Sweet osmanthus (Osmanthus fragrans), a member of the Oleaceae family, is renowned as one of the ten most famous traditional flowers in China. It is highly valued for landscaping and is extensively utilized in food, cosmetics, and traditional Chinese medicine (Fu et al., 2022; Shi et al., 2024; Zhou et al., 2024). Despite its adaptability to various environmental conditions, O. fragrans is increasingly subjected to abiotic stresses, such as extreme temperatures, drought, and salinity (Li et al., 2020; Zhu et al., 2022; Yue et al., 2024b), and metal ion pollution, all exacerbated by global climate change and environmental degradation (Dang et al., 2022, 2024; Yue et al., 2024a). To address these challenges, it is crucial to investigate the molecular mechanisms underlying plant responses to these stress conditions. Currently, molecular biology research on O. fragrans predominantly focuses on functional gene analysis to elucidate its growth, development, and resistance mechanisms. However, no studies have yet been reported on long non-coding RNAs (lncRNAs) in this species.

LncRNAs are a class of RNA molecules longer than 200 nucleotides that do not encode proteins. Recently, lncRNAs have garnered considerable attention for their crucial roles in regulating gene expression (Prabakaran, 2016; De et al., 2024). Unlike traditional protein-coding genes, lncRNAs regulate gene expression through various mechanisms, including the modulation of chromatin structure, transcription, RNA processing, and translation (Bridges et al., 2021; Cui, 2023). In plants, lncRNAs are essential for developmental regulation (Li et al., 2015) and stress responses (Wu et al., 2019; Jia et al., 2023; Jin et al., 2024), encompassing responses to abiotic stress, pest and disease defense, and hormonal regulation.

Gene expression analysis is a widely used technique for investigating gene function, providing insights into the molecular processes that govern plant organ development and stress resilience. Techniques such as microarray, northern blotting, semiquantitative reverse transcription-PCR (RT-PCR), and quantitative reverse transcription-PCR (qRT-PCR) are commonly employed in these studies. Among these, qRT-PCR is particularly noteworthy for its high sensitivity, specificity, speed, and throughput, making it an essential tool for the relative quantification of gene expression (Hu et al., 2009; Zhang et al., 2018; Ni et al., 2019). The accuracy of qRT-PCR depends heavily on the stability of housekeeping genes used for data normalization. Key factors influencing this accuracy include primer specificity, RNA quality and integrity, reverse transcription efficiency, amplification efficiency (E value), and the quantity of initial material (Klein, 2002; Fleige and Pfaffl, 2006). Therefore, selecting stably expressed housekeeping genes as reference genes (RGs) is essential for error correction and standardization.

Typically, genes involved in fundamental cellular functions are selected as RGs, including 18S rRNA (18S) (Zhang et al., 2021b), actin (ACT) (Liu et al., 2016; Zhang et al., 2021b), cyclophilin (CYP) (Li et al., 2024), elongation factor 1 beta (EF1B) (Liu et al., 2016), glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Wang et al., 2024a), ras-related nuclear protein 1 (RAN1), α-tubulin (TUA) (Li et al., 2024), and ubiquitin (UBQ) (Tang et al., 2023). However, increasing experimental evidence indicates that RGs may not be as stable as previously believed under various conditions, and ideal RGs consistently expressed across all experimental conditions are rare (Jain et al., 2006; Løvdal and Lillo, 2009). The stability of RG expression can vary significantly across different species, stresses, and tissues. For example, CYP, ACT, UBC, and 18S are reliable for studying gene expression in Chinese cedar (Cryptomeria fortunei) under abiotic stress, hormone treatments, and various tissues (Zhang et al., 2021b). In contrast, ACT exhibits high instability in wild barley (Hordeum brevisubulatum) under diverse stress treatments, including low and high temperatures, salt, drought, abscisic acid (ABA), gibberellic acid 3 (GA₃), and ethephon (Zhang et al., 2018). Similarly, 18S is the most unstable RG in hardy rubber tree (Eucommia ulmoides) under abiotic stress and across various tissues (Ye et al., 2018), and CYP2 exhibits high instability in switchgrass (Panicum virgatum) under cadmium, mercury, chromium, and arsenic stresses (Zhao et al., 2020). Therefore, evaluating the expression stability of RGs under specific experimental conditions is essential before their application. Several statistical algorithms, including delta-Ct (Silver et al., 2006), geNorm (v3.5) (Vandesompele et al., 2002), NormFinder (v0.953) (Andersen et al., 2004), BestKeeper (v1.0) (Pfaffl et al., 2004), and Refinder, have been developed to assess and rank the stability of candidate RGs. However, these methods have limitations in terms of applicability and accuracy, particularly for lncRNA expression studies. Since lncRNA expression levels are generally lower than those of mRNA (Wu et al., 2014; Chen et al., 2018), using protein-coding genes as internal RGs may not accurately reflect lncRNA expression levels. To date, no systematic evaluation of RGs for lncRNA studies in O. fragrans has been reported. This gap highlights the necessity of this study to identify reliable RGs for accurate lncRNA expression analysis in O. fragrans.

