Evaluation of suitable reference genes for normalization of quantitative reverse transcription PCR analyses in Clavibacter michiganensis

Abstract Clavibacter michiganensis, the causal agent of bacterial canker of tomato, is a Gram‐positive bacterium and a model for studying plant diseases. The real‐time quantitative reverse transcription PCR (real‐time qRT‐PCR) assay is widely used to quantify gene expression in plant pathogenic bacteria. However, accurate quantification of gene expression requires stably expressed reference genes that are consistently expressed during the experimental conditions of interest. The use of inappropriate reference genes leads to a misinterpretation of gene expression data and false conclusions. In current study, we empirically assessed the expression stability of six housekeeping genes (gyrB, rpoB, tufA, bipA, gapA, and pbpA) of C. michiganensis under five experimental conditions using two algorithms, geNorm and NormFinder. C. michiganensis expressed gyrB, bipA, and gapA stably when growing in nutrient‐rich broth (TBY broth and modified M9 broth). We concluded that pbpA, tufA, and gyrB were suitable reference genes in C. michiganensis—tomato interaction studies. We also recommended bipA and rpoB to be used to study bacterial gene expression under nutrient‐poor conditions. Finally, gyrB, pbpA, and rpoB can be used to normalize the quantification of C. michiganensis gene expression while the bacterium is in the viable but nonculturable (VBNC) state. This study identified the most suitable reference genes depending on the experimental conditions for calibrating real‐time qRT‐PCR analyses of C. michiganensis and will be useful in studies that seek to understand the molecular interactions between C. michiganensis and tomato.


| INTRODUC TI ON
Species within Clavibacter genus are xylem colonizing Gram-positive bacteria. Based on average nucleotide identity from whole genome analyses, Clavibacter michiganensis subspecies have been to elevate to the species level recently (Li et al., 2018;Tambong, 2017). Specific Clavibacter species can infect different hosts, and Clavibacter michiganensis (also known as C. michiganensis subsp. michiganensis) is the causal agent of bacterial canker of tomato (Solanum lycopersicum) (Davis, Gillaspie, Vidaver, & Harris, 1984;Strider, 1969). This bacterium has become a model for studying the mechanisms of pathogenicity of Gram-positive bacteria due to its potential risk of causing substantial economic losses in tomato production and the availability of its whole genome sequences (Gartemann et al., 2008;Thapa et al., 2017). Relative to most economically important phytopathogenic bacteria that are in the Proteobacteria phylum, less attention has been paid to C. michiganensis with regards to molecular host-pathogen interactions (Mansfield et al., 2012). The mechanisms of pathogenicity for C. michiganensis are substantially different to those Proteobacteria members (Eichenlaub & Gartemann, 2011).
The genome of C. michiganensis strain NCPPB382 was published in 2007 and provides a valuable resource for exploring its pathogenesis (Gartemann et al., 2008). Studies of the molecular biology of bacterial canker of tomato can ultimately improve disease management.
The survival and transmission of C. michiganensis on/in the seed or in the field is important for understanding and managing tomato bacterial canker (Fatmi & Schaad, 2002). Specifically, C. michiganensis cells on tomato seeds may be induced into the viable but nonculturable (VBNC) state by a range of conditions including bactericide (copper agent) treatments, dry conditions, and lack of available nutrients and may pose a risk for tomato production (Jiang et al., 2016;de León, Siverio, López, & Rodríguez, 2011). The VBNC state of some Gramnegative phytopathogenic bacteria has been shown to be a survival strategy in response to harsh environmental conditions (Oliver, 2010). However, there are very few studies of the molecular mechanisms involved in the induction and maintenance of VBNC state of phytopathogenic bacteria.
Real-time quantitative reverse transcription polymerase chain reaction (real-time qRT-PCR) has become a powerful technique for RNA quantifying and the expression of specific genes (Die & Román, 2012). This assay was used to study the interactions between C. michiganensis-tomato plants, and the results revealed that the difference of gene expression between systemic and located infection of C. michiganensis, while some genes located on the pathogenicity island of C. michiganensis were involved in the suppression of tomato basal defenses (Chalupowicz et al., 2017(Chalupowicz et al., , 2010. qRT-PCR was also used to validate the expression patterns obtained by cDNAamplified fragment length polymorphism (AFLP) and identified five candidate genes for developing C. michiganensis-tolerant tomato cultivars (Lara-Ávila, Isordia-Jasso, Castillo-Collazo, Simpson, & Alpuche-Solís, 2011). Moreover, real-time qRT-PCR was used to detect VBNC cells of bacteria, because sustainable gene expression is a reliable indicator for cell viability (Oliver, 2010;Liu, Wang, Tyrrell & Li, 2010). For example, the expression of housekeeping genes, such as 16S rRNA and rpoS (RNA polymerase and sigma factor), was quantified to reveal the variation in viability of bacterial cells (González-Escalona, Fey, Höfle, Espejo, & A Guzmán, 2006).
Obtaining accurate gene expression results by real-time qRT-PCR is difficult in some cases, due to the shortage of stably expressed reference genes for data normalization and quantification.
Despite the comments above, the empirical validation of C. michiganensis reference genes for use in real-time qRT-PCR analyses received few attention. The expression stability of 3 housekeeping genes, gyrA, bipA, and qcrA, was estimated by geNorm under experimental conditions and as a result, gyrA and bipA were used as reference genes in the study of interactions between tomato and C. michiganensis (Chalupowicz et al., 2017(Chalupowicz et al., , 2010. Since only three housekeeping genes were evaluated under one experimental condition, it is unlikely that these genes will be suitable for normalizing gene expression studies conducted in other experiments.
In current study, the expression stabilities of six housekeeping genes from C. michiganensis were tested under five different conditions. The results provide information on the most suitable reference genes for gene expression studies in C. michiganensis and other phytopathogenic bacteria.

