Selection of reference genes for normalization of mitochondrial gene expression by qRT-PCR in different potato tissues and during anther development

Abstract

Background

Potato is the most widely grown tuber crop worldwide and a staple food in many countries: it has become the focus of many molecular breeding studies. One topical area is breeding potato seeds, especially advancing male sterile plants, focusing on developing cytoplasmic male sterility (CMS) as a breeding tool. A major obstacle has been the identification of mitochondrial genes for CMS. Quantifying the expression of candidate CMS genes is a critical aspect needed for the validation of gene expression levels for all organisms, and quantitative real-time polymerase chain reaction (qRT-PCR) is a powerful tool for this purpose. However, selecting appropriate internal control genes for normalization of mitochondrial gene expression presents specific challenges. The aim of this study was to identify suitable reference genes best suited for analysis of mitochondrial gene expression in different tissues and developmental stages of potato, particularly in developing anthers.

Results

We assessed the expression of eighteen candidate internal control genes, including four previously studied nuclear reference genes and fourteen mitochondrial candidate reference genes. By studying gene expression in a range of tissues, the genes nad1 and nad2 were found to be the most stable reference genes, since they were expressed most consistently using four different analytical tools, GeNorm, Delta Ct, Bestkeeper and NormFinder. In contrast, expression levels of the conventional nuclear reference genes were more variable. The comprehensively ranked first candidate gene, nad2 is proposed as the preferred choice as a reference gene, especially when studying different stages of anther development. Notably, actin, the most widely used marker expression gene, worked well in some cases, but there was significant variation in its rankings, for example, using the Bestkeeper tool it was ranked sixth.

Conclusions

The results indicate that nad1 and nad2 respectively were the most stably expressed marker genes in 8 different tissues and stages of anther development. This study provides valuable support for future research on mitochondrial gene expression in potato, specifically for identifying patterns of expression of CMS genes, and can be a valuable tool to quantify gene expression for other Solanaceae species.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-025-07301-3.

Background

With the increasing world population, demand for food is also increasing. Hybrid breeding is a strategy that can help increase crop yields, and has been used successfully to increase the yields of crops such as maize and rice [1, 2].

The potato (Solanum tuberosum L.) is a vital tuber crop, and is a staple food for approximately 1.3 billion people worldwide [3, 4]. Unlike most seed crops, cultivated potato is normally propagated clonally through tubers [5]. However, tuber propagation has the disadvantage of exposing plants to various diseases, especially viruses, during the bulking up of seed potatoes. Additionally, most cultivated potato species exhibit tetrasomic inheritance, complicating the breeding process. To address these challenges, several research groups advocate re-inventing potato as an inbred-line-based diploid crop. This approach aims to simplify the genome and promote propagation by seeds rather than tubers, so reducing the opportunity to accumulate diseases [6–8]. In this context, Zhang et al. [3] used genome design to develop pure, fertile diploid potato lines, resulting in uniform and vigorous F₁ hybrids. This innovation can transform potato breeding from a slow, non-accumulative process into a rapid, iterative one. Importantly, this research provides inbred-lines, suitable materials, for potato breeding, heralding a ‘green revolution’ in the potato industry and facilitating the advance of hybrid potato breeding through seed propagation [9].

Hybrid potato breeding involves significant work, particularly the emasculation of female parental lines for cross-pollination. Manual emasculation, mechanical emasculation or chemical treatments used in hybrid breeding programs are costly, inefficient, and chemical emasculation agents may be environmentally damaging [10]. These methods may also allow some self-pollination, reducing hybrid seed purity and so reducing seed hybrid quality. The availability of male-sterile plants is addresses these issues. In addition, male sterility is crucial for studying heterosis in hybrid crops, and to enable large-scale production of hybrid seeds [11]. Cytoplasmic male sterility (CMS), a maternally inherited inability to produce functional pollen, occurs in over 200 species of higher plants. CMS lines lack functional ‘restorer-of-fertility’ genes (Rf) in the nucleus [12, 13]. At least 29 cm genes have been identified in 13 crop species [10, 13, 14]: these genes are often chimeric, resulting from the recombination of mitochondrial genomes [15–17].

