In most plants, centromeric DNA contains highly repetitive sequences, including tandem repeats and retrotransposons; however, the roles of these sequences in the structure and function of the centromere are unclear. Here, we found that multiple RNA sequences from centromeric retrotransposons (CRMs) were enriched in maize (Zea mays) centromeres and back spliced RNAs were generated from CRM1. We identified three types of CRM1-derived circular RNAs with the same back-splicing site based on the back-spliced sequences. These circular RNAs bound to the centromere through R-loops. Two R-loop sites inside a single circular RNA promoted the formation of chromatin loops in CRM1 regions. When RNAi was used to target the back-splicing site of the circular CRM1 RNAs, the levels of R-loops and chromatin loops formed by these circular RNAs decreased, while the levels of R-loops produced by linear RNAs with similar binding sites increased. Linear RNAs with only one R-loop site could not promote chromatin loop formation. Higher levels of R-loops and lower levels of chromatin loops in the CRM1 regions of RNAi plants led to a reduced localization of the centromeric H3 variant (CENH3). Our work reveals centromeric chromatin organization by circular CRM1 RNAs via R-loops and chromatin loops, which suggested that CRM1 elements might help build a suitable chromatin environment during centromere evolution. These results highlight that R-loops are integral components of centromeric chromatin and proper centromere structure is essential for CENH3 localization.

Raw images in PDF format:

Fig. 2A The 354-nt RNA was stable after an RNase R treatment. Cenh3 mRNA was used as a linear RNA control. The black arrows in the right panel show the positions of primers used. Fig. 2E Divergent PCR showed that the 607-nt and 277- to 296- nt RNAs were stable after RNase R treatment. The right panel shows the compositions of the amplified sequences mentioned in Fig S2E, 2C, and 2D.

Fig. 3F 3C-PCR confirms the potential ligations of chromatin loops after DpnII digestion. The left panel shows the PCR results in the undigested, unligated samples and 3C samples under potential ligation forms. The right panel shows the sequences from the bands on the left, including the expected sequences, the first and the second part of the expected sequences, and the amplified sequences.

S1D Fig. RT-PCR analysis of the 354-nt RNA without and with reverse transcription.

S3C Fig. Detection of the R-loop structure by T7 endonuclease I digestion and subsequent ligation. The red arrows show the shortened sequences.

S3F Fig. RNA-85, RNA-269 and RNA-85+269 were sensitive to RNase R treatment. The right panel shows the positions of the primers.

S5 Fig. The sense strand transcribed CRM1 RNA can be spliced into the 354-nt-like back-spliced RNA after being transformed into oat (A), rice (B), wheat and sorghum (E), and soybean (H) protoplasts.