| Bacterial strains and culture conditions
Clavibacter michiganensis strain BT0505 (isolated from tomato field in Inner Mongolia autonomous region of China in 2005) was grown in TBY broth (5 g/L yeast extract, 5 g/L NaCl, and 10 g/L tryptone) at 28°C with shaking (140 rpm) for 22 hr (to an OD 580nm of ca. 1.0). The cells were harvested by centrifugation at 14,000 g for 3 min and washed three times with a 0.85% (w/v) NaCl solution before treatment.
Five experimental conditions were used for bacterial RNA extraction and candidate genes expression. TBY broth was used as a standard medium, modified M9Cmm-minimal (mM9) medium (11.28 g/L 5 × M9 minimal salts, 2 mM MgSO 4 ·7H 2 O, 0.01 mM CaCl 2 ·2H 2 O, 0.5 mg/L Vitamin B1, 0.5 mg/L nicotinic acid, and 3.96 g/L glucose) was a basic medium with less nutrients, and tomato seedling homogenate (TSH) medium was used to mimic the natural host environment and the final concentration of TSH was adjusted to 10% (Flügel, Becker, Gartemann, & Eichenlaub, 2012). Clavibacter michiganensis cells at late log phase were added to TBY broth, mM9 medium, and TSH medium at the ratio of 1:100, respectively, and the final titer of bacteria was ca. 10 7 CFU/ml. The growth curve of C. michiganensis in these three media was measured with a Bioscreen C Pro Automated Microbiology Growth Curve Analysis System (Oy Growth Curve Ab Ltd.) at 28°C with continuous shaking and calculated by Prism 6 (GraphPad Software). Time points representing log, stationary, and decline phases were selected for RNA extraction according to the growth curves.
The other two experimental conditions were 0.85% NaCl solution supplemented with and without 50 μM of CuSO 4 . These conditions represent a VBNC induction condition and a starvation treatment, respectively. Clavibacter michiganensis cells at late log phase were added into 0.85% NaCl solution with and without 50 μM CuSO 4 to an OD 580nm of 0.45 (ca. 10 8 CFU/ml), respectively. Cells were incubated at 28ºC without shaking and harvested samples at 12, 24, and 48 hr, respectively. The culturable cells in the starvation treatment were assessed by spreading serial dilutions on TBY agar incubating at 28°C for 3 days and counting the colonies. In contrast, the VBNC cells were counted as described previously reported method (Jiang et al., 2016). This experiment included two biological replicates of each treatment, and each biological replicate included three technical replicates in survival curves analysis. During real-time qRT-PCR analyses, each treatment had three biological replicates.