CMS has been studied extensively in monocotyledonous plants, particularly rice. The major CMS types in rice include the wild abortive type (CMS-WA), the Boro II type (CMS-BT), and the Honglian type (CMS-HL). The BT-type was the first CMS system identified in rice [18, 19]. Genetic transformation experiments later confirmed orf79 as the CMS-causing gene in BT-type rice, and the restorer genes Rf1a and Rf1b were found to restore fertility by suppressing orf79 expression [20]. In the HL-type system, the CMS-associated gene is orfH79, which shares a high sequence similarity with orf79, differing only by five SNPs in the coding region, leading to five amino acid substitutions in its translated protein [21, 22]. For the WA-type, the sterility-inducing gene is WA352, is composed of four DNA fragments (284s, cs3, cs2, and cs1) and encodes a 352-amino acid protein [23]. Although some CMS genes have been identified in other plants, they have rarely been reported in potato.

Quantitative real-time PCR (qRT-PCR) is still one of the most widely used methods to assess gene expression. A crucial aspect of this method is normalization, which mitigates technical variability arising from factors such as differences in sample size, pipetting inaccuracies, and sample quality [24]. Current evidence suggests that no gene can be universally used as a reference gene, emphasizing the need for systematic validation of reference genes [25, 26]. For instance, GAPDH (glyceraldehyde-3-phosphate dehydrogenase), a commonly used reference gene and is referred to as “classical.” While it provides good results in many studies, it is not recommended in others due to variability of expression caused by varying experimental factors [26]. This variability highlights the necessity of validating candidate reference genes under specific experimental setups and using multiple analytical tools to ensure accurate normalization.

In addition, reference genes chosen for normalization in expression studies should take into account the origin of the mRNA type to minimize bias resulting from differences in extraction efficiency, reverse transcription, or PCR amplification [27]. Historically, most reference genes have been nuclear genes, which are unlikely to be useful for mitochondrial gene expression studies. Alternatively, reference genes which target microRNAs (miRNAs) have also been studied [27]. In the tea plant (Camellia sinensis), miR159a was found to be the best single reference gene from the bud to the fifth leaf, 5S rRNA was the most suitable gene in different organs, and miR6149 was the most stable gene when leaves were attacked by Ectropis oblique [28]. In sweet potato (Ipomoea batatas L.), a combination of miRn60 and miR482 was used as reliable reference genes across four tissues and two cultivars under drought and salt stress treatments [29].

Investigating expression patterns of mitochondrial genes could provide clues to identify CMS genes. Because of the relative cost of sequencing, assembly and annotation of mitochondrial genomes, qRT-PCR is still used widely as a sensitive technique for quantifying levels of mitochondrial gene expression. Its accuracy depends on the reference genes used for data normalization. It is essential to use a reference gene with stable expression and wide applicability for measuring relative patterns of CMS gene expression. However, no mitochondrial reference genes have been reported so far for potato.

In this study, we aimed to identify reliable mitochondrial reference genes for normalization of qRT-PCR data in potato, focusing on the characterization of CMS genes. Fourteen mitochondrial genes, including four types of adenosine triphosphate (ATP1, ATP4, ATP6, ATP9), apocytochrome b gene (cob), subunit 2 of cytochrome c oxidase (cox2), five types of NADH dehydrogenase subunit (nad1, nad2, nad3, nad5, nad6), genes for three cytoplasmic ribosomal protein (rps3, rps4, rps19), and four previously used nuclear reference genes - Elongation factor 1-alpha (EF1-α), Actin, exocyst complex component (Sect. 3), and tubulin, were selected as candidate reference genes. The stability of their expression was determined in eight different tissues - anthers, roots, stems, leaves, tubers, petals, stigmas, stolons, including anthers at different developmental stages, and systematically evaluated using the programs GeNorm, NormFinder and BestKeeper, and Delta Ct. A comprehensive analysis of reference gene stability resulted in identification of the most stable reference gene(s) for corresponding experiments.

Materials and methods

Plant materials

A diploid potato line derived from code number BS 278 provided by the United States Department of Agriculture (USDA) was used in this study. The plants were grown in a growth chamber under long-day conditions (16 h light/8 h dark at 25 °C) to produce seedlings and flowers. Plant tissues were collected and divided into two groups. The first combined eight different fresh tissues, including petal, stigma, stem, root, tuber, leaf, stolon, and anthers. To study reference genes during anther development, six anther developmental stages were collected, primarily covering tapetum development and microspore release (Fig. 1). All these tissues were frozen immediately in liquid nitrogen and stored at − 80 °C until RNA extraction and analysis of target gene expression.

