Pache.co Information Technologies

The di locus maps with a gene transcribing a miR172 precursor sequence

In the attempt to more precisely delimit the di locus position on chr 2, linkage disequilibrium (LD) pattern around the 23 – 26.5 Mb genomic interval enclosing the two associated GWAS signals was inspected in 21 Oriental DF accessions, using about 4K SNPs extracted from whole-genome sequencing (WGS) data [21]. The LD map showed a complex pattern characterized by spaced and relatively short regions with long-range high LD levels, defining an interval of about 2.4 Mb, approximately comprised between SNP1456516 (24,277,038) and SNP1501405 (26,628,544) (Supplementary Figure S1). At least 381 gene models were annotated in this interval, based on a peach reference transcripts v2.1 annotations (Supplementary Table S1). Homology-based searches of genes putatively involved in flower tissues differentiation, growth and/or development allowed the identification of a predicted gene (Prupe.2G237700), transcribing a microRNA precursor sequence belonging to the miR172 family, previously classified as ppe-miR172e [28] and hereafter named ppe-miR172d (as currently annotated on www.mirbase.org, accession: MI0026098). Indeed, an increasing body of knowledge around the role of the euAP2/TOE-miR172 module in the regulation of petaloidy in various species [27] makes Prupe.2G237700 a prime candidate for the recessive DF trait. Firstly, the sequence of the miR172d primary transcript was annotated by mapping RNA-seq reads from peach dormant buds of various stages and carpel tissues. The expressed pri-miR172d sequence has no introns, spanning ~2.0 Kb from 25,860,117 to 25,861,987 bp, with the predicted pre-miR172d precursor sequence (119 nt) starting at 25,860,595 (Fig. 1).

Genomic structure of the micro-RNA miR172d encoding transcript Prupe.2G237700. The pri-miRNA sequence was identified by mapping transcriptome data from various peach flower tissues (petal, stamen, carpel). The predicted pre-miR172d stem-loop hairpin structure is also shown

Full size image

Genomic structure of the micro-RNA miR172d encoding transcript Prupe.2G237700. The pri-miRNA sequence was identified by mapping transcriptome data from various peach flower tissues (petal, stamen, carpel). The predicted pre-miR172d stem-loop hairpin structure is also shown

Three independent mutant alleles of Prupe.2G237700 are associated with the recessive double-flower trait

Several primer pairs covering the entire pri-miR172d sequence plus ~1 Kb upstream and downstream were designed to analyze allelic polymorphisms in various ornamental materials, including main commercial cultivars, and accessions or selections retrieved from reliable germplasm collections of INRAE-GAFL (France), University of Milan (Italy), USDA-ARS (Kearneysville, West Virginia, and Byron, Georgia) and Clemson University (South Carolina) (the accession list is provided in Table 1). In agreement with the recessive inheritance of the di trait, causal variant(s) were assumed to be homozygous in DF genotypes. Interestingly, primers 2G237700_F1 and 2G237700_R2 resulted the most informative, as PCR showed a diverse pattern of amplifications in the various accessions (Fig. 2): in single-flower (SF) accession ‘Crimson Rocket’ the wild-type (Di) amplicon of 563 bp (SF ‘Lovell’ peach genome) was observed, as well as in two DF selections ‘KV045590’ and ‘KV021779’; another SF accession, ‘Redleaf Pillar’, showed the 563 bp amplicon alongside a second PCR product of about 5.5 Kb. This PCR product, about 5.0 Kb longer than the wild-type (Di) allele, was exclusively obtained in DF accessions ‘Hokimomo op’, KV872615, ‘Compact Pillar’, ‘Peppermint Stick’, S10321 ‘Hua 5-15’, S10322 ‘Hua 5-25’ and ‘Okinawa’. Interestingly, the other four DF accessions (S7258 ‘clone IRAN 6:59’, ‘Kikoumomo D’, ‘G. Biffi fiore doppio’ and ‘Helen Borchers’) showed an amplicon of ~1.7 Kb (suggestive of an insertion of ~1.2 Kb compared to the Di allele). Finally, three additional DF accessions showed no amplification (among them ‘Klara Meyer’). Sanger sequencing of the 1.7 Kb amplicon revealed a 1,198 bp insertion without LTR elements (named di² located at position +297 (Pp02:25,860,414) from the pri-miR172d transcription initiation site (Supplementary File S1). BLAST searching showed no clear similarity with known genes or transposable elements, although identical sequences are spread across several peach chromosomes. Instead, a long-read Oxford Nanopore Technology (ONT) approach was used to sequence the miR172d allelic variants in ‘Klara Meyer’ and ‘S10321 Hua 5-15’, as representative of the genotypes respectively yielding no amplification and the 5.5 Kb amplicons with 2G237700_F1 and 2G237700_R2. Whole-genome coverage was ~10 and ~20x, respectively for ‘Klara Meyer’ and ‘Hua 5-15’, with estimate read length N50 of 32.5 and 21.2 Kb. After aligning to the Peach V2.0 reference genome, two different allelic variants were discovered at the miR172d locus: a ~9.5 Kb deletion (Pp02:25,855,441 to Pp02:25,864,957) encompassing the whole miR172 miRNA precursor sequence including its flanking DNA regions, named di^Δ; a ~5.0 Kb LTR insertion (hereafter di¹), located at position +221 (Pp02:25,860,344) from the pri-miR172d transcription initiation site (Fig. 3).

Molecular analysis of sequence variants of the miR172d gene. A Genomic structure of Prupe.2G237700 (purple box) and position of pre-miR172d encoding sequence (red segment). Arrowheads indicate the positions of the primers used for PCR analysis. B Phenotype of double-flower accessions used in this study. Left to right, Top: ‘KV872615’, ‘Compact Pillar’, ‘S7258 cl. IRAN 6:59’, ‘S10321 Hua 5-15’; Bottom: ‘Klara Meyer’, ‘S10322 Hua 5-25’, ‘Kikoumomo D (KV044770)’, ‘Hokimomo op’ (trees and flowers photos were provided by C. Dardick and V. Signoret). C PCR analysis of the single and double-flower accessions listed in Table 1

ONT long-sequencing reads alignment at di locus in ‘S10321 Hua 5-15’ and ‘Klara Meyer’ accessions. At the top: reads breakpoint due to a ~5.0 Kb LTR insertion (di¹) located at position +221 (Pp02:25,860,338) from the pri-miR172d transcription initiation site. At the bottom: a ~9.5 Kb deletion (di^Δ) from (Pp02:25,855,441 to 25,864,957), encompassing the entire pri-miR172 precursor sequence including its flanking DNA regions in ‘Klara Meyer’

The di¹ insertion showed homology with a putative FRS5-LIKE gene of the FARS (FAR-RED IMPAIRED RESPONSE1 RELATED SEQUENCE) transcription factors family derived from Mutator-like element (MULE) transposases [29]. Validation of the three allelic variants of the di locus was performed through PCR analyses. Specific primer combinations confirmed the presence of the homozygous di¹ insertion found in ‘Hua 5-15’ also in other six DF accessions: ‘Houki Momo op’, ‘KV872615’, ‘Compact Pillar’, ‘Peppermint Stick cl.’, ‘Hua 5-25’ and ‘Okinawa’ (Supplementary Figure S2); consistent with the recessive inheritance of the di trait, the di¹ allele was amplified in the heterozygous SF accession ‘Redleaf Pillar’. Finally, the two remaining DF selections ‘KV045590’ and ‘KV021779’ showing the wild-type Di/Di genotype (i.e. 563 bp amplicon) both resulted heterozygous (Di2/di2) for the PETALOSA variant (underlying the dominant inherited Di2 allele) [24] (Table 1). In addition to the panel of ornamental cultivars, the absence (in homozygosis) of any of the three identified di variant(s) in single-flower accessions was further confirmed in a genetically diversified panel of 151 accessions from the PeachRefPop collection [30] (data not shown).

Thus, at least three distinct mechanisms acting on the miR172d gene underpin the recessive double flower trait in peach: di¹ (a ~5.0 Kb insertion), di² (a ~1.2 Kb insertion) and di^Δ (a ~9.5 Kb deletion) (Fig. 4).

Structural variants identified at the gene encoding micro-RNA miR172d, Prupe.2G237700. The wt genomic structure of Prupe.2G237700 is represented by a light lilac box and the red segment indicates the position of the pre-miR172d encoding sequence. Insertions in alleles di¹ and di² are represented by a dark purple box, with LTR regions marked in lighter purple in di¹. The genomic region deleted in di^Δ is indicated by the dotted/shaded region

Full size image

Structural variants identified at the gene encoding micro-RNA miR172d, Prupe.2G237700. The wt genomic structure of Prupe.2G237700 is represented by a light lilac box and the red segment indicates the position of the pre-miR172d encoding sequence. Insertions in alleles di¹ and di² are represented by a dark purple box, with LTR regions marked in lighter purple in di¹. The genomic region deleted in di^Δ is indicated by the dotted/shaded region

