It does not hurt to double up: the role of translation in inter-subspecific hybrid rice

Early plant breeders thought that self-pollination and inbreeding would eventually lead to ‘pure’ and superior plants (Curry, 2023). So, they inbred maize generation after generation, but against their expectations the plants became smaller and smaller, and the yield decreased. In the early 20th century, the plant geneticist George Harrison Shull crossed two maize inbred lines to create F1 hybrids (Crow, 1998). Their yield exceeded the ones of the lines they were derived from. Shull named this phenomenon ‘heterosis’, as shorthand for ‘stimulation of heterozygosis’. Nowadays, nearly all maize production in the United States relies on hybrids (Duvick, 2005).

In contrast to maize, rice is a self-pollinating crop. Only the discovery of a naturally pollen-aborting wild rice in the 1970s promoted the commercial development of hybrid rice breeding because the pollen-aborting variety could be used as female in the cross (Gu & Han, 2024). Hybrid rice has significantly contributed to addressing global food security. For Jianbo Wang, professor at the State Key Laboratory of Hybrid Rice in Wuhan, this has always been the driving force behind his dedication to hybrid rice research.

Heterosis has been explored from different angles, but its genetic foundation remains poorly understood. Because translational regulation is important for different abiotic stresses, such as chilling, drought and salinity (Lei et al., 2015), the group decided to analyse whether translation might be important for heterosis. Therefore, for the highlighted study, the PhD student in Wang's group, Zengde Xi, and the team set out to examine genome-wide translational dynamics in the hybrid rice line ZY19 (Xi et al., 2025). ZY19 is a cross between parents of the two subspecies indica and japonica (Z04A and ZHF1015), which generally shows higher yield than intra-subspecific hybrids (indica × indica or japonica × japonica), making it particularly interesting for the study.

The authors sampled flag leaves of ZY19 and its parental lines and subjected them to RNAseq and ribosome profiling (Ribo-seq) (Figure 1a). Ribo-seq is a deep sequencing technology that captures ribosome-protected mRNA fragments and thereby actively translated mRNA (Ingolia et al., 2009). They found that around 80% of genes were expressed in all three lines at both the transcriptome and translatome levels (Figure 1b). The authors then identified differentially expressed genes (DEGs) between ZY19 and its parents by comparing gene expression at the transcriptional level. Of those DEGs, over 60% were more highly expressed in ZY19. However, fewer than 10% of the DEGs overlapped between the transcriptome and translatome, suggesting discrepancies between transcription and translation.

The translation efficiency (TE), that is the ratio of normalized Ribo-seq read abundance to normalized RNA-seq read abundance, showed that over 1,000 TE genes were significantly different between ZY19 and its parents (Figure 1c). More highly expressed genes were linked to growth-related GO terms (e.g. transcription, metabolism), while ones less highly expressed were related to stress responses. A negative correlation between mRNA abundance and TE suggests a translational buffering effect, which could conserve energy in ZY19 by minimizing unnecessary translation from misregulated transcripts. For example, OsFLO19, a gene-promoting grain filling and yield (Lou et al., 2021), had a different TE in ZY19 than in the parents.

Some studies suggested that additive gene expression through transcription contributes to heterosis, that is the expression levels in the hybrids equal the mid-parent value (Guo et al., 2006). This could benefit hybrids by reducing negative effects of either under-expression or over-expression. In ZY19, the authors found strong additive effects at both the transcriptional and translational levels. A small proportion of genes switched from additive to non-additive or vice versa during translation. Most non-additive genes (60%) had a lower expression level than the mid-parent value at the transcriptional level, but at the translational level, most non-additive genes showed expression above the mid-parent value, suggesting a higher expression of non-additive genes during translation.

Allele-specific expression, that is the preferential expression of alleles from different parental lines, has also been linked with heterosis (Botet & Keurentjes, 2020). The authors found that more allele-specifically expressed genes favoured the Z04A allele at the mRNA level; but no significant bias was observed in TE. Parental expression patterns influenced allele-specific expression, but this effect was diminished at the translational level.

Cis- and trans-regulation of gene expression has also been implicated in heterosis (Sun et al., 2023). Cis-regulation refers to regulatory elements of a gene that are located on the same DNA molecule as the gene itself and directly influence its expression, such as upstream open reading frames (uORFs) and downstream open reading frames (dORFs). Trans-regulation refers to regulatory factors that are not located on the same DNA molecule as the target gene but instead regulate gene expression by reaching the target site through mechanisms such as diffusion or transport, such as microRNAs (miRNAs). By comparing which genes were cis- and which were trans-regulated, the authors found that cis- and trans-regulation tended to function independently in transcription; however, they were more likely to act together in translation. Furthermore, genes with either uORFs or dORFs had significantly lower TE than those without uORFs in the hybrid or its parental lines, suggesting that both types of ORFs may repress mRNA translation.

Post-transcriptional regulatory mechanisms, such as N6-methyladenosine and miRNAs, play key roles in controlling gene expression linked to crop yield, as these elements affect grain development, size and number by modulating gene activity at the transcriptional level (Yu et al., 2021). Here, the authors expand these mechanisms by demonstrating that they also repress TE on a global scale. They found several yield-related genes affected by these regulators, influencing traits such as grain weight, panicle number and tillering. For example, OsCIPK1, which contains an uORF, is connected to the gibberellic acid pathway to regulate grain size and weight (Zhