Characterization of MITEs in the promoter region of the CitRWP gene controlling nucellar embryony
Previous studies (Wang et al. 2017; Shimada et al. 2018) have shown that transposition of an MITE in the promoter region of the CitRWP gene is responsible for nucellar embryony in Citrus species. In both studies, molecular markers were independently developed to genotype CitRWP using primer pairs flanking the inserted MITEs. However, when these molecular markers were applied to yuzu accessions, PCR products amplified from the mutant allele containing the MITE were hardly detected in any accessions (Supplementary Fig. 1). To improve robustness of molecular markers, nucleotide sequences flanking the MITE were retrieved from two previous studies (Wang et al. 2017; Shimada et al. 2018). Although two sequences were derived from mandarins, both sequences were highly diverged from each other, sharing only 73.5% sequence identities. In addition, positions of predicted TIRs and TSDs of the MITE also varied between the two studies (Supplementary Fig. 2.).
To determine exact boundaries of the MITE sequence, an MITE family showing the highest homology was identified among 110 citrus MITE families isolated from the previous study (Liu et al. 2019). DTM10, an MITE family belonging to the Mutator superfamily, showed 91.7% nucleotide sequence identities with the MITE in sweet orange CitRWP, suggesting that the MITE in CitRWP might be a member of DTM10. No other MITE families showing any significant homologies were found. Frequencies of full-length DTM10 elements in genomes of Citrinae species were examined to assess the origin of this element. At least 80 copies of full-length DTM10 were identified from each of 10 available genome sequences of Citrinae species (Fig. 1, Supplementary Table 3), implying that DTM10 might have existed in a common ancestor of Citrinae species.
For detail analysis of DTM10, flanking sequences of 114 DTM10 elements found in sweet orange genome were analyzed to identify reliable TSD and TIR sequences. Fifty-seven elements contained intact TSD sequences. Most of them contained 9-bp TSDs except for six elements harboring 10-bp TSDs (Supplementary Table 4). A consensus sequence of DTM10 was inferred by aligning 57 elements containing intact TSDs (Supplementary Table 5). The 192-bp consensus DTM10 sequence contained 56-bp identical TIR sequences. A phylogenetic tree of 114 sweet orange DTM10 elements implied that this element might have been amplified twice during evolution (Fig. 2A). The DTM10 element in CitRWP designated as DTM1-Cs39 was assumed to be amplified during a more recent burst (Fig. 2A). In addition, similar amplification patterns of DTM10 were observed in mandarin, pummelo, and lemon (Fig. 2). However, a somewhat different amplification pattern was observed in the distantly related poncirus genome (Fig. 2B). The relative position of DTM10-Cs39 in the poncirus tree was also different from those of Citrus species. These results imply that DTM10-Cs39 might have transposed in the CitRWP gene of a common ancestor of Citrus species.
Analysis of MITEs and their flanking sequences of the CitRWP gene in yuzu
Interestingly, two inverted DTM10 elements were identified in promoter regions of sweet orange and lemon CitRWP genes (Fig. 3) compared with two previous reports (Wang et al. 2017; Shimada et al. 2018) in which only one DTM10 was identified (Supplementary Fig. 2). Based on sweet orange CitRWP sequences, primer pairs flanking the MITEs were designed to amplify corresponding yuzu sequences. Wild-type CitRWP alleles of two yuzu varieties could be easily amplified using wild-type allele-specific primers. However, it was difficult to obtain the mutant CitRWP allele containing MITEs due to low efficiencies of PCR amplification and Sanger sequencing of PCR products. A strong stem-loop structure was expected to be formed between two inverted DTM10 elements in sweet orange (Supplementary Fig. 3). Due to a strong intramolecular structure, sudden drop of signals in Sanger sequencing was frequently observed near the MITE insertion (Supplementary Fig. 4).