| RNA isolation and cDNA synthesis
To extract RNA, 1 ml of C. michiganensis cells (OD 580nm = 1.0) was harvested by centrifugation at 14,000 g for 5 min at 4°C at each time point in different treatment and stored at −80°C immediately.
Total RNA was extracted using the SV Total RNA Isolation System (Promega Corporation) according to the manufacturer's instructions with minor modifications. Cells were resuspended in 100 μl freshly prepared 1 × TE buffer containing 50 mg/ml lysozyme and incubated at 37°C for 10 min, then the manufacturer's protocol was followed. RNA was eluted in 70 μl of nuclease-free water and kept at −80°C. Total RNA concentration and purity were determined using a NanoDrop 2000 (Thermo Fisher Scientific), and RNA integrity was tested by running 1% (w/v) native agarose gel in 1 × TAE buffer under 6 V/cm for 15 min. Two bands will be observed on the gel after EB staining, the intensity of 23S rRNA band should be twice as strong as the 16S rRNA band if the integrity of RNA sample is good.
The concentration of all RNA samples was adjusted to 100 ng/μl and used for following experiment.

| Oligonucleotide primer design
The DNAMAN 6 software (Lynnon Biosoft) was used to design primers based on the genome sequence of C. michiganensis strain NCPPB382 to amplify the corresponding sequences of candidate TA B L E 1 Clavibacter michiganensis genes and oligonucleotide primers used for real-time qRT-RCR analyses RT-R: AACGCAAGGAGAAGGACG a Amplification efficiency, calculated as E = (10 1/slope −1)*100, where the slope was obtained from the cDNA dilution ratio -C t value standard curve.
genes from strain BT0505. Oligonucleotide primers were synthesized by Sangon Biotech (Shanghai, China), and the amplicons were sequenced by Life technologies (Beijing, China). Based on the sequence data, eight pairs of primers were selected for real-time qRT-PCR and used for gene expression analyses (Table 1). The amplicon size and the annealing temperature ranged from 50 to 210 bp and 55 to 60°C, respectively. The specificity of each primer pair was determined by SYBR Green-based real-time PCR followed by electrophoresis in 2% (w/v) agarose gel and melt curve analyses. The amplification efficiency of each primer pair was obtained from the cDNA dilution ratio-C t value standard curve. The curve was generated with fourfold diluted cDNA template and the slope of the curve was used in the efficiency equation: E = (10 1/slope -1)*100. The primer pair was chosen when the efficiency was between 90% and 110% and the correlation index R 2 was >0.99.

| Conventional PCR and qPCR analyses
The primers listed in Table A1 in Appendix were used to amplify the six candidate reference genes and two validating genes. Clavibacter michiganensis cells suspended in sterile water at a final titer of ca. were distinct for different primer sets, Table A1 in Appendix), while the gyrB5754F/gyrB8082R and the celA17558F/celA19998R assays were conducted as previously described (Luo, 2008).

| Data analysis and normalization of pathogenicity genes
Software geNorm and NormFinder were used to evaluate the expression stability of 6 reference gene candidates. geNorm estimates the most stably expressed reference genes through a stepwise exclusion or ranking process, and the normalization factor is the geometric mean of the most stable reference genes (Vandesompele et al., 2002). NormFinder is an add-in tool for Excel, which considers the intragroup and intergroup variations of the sample sets. It will show the best reference gene or the best combination of two genes regards to expression stability (Andersen et al., 2004).

| Selection of reference gene candidates and their qRT-PCR primers
The reference gene candidates used in this study were selected These genes had different functions to avoid problems associated with coregulation. The celA (cellulase) and chpC (serine protease) genes, which are important for C. michiganensis pathogenicity, were selected as validating genes and used to evaluate the applicability of reference genes.
Six primer pairs that amplified pbpA, gapA, bipA, tufA, rpoB, and chpC genes (  Table A2 in Appendix. Real-time qRT-PCR was optimized for each primer pair (Table 1) using cDNA samples. The primer pair rRNA-I/rRNA-O designed in this study was used to confirm the absence of gDNA. The 516-bp amplicon included the space between rRNA and tyrS (tyrosyl-tRNA synthetase) sequences which was not in the same transcript. All cDNA samples were negative for the 516-bp amplicon, indicating that the cDNA samples were free of gDNA ( Figure A2 in Appendix).
Melt curve analyses showed a single melting peak for each primer pair in qPCR with the C. michiganensis cDNA ( Figure A1 in Appendix).
This indicated that there was no nonspecific amplification with the primer pairs. Amplification efficiency values ranged from 91% to 101% (Table 1) confirmed that these 6 primer pairs were suitable for evaluating the expression stability of the reference gene candidates.