In the initial stages of our experiment, our research team performed high-throughput sequencing of lncRNAs during the flower opening and senescence stages of O. fragrans. Using the fragments per kilobase of transcript per million mapped reads (FPKM) method to calculate expression levels, we selected 10 highly expressed and stable lncRNAs as candidate RGs ( Supplementary Table S1 ), along with 7 common used internal RGs. We then employed qRT-PCR to analyze the expression stability of these lncRNAs under various conditions, including abiotic stress (low temperature, drought, and salt), hormone treatments (ABA, methyl jasmonate (MeJA), and ethephon), metal ion treatments (Fe²⁺, Al³⁺, and Cu²⁺), different tissues (stem, root, seed, leaf, and flower), and during flower opening and senescence stages. This analysis utilized the delta-Ct, geNorm, NormFinder, BestKeeper, and RefFinder systems, and identified appropriate internal RGs for each condition, providing reliable RGs for quantifying lncRNA expression in O. fragrans under different environmental conditions. This study establishes a scientific foundation and technical support for research on stress resistance in O. fragrans and its applications. Additionally, these findings provide a reference for lncRNA functional studies in other plants, contributing to advancements in plant stress resistance breeding and improving environmental adaptability.

Gene expression analysis is essential for understanding gene function, particularly in biological research (Bourgeois et al., 2016; Torres-Rodríguez et al., 2024). Among the various methods available for detecting gene expression levels, qRT-PCR is a widely employed technique for examining gene expression patterns (Wei et al., 2020; Zhang et al., 2021b). However, achieving accurate qRT-PCR results can be challenging due to limitations in primer design flexibility and the lack of suitable RGs for standardizing gene expression. Previous studies have indicated that lncRNAs are generally expressed at lower levels compared to mRNAs (Wu et al., 2014; Chen et al., 2018). Consequently, using protein-coding genes as internal RGs for studying lncRNA expression may not accurately reflect their true expression levels. The use of common RGs without rigorous screening can compromise the accuracy of quantitative analyses and potentially lead to erroneous conclusions. Therefore, it is essential to meticulously screen RGs before their use to ensure the precision of qRT-PCR analyses (Huang et al., 2014; Wei et al., 2020; Zhang et al., 2021b; Wang et al., 2024a). Currently, there are no reports on lncRNA in O. fragrans. To enhance the accuracy of gene expression studies, we systematically screened for reliable internal RGs to normalize lncRNA expression across various tissues of O. fragrans and under different stress conditions. This approach aims to improve the reliability of gene expression studies and provide more accurate insights into gene function in this species.

In this study, the E values for the candidate RG primers ranged from 91.216% to 111.998%, with an R² ≥ 0.973 ( Table 1 ). These results indicate that the primers used for screening the RGs exhibit high accuracy, efficiency, and sensitivity. Additionally, the average Ct values of the candidate RGs ranged from 10.556 (18S) to 30.580 (lnc00174850) ( Figure 2 ). Among these, lnc00031789 exhibited the greatest variation in Ct values (15.664 Ct), ranging from 22.301 to 37.965, while 18S showed the smallest variation (3.820 Ct), with maximum and minimum Ct values of 12.496 and 8.676, respectively ( Figure 2 ). These findings align with previous studies on species such as C. fortunei (Zhang et al., 2021a, b), gerbera flowers (Gerbera hybrida) (Li et al., 2022), dogbane (Apocynum venetum) (Li et al., 2024), tomato (Solanum lycopersicum) (Løvdal and Lillo, 2009), soybean (Glycine soja) (Liu et al., 2016), and arrowheads (Sagittaria trifolia) (Tang et al., 2023), where candidate RGs exhibited varying expression levels across different test materials. Although the expression level of 18S was relatively stable across all samples, it was not consistent under different stress treatments or in different tissues ( Figure 2 ). Therefore, the selection of appropriate RGs for normalizing lncRNA expression under specific experimental conditions is crucial.

We employed four commonly used algorithms (delta-Ct, geNorm, NormFinder, and BestKeeper) to evaluate and identify stable RGs. The results revealed that the top five genes selected by the different algorithms were generally consistent ( Supplementary Figure S1 ). For example, under low-temperature stress, the algorithms consistently identified GAPH, lnc00249739, and lnc00265419 as relatively stable genes ( Supplementary Figure S1 ). Similarly, under salt stress, abiotic stress, MeJA treatment, Fe²⁺ treatment, and across all samples, at least three of the top three genes identified by these algorithms were the same ( Supplementary Figures S1B, D, F, K, O ). However, the stable genes identified by different algorithms varied significantly under each experimental condition ( Figures 3 – 7 ). For instance, under low-temperature stress, the delta-Ct and NormFinder algorithms identified lnc00042194 and GAPH as the most stable genes, whereas BestKeeper and geNorm identified lnc00239991 and lnc00249739, and lnc00265419 and lnc00249739, respectively ( Figures 3 – 6 ). This variation is consistent with findings in other plant species (Vandesompele et al., 2002; Zhang et al., 2018, 2021a, b; Wang et al., 2024a). We speculate that these discrepancies arise from differences in the calculation methods used by the algorithms and their varying sensitivities to co-regulated candidate genes. Therefore, it is essential to comprehensively consider the results from these algorithms to select the most appropriate RGs for practical applications.