In addition to such inhibitory effects on PCR amplification and sequencing, polymerase jumping, an enigmatic phenomenon, was observed during sequencing of mutant CitRWP alleles. In the case of a yuzu variety, Harim12, the wild-type allele contained 42-bp deletion on the 5’ upstream region of the MITE insertion (Supplementary Fig. 5). By designing a forward primer in this deleted region, the mutant allele was specifically amplified in Harim12. However, a PCR product with a size similar to that of wild-type allele was amplified together with the mutant allele (Fig. 4A). Sequence analysis of the small-sized PCR product showed that a mutant allele sequence without MITE insertion was amplified, whereas the large-sized PCR product was a mutant allele containing the MITE insertion (Fig. 4B). Polymerase jumping occurred between two 6-bp repeats (‘CAATAA’) positioned at borders of the stem-loop structure (Fig. 4C).
Although sequencing of the large-sized PCR product containing the MITE insertion was difficult due to sudden drop of signals, some primers could barely read through MITE sequences. Almost complete sequence of the mutant CitRWP allele of yuzu was obtained except for sequences between two inverted DTM10 elements. A complete sequence of the mutant allele was obtained from ‘Nianju’ mandarin whose CitRWP genotype was homozygous mutant (Fig. 3A). Like sweet orange CitRWP, two inverted DTM10 elements were identified in promoter regions of yuzu and mandarin CitRWPs (Fig. 3A). While two intact DTM10 elements were identified in yuzu and mandarin, a 161-bp deletion spanning the second DTM10 element was observed in sweet orange (Fig. 3). To confirm the presence of this 161-bp deletion in sweet orange, PCR amplification and sequencing were performed. However, deletion was not found in the sweet orange CitRWP analyzed in this study.
Sequence analyses of wild-type and mutant CitRWP alleles of diverse Citrus species and development of complementary molecular markers for genotyping of the CitRWP gene
Based on sequence alignment of available wild-type and mutant alleles of CitRWP, forward primers specific to wild-type and mutant alleles were designed, respectively, and a common reverse primer was designed to cover full-length coding regions of CitRWP (Supplementary Fig. 6). Regarding mutant alleles, polymerase jumping was observed in all analyzed Citrus species as shown in yuzu. A total of four different mutant alleles were identified from sweet orange, mandarin, yuzu, and kumquat. They were designated as CitRWP-M1 to CitRWP-M4, respectively. SNPs in MITE sequences of the first three mutant alleles were identified (Fig. 3A). However, genic sequences of these mutant alleles were identical to each other. Interestingly, three MITE elements were tandemly positioned in the promoter region of CitRWP-M4 identified from kumquat (Fig. 3B), although the position of MITE insertions was identical to those of three other mutant alleles. In addition, five SNPs and one InDel were found in the genic region of CitRWP-M4 compared with other mutant alleles.
Meanwhile, 15 different wild-type alleles were obtained from diverse Citrus species. Phylogenetic analysis of wild-type and mutant CitRWP alleles showed that mutant alleles seemed to be diverged from citron-related species (Fig. 5). However, pairwise genetic distances indicated that these mutant alleles showed similar genetic distances with citron, mandarin, and pummelo, respectively (Supplementary Fig. 7), implying that these mutant alleles might have originated from a common ancestor of Citrus species.
To develop universally applicable molecular markers, all available sequences were aligned and a pair of primers with Tm higher than 65°C were designed based on conserved regions flanking the MITE insertion (Supplementary Fig. 8). In addition, CAPS and HRM markers were developed based on the SNP unique to mutant alleles. Three SNPs were unique to mutant alleles (Supplementary Table 4). In the case of the MITE-flanking marker (CitRWP-MK1), PCR products amplified from mutant alleles showed low intensity and multiple bands, although polymerase jumping was not observed (Fig. 6B). As complementary molecular markers, CAPS and HRM markers were developed based on the SNP unique to mutant alleles (Fig. 6). Three genotypes were clearly distinguished by these two markers (Figs. 6B, 6C). A total of 241 Citrus accessions were analyzed using three molecular markers. Genotyping results of three markers were identical to each other (Supplementary Table 1).