| Expression stability of the reference gene candidates in cultural conditions
The expression stability of six candidates of C. michiganensis was evaluated in three different experimental conditions at the log, stationary, and decline growth phases. C. michiganensis grew fastest in TSH medium, followed by TBY broth and mM9 medium at the log phase. However, the greatest biomass was obtained in mM9 medium at the stationary phase ( Figure 1a). The expression levels of the six candidate genes were presented as the raw C t values.  (Figure 2b). The results show that these genes displayed lower expression levels, but greater expression stability in mM9 medium although the differences were not significant according to t test (p < .05). In geNorm analyses, gyrB (0.454), bipA (0.454), and gapA (0.622) fit the cut-off limit of M ≤ 0.7, but the lowest V value was 0.227 (V 2/3 ), which was greater than the threshold for V (≤0.15) ( Figure 3b). The ranking of the genes based on expression stability was different between NormFinder and geNorm. According to NormFinder, the most stably expressed gene was gapA and the least stably gene was pbpA (Figure 3b). Based on geNorm, gyrB, bipA, and gapA were the most suitable reference gene candidates despite the fourth ranking of bipA by NormFinder.
The expression stability of C. michiganensis reference genes in TSH medium was different to that in TBY broth and mM9 medium.  Figure 2c). Interestingly, pbpA, which was the least stably expressed gene in TBY and mM9 media, was the most stably expressed gene in TSH medium (Figure 3c). Analyses by both geNorm and NormFinder showed the same rankings for expression stability of the candidate reference genes. pbpA and gapA were the most and least stably expressed genes, respectively.
F I G U R E 1 Population growth of Clavibacter michiganensis in five experimental conditions. (a) Three cultural conditions, TBY broth (purple), modified M9Cmm-minimal medium (mM9, blue), and tomato seedling homogenate (TSH, orange). The curves were generated using a BioScreen C Pro Automated Microbiology Growth Curve Analysis System. Time points on each curve represented the log phase, stationary phase, and decline phase from left to right. (b) Survival curve of Clavibacter michiganensis in 0.85% NaCl solution (triangle) and 0.85% NaCl solution supplemented with 50 μM CuSO 4 (circle). Culturable cell counts were determined by the plating method, and viable cells counts were calculated by flow cytometry method. Error bars represent the standard deviation of two biological replicates The top three stably expressed genes, pbpA, tufA, and gyrB, were suitable for transcript normalization in vitro according to the criteria described above. Combined all results, the gyrB was an optimal reference gene for culture-based research. Moreover, the bipA and gapA were also recommended for in vitro research, such as incubation in TBY broth and mM9 medium. However, pbpA, tufA, and gyrB were an appropriate reference gene set in C. michiganensis-tomato interaction research (Table 3).