We employed the geometric mean of ranks method in conjunction with RefFinder to conduct a comprehensive analysis of RG stability. Based on the optimal number of RGs determined by geNorm analysis, we identified the most stable RG combinations. Notably, the results from the geometric mean of ranks method were consistent with those obtained from RefFinder across various treatments and tissues ( Figure 7 ; Supplementary Table S3 ). Specifically, the most stable RG combinations identified under different conditions were as follows: lnc00249739 and lnc00042194 for cold stress; lnc00042194 and lnc00174850 for drought stress; lnc00239991 and lnc00042194 for salt stress; lnc00239991, lnc00042194, lnc00067193, and lnc00265419 for abiotic stress; lnc00265419 and lnc00249739 for MeJA treatment; lnc00229717 and lnc00044331 for ethephon treatment; lnc00265419 and lnc00239991 for hormone treatment; lnc00067193 and 18S for Cu²⁺ treatment; lnc00239991 and lnc00067193 for metal ion treatments; lnc00229717 and RAN1 for flower opening and senescence processes; lnc00239991, lnc00042194, lnc00067193, TUA, UBQ4, and RAN1 for various tissues; and lnc00239991, lnc00042194, lnc00265419, and UBQ4 for all samples ( Figure 7 ; Supplementary Table S3 ). These findings suggest that the analysis conducted in this study is accurate. It is noteworthy that the optimal RG combinations varied across different treatments and tissues. Specifically, under ABA treatment, lnc00239991 and 18S (instead of lnc00229717) were the most stable RGs; under Al³⁺ treatment, lnc00087780 and lnc00265419 (instead of lnc00249739) were optimal; and under Fe²⁺ treatment, lnc00229717 and ACT7 (instead of lnc00249739) were the most stable RGs ( Figures 7E, I, K ; Supplementary Table S3 ). Similar variations in RG stability have been reported in other plants (Zhang et al., 2021a, b). We speculate that these differences may result from varying environmental conditions, tissue types, and treatment methods, which impact gene expression stability and lead to different RG combinations exhibiting optimal stability under distinct circumstances.

Although no single RG consistently emerged as the best choice across all experimental conditions, lnc00239991, lnc00042194, lnc00265419, and UBQ4 were identified as good candidates for normalizing lncRNA expression in O. fragrans. Notably, UBQ4 was identified as the optimal RG in a qRT-PCR study of rice seedling leaves under salt stress (Moraes et al., 2015) and for perennial ryegrass (Lolium perenne) under various abiotic stresses (Huang et al., 2014). In contrast, UBQ10 was found to be the most unstable RG in sudangrass (Sorghum sudanense) under NaCl stress (Wang et al., 2024b). These discrepancies can be attributed to the specific expression patterns of RGs and their varying suitability across different conditions. Furthermore, the lncRNAs lnc00239991, lnc00042194, and lnc00265419 have rarely been reported as internal RGs in plants, highlighting the generally lower expression levels of lncRNAs. Therefore, selecting protein-coding genes as internal RGs to study lncRNA expression may not always accurately reflect their true expression levels of lncRNAs.

Numerous studies have demonstrated that selecting appropriate RGs is crucial for gene expression research using qRT-PCR technology. Inappropriate RG selection can lead to erroneous conclusions. In this study, we utilized both stable and unstable RGs to examine the expression patterns of target lncRNAs across various stress conditions, different tissues, and during flower opening and senescence. Our results showed that using inappropriate candidate RGs can lead to the underestimation or overestimation of lncRNA expression or misinterpretation of expression trends ( Figures 8 – 10 ). This highlights the critical importance of selecting appropriate RGs for accurate molecular biological analyses.

In summary, this study represents the first systematic and comprehensive investigation into the selection and evaluation of 17 candidate RGs for normalizing lncRNA expression in O. fragrans under abiotic stress conditions, as well as hormone and metal ion treatments. Although no single RG was proved to be optimal across all experimental conditions, lnc00239991, lnc00042194, lnc00265419, and UBQ4 emerged as reliable choices for lncRNA expression studies in O. fragrans. The findings facilitate accurate and thorough analysis of lncRNA expression across various conditions. This research provides a solid foundation for further studies exploring the roles of lncRNAs in growth and development, as well as their responses to abiotic stress, hormone treatments, and metal ion treatments. Furthermore, it provides valuable insights that could drive advancements in crop improvement and plant biotechnology by enhancing our understanding of lncRNA functions. Future research could explore the molecular mechanisms through which these lncRNAs influence plant stress responses, offering potential strategies for improving stress tolerance in O. fragrans and other crops.