Fig. 1.

Samples collected at different stages of anther development

RNA extraction and cDNA synthesis

RNA extraction was done using TRIzol™ Reagent (Code No 15596018, Thermo Fisher Scientific, USA) following the manufacturer’s guidelines. The quality and quantity of extracted RNA was assessed using a Nanodrop Spectrophotometer. To generate the first-strand cDNA, a HiScript III 1 st Strand cDNA Synthesis Kit (+ gDNA wiper) (Code No R312-02, Vazyme, Nanjing, China) was used with 1 ug of total RNA. Both oligo (dT) primers and random primers were used in each PCR reaction mixture. Before obtaining first-strand cDNA, genomic DNA contamination was removed from isolated RNA using DNase provided in the kit. The cDNA was stored at − 20 °C until use.

Reference gene selection

Based on previous studies [30], the four nuclear reference genes and 14 mitochondrial genes selected were: Sect. 3, EF1-α, Actin, tubulin, and ATP1, ATP4, ATP6, ATP9, cob, cox2, nad1, nad2, nad3, nad5, nad6, rps19, rps3 and rps4. . Target-specific primers of 14 mitochondrial genes were designed using Primer-BLAST [31] with a melting temperature (Tm) between 57 °C and 63 °C, primer length of 19–21 nucleotides, and amplicon size between 78 and 126 bp. The names and sequences of primers of candidate reference genes are provided in Table 1. Primer specificity was confirmed through agarose gel, Sanger sequencing and melting curve analysis of qRT-PCR reactions.

Table 1.

Candidate reference genes and primer sequences

Gene IDs	Primer sequence (5’−3’)	Primer Length (bp)	GC content (%)	Product Length (bp)	R 2	E (%)	Resources
ATP1	F: TGGTCTCAGTTGGGGATGG	20	55.0	86	0.999	111.4	This study
	R: ACACCGCTGGCAAATTCAAC	21	47.6
ATP4	F: CGGTGTAGCTCGAAAGCAGA	21	52.4	84	0.999	101.6	This study
	R: AAGAGAATCCCCCACCCGAA	21	52.4
ATP6	F: TTCGTGCTGAACCCGGTAAA	21	47.6	86	0.999	104.3	This study
	R: AAAGTGACCGAGATGCGAGG	21	52.4
ATP9	F: CTTCAGCGGGAGCTGCTATC	21	57.1	78	0.997	103.7	This study
	R: CCAATGATGGATTTCGCGCC	21	52.4
cob	F: TGGGTTCTCCGTGGACAATG	21	52.4	99	0.999	100.5	This study
	R: GCGGCCAGATGAAGAAGACT	21	52.4
cox2	F: CTCGTCCCATACCTTCTGCC	21	57.1	100	0.992	104.8	This study
	R: TCTCACCCAGCCCTACCTAC	21	57.1
nad1	F: TATGGGTCCGTGCAGCATTT	21	47.6	108	0.998	98.0	This study
	R: CACCAGAAACGGGGACTACC	21	57.1
nad2	F: GGCTAACGGGGGTATTCCTG	21	57.1	125	0.999	100.0	This study
	R: TAGCATTACGGCAAACCCGT	21	47.6
nad3	F: AGTGATCAGCCCGCTAGTTTC	22	50.0	123	0.992	105.0	This study
	R: GCATCACCGGAAGGATCGAA	21	52.4
nad5	F: AAAGGGAACGAGGAGGCAAG	21	52.4	87	0.994	96.7	This study
	R: ATTCCTGAGTGCAGGTTCGG	21	52.4
nad6	F: ACGGTTTATGCCGGAAAGGT	21	47.6	126	0.997	98.9	This study
	R: AGCCCCAATCATGGCTACTA	21	47.6
rps19	F: CGGAATTCGTTGATTGCTCCG	22	50.0	126	0.998	92.2	This study
	R: TCGAAGGTCTTCGTTTCCGTG	22	50.0
rps3	F: GTGCTTCTCCGATTGCTCAAG	22	50.0	122	0.992	101.2	This study
	R: CCCCTCCACCCCCTTTTTC	20	60.0
rps4	F: TCAAGCAAGGCAGCCGATAA	21	47.6	124	0.996	103.1	This study
	R: GCGGGTTCTCGCATCATTTT	21	47.6
Actin	F: AGGAGCATCCTGTCCTCCTAA	22	50.0	180	0.998	96.9	[30]
	R: CACCATCACCAGAGTCCAACA	22	50.0
EF1-α	F: GATGGTCAGACCCGTGAACA	21	52.4	106	0.997	106.2	[30]
	R: CCTTGGAGTACTTCGGGGTG	21	57.1
Sect. 3	F: GCTTGCACACGCCATATCAAT	22	45.5	160	0.960	107.0	[30]
	R: TGGATTTTACCACCTTCCGCA	22	45.5
tubulin	F: GGGAATAACTGGGCGAAAGGT	22	50.0	134	0.994	92.3	[30]
	R: CCTCCACCAAGTGAGTGACAA	22	50.0