| Expression stability of the reference gene candidates in oligotrophic conditions
The expression stability of six C. michiganensis reference genes was evaluated in 0.85% NaCl solution and 0.85% NaCl solution supplemented with 50 μM CuSO 4 , which represent starvation and VBNC-inducing conditions, respectively. In 0.85% NaCl solution, the population of culturable cells decreased slightly, from 10 8 CFU/ml to 10 7 CFU/ml within 4 days (Figure 1b). Under the presence of copper ions, however, the culturable cells decreased below the limit of detection (0.1 CFU/ml) at 12 hr postinduction, and the viable cells remained constantly at 10 7 cell/ml by 2 d after induction, suggesting that the cells were in VBNC state (Figure 1b).
For the two oligotrophic environments, the Ct values of the six candidate genes were 23. 97-33.80, 24.36-36.51, 17.85-28.47, 24.20-31.46, 13.86-30.42, and 26.40-36.92 for gyrB, pbpA, gapA, bipA, tufA, and rpoB, respectively. Compared to the 0.85% NaCl solution, the standard deviations of C t values for expression of all genes except rpoB were smaller in the copper-stress environment ( Figure 2d,e). Based on the geNorm analyses, the gyrB and tufA genes did not meet the threshold for acceptance (M ≤ 0.7) in 0.85% NaCl solution and copper-stress conditions, respectively ( Table 2). The V 2/3 value for bipA and rpoB (V 2/3 = 0.09) and the V 3/4 value of gyrB, pbpA, and rpoB (V 3/4 = 0.14) were less than 0.15 in 0.85% NaCl solution and the copper-stress condition, respectively ( Table 2). The combination of bipA and rpoB and the combination of gyrB, pbpA, and rpoB were sufficiently stable to serve as normalization factors for measuring C. michiganensis gene expression in starvation and VBNC studies, respectively. NormFinder ranked rpoB and gyrB as the most stably and most unstably expressed genes in 0.85% NaCl solution, respectively. In the copper-stress condition, geNorm and NormFinder were in agreement with regards to the rank of six genes (Table 2). These results indicated bipA and rpoB were consistently found to be the most suitable reference genes for C. michiganensis gene expression studies in starvation conditions, and the combination of gyrB, pbpA, and rpoB was recommended for VBNC research (Table 3).

| Validation of reference genes
In order to validate the selected reference genes, the relative expression level of two C. michiganensis pathogenicity genes, celA, and chpC, were evaluated in TSH medium. Clavibacter michiganensis cells collected at the log phase in TBY broth were used as control. In TSH medium, the expression of celA and chpC were normalized using the three most stable reference genes (pbpA, tufA, and gyrB) and the least stable gene (gapA), respectively. celA expression increased over time. In contrast, chpC expression was down-regulated initially then increased. Normalized of gene expression with the most suitable reference genes (pbpA, tufA, and gyrB) or gapA, revealed that expression of both celA and chpC increased over time in TSH medium ( Figure 4). However, comparing with the three most suitable reference genes, the expression of chpC and celA are higher when only gapA gene was used for normalization. Moreover, the difference at 105 hr was significant according to t test (p < .05). These results indicate that unsuitable reference genes may lead to an overestimation of pathogenicity gene expression in C. michiganensis.  (Takle et al., 2007). Hence, it is critical to screen appropriate reference genes under the experimental condition to confirm stable gene expression.

