Selection and Validation of Reference Genes for RT-qPCR Analysis of Lipid Biosynthesis-Related Genes in Perilla frutescens

Perilla frutescens is a specialty oilseed crop rich in polyunsaturated fatty acids, particularly α-linolenic acid (ALA), and the molecular regulation of seed lipid metabolism is an active area of research. Reverse-transcription quantitative PCR (RT-qPCR) is essential for investigating the molecular mechanisms underlying seed oil biosynthesis; however, systematically validated reference genes are lacking for this species. Based on P. frutescens transcriptome data, seven commonly used housekeeping genes—PfActin, Pf18S (18S pre-ribosomal assembly protein GAR2-like), PfEF-1α, PfCYP, PfTUA, PfTUB, and PfGAPDH—were initially selected. After PfGAPDH was excluded because of its extremely low abundance and unstable expression, the remaining six candidates were evaluated in vegetative organs (leaves, stems, and roots), seeds at different developmental stages, and leaves subjected to PEG 6000 treatment. Expression stability was independently assessed using geNorm, NormFinder, and BestKeeper, and the resulting stability values were integrated with RefFinder. Pf18S showed the highest overall stability in the complete sample set and developing seeds. Under osmotic stress, PfTUB was the most stable single reference gene, and geNorm identified PfTUB + Pf18S as the optimal two-gene combination. In vegetative organs, PfTUB ranked first and PfTUA second, whereas PfEF-1α was among the least stable genes in most sample sets. The recommended two-gene combinations were PfTUA + PfTUB for vegetative organs, Pf18S + PfTUA for developing seeds, and PfTUB + Pf18S for osmotic stress. Biological validation showed that normalization to Pf18S reproduced spatiotemporal expression patterns of six key lipid-biosynthesis genes that were consistent with the dynamic accumulation of total seed oil and ALA. These results support a tiered normalization strategy using Pf18S for preliminary screening and condition-specific reference-gene pairs for precise quantification.