F forward primer, R reverse primer, E PCR efficiency, R²regression coefficient.

Real-time polymerase chain reaction

Real-time PCR was performed using SYBR Green Premix Pro Taq HS qPCR Kit III (High Rox Plus) (Code No AG11738, Accurate Biotechnology, Hunan, China). To assess the specificity of the primers, cDNA was synthesized from different tissue samples, each using 1000 ng of total RNA. Equal volumes of each cDNA sample were pooled to create a mixed cDNA template. This template was then serially diluted five times (5⁰, 5¹, 5², 5³, 5⁴, 5⁵) using nuclease-free water, and these dilutions were used for melt curve analysis. A negative control using distilled water instead of cDNA was included at every stage of the experiment. The respective qRT-PCR efficiencies (E) (Table 1) for each gene were calculated based on the slope. Each reaction consisted of a 10 µL volume, and two-step PCR was used. qRT-PCR was performed on an Applied Biosystems StepOnePlus™ Real-Time PCR System, using the following thermal cycling conditions: initial denaturation at 95 °C for 30 s, followed by 40 cycles of 95 °C for 5 s and 60 °C for 1 min. A melt curve analysis was conducted by holding at 95 °C for 15 s, cooling to 60 °C for 1 min, and then increasing the temperature to 95 °C at a ramp rate of 0.3 °C/s with a final 15 s hold. Each sample included three biological replicates, and each biological replicate was done in triplicate for technical replication.

Data analysis

To identify the most stable gene expression values, GeNorm [32], BestKeeper [33], NormFinder [34] and Delta Ct [35] tools were used. BestKeeper and Delta Ct directly employ the Ct value for stability analysis, avoiding an additional conversion step. These two tools estimate the standard deviation values (SD) and variation coefficient (CV) of each reference gene based on their Ct values. The reference genes with the lowest SD and CV are considered the most stably expressed. In contrast, GeNorm and Normfinder tools transform raw Ct values into relative quantities through the formula 2^−ΔCt (ΔCt = each corresponding Ct value − lowest Ct value). GeNorm introduces expression stability values (M) to investigate the most stable reference gene. As for GeNorm, the stability value (SV) is applied to NormFinder to estimate expression variation among the tested candidate reference genes. The final ranking of candidate reference genes was determined by calculating the average rank across the four algorithms. Reference genes were then ranked in ascending order based on these mean values, with the lowest average rank assigned the highest final rank.

Results

Tissues from eight stages of potato growth and development were analyzed, together with six stages of anther development, to identify the most stable reference genes for study of expression patterns of mitochondrial genes, especially CMS genes. PCR efficiency and correlation coefficient data for the candidate reference genes are summarized in Table 1. The results indicate that PCR efficiency for the candidate genes tested ranged from 92.3 to 111.4%, with an R² between 0.992 and 0.999. According to the information for publication of quantitative real-time PCR experiments (MIQE) guidelines, PCR efficiency should be in the order of 80% ≤ E ≤ 120% to ensure the reliability of experiments [36]. The additional test of melt curve analysis is usually done to confirm the specificity of primer annealing. All 18 primer pairs produced single, specific bands on agarose gel using pooled DNA from all listed samples as shown in Fig. 2-a, and the PCR products were confirmed to match the target sequences by Sanger sequencing. Also, as illustrated in Fig. 2-b, the melt curve exhibited a singular sharp peak, indicating that PCR amplification was specific for each of the genes studied.

Fig. 2.

