB and GAPDH are the most commonly used reference genes, they have not been validated [16]. Furthermore, studies on reference gene expression in insects have found that the traditional genes show low stability across various biotic and abiotic conditions [16], and the most frequently used reference genes might not be stably expressed under different experimental conditions, causing a high potential for result misinterpretation. Therefore, it is necessary to perform systematic validation of reference genes to maximize the accuracy of PCR analysis and the reliability of published results [17,18]. Ribosomal protein genes are considered good reference genes because they are expressed in all cell types for the synthesis of new ribosomes [19][20][21]. Indeed, ribosomal protein genes (ribosomal protein L genes and S genes) exhibit high stability in different tissues of insects [22][23][24][25][26][27][28][29]. Our next study will investigate the gene expression of different tissues in E. scrobiculatus; therefore, the screening of reference genes is necessary in different tissues.

Insects
Adults of E. scrobiculatus were collected from A. altissima individuals in Wutongshu Town, Lingwu City, Ningxia Province and kept in the laboratory at 25±1˚C for 24 hours under 70 ± 5% humidity and a 16:8-h light/dark (L:D) photoperiod.

Sample preparation
The stability of candidate genes in E. scrobiculatus was assessed in different tissues of adult males and females. The various tissues examined (10 tissues) included the Malpighian tubule, hindgut, antenna, head, foregut, leg, male genitalia, female genitalia, wing and midgut. Each tissue sample was collected from the same 20 individuals. All samples were frozen at -80˚C until RNA extraction was performed.

RNA extraction and cDNA synthesis
Total RNA was extracted from samples using TRIzol (Aidlab Biotechnologies, China) following the manufacturer's instructions. The RNA concentration was quantified by measuring absorbance at 260 nm using a spectrophotometer (NanoDrop 8000, Thermo Fisher Scientific, USA). RNA purity was assessed at absorbance ratios of OD260/280 and OD260/230, and its integrity was determined by 1% agarose gel electrophoresis. Single-stranded cDNA was synthesized using 1 μg of RNA from various samples with a reverse transcription system (Takara, Japan) according to the manufacturer's recommendations. The cDNA was stored at -20˚C until further use.

Candidate reference genes and primer design
Primers for subsequent RT-qPCR of the selected candidate reference genes were designed using Primer3Plus software (Table 1). Reactions were performed using StepOnePlus (Applied Biosystems). cDNA was amplified using 2× TB Green™ Premix Ex Taq™ II (TaKaRa, Japan) according to the manufacturer's protocol in final reaction volumes of 20 μL. Each reaction contained 2 μL cDNA as a template, 0.8 μL forward and reverse primers (10 μM), 10 0078L 2× TB Green Premix, 0.4 μL ROX dye, and 7 μL ddH2O. PCR involved 95˚C for 1 min and 40 cycles of 95˚C for 10 s, 57˚C for 10 s and 72˚C for 20 s, followed by melting curve analysis (from 60 to 95˚C) to ensure the specificity of the amplified product. A dissociation curve analysis was performed for each reaction to confirm the specificity of the amplification. RCR products were cloned and sequenced to confirm the amplification of the correct products. Relative standard curves for the transcripts were generated using serial dilutions of cDNA (1/5, 1/25, 1/ 125, 1/625, and 1/3125), and the corresponding RT-qPCR efficiencies (E) were calculated using the following equation: E = (10[−1/slope] − 1) × 100. Reactions for three technical replicates were performed independently.

