Plant Science & Agricultural Biology

The Nuclei Never Merged, and a Plant Grew Anyway

A single amino acid change in a cotton gene delays the moment sperm and egg nuclei fuse. Embryos develop anyway, and some carry only the father's genome.

BioBot
·
November 19, 2026
·
5 min
Article hero

The slow step in crop breeding is not making the cross. It is making the result uniform. Offspring of a cross carry two different versions of every gene, and a breeder spends generation after generation of self-pollination grinding that variation out. The shortcut is the doubled haploid: get an embryo to develop from one parent's chromosomes alone, double them chemically, and a fixed line arrives in a single step. Almost every system that does this returns the mother's genome and throws the father's away.

Cotton has an exception that has gone unexplained since the 1960s. In Pima cotton, a trait called semigamy produces haploid seedlings at frequencies reaching 61% in the classical Pima S-1 variety, and it produces them from either parent. New work traces that behaviour to a single gene, GEX1, and shows that what fails is not fertilisation but the step just after it, when the two parental nuclei are meant to merge.

Why it matters: An inducer that works in both directions returns two fixed lines from one cross rather than one, cutting the cost of the slowest stage of variety development. It also pins down a step in sexual reproduction that had been treated as automatic.

Li et al., at the Institute of Cotton Research of the Chinese Academy of Agricultural Sciences, with collaborators at the Institute of Genetics and Developmental Biology in Beijing and the Leibniz Institute of Plant Genetics and Crop Plant Research in Germany, report the work in Science Advances.

Fertilisation has two steps, and only one failed

Double fertilisation entails three nuclear fusions inside the embryo sac: the two polar nuclei join to form the central cell nucleus, one sperm nucleus joins the egg to make the zygote, and the second joins the central cell to start the endosperm. Each is preceded by plasmogamy, the fusion of outer membranes that delivers a sperm nucleus into position. Karyogamy, the fusion of the nuclear envelopes themselves, follows so dependably that the two are usually discussed as one event.

In the semigamy mutant they come apart. Pollen was normal and stained as viable, and the embryo sac formed correctly. The failure appeared only at karyogamy: where wild-type ovules had fused their polar nuclei and begun forming free endosperm nuclei, mutant ovules at the same stage had none, and the sperm and egg nuclei sat adjacent but unmerged. The two then divided independently, producing four daughter nuclei spread across what would become the apical and basal regions of the embryo. An embryo built from one of those lineages carries one parent's chromosomes and nothing else.

One gene in a 4.85-megabase window

The mapping is unusually clean. Bulked segregant analysis on thirty plants with the trait and thirty without gave a single peak on chromosome A12 and no secondary peak anywhere else in the genome. The interval spans about 4.85 megabases and holds 135 annotated genes. Across all of them, exactly one carried a nonsynonymous change in its coding sequence: a substitution at position 443 swapping threonine for isoleucine at residue 148. That gene is the cotton version of GEX1, whose yeast counterpart is required for nuclear envelope fusion.

The change sits in a cysteine-rich domain but spares all six conserved cysteines, and structural prediction suggests the domain keeps its shape, so the authors propose a disturbance of local interactions rather than a broken protein. Expression matched the role, highest in mature pollen and in early endosperm, with a promoter reporter active in sperm cells. A survey of cultivated and wild cottons found no other nonsynonymous variant in this gene, making the semigamy allele a one-off.

The knockout works, and the gap it leaves

Correlation of this kind is suggestive rather than sufficient, so the group knocked the gene out with CRISPR-Cas9 in an upland cotton background. The edited plants reproduced the cytology: only 48.91% of their ovules had reached the free-endosperm stage at which nearly all wild-type ovules had arrived. They also produced haploids from both parents, at roughly 1.2% maternal and 0.3% paternal in outcrosses. Used as a pollen donor rather than as the mother, neither the natural mutant nor the knockout induced anything, which places the defect on the female side of the cross.

The knockout's self-pollinated haploid rate, about 1%, is more than twenty times below the natural mutant's figure above 21%. A complete loss of function is therefore worse at the job than a single amino acid change, which is a useful warning against reading this gene as a simple switch.

What the study can't say yet

The paper calls the gene the leading candidate rather than the proven cause, and the distinction is real: the specific threonine-to-isoleucine substitution was never introduced into a normal plant to see whether it alone produces semigamy. What was tested is a frameshift knockout, which produces a related but quantitatively different phenotype. Fertility is not free either. Bolls on the mutant held aborted ovules, the proportion of properly developed seeds per boll fell from above 95% to below 90%, and the absolute number of viable seeds dropped.

Nothing here follows a doubled haploid line through to a field trial; chromosome doubling and breeding deployment are proposed, not demonstrated. Paternal haploids, the genuinely novel product, arrive at a fraction of a percent. On mechanism the authors are careful about what the result establishes: These observations position nuclear fusion not merely as a downstream event of fertilization but as a key regulatory checkpoint safeguarding biparental inheritance during sexual reproduction. Why the same gene yields mostly maternal haploids in maize and both kinds in cotton is left open.

Quick questions

What is a haploid plant? One carrying half the usual chromosome complement. In this study the haploids were smaller, with visibly reduced flowers, but doubling their chromosomes gives a plant that is homozygous at every position.

Why does getting haploids from the father matter? Existing inducers almost always discard the father's contribution, so recovering both sides of a cross as separate fixed lines doubles what one pollination yields.

What is the one-line takeaway? A single amino acid change delays the fusion of sperm and egg nuclei in cotton, and embryos that start dividing before that fusion inherit from one parent only.

Sources

Li et al. "GEX1-mediated semigamy induces both maternal and paternal haploids in cotton." Science Advances, 2026. doi.org/10.1126/sciadv.aei1433

PubMed PMID: 43024697.

Image: Gossypium barbadense in flower, Enchanting Gardens of Kula, Maui. Forest and Kim Starr, CC BY 3.0, via Wikimedia Commons.

Comments

Comments

Stay current on biology.

Weekly research updates, breakthrough summaries, and new articles — straight to your inbox. Free, always.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.