Evolutionary Diversity and RNA Editing Patterns in Soybean Chloroplast Genomes

Soybean ( Glycine max ) has five recognized cytoplasmic types (G1–G5) defined primarily by mitochondrial variation, yet the chloroplast genome (cpDNA) architecture and variation of these cytoplasmic types remain poorly characterized. Here, we de novo assembled and analyzed ten representative soybean cpDNAs, two for each type. All cpDNAs exhibited conserved quadripartite structures, with sizes ranging from 152,217 to 152,227 bp. A total of 23 indels and 77 SNPs were identified within and across cytoplasmic types, with indels restricted to the large single-copy (LSC) region. Among these, 24 SNPs and one indel occurred in coding regions, including 14 non-synonymous variants that may contribute to chloroplast functional diversification. Chloroplast-based phylogenomics supported a recent G1/G3 divergence, confirmed the introgression of G4 and G5 from Glycine soja , and indicated that semi-wild soybean ( Glycine gracilis ) is a hybrid of G. max and G. soja , providing cpDNA-based evidence consistent with mitochondrial-defined cytoplasmic classifications. RNA-editing sites and levels in G1-type cpDNA were assessed using publicly available RNA-seq datasets derived from eight major tissues of cultivar Lincoln (Lin), revealing 49 editing sites and tissue-specific editing patterns. RT-PCR/Sanger-sequencing analysis using random-hexamer-primed cDNA further validated 23 representative sites. Both approaches consistently showed that housekeeping genes and atpF are constitutively edited at a relatively high level, whereas edited transcripts of photosynthesis-related genes in non-photosynthetic tissues were mostly under-represented in poly(A)-enriched RNA-seq libraries. Finally, four chloroplast CAPS markers were developed to identify the male-sterile-associated G5-type cytoplasm. This study provides high-quality cpDNA references, deepens our understanding of cpDNA evolution and regulatory mechanisms, and offers valuable resources for soybean improvement.