Agarose gel electrophoresis and melting curve analysis of candidate reference genes in experimental samples. a, Representative agarose gel electrophoresis of PCR products for each candidate reference gene. b, Melting curve analysis of 18 candidate reference genes

Analysis of expression stability of selected reference genes

Absolute Ct values in qRT-PCR reflect transcript abundance, and the results obtained show the stability of expression of the candidate reference genes. As shown in Fig. 3, overall, the mean Ct values of the reference genes ranged from 17.96 to 24.90 in all groups. This analysis shows that the most highly expressed candidate reference gene was nad1, which is involved cellular energy metabolism [37], with a mean Ct of 17.96 in 8 different tissues, and 18.27 in anther developmental stages. The three nuclear reference genes, actin, Sect. 3 and tubulin, had a relatively lower expression level: Sect. 3 had the highest mean Ct value in all eight stages of potato development, and in the stages of anther development studied. The Coefficient of Variance (CV) as anther standard was used to assess the variability of experimental replicates and further visualize the fluctuation in the expression of reference genes. As shown in Fig. 3-a, in group1, nad3 had the lowest expression, CV of 2.96%, closely followed by ATP4 at 3.53%, nad5 at 3.54% and ATP1 at 3.57%. Interestingly, the two conventional nuclear reference genes had fairly high CVs (tubulin at 6.40% and EF1-α at 5.68%). However, in group 2 (Fig. 3-b), the nuclear candidate reference genes displayed better stability (actin at 1.02%, Sect. 3 at 1.38% and EF1-α at 5.68% CV), with tubulin the least stable at 5.00% CV.

Fig. 3.

Expression levels of candidate reference genes in experimental samples. a, Expression levels of candidate reference genes in Group (1) b, Expression levels of candidate reference genes in Group (2) Expression data are displayed as Ct values for each reference gene in all samples. The line across each individual box indicates the median value, and the box limits indicate the 5th and 95th percentiles. ‘×’ represents the maximum and minimum values. Points represent the average. CV calculated using the formula: CV (%)= [Standard Deviation/Mean (Average Ct)]×100%

Four reliable statistical analytic tools, NormFinder, Detal Ct, GeNorm and BestKeeper were applied to evaluate the stability of gene expression. These tools were employed to validate the results and gain a deeper understanding of the dataset. By applying these statistical tools, we evaluated the stability of mitochondrial genes as reference genes during anther development, providing crucial insights for the relatively quantitative analysis of CMS genes, thereby enhancing the identification of CMS-associated genes.

GeNorm analysis

GeNorm analysis determines the normalization value based on the geometric mean of various candidate reference genes and mean pairwise variation of each gene from all the reference genes in a given set of samples. To determine the optimal number of reference genes, pairwise variation (Vn/Vn + 1) was calculated. For both experiment sets, the V2/3 values were below the recommended threshold of 0.15, indicating that two reference genes are sufficient for accurate normalization (Fig. 4). To identify the most stable reference genes, the internal control gene-stability measure stability value (M) was defined as the average pairwise-to-decreasing expression stability. A lower M value indicates higher stability, with a recommended cut-off of less than 1.5 [30, 32, 36]. We analyzed our data and found that all 18 candidate reference genes exhibited high expression stability with low (< 0.9) M values among the two experimental sets, and these were much lower than the default threshold of 1.5 (Fig. 5). More specifically, in the first group of sample tissues, the results indicate that the combination of two mitochondrial genes nad5 and nad6 were the most stable reference genes with M values of 0.276, followed by nad2, at nearly 0.4, and the third, cob, with M values 0.462. Interestingly, the M values of four nuclear candidate genes were either the highest M (tubulin at 0.836) or a significantly higher M (Actin at 0.64, Sect. 3 at 0.675, EF1-α at 0.709). In the group of samples from anthers, cob and nad1 exhibited the lowest M value, and tubulin had the highest M value, indicating that cob and nad1 were most stably expressed and tubulin the least.

Fig. 4.

Pairwise variation (Vn/Vn + 1) analysis of the optimal number among 18 candidate reference genes in different experimental sets

Fig. 5.