| D ISCUSS I ON
Here, we empirically assessed the expression stability of six reference genes in C. michiganensis under five experimental conditions.
Six housekeeping genes, gyrB, pbpA, gapA, bipA, tufA, and rpoB, were evaluated in this study. The gyrB gene, encoding DNA gyrase, is a common reference gene especially for Gram-positive bacteria and is stably expressed in nutrient-rich media or under certain stress conditions (Carvalho et al., 2014;Crawford, Singh, Metcalf, Gibson, & Weese, 2014;Duquenne et al., 2010;Sihto, Tasara, Stephan, & Johler, 2014). The gapA gene, encoding glyceraldehyde-3-phosphate dehydrogenase, is a key gene for metabolism and has been used to normalize the expression of virulence genes (Bjelland et al., 2013;Kjeldgaard, Henriksen, Cohn, Aabo, & Ingmer, 2011). We observed that M value of gyrB and gapA was 0.82 (data not shown), which suggests that these genes are suitable for some studies (Botteldoorn et al., 2006;Mafra et al., 2012). These two genes can be used as reference genes for C. michiganensis when the bacteria are cultured under most common conditions, such as TBY broth and other nutrient-rich media.
The pbpA gene, which is responsible for synthesizing the penicillin-binding protein, expressed stably under different conditions, including interaction with host cells, nutrient starvation, and VBNC induction. Previously, pbp5, which is a homolog of pbpA, was successfully used as a reference gene for studying the viability of VBNC Enterococcus faecalis cells (Lleò et al., 2001). bipA and gyrA were reported as reference genes in C. michiganensis-host interaction study (Chalupowicz et al., 2017(Chalupowicz et al., , 2010. In current study, we showed that pbpA and tufA were better reference genes than bipA in TSH medium, which is a mimic of natural host-pathogen interaction condition ( Figure 3c). Interestingly, tufA and rpoS genes were constitutively expressed in VBNC Vibrio vulnificus cells (Smith & Oliver, 2006). rpoB is a homolog gene of rpoS in C. michiganensis, and its M value was less than 0.5 under nutrient starvation and VBNC-inducing conditions, indicating that rpoB is a suitable reference gene under stressful conditions. On the other hand, tufA expression was less stable than rpoB when C. michiganensis was grown under stress conditions. Furthermore, the commonly used reference gene, 16S rRNA, was ruled out in our study. Since rRNA accounts for more than 95% of bacterial total RNA, and the expression level of rRNA is significantly higher than that of mRNA (Peano et al., 2013). Additionally, rRNA is more stable than mRNA, and its quantity is not comparable to that of mRNA (Deutscher, 2006;Peano et al., 2013). The use of rRNA as reference gene for a human tissue biopsy was also found to be inappropriate (Tricarico et al., 2002).
The expression of C. michiganensis reference genes tested in this study varied among the experimental conditions, confirming that standard housekeeping genes should be screened before use for normalization of gene expression. Based on t test (p < .05), we observed significant difference when normalizing the expression of celA and chpC with different normalization factors, the three most stable reference genes (pbpA, tufA, and gyrB) and the most variable gene (gapA) at 105 hr in TSH medium (Figure 4). Similar results were reported in studies using reference genes of melon and citrus (Kong et al., 2014;Mafra et al., 2012). The use of inappropriate reference genes led to an increasing fold change of WRKY70, which was upregulated 35-fold compared to the most stable reference genes (Mafra et al., 2012). These results indicate that inappropriate reference genes can lead to a significant misinterpretation of data.
geNorm and NormFinder ranked the six C. michiganensis reference genes differently in all experimental conditions but the TSH medium and VBNC state. Other studies also reported that the use of different software could lead to different rankings of reference genes (Cruz et al., 2009;Kong et al., 2014;Mafra et al., 2012). These differences may be caused by the use of different statistical algorithms (Andersen et al., 2004). NormFinder ranked the most stable genes with minimal inter-and intra-group variation and geNorm selected the top two genes with the highest similarity in expression and the lowest intragroup variation (Andersen et al., 2004;Vandesompele et al., 2002). Thus, coregulated genes should be excluded when using geNorm to screen reference genes.

| CON CLUS ION
In summary, we systematically screened C. michiganensis reference genes for use in real-time qRT-PCR. The gyrB and gapA genes stably expressed in most of the experimental conditions except in 0.85% F I G U R E 4 Relative expression levels of two Clavibacter michiganensis pathogenicity genes in TSH medium. The sample collected at the log phase in TBY broth was used as control. The normalization factor (NF) was calculated as the geometric mean of C t values of three most stable genes (pbpA, tufA, and gyrB) in tomato seedling homogenate. The relative expression levels of celA and chpC were normalized to NF (solid line, blank symbol) and the least stable gene gapA (dash line, dark symbol), respectively. The fold change in gene expression was calculated as-ΔΔC t . Error bars show standard deviation calculated from three biological replicates. Asterisks represent significant differences between the relative expression levels of pathogenicity genes normalized to NF and gapA according to a two-sided one sample t test (p < .05) NaCl solution and TSH medium. This highlights the need to empirically assess the suitability of reference genes under different experimental conditions before use. We found that gyrB, bipA, and gapA were suitable for artificial nutrient conditions. On the other hand, pbpA, tufA, and gyrB were suitable for use as reference genes in pathogen-host interaction studies. The combination of bipA and rpoB and the combination of gyrB, pbpA, and rpoB would be useful for nutrient starvation and VBNC experiments, respectively. Because of the similarity of C.
michiganensis and other phytopathogenic bacteria, this study provides useful information on the reference genes selection for other researches, especially for studies related with plant pathology.

CO N FLI C T O F I NTE R E S T S
The authors declare no conflict of interests.

E TH I C A L A PPROVA L
None required.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data are included in the main manuscript and in the appendices.
Raw data and materials are available on request.