Pre-miR172d expression is abolished in homozygous di
¹, di
² and di
^Δ genotypes

The expression profile of pre-miR172d was investigated using qRT-PCR in pooled bud flower tissues from two developmental stages and three di variants: ‘Okinawa’ (di¹/di¹), ‘clone IRAN 6:59’ (di²/di²) and ‘Klara Meyer’ (di^Δ/di^Δ). The pre-miR172d was only expressed in the SF control accession ‘Bounty’ (Di/Di), being undetectable in all the three DF accessions (Fig. 5). These findings were clearly consistent with the complete deletion of miR172d sequence in the di^Δ allele; in case of di¹ and di² alleles, transposon-induced gene silencing mechanisms could be the most probable explanation, although the exact mechanism(s) leading to their miss-expression should be more specifically addressed.

Expression pattern of pre-miR172d in pooled flower bud tissues from Baggiolini stage B (flower bud swelling) to C1 (flower buds apparent) from ‘Bounty’ (single-flower), ‘Klara Meyer’ (double-flower, di^Δ), ‘S7258 cl. IRAN 6:59’ (double-flower, di²) and ‘Okinawa’ (double-flower, di¹) genotypes. Actin (Prupe.6G163400) was used as a reference to normalize expression data

Full size image

Expression pattern of pre-miR172d in pooled flower bud tissues from Baggiolini stage B (flower bud swelling) to C1 (flower buds apparent) from ‘Bounty’ (single-flower), ‘Klara Meyer’ (double-flower, di^Δ), ‘S7258 cl. IRAN 6:59’ (double-flower, di²) and ‘Okinawa’ (double-flower, di¹) genotypes. Actin (Prupe.6G163400) was used as a reference to normalize expression data

Distribution of allelic variants in DF varieties of germplasm collections

The availability of phenotypic and genotypic information on peach ornamental accessions across important Chinese collections, allowed exploring the allelic assortment at both di and Di2 loci. The presence of allelic variants (and co-segregation with flower phenotype) was assessed through a sequencing-based method (Supplementary Figure S3 and Materials and Methods section) using public WGS data (retrieved from NCBI SRA database) of 41 peach accessions with clear flower trait (double or semi-double) and sufficient genome coverage (i.e. > 10x), and several SF accessions as control. Among the DF accessions, 21 were homozygous di¹/di¹ (including the dwarf DF ‘Bonanza peach’, probably erroneously indicated as SF), 7 homozygous di²/di² and ‘Ju Hua Tao’ carried the di¹/di² allelic combination (Supplementary Table S2); in this last case, the concomitant presence of di¹ and di² alleles suggests that both mutations are functionally equivalent; surprisingly, the di^Δ variant seems to be absent in Chinese germplasm collections, suggesting the hypothesis that this allele originated after the arrival of peach in Europe (tracing back to the Greek-Roman era). Apart from the di locus, 7 DF accessions resulted heterozygous for the PET-Di2 variant (Supplementary Table S2). Interestingly the DF accession ‘Hua Yu Lu’ was homozygous for Di and di2 alleles but showing a heterozygous SNP within the core miR172 recognition sequence at the three-prime end portion of the coding sequence of the PET gene (Supplementary Figure S4).

Total RNA was collected from pooled flower bud tissues from Baggiolini stage B (flower bud swelling) to C1 (flower buds apparent). RNA was treated with DNase I (Takara) and extracted using TRIzol (Thermo Fisher Scientific) followed by phenol extraction; 1 μg of total RNA was reverse-transcribed using random hexamers and SuperScript III RT (Thermo Fisher Scientific). According to Zhu et al. (2012), specific primers for pre-miR172d amplify a 90 bp region including mature miR172d, located 84 nt from the 5’-prime end of the precursor sequence (Supplementary File S1): pre-miR172d_For (AGCATCATCAAGATTCACAATTTC) and pre-miR172d_Rev (TGCCGCTGCAGCATCATCAAGA). Real-time PCR reaction were carried out using Power SYBR Green qPCR Supermix (Thermo Fisher Scientific). Each reaction was performed in triplicate on QuantStudio 3 Real-Time PCR Systems (Thermo Fisher, United Kingdom) using the following cycling conditions: 95°C for 10 min, 40 cycles at 95°C for 20 s, 56°C for 25 s, and 72°C for 30 s. Each thermal cycle was followed by a melting curve stage, with temperatures ranging from 60°C to 95°C. Pre-miR172d relative expression was calculated using ∆C_t method [55], using Actin (Prupe.6G163400) as reference gene [56]. RT-PCR products were run on 2.0% agarose gels with Midori staining and visualized under UV light. Three to four biological pool replicates were analyzed for each sample.