Constancy analysis of candidate reference genes
The Ct value is the number of amplification cycles that are elapsed when the fluorescence signal of the amplified product reaches the set threshold during PCR amplification. Data analyses were analyzed independently for each adult tissue of Ct values. The stability of the reference gene means that the expression levels should be approximate and no significant differences were observed under various types of tissues and various experimental conditions. The stability of the 18 candidate reference genes was evaluated using geNorm, NormFinder, and Best-Keeper as well as the delta Ct method in Microsoft Excel.
The geNorm algorithm determines an expression stability value (M) for each gene and then compares the pairwise variation (V) in this candidate reference gene with that in other tested candidate reference genes. Candidate reference genes with lower M-values have more stable expression. If the pairwise variation (V n /V n+1 ) between sequential normalization factors is less than 0.15, then it is necessary to use n genes as reference genes for that particular experimental condition [30]. NormFinder uses a model-based method to estimate the variation in expression of candidate reference genes, assigning a stability value to each candidate reference gene, whereby candidate reference genes with lower values are more stable [31]. BestKeeper calculates the standard deviation (SD) and stability value (SV) of candidate reference genes based on raw data (CT values), and those with low index scores are considered to be highly stable [32]. The delta Ct method calculates the mean SD by pairwise comparisons; a lower SD indicates a more stable gene [33]. Finally, RefFinder, a web-based comprehensive algorithm used to evaluate and screen candidate reference genes, integrates four computational programs (geNorm, NormFinder, BestKeeper, and delta Ct) to rank candidate reference genes. RefFinder is a comprehensive program that ranks candidate reference genes based on the rankings from four algorithms; each candidate reference gene is assigned an appropriate weight to determine a comprehensive final ranking (http://150.216.56.64/referencegene.php?type= reference).

Validation of a selected reference gene
To further confirm the reliability of candidate reference genes, the relative expression profiles of GH45 were determined in ten tissues and normalized to the two most stable candidate reference genes (RPL13 and RPS3) and the least stable gene (actin-5C). The relative quantity of the target gene was calculated using the 2 −ΔΔCt method [34]. To determine the effect on the relative expression of the target gene GH45 when different reference genes were used, we performed an independent sample t test. Primers used for amplifying and quantifying GH45 in E. scrobiculatus are provided in S2 Table. 3. Results

PCR amplification of candidate reference genes
Based on the results of the dissociation curves, a single peak but no signal in the negative controls for each reaction were obtained, suggesting that each gene was specifically amplified. Candidate reference gene sequences were amplified correctly by gel electrophoresis, and all the fragments were cloned and sequenced. The standard curves of candidate genes revealed correlation coefficient (R 2 ) values for all primer pairs ranging from 0.99605 to 0.99983, and the PCR efficiency determined by the standard curve ranged from 85.85 to 100.19% (S1 Fig, Table 1).

Expression profiling of candidate reference genes in E. scrobiculatus
To obtain an overview of transcript abundance, expression levels of the 18 candidate reference genes were investigated in different tissues of E. scrobiculatus. The variable Ct value of all candidate reference genes across different tissues showed different ranges of expression and different expression patterns for the 18 genes across various experimental groups (Fig 1). The Ct values for the 18 genes were used to estimate the stability of gene expression among the tissue samples ( Table 1). The mean Ct values of the candidate reference genes ranged from 22.49 for ribosomal protein L10a (RPL10a) to 29.97 for arginine kinase (AK), which exhibited the most and least abundant transcripts, respectively (Fig 1).

Expression stability and ranking of candidate reference genes
Four statistical algorithms were used to rank the candidate reference genes based on expression stability across different tissues. From the geNorm analysis, the pairwise variation value of V2/3 was less than the cut-off value of 0.15 (Fig 2). Thus RPS3, RPL36 and RPL14 were ranked the most stable genes based on geNorm tissue-specific expression profiling (Fig 3).
According to the rank order assigned by NormFinder, RPL13, RPL27 and RPS3 were the suitable genes for tissue-specific experiments (see Table 2). BestKeeper considered ACTIN, RPL13 and RPS11 to show the most suitable expression, which was rated as the best reference gene (Table 2). Lastly, the delta Ct method selected RPL13, RPS3 and RPL27 as the most suitable reference gene among the various tissues (Table 2).

Validation of the recommended candidate reference genes
In our study, the comprehensive rankings from RefFinder showed that RPL14 and RPS3 had the greatest stability in various tissues, whereas actin-5C showed the lowest stability. The normalization applicability of these candidate reference genes alone and in combination was tested to obtain expression profiles for GH45. Similar results were obtained using either one (RPL13 or RPS3) or two (RPL13-RPS3) candidate reference genes for normalization ( Fig 5). However, when actin-5C was used as a reference gene to standardize GH45, GH45 gene expression showed a huge difference in standardization with two other genes as references. Similar expression levels of GH45 were observed when normalized using RPL13, RPS3 and RPL13-RPS3 in different tissues. However, the expression levels of GH45 normalized using actin-5C and other candidate reference genes were significantly different in those tissues (P<0.05, Fig 5). These results indicated that the expression levels of genes are significantly Selection of reference genes of Eucryptorrhynchus scrobiculatus different when unstable reference genes are used to normalized qRT-PCR data in different tissues. The results further verified that appropriate selection of candidate reference genes is necessary for the investigation of gene expression levels.

