Persimmon study traces natural astringency loss to two types of gene damage

Persimmon’s powerful astringency comes from its accumulation of proanthocyanidins, also called condensed tannins. In cultivars classified as pollination-constant and non-astringent, or PCNA, that accumulation stops early in fruit development. Researchers have long linked this commercially important trait to a genetic region called ASTRINGENCY, or AST, without identifying the responsible gene.

A bioRxiv preprint by Yosuke Fujiwara and colleagues now identifies TANNIN, abbreviated DkTAN in persimmon, as the likely answer. The study combines long-read DNA sequencing, inheritance analysis and RNA measurements to argue that two distinct mutation types disable the gene. The work has not been peer reviewed.

A six-copy genetic puzzle

Cultivated persimmon is hexaploid: it has six sets of chromosomes. The PCNA trait studied here is recessive, requiring recessive versions at all six copies of the AST locus. Resolving which sequences belong to which inherited copy is therefore central to understanding the trait.

The team began with the non-PCNA cultivar ‘Taiten’, using long DNA reads and information from its parents to reconstruct inherited sequence versions, known as haplotypes, across the previously mapped AST region. The assembly did not recover every copy separately; two highly similar versions were treated as one group for subsequent analysis.

Comparisons with other cultivars and offspring from a breeding cross concentrated the strongest candidate DNA differences within one gene. Related to an uncharacterized Arabidopsis gene, it had no established biochemical role. The researchers named it TANNIN.

The difference was in the RNA instructions

Total DkTAN expression did not distinguish PCNA from non-PCNA fruit. Instead, the important difference emerged in the structure of its RNA transcripts—the intermediate instructions used to make proteins.

The researchers sequenced gene-derived RNA from immature fruit across 25 PCNA cultivars and 106 non-PCNA cultivars, alongside six related Diospyros accessions. In PCNA cultivars, an average of about 1.2% of the reads were predicted to encode full-length DkTAN proteins. In every non-PCNA cultivar examined, at least 25% did so.

A similar distinction appeared among ten offspring selected from the ‘Fuyu’ × ‘Yoshidagosho’ breeding population. The proportion of full-length transcripts also increased with the number of dominant AST copies, strengthening the link between the gene and the inherited trait.

These measurements concern predicted protein products, not direct tests of protein activity. The researchers classified transcripts as functional primarily from their predicted length and absence of premature stop signals.

Two different routes to disruption

DNA and RNA comparisons pointed to two separate mutation types. One is an approximately 1,500-base deletion that removes exon 7 and part of exon 8—sections normally retained in the protein-coding message. The other is a single-base change near the splice site of exon 3, associated with abnormal processing of the RNA.

Both mutation types were linked to transcripts predicted to produce shortened proteins. Crucially, they occurred in different inherited sequence groups. A cultivar could therefore carry different defective versions across its six copies, rather than needing six identical copies of one mutation.

To examine that arrangement, the team enriched the DkTAN region and sequenced it in 21 PCNA and four non-PCNA cultivars. All reconstructed DkTAN sequences from the PCNA cultivars carried one or the other disruption. The authors interpret this as accounting for the absence of intact gene copies across all six alleles in those cultivars.

Strong genetic evidence, an unresolved mechanism

The convergence of mapping, inheritance, transcript abnormalities and two distinct disruptive mutation classes provides substantially stronger evidence for causality than a simple cultivar association. Nevertheless, the study does not report a targeted gene knockout that creates the trait, or restoration of an intact gene that reverses it. Such experiments would directly test the proposed causal relationship.

The evolutionary reconstruction also remains an inference. Sequence relationships suggest that the two mutation classes arose independently and later accumulated within hexaploid genomes. They do not establish precisely when those events occurred. A diversity analysis found no evidence of a local selective sweep, but compared only nine PCNA and four non-PCNA cultivars.

For breeders, identifying the likely gene and its disruptive variants offers a basis for more precise selection, rather than relying solely on nearby markers. The study does not yet demonstrate improved breeding outcomes. It also leaves open what TAN actually does: its proposed involvement in persimmon’s exceptional tannin accumulation and intracellular storage remains a question for future experiments.