Truncated sequence reads at the immediate borders of the putative deletion region were used to perform a BLAST search of the Hs sequence reads to identify untruncated readsequences. Identifed sequence reads from the two sides of the deletion region overlapped allowing us to assemble the Hs sequence in this region . This showed that the lesion was in fact a clean deletion of the region deficient in read sequences with no inserted DNA and no sequence duplication at the deletion borders . The deleted region included the entire INO gene and part of the first coding region of another putative gene encoding a possible UDP-N-acetylglucosamine-N-acetylmuramyl-pyrophosphoryl-undecaprenol N-acetylglucosamine protein. The genomic region containing INO in wild-type and the corresponding region from Hs were deposited in GenBank as accessions ON248606 and ON248607, respectively. These primers were used for PCR on DNA from Hs and Ts lines. Both lines produced a fragment with migration consistent with a length consistent with the 456 bp expected for the deletion . Further, best grow pots because the region between the primers in wildtype is too large to be amplified in standard procedures, this procedure provides a positive assay for the presence of the deletion.
In combination with primers that will amplify a region of the wild-type INO gene, this would enable a codominant test for the wild-type and seedless alleles of the gene in a single reaction. To test this, we modeled amplification from a “heterozygote” by mixing DNA from Hs and wild-type A. squamosa. We found that the amplification with the deletion-specific primers and LMINO1/2 primers in a single reaction could clearly differentiate the homozygous seedless, homozygous wild-type and heterozygous lines . Subsequent work showed that AsINODel primer PCR on all three seedless isolates produced comigrating fragments . The sequences of the PCR products were determined for all three seedless lines and were aligned with the assembled deletion region sequence . The sequence of the deletion junction was confirmed in the sequence from Hs, and an identical deletion was present in the same position in the sequences from mutants Bs and Ts , consistent with all lines producing PCR products of the same size. Further, the presence of identical deletions in all three lines indicates a single origin for the INO deletion among the isolates.To evaluate the use of the AsINODel primers together with the LMINO1/2 pair as codominant markers for plant breeding, lines M1, M2, and M3 were used as wild-type for INO gene and the presence of seeds, and the mutant Bs for the INO deletion and the absence of seeds. The genotype Hs was used as a positive control for the marker AsINODel. After PCR reactions, the combined primer pairs amplified a single fragment of 350 bp from the wild-type parents, representing the presence of the INO gene, and a 456 bp fragment from the mutants Bs and Hs, corresponding to the deletion junction fragment .
As expected, the wild-type fertile parents were homozygous for the presence, and the mutant seedless parent Bs was homozygous for the deletion of the INO gene. In generation F1 both INO gene and deletion region bands were present in all 32 genotypes evaluated, consistent with the expected heterozygous state . The codominant markers were also used in genotyping of individuals of the F2 and backcross generations with some individuals selected to illustrate the discriminatory capacity of markersin Fig. 6 A, B and C. In generation F2, analysis of 145 plants gave results consistent with the expected ratio of 1:2:1 for homozygous INO: heterozygous: homozygous ino deletion . The expected ratios of genotypes were also observed for the sixty-one BCBs plants and the fifty-six BCM plants, where ratios close to 1:1:0 and 0:1:1 were observed, respectively, . The χ2 test on each of these generations showed probabilities consistent with a monogenic segregation of the wild-type and mutant alleles.Fruit was obtained from sixty-three of the genotyped segregating plants. In every case where a band corresponding to the wild-type INO gene was present the seeded phenotype was observed. Similarly, all of the fifteen plants homozygous for the deletion allele exhibited the Bs mutant seedless phenotype. Thus, there was 100% correlation of the seeded and seedless phenotypes with the wild-type containing and homozygous ino deletion genotypes, respectively. To obtain more information to further characterize the linkage between the molecular and visible phenotypes we examined the ovules of plants that were fowering but not yet producing fruit where the ino mutant ovules could be differentiated from wild-type . This determination was done on ninety-five plants and once again there was a complete correspondence between the molecular genotype and the presence of wild-type ovules or aberrant ovules incapable of forming seeds .