Stability values of candidate reference genes calculated by GeNorm and NormFinder in two groups. a, Group 1: different tissue types. b, Group 2: different anther development stages

NormFinder analysis

NormFinder calculates the SV using an ANOVA-based model to estimate expression variation among the tested candidate genes. A higher SV indicates lower stability. By considering both intra-group and inter-group variations, NormFinder ranks the candidate genes based on their stability [34, 38]. The outcome of the NormFinder analysis for the 18 candidate reference genes is provided in Fig. 5. There was a significant gap between the top and bottom of SV in the first group of different tissues. The results highlight nad1 as the most stable gene, with a stability value of 0.216, surpassing the others. ATP1 and rps19 had close stability values, 0.241 and 0.249, respectively. In contrast, the four nuclear candidate reference genes exhibited poorer expression stability in all of these samples. However, in the second group, the overall SV for each candidate gene was relatively low, with the lowest SV of 0.282 for tubulin. The distinct difference from the first group is that the nuclear gene Actin displayed the minimum SV with 0.074, indicating that Actin was most stably expressed.

Delta Ct (ΔCt) analysis

The ΔCt method is used to identify useful reference genes, by comparing the relative expression of ‘pairs-of-genes’ in each sample. Genes remaining constant in distinct samples are regarded as expressed stably. However, fluctuations in ΔCt values suggest that one or both genes may have variable expression. Introducing a third, fourth, or even fifth gene into the comparisons allows a deeper analysis, revealing which pairs demonstrate less variability. In turn, this helps identify the gene(s) with stable expression across the tested samples, enabling a process of ranking or discarding based on the results obtained. Expression stabilities are determined from the mean SDs [35]. Data analysis using the ΔCt method suggested that the five most stable genes in the eight tissue samples under normal conditions were nad1 > nad2 and rps19 > ATP1 > cob, which were all mitochondrial genes (Fig. 6). The rank of all the nuclear candidate reference genes was lower. However, for the different anther development stages, Actin, a potential nuclear candidate reference gene, exhibited the smallest mean SD, together with nad2 and ATP9. Because of their similar expression stability, the ΔCt method these three genes shared the same ranking. To ensure reliability, we further validated the selection using additional statistical tools. However, tubulin was ranked last. These results are similar to those from GeNorm and NormFinder - tubulin exhibited the poorest stability among the two experiment sets.

Fig. 6.

Mean SDs of candidate reference genes calculated by ΔCt in two groups

BestKeeper analysis

BestKeeper evaluates the stability of candidate genes by analysing several parameters, including SD, CV, and Pearson correlation coefficient (r). Genes exhibiting the highest r value and an SD > 1 are considered to be the most stably expressed. The number of candidate genes that can be analysed at one time is limited to ten candidate genes by BestKeeper. Hence, in the first group, eight candidate genes, including all four nuclear candidate genes and rps4, rps3, ATP4 and ATP9, were eliminated by the results from GeNorm, NormFinder and Delta Ct, as they consistently ranked the lowest in all three software tools (Table 2). nad1 was considered the most stable reference gene in the eight different tissues, with the highest r value at 0.952 and SD at 0.64. nad2 was placed second, demonstrating strong stability in all evaluations with rps19 third, followed by cox2. ATP6 occupied the fifth position in terms of stability across the tools. Using the same analytical approach for the second group of samples, the eight candidate genes - cox2, ATP4, ATP6, nad3, nad5, rps3, rps4, and tubulin were removed from the BestKeeper analysis. nad1 was also the most stably expressed in the different anther tissues/developmental stages, with the highest r value at 0.924 and SD at 0.39, followed by cob and nad2. Notably, the ranking of Actin decreased, and it was placed sixth using the BestKeeper tool.

Table 2.

The stability of candidate reference genes evaluated by bestkeeper

Gene IDs	Group 1: Different tissue types			Group 2: Different anther developmental stages
	r	SD	CV (%)	r	SD	CV (%)
ATP1	0.813	0.51	2.60	0.790	0.29	1.53
ATP4	-	-	-	-	-	-
ATP6	0.874	0.68	3.70	-	-	-
ATP9	-	-	-	0.684	0.24	1.25
cob	0.847	0.50	2.61	0.909	0.33	1.74
cox2	0.875	0.70	3.11	-	-	-
nad1	0.952	0.64	3.54	0.924	0.39	2.11
nad2	0.897	0.55	2.86	0.836	0.25	1.26
nad3	0.800	0.50	2.28	-	-	-
nad5	0.776	0.50	2.49	-	-	-
nad6	0.849	0.56	2.89	0.769	0.37	1.89
rps19	0.894	0.68	3.34	0.700	0.28	1.42
rps3	-	-	-	-	-	-
rps4	-	-	-	-	-	-
Actin	-	-	-	0.703	0.20	0.88
EF1-α	-	-	-	0.449	0.20	1.05
Sect. 3	-	-	-	0.655	0.22	0.89
tubulin	-	-	-	-	-	-

