Introduction

Shiraia bambusicola is an important and valuable macrofungus in the medical and food industries. It is noteworthy for its hypocrellins, the main secondary metabolites of S. bambusicola, whose use has been proposed for disease treatments that involve anti-clinical strains and anti-inflammatory and anti-viral activity (Su et al. 2011; Jiang et al. 2011; Zhou et al. 2009). Currently, to break the bottleneck of product yield and improve the understanding of biosynthetic pathways, the analysis of functional genes is attracting increasing attention (Deng et al. 2016).

Quantitative real-time reverse transcription PCR (qRT-PCR) is generally regarded as a convenient and efficient tool to analyse gene expression, but the RNA quantification, the reverse transcription reaction efficiency and other uncontrolled factors may limit the accuracy and stability of the final results (Huang et al. 2014; Hao et al. 2014). Thus, it is necessary to apply reliable reference genes to normalize the data.

A large number of research papers have been published on reference genes under different stresses or from different organs in plants (Warzybok and Migocka 2013; Lin et al. 2014), and similar works in filamentous fungi are gradually being conducted (Zampieri et al. 2014; Zhou et al. 2011). It is striking that most of the applied reference genes for fungi were directly copied from the existing results in plants or animals, such as actin, tubulin and 18S rRNA (Fang and Bidochka 2006; Yan and Liou 2006). However, the divergent regulatory mechanisms and environmental stresses suggested that these reliable reference genes might not be suitable for the analysis of fungal gene expression. For example, the classical reference genes elongation factor (EF-1) and beta-tubulin (β-tubulin), which are widely used in plants, were not appropriate in Hemileia vastatrix (Vieira et al. 2011); another reliable reference gene, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), used in several qPCR studies for the normalization of data, was not stable in Heterobasidion annosum (Raffaello and Asiegbu 2013). Thus, it is necessary to systemically rescreen the candidates in novel species from the fungal kingdom.

According to the current literature, exploring putative reference genes from the transcriptomic data is a feasible and easy assay in fungal species (Llanos et al. 2015; Nadai et al. 2015; Steiger et al. 2010). Likewise, there is a series of popular Excel-based software programs, such as geNorm, NormFinder and BestKeeper, that have been successfully used in previous studies (Vandesompele et al. 2009; Andersen et al. 2004; Pfaffl et al. 2004).

Thus far, research on S. bambusicola has focused on strain mutagenesis and product fermentation (Song et al. 2015). However, since the genome and transcriptome were published (Yang et al. 2014; Zhao et al. 2016), gene function and molecular regulation have become research hotspots. To make qRT-PCR results more reliable and the application of reference genes more widespread, we comprehensively rescreened the candidate reference genes for S. bambusicola under different light conditions and in different media.

Materials and Methods

Fungal strain and culture conditions

Strain zzz816 of Shiraia bambusicola was deposited in the China General Microbiological Culture Collection Centre (CGMCC, No: 3135). Fungal isolates adhered to agars were cultured in potato dextrose agar (PDA) medium (potato 200 g/L, dextrose 20 g/L, agar 15 g/L) at 26 °C for 7 days, and then the fresh mycelia were inoculated onto different media (rice medium: rice 800 g/L, PD medium: potato 200 g/L, dextrose 20 g/L, and Czapek–Dox medium: NaNO3 3 g/L, K2HPO4 1 g/L, MgSO4⋅7H2O 0.5 g/L, KCl 0.5 g/L, FeSO4 0.01 g/L, sucrose 30 g/L) under three different light conditions, namely, 24 h of continuous darkness (all-dark), 24 h of continuous lighting with a light intensity of 1500 lx (all-light), and 12: a 12 h light photoperiod with 1500 lx light intensity (12-h light/dark).

Finally, each sample was collected at different time points: 2, 4 and 6 days.

RNA isolation and cDNA synthesis

The total RNA of each sample was extracted from frozen mycelia using an improved method in our laboratory (Song et al. 2015), and the total RNA was dissolved in RNase-free water.

The first strand of cDNA was synthesized by reverse transcribing 1 μg of RNA with TransScript ® All-in-One First-Strand cDNA Synthesis SuperMix for qPCR (One-Step gDNA Removal; TransGen, China). The quantity and quality of the total RNA extracted was determined using an Eppendorf BioPhotometer plus (Eppendorf, Ger), and the cDNA samples were stored at −20 °C.

Primer design and quantitative real-time PCR

Primers were designed using Primer3 software (http://primer3.ut.ee/), and the specificity of the product was verified using 1% agarose gel electrophoresis and melting curves. The efficiency of the validated primer pairs remained at approximately 100% (Table 1). The qRT-PCR reaction was performed in LightCycler®480 multiwell plates (Roche Applied Science, Indianapolis, IN, USA) with a LightCycler®480II/96 (Roche Applied Science, Indianapolis, IN, USA) real-time PCR system using LightCycler®480 SYBR Green I Master Mix (Roche Applied Science, Indianapolis, IN, USA). The reactions were performed according to the recommendations of the manufacturer: 95 °C for 5 min for initial denaturation, followed by 45 thermal cycles of 10 s at 95 °C, 10 s at 60 °C and 20 s at 72 °C. The melting curve was performed with slow heating from 65 to 97 °C with continuous measurement of fluorescence 5 acquisitions per 1 °C. All reactions were performed with three biological and two technical replicates with negative controls. The qRT-PCR data were directly analysed using the “second derivative maximum” function in the LightCycler®480 Software Version 1.5 (Roche Applied Science, Indianapolis, IN, USA).