If the ino mutant gene is only linked to the seedless mutant allele, and not causally responsible for it, then recombination between the molecular and phenotypic markers would be possible. Among segregating F2 plants, a single recombination event could be observed in the wildtype INO containing chromosome of heterozygous progeny, or in either chromosome of homozygous ino progeny. The same would be true of chromosomes deriving from the F1 plants in the BC Bs progeny. In these cases, recombination would lead to a switch of the seedless/seeded phenotype and breakage of the 100% cosegregation . Seed or ovule phenotypic data were produced for 57 informative heterozygous plants and 36 homozygous mutant plants . Thus, no recombination events were observed between the INO locus and the Bs gene in 114 chromosomes. This allows an initial limit on the maximum genetic distance between ino and Bs. A genetic distance of 3.5 cM would lead to a prediction of 3.5% recombination, where 0% recombination was observed. The comparison of these observed and expected result using the χ2 test produces χ2=3.99 and a corresponding P-value of 0.046. Thus, a genetic distance of 3 cM can be rejected at the 5% probability level . In addition, within the 587 kb of sequence surrounding the INO gene, we find that the 16 kb deletion is the only significant difference between wild-type and Hs in this region . Together these observations are consistent with the INO gene deletion being the lesion responsible for the seedless phenotype.Of the 67 SSR primers used, 41 were amplified under the tested conditions, 12 presented a null amplification pattern with no bands evident, and the remaining 14 did not present a clear band pattern in the tested samples . Among the primers that amplified, 38 generated a total of 63 monomorphic alleles; only three primers generated specific polymorphisms among the genotypes evaluated. The LMCH 3 primer generated two alleles between 200 and 300 bp for the accession M2 genotype and the seedless genotype, respectively. The primer LMCH 39 amplified five alleles in total. Three generated a band only in genotype M2, distinguishing it from seedless genotypes. Similarly, the primer LMCH 137 amplified five alleles. However, in this marker the seedless genotypes presented two specific bands differing from genotype M2, displaying amplification of two monomorphic bands common to all individuals. The genetic distance between genotypes, plants in pots ideas estimated by the complement of the similarity matrix generated by the Jaccard index, ranged from 0 among genotypes Bs, Ts, and Hs, and 0.0933 between the M2 with Bs, Ts, and Hs. Thus, these markers differentiated between the mutant lines and the wild-type line, but could not demonstrate any divergence between the three mutant accessions.Three seedless stenospermocarpous sugar apple varieties, Bs, Ts, and Hs, from difering global locales have been described [Lora et al. ; Santos et al. ; and this work]. The seedless phenotype results from an ovule deffect indistinguishable from the ovule effects of the inner no outer mutation in Arabidopsis , and was associated with a deletion of this locus in the Ts accession . On this basis, Lora et al. hypothesized that the deletion was the cause of the seedless trait, but segregation data were not available.
We show that the three seedless accessions all carry identical 16 kb deletions of the region including INO. The identical sequence of the deletion region in the three accessions indicates that the mutant allele arose as a single deletion event in a common ancestor and propagated globally. The deletion was not associated with any apparent repeated sequence at the deletion junction, but was a precise excision of the deleted region. We tested the causal relationship between the deletion and seedlessness by examining segregation of the seedless trait and the deletion. We demonstrated that seedlessness is a single locus recessive trait and that the fertile/seedless phenotypes cosegregated 100% with the presence/absence of the INO gene among progeny plants. Our data demonstrate that the ino deletion and the seedless trait are separated by less than 3.5 cM. We have determined the sequence of 587 kb surrounding the deletion region, and based on prior measures of the relationship between genetic and physical mapping distances this would constitute from three to thirty centimorgans and so would include the seedless mutation. Outside of the deletion that includes the INO gene, we found no other significant differences from wildtype in the sequenced region surrounding the INO locus. Together with the clear expectation of an ino gene deletion causing the observed ovule phenotype, these data support the ino deletion being the sole cause of the seedless trait. It is possible that the deletion and seedless trait was retained in multiple lines by selection despite significant outcrossing, or alternatively that all lines with this deletion are vegetatively propagated clones. Genotyping through the use of SSR markers failed to detect polymorphism among our available seedless accessions but did show a low level of polymorphism when compared to the wild-type M2 . These data, therefore, do not rule out a possible spread by vegetative propagation. The utilized SSR markers were developed for A. cherimolia and no specific SSR primers were available for A. squamosa and some outbreeding could have been missed in this analysis. But with the current data, the simplest explanation is that a single mutational event was dispersed to different continents through vegetative propagation. Tracing a temporal and geographical route of this dispersal is difficult since the species has been widely developed and naturalized. It is believed that the primary center of origin of A. squamosa is in the lowlands of Central America. Historical data indicate accessions to Mexico by the natives of the region . According to these authors, after their arrival, the Spaniards were responsible for spreading the seeds to the Philippines. By 1590, the species had been introduced into India. In Brazil, 1626 is the first record of introduction of the species. By the early seventeenth century the species was already widespread in Indonesia, China, Australia, Polynesia and Hawaii. In 1955, the Cuban seedless variety had been introduced in the state of Florida in the United States. The mutant Brazilian accession Bs was first described in 1940 in the state of São Paulo . We obtained Hs from a commercial nursery in Hawaii. The nursery reports obtaining the line from the Philippines. Their source in the Philippines reports that the line was brought to the Philippines from an unknown location in the 1920s, but documentary evidence of this was not found. Information on the origin of Ts was not available. Lora et al. designed primers within, or closely fanking the transcribed region of INO for detection of the wild-type locus, and demonstrated amplification from wild-type A. squamosa, A. cherimola, and four other species that cover the phylogenetic range of the genus Annona. These primers were confirmed to function as expected in our A. squamosa varieties . Notably the primer pairs only allow detection of the wild-type locus and a homozygous mutant could only be detected by the absence of amplification. According to Singh and Singh , markers that can directly identify the traits of interests should be more efcient in incorporating and monitoring genes in breeding programs. Our determination of the sequence of the region from which INO has been deleted enabled formulation of primers for direct detection of the mutant allele allowing codominant detection of both alleles in wild-type and seedless mutant parents and to differentiate fertile heterozygotes from homozygous fertile plants in a single reaction. The improvement of perennial plants, especially fruit species, is a challenging task due to a long juvenile period and seed dormancy, among other factors, which translates into a relatively long generation period .