Long noncoding RNAs, or lncRNAs, are RNA molecules longer than 200 nucleotides that carry no protein-coding potential. Transcribed largely by RNA polymerase II and typically capped, spliced, and polyadenylated like messenger RNAs, these molecules were once dismissed as transcriptional noise. Research over the past two decades has overturned that view, showing that lncRNAs regulate gene expression in diverse ways, acting on neighboring genes in cis or on distant genes in trans, and participating in chromatin remodeling, transcriptional control, and posttranscriptional regulation. In plants, lncRNAs have been implicated in everything from flowering time and fruit development to anthocyanin accumulation and stress responses. The classic examples come from Arabidopsis, where the lncRNAs COOLAIR and COLDAIR, transcribed from the FLOWERING LOCUS C locus, are essential for the epigenetic silencing of that key floral repressor during vernalization. Despite this growing appreciation, lncRNAs in radish had never been systematically characterized in the context of flowering, leaving a significant gap in the crop’s molecular biology.

The research team, led by Xiaobo Luo and Wanping Zhang, addressed this gap by studying a late-flowering radish inbred line designated “XHT,” which requires approximately 185 days to flower. The plants were grown under carefully controlled conditions in an artificial climate chamber, subjected to 30 days of vernalization, and sampled at two critical points: the vegetative growth stage and the flowering stage. From these samples, the researchers performed strand-specific whole-transcriptome RNA sequencing on an Illumina HiSeq 2000 platform and small RNA sequencing on a HiSeq 2500, generating hundreds of millions of reads that were mapped to the radish reference genome. To identify high-confidence lncRNAs, the team applied a battery of computational filters, excluding transcripts shorter than 200 base pairs, those with low expression, and any transcripts showing evidence of protein-coding potential as assessed by the Coding-Noncoding Index, the Coding Potential Calculator, Pfam protein domain searches, and phylogenetic codon substitution frequency analysis.

The scale of the resulting dataset was impressive. From more than 50,000 assembled transcripts, the researchers identified a total of 8,828 lncRNAs, of which 7,299 were already annotated on the radish genome and 1,529 were entirely novel. Comparative analysis revealed that radish lncRNAs tend to have fewer exons, shorter sequence lengths, and shorter open reading frames than protein-coding messenger RNAs, consistent with patterns observed in other plant species such as Brassica rapa and tomato. Critically, 263 lncRNAs were differentially expressed between the vegetative and flowering stages, with 165 upregulated and 98 downregulated as the plants transitioned to reproduction. Hierarchical clustering confirmed that these differentially expressed lncRNAs showed strong stage-specific expression patterns, hinting at their involvement in the floral transition.

The messenger RNA side of the analysis was equally revealing. A total of 5,315 differentially expressed genes were detected between the two stages, 3,190 of them upregulated and 2,122 downregulated at flowering. Gene Ontology enrichment analysis placed these genes into 52 significant categories, with biological process terms tied to photosynthesis, cellular carbohydrate metabolism, reproductive processes, hormone responses, and pollination. K-means clustering sorted the genes into six expression clusters, and within these clusters the researchers found many of the canonical players in flowering regulation. Genes such as FLOWERING LOCUS T, EARLY FLOWERING 3, SPL15, SUPPRESSOR OF OVEREXPRESSION OF CO 1, SWEET11, and SWEET12 appeared among the differentially expressed set, as did FLC-like genes, PHYTOCHROME-INTERACTING FACTOR 4, CONSTANS-like 1, EMBRYONIC FLOWER1, and VERNALIZATION INSENSITIVE 3. The presence of these well-characterized flowering pathway genes validated the experimental approach and provided a rich framework for building regulatory networks.

On the small RNA front, sequencing yielded roughly 35 and 38 million clean reads from the vegetative and flowering samples respectively, from which the team identified 134 microRNAs, including 74 known and 60 novel species. Of these, 38 microRNAs were differentially expressed between the two stages. MicroRNAs are roughly 22-nucleotide RNAs that silence their targets by directing cleavage of complementary transcripts or blocking their translation, and several of the differentially expressed species, including miR156, miR399, miR172, and miR169, are established regulators of flowering time in plants.

Perhaps the most valuable contribution of the study lies in the regulatory networks constructed from these three RNA classes. By analyzing protein-coding genes located within 100 kilobases of differentially expressed lncRNAs and calculating Pearson correlation coefficients between their expression profiles, the researchers identified 84 lncRNAs with potential cis-regulatory effects on 74 messenger RNAs and 107 lncRNAs with potential trans-regulatory effects on 96 messenger RNAs. In parallel, target prediction identified 85 messenger RNAs as putative targets of 31 differentially expressed microRNAs and 87 lncRNAs as putative targets of 32 microRNAs, many showing the negative correlations expected of genuine microRNA-target relationships. From this data, five mRNA-lncRNA-miRNA regulatory networks centered on flowering time emerged, involving miR156a-5p, miR399b, a novel microRNA designated novel-23, miR164c-5p, and miR165a-5p.

Among these, the miR156 module stood out and became the focus of functional validation. In Arabidopsis, miR156 and its SPL transcription factor targets are central controllers of the juvenile-to-adult transition and floral induction. The radish data showed that miR156a-5p was negatively correlated with SPL10 and SPL15 as well as with 12 lncRNAs. To test whether these relationships hold up experimentally, the researchers performed transient overexpression experiments in radish cotyledons using an Agrobacterium-mediated system with a visible RUBY reporter marker. When miR156a-5p was overexpressed, the expression levels of RsSPL10, RsSPL15, and two lncRNAs, RsLinc1162 and RsLinc214, dropped significantly. Conversely, overexpressing the lncRNA RsLinc214 increased RsSPL10 and RsSPL15 expression while suppressing miR156a-5p, suggesting a feedback relationship. The decisive evidence came from dual-luciferase reporter assays in tobacco leaves: co-expression of miR156 with reporter constructs carrying RsSPL10 or RsSPL15 significantly suppressed luciferase activity, demonstrating that miR156 directly targets and inhibits both genes.

Quantitative reverse transcription PCR provided additional confirmation across the broader dataset. Four messenger RNAs, including RsSOC1, RsFT, RsIQM-1, and RsRLF3, were significantly upregulated at flowering, while three lncRNAs, RsLINC84, RsLINC502, and RsLINC299, rose steadily through six vernalization time points and into the flowering stage. The expression of miR399b increased after 20 days of vernalization and peaked at 25 days, whereas miR169b-3p and miR156a-5p remained elevated throughout vernalization and declined only after flowering. All of these patterns matched the sequencing data closely, lending confidence to the network models.

For the vegetable industry, the implications are potentially significant. Premature bolting in winter and spring crops causes fleshy roots to lose quality and market value, and although prior work had identified flowering-related genes and quantitative trait loci in radish, the molecular mechanism of the floral transition remained obscure. By cataloguing thousands of stage-specific transcripts and validating a key miR156-SPL regulatory module, the study provides both a theoretical foundation and a practical set of candidate genes and RNAs that breeders and molecular biologists can now pursue. Markers based on these loci could accelerate the development of radish varieties with more stable flowering behavior, and further dissection of the newly discovered lncRNAs and novel microRNAs may reveal additional levers for controlling bolting. The study also demonstrates how integrating messenger RNA, microRNA, and lncRNA sequencing can illuminate crop traits that single-omics approaches miss, offering a template for similar investigations in other root vegetables across the cruciferous family.