Ordering of reference genes from the results of the four analytical tools

A comprehensive comparative ranking of the effectiveness and stability of candidate reference genes is provided in Fig. 7, which combines the ranking results from the four tools, and then ranked in ascending order based on these mean values. Despite some differences in the results obtained from these tools, three genes nad1, nad2 and rps19 emerged as stably genes expressed in different tissues, with a consistent ranking from 1 to 3 based on NormFinder, BestKeeper and Detal Ct analysis. In particular, the mitochondrial gene, nad1, was ranked first using three tools. In contrast, four previously reported nuclear reference genes from the first group were less stable as shown by NormFinder, GeNorm and Detal Ct analysis. Similarly, there was a fluctuation in their rank in tissues of different stages of anther development. In contrast, the rank of nad2 was consistent and was always in the top three for all analytical tools. It also was ranked at the top in the second group, although for this group, Actin could also be considered to be useful, since in different anther developmental stages, it ranked second in NormFinder and Detal Ct analyses.

Fig. 7.

Comprehensive ranking of candidate reference genes using the GeNorm, NormFinder, Delta Ct, and BestKeeper tools. a, Comprehensive ranking for the group including different potato tissues. b, Comprehensive ranking for the group of different anther developmental stages. ‘-’ represent not available to rank

Discussion

qRT-PCR is a widely used method to examine gene expression patterns [39]. Ensuring the precision of experiments relies on the qRT-PCR assay itself, including primer design and length, and the PCR conditions, but most importantly, on selecting reliable internal controls. It requires an appropriate reference gene to normalize the target transcript levels. While reference genes generally exhibit stable expression under standard conditions, they are involved in essential cellular processes, and their expression can fluctuate under specific circumstances, such as environmental stress or during different developmental stages [26, 40]. Additionally, the availability of suitable reference genes for accurate normalization in mitochondrial genes remains limited. In eukaryotes, poly(A) tails are present on almost every mRNA [41], whereas in plant mitochondria, polyadenylated mitochondrial transcripts are rare and unstable because polyadenylation of mRNAs in chloroplasts serves as an RNA degradation signal [42, 43]. Reverse transcriptase PCR (RT-PCR), a step before qRT-PCR is used to synthesize cDNA, but RT-PCR is usually undertaken with Oligo (dT) primers which target mRNAs with poly(A) tails. The synthesis of mitochondrial mRNAs and their patterns of expression has rarely been studied. Thus, selecting reference genes tailored for measuring mitochondrial gene expression, based on their stability and consistency of expression under different experimental conditions, is required to ensure accurate and dependable analysis of gene expression. While our current study focused on the evaluation of suitable reference genes, these results may provide useful information for future studies investigating mitochondrial gene expression, where appropriate normalisation remains critical.

Mitochondrial genome rearrangements and substoichiometric shifting (SSS) have been implicated in the expression of cytoplasmic male sterility (CMS) genes in several plant species [10, 13], highlighting the importance of accurate mitochondrial gene expression analyses in CMS studies. Additionally, as previously mentioned, qRT-PCR analysis is widely used to study gene-expression patterns, therefore it can be used to identify CMS genes in potato. However, accurate qRT-PCR analysis of mitochondrial genes is challenging because of limited flexibility of primer design and the lack of appropriate reference genes to standardize gene expression levels. If a common internal reference gene is selected without screening, it is likely to reduce the accuracy of quantitative analyses, and could result in incorrect conclusions. In this context, the identification of stable reference genes in different tissues and developmental stages may help facilitate future studies on mitochondrial gene expression related to CMS in potato. To improve the accuracy of gene-expression analyses, we studied and selected reliable internal control genes systematically for normalization of mitochondrial gene expression in a range of potato tissue and anther developmental stages.

The E values of the eighteen candidate reference gene primer pairs ranged from 92.3 to 111.4%, with R² values were 0.960–0.999 (Table 1). These results demonstrate the high accuracy, efficiency, and sensitivity of the primer pairs used for reference gene selection. In addition, the mean Ct values of the candidate reference genes ranged from 17.964 (nad1) to 24.717 (Sect. 3) (Fig. 3). Notably, the mean Ct values for four nuclear candidate reference genes were not consistent in potato under abiotic stress [26], suggesting that no reference gene is expressed equally under different conditions. Therefore, it is evident that selecting appropriate reference genes for normalization of mitochondrial genes under specific experimental conditions is required.