Discussion
Many biological processes are closely related to the expression level of genes, and detection of gene expression levels is very important. Due to its high accuracy, sensitivity, specificity, and rapidity, RT-qPCR has been widely used for validating the results of gene expression profiles [35]. However, RT-qPCR requires normalization by reference genes to offset variation among samples processed from different tissues [36]. In recent years, with the gradual deepening of insect molecular biology research, it is becoming increasingly necessary to select appropriate reference genes to evaluate many biological processes or important functional genes in insects [2,3,22,24,29,[37][38][39][40][41][42][43][44][45][46][47]. Regardless, there is no study to date on reference genes in E.  scrobiculatus. Herein, we attempt to identify the most suitable reference genes for RT-qPCR analyses of samples from different tissues of this pest. Comparative analyses revealed some differences among the different statistical methods with regard to the rankings of the stability of the 18 candidate reference genes ( Table 2). The   pairwise correlations to assess stability among reference genes [48]. The delta Ct method analyzes the relative expression of gene pairs to identify suitable reference genes [33]. In our study, the delta Ct method results were similar to those of geNorm and NormFinder, and all stable reference genes were ribosomal protein genes, which in previous studies were identified as being highly stable between different insect tissues [35]. However, the rankings by Best-Keeper differed greatly from those of the other three algorithms, similar to the results of previous reports [44,49]. We also utilized the web-based comprehensive tool RefFinder, which integrates the four algorithms, to rank the overall stability of the 18 reference genes. Overall, our results suggest that it is necessary to use different statistical methods to assess the stability of reference genes in different tissues. β-Actin and GAPDH are usually selected as reference genes for RT-qPCR, but under many conditions, these two genes are not always ideal or even suitable reference genes [16]. Therefore, we selected three genes, namely, ACTIN, β-actin, and actin-5C, as candidate reference genes. However, ACTIN, β-actin, and actin-5C were mainly the most unstable candidate reference genes according to our results, and the stability of GAPDH was also low. In contrast, RPL13 showed the most stable expression in different tissues. Recent studies of potential reference genes in insects have found that ribosomal protein genes are among the most stable, especially between different tissues [35]. This result is consistent with that of reference gene . TUB belongs to the small globular family of proteins and is the major building block of microtubules in most eukaryotic cells. α-TUB and β-TUB were used in our study, and their rankings were very similar among the four methods. AK, a type of phosphagen kinase that is widely distributed in various tissues of invertebrates, exhibits high stability between different tissues of Drosophila suzukii and Bombus lucorum [50,51]. Eukaryotic translation initiation factor (EIF) is involved in the initiation of eukaryotic translation and is the most stable gene among different tissues of Bombyx mori [52]. UBC and EF1-α are also commonly used as reference genes [22,35], but in our study, these genes were not as stable as were ribosomal protein genes. Ribosomal protein genes were found to be the most stable reference genes in different tissues in E. scrobiculatus, consistent with the results of a previous study [35].
Glycoside hydrolase family 45 is a known insect endogenous cellulolytic family [53,54], and in this study, reference genes were validated by determining the expression profiles of the target gene GH45. As the expression level of GH45 was significantly different when normalization was based on RPL13, RPS3 and two reference genes (RPL13-RPS3) and on actin-5C, the normalization results based on actin-5C did not accurately reflect the expression level of GH45. These results are necessary for the selection of reference genes, suggesting that it is important to standardize the expression of target genes in E. scrobiculatus using a stable reference gene.
Overall, this is the first study to evaluate candidate reference genes for gene expression analyses in E. scrobiculatus. The expression profiles of 18 candidate reference genes in different tissues were investigated using four algorithms and RefFinder. We further confirmed that the reference genes commonly used may not be suitable for RT-qPCR normalization; consequently, reference genes must be verified before use. Nevertheless, we hereby recommend that RPL13 is the most stable reference gene for various tissues. The findings of this study not only provide stable reference genes for quantification of gene expression in E. scrobiculatus but also lay the foundation for transcriptomics and functional gene research on this pest.