Table 1 Description of candidate reference genes, and the details of primers and amplicons
Full size table

Statistical analyses

To select suitable reference genes, three software packages were used to calculate the stability: NormFinder, geNorm, and BestKeeper. An additional web-based tool, RefFinder, (http://fulxie.0fees.us/) was applied to integrate and rank all candidate reference genes (Xie et al. 2012).

To calculate the PCR amplification efficiencies (E) and correlation coefficients (R2) of each primer pair, standard curves were prepared using a 10-fold serial diluting plasmid, into which the reference gene was cloned in PEASY-T3® (TransGen, China).

The efficiency (E) was calculated according to the equation E = (10(−1/slope)−1) × 100% (Radonić et al. 2004).

Results

Strategy for selecting reference gene candidates

In this study, twelve candidate reference genes appeared in the stabilization assay, and nine of them were involved in this test for the first time. The novel ones were sought out directly from the public transcriptome data sets (Zhao et al. 2016), and the selection was based on the ranking of the expression levels of each gene, expressed as reads per kilobase per million (RKPM). For the whole data set, we evaluated the coefficient of variation of RKPM, and the ones with relatively smaller coefficients were considered the genes of interest, including GTP-binding protein SAS1 (SbSAS1), 4A/4B type thioredoxin-like protein (SbTRX), t-SNARE (SbtS), RNA binding protein-like protein Jsn1 (SbJsn1), conserved hypothetical protein (SbCHP), LAlv9 family protein (SbLAlv9), Pkinase-domain-containing protein (SbPKI), FYVE-domain-containing protein (SbFYVE) and vacuolar protein sorting 55 (SbVAC55). Other three genes, ubiquitin-activating enzyme (UBI), vacuolar ATPase subunit 1 (VAC) and transcription factor TFIIIC (TFC), were generally used to normalize the qRT-PCR data and appeared to be especially reliable as reference genes in the preliminary study for Shiraia bambusicola (Song et al. 2015).

Before the analysis of expression stability, the gene-specific amplification of these genes was confirmed by the single-peak melting curves of the qRT-PCR products, and the primers provided a good reaction efficiency ranging between 96 and 102% (Table 1).

The crossing point (CP) values were collected from all tested samples under different conditions and are shown in the box-plot (Fig. 1). The value of gene expression studied indicated a compact distribution and a limited range between 21 and 33. Among the genes, TFC displayed a lower expression variation, which mainly depended on the different media (Song et al. 2015).

Fig. 1
figure 1

The range of CP values of 12 reference gene candidates for all samples. Each box indicates the 25th and 75th percentiles, and the caps represent the maximum and minimum values. The median is shown by the line across the box

Full size image

Expression stability analysis

Three different applets, geNorm, NormFinder, and BestKeeper, were applied to measure and rank the stability of candidate reference genes. The program geNorm (Steiger et al. 2010) classifies genes according to the control gene stability measure (M value), which represents the average of the pair-wise variation of a gene with all other control genes. NormFinder (Vandesompele et al. 2009) examines the stability of each single candidate gene independently and not in relation to the other genes. The results from geNorm and NormFinder can be compared easily because they both use raw data (relative quantities) as input data. BestKeeper (Andersen et al. 2004) is another Excel-based tool that determines the optimal reference genes using a pair-wise correlation analysis (Pearson correlation coefficient) of all pairs of candidate genes. It uses CP values (instead of relative quantities) as input and employs a different measure of expression stability from geNorm and NormFinder. However, this software is not able to analyse more than ten reference genes together, and thus the first ten ranking genes should depend on geNorm and NormFinder. The rankings of these software programs relied on different algorithms and were expected to lead to distinct outputs. Therefore, another software program, RefFinder (Radonić et al. 2004), was used to integrate the currently available major computational programs (geNorm, NormFinder, BestKeeper, and the comparative ΔΔCt method) to compare and rank the tested candidate reference genes. Based on the rankings from each program, it could assign an appropriate weight to an individual gene and calculate the geometric mean of their weights for the overall final ranking.