Interestingly, we observed that nuclear candidate reference genes such as Actin and EF1-α exhibited greater expression stability in Group 2 (different anther developmental stages) compared to Group 1 (different tissue types). This may be due to the biological specificity and relative homogeneity of the samples in Group 2. During anther development, especially when carefully staged, cellular composition changes in a coordinated and gradual manner. This developmental continuity can result in more consistent expression of housekeeping nuclear genes such as Actin and EF1-α, which are involved in essential cellular processes like cytoskeleton maintenance and translation. In contrast, the tissues included in Group 1 differ substantially in structure, function, and cellular composition, which may contribute to greater variability in nuclear gene expression. Previous studies have demonstrated that reference gene stability is highly context-dependent, and nuclear genes often show higher stability within a single organ across developmental stages than across multiple tissue types [44]– [45]. These findings are consistent with our observation that nuclear genes displayed greater expression stability during anther development.

Four well-established algorithms (Delta Ct, GeNorm, NormFinder and BestKeeper) were used to identify stably expressed reference genes [46]. The operating principle of NormFinder is similar to that of the GeNorm program; however, GeNorm has the added capability to identify optimal reference gene combinations and determine the ideal number of reference genes. In contrast to GeNorm and NormFinder, BestKeeper and Delta Ct software directly make calculations using Ct values [47]. The results of NormFinder and Delta Ct were generally similar in this study (Fig. 7) as the top two candidate reference genes (nad1 and nad2) selected by these two algorithms remained consistent in all tissues studied, whereas four nuclear candidate reference genes did not meet the required criteria in eight different tissue types of potato. Overall, nad2 was identified as the best candidate reference gene for different anther developmental stages, although Actin also performed well, albeit with more variation. These result show that selection of reliable reference genes requires analysis of expression under different conditions or treatments. It should be clearly stated that, although variations among the four widely used analytical tools (GeNorm, Delta Ct, BestKeeper, and NormFinder) were observed due to differences in their underlying algorithms, the overall results consistently identified suitable reference genes, underscoring the robustness of the selection.

Conclusions

This is the first detailed study to select the best candidate mitochondrial reference genes which can be used in qRT-PCR expression studies in different tissues and developmental stages of potato to normalize gene expression. The results show that the most suitable reference gene is nad1 for expression studies in eight potato tissues, whereas nad2 is the most appropriate gene for normalizing expression in developing anthers. Overall expression of nad2 was most stable in all of the tissues studied.

Abbreviations

(CMS)

Cytoplasmic Male Sterility

(qRT-PCR)

quantitative Real-Time Polymerase Chain Reaction

(MIQE)

Minimum Information for publication of Quantitative real-time PCR Experiments

(ATP)

Adenosine Triphosphate

(cob)

apocytochrome b gene

(cox2)

subunit 2 of cytochrome c oxidase

GAPDH

Glyceraldehyde-3-Phosphate Dehydrogenase

(nad)

NADH dehydrogenase subunit

(rps)

cytoplasmic ribosomal protein

(EF1-α)

Elongation Factor 1-alpha

(sec3)

exocyst complex component

(Tm)

melting Temperature

(E)

PCR efficiency

(CV)

Coefficient of variance

(ΔCt)

Delta Ct

(SD)

Standard Deviation

Authors’ contributions

QL and LY contributed equally to this article. Study conception and design: LY, QL, JX; data collection: JX, LY; analysis and interpretation of results: QL, JX, MJ; draft manuscript preparation: MJ, QL. All authors reviewed the results and approved the final version of the manuscript.

Funding

This work was funded by a grant to Li Yuan from the China Postdoctoral Science Foundation ID: 2024M753583, and a Murdoch University International PhD Scholarship to QL.

Data availability

Ct values, together with the raw outputs from the four stability analysis algorithms (BestKeeper, ΔCT, NormFinder, and geNorm), are provided in Supplementary datasets in an Excel file: Group 1 (different tissue types) and Group 2 (different anther development stages). The qPCR amplicons are less than 200 bp and so do not have NCBI numbers: these data can be obtained from author Li Qing on request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

All authors consent to this publication.

Competing interests

The authors declare no competing interests.