Using geNorm, SbLAlv9, SbJsn1, SbCHP and SbSAS1 displayed the highest reliability overall (Table 2). In the rice medium with difference light conditions, SbLAlv9, SbSAS1, SbJsn1 and TFC had a better effect than other candidate reference genes. In the PD medium, SbLAlv9, SbJsn1, SbtS, and SbSAS1 ranked at the top positions, and in the Czapek-Dox medium, SbJsn1, VAC, SbLAlv9 and SbFYVE were found to be the ideal reference genes. In the different light conditions, the best reference candidates were divided into three groups, including SbLAlv9, SbJsn1, VAC and SbCHP under all-dark conditions, SbLAlv9, SbJsn1, SbCHP, SbtS under all-light conditions and SbLAlv9, SbJsn1, SbSAS1, SbtS under the 12-h light/dark conditions. SbFYVE and SbPKI have been classified as the least reliable reference genes in most of the conditions. Furthermore, geNorm provided an output allowing a set of reliable normalization for the pairwise variation (Vn/n+1) to help to determine the optimal number of reference genes (Fig. 2). Two reference genes were sufficient for most of the conditions, but the continuous darkness required a third reference gene. Four reference genes were suggested to ensure accurate all-round analysis of the nutrient and light conditions.

Table 2 Ranking of candidate reference genes calculated by geNorm according to different expression conditions
Full size table
Fig. 2
figure 2

The geNorm-based results of the pairwise variation analysis. A sufficient number of genes n can be used for reliable normalization when Vn/n + 1<0.15

Full size image

As shown in Table 3, SbLAlv9, SbJsn1 and SbSAS1 were identified as the best reference candidates by NormFinder, and SbCHP, SbtS and VAC displayed good properties for certain nutrient and light conditions. SbFYVE and SbPKI were ranked as the most unstable genes, and this result was also in agreement with the geNorm calculation.

Table 3 Ranking of candidate reference genes calculated by NormFinder according to different expression conditions
Full size table

In contrast to geNorm and NormFinder, the BestKeeper algorithm is based on the coefficient of variance (CV) and the standard deviation (SD) calculated by the average CP value of each reaction. Moreover, we removed SbFYVE and SbPKI due to the higher variations based on geNorm and NormFinder. (Table 4). SbVAC55, SbSAS1, UBI and SbTRX were the four best reference genes among all the samples. Based on the different points, SbSAS1, TFC, UBI and SbTRX were the most stable on the different media, and SbVAC55, UBI, SbSAS1 and SbtS were identified as suitable reference genes for the different light conditions, including all-dark, all-light and 12-h light/dark.

Table 4 Ranking of candidate reference genes were calculated by BestKeeper according to different expression conditions
Full size table

To integrate the results obtained from different algorithms, RefFinder was used to perform the final ranking. As illustrated in Table 5, SbLAlv9 was singled out as the top candidate in the global consideration of different nutritional and light conditions, followed by SbJsn1, SbSAS1 and SbVAC55. Nevertheless, under specific conditions, the gene order would be altered, and the reliability of proper reference genes might be reduced. Thus, we propose two different methods—one medium with different light conditions and one light condition with different media—to improve the utilization of reference genes in practical experiments.

Table 5 Ranking of the candidate reference genes calculated using RefFinder under different conditions
Full size table

Application in detailed conditions

Although the appropriate reference genes had been presented as in Table 5, it was still inconvenient to search for a target effectively for one particular condition. To optimize the application of the ideal reference genes, we reprocessed the data and characterized the reference genes as a two-way table (Table 6). Therein, we could find the suitable reference genes more easily and accurately. For example, under continuous darkness (Table 5) and limited to rice medium, SbVAC55 or SbLAlv9 was sufficient for qRT-PCR analysis (Table 6).

Table 6 The most stable reference genes under different nutritional and light conditions
Full size table

Discussion

Of all the candidate reference genes, three familiar genes (UBI, VAC and TFC) had been approved previously (Song et al. 2015), and another nine (SbSAS1, SbTRX, SbtS, SbJsn1, SbCHP, SbLAlv9, SbPKI, SbFYVE, SbVAC55) first emerged for the reliability of gene expression. Three new genes, SbLAlv9, SbJsn1 and SbSAS1, indicated appropriate properties for all the nutrient and light conditions. SbtS and SbVAC55 also separately showed high reliability for certain conditions. In contrast, only one old reference gene, VAC, displayed good activity for Czapek–Dox medium or all-dark conditions. As illustrated by these results, the exploration of new reference genes could indeed provide the more reliable genes and improve the assurance of the stability of gene expression in S. bambusicola. Furthermore, the discovery of novel reference genes might contribute to the use of reference genes in other species.

We chose typical nutrient and light conditions in this study that generally cover the culturing conditions of S. bambusicola or other species. Therefore, the reliable reference genes in this work were anticipated to be broadly used in the gene expression analysis of S. bambusicola and could also provide benefits for other fungal species. Furthermore, a table (Table 6) was constructed to facilitate the application of these results, with a two-way form to facilitate finding the stable reference genes for specific conditions. In this work, we tried our best to expand the scope of application and make the process more convenient.

In conclusion, nine novel candidate reference genes were introduced for the first time, and several of them (SbLAlv9, SbJsn1, SbSAS1 and SbVAC55) were validated as ideal ones. In addition, this work provided a set of reference genes under different nutrient and light conditions, and these housekeeping genes are expected to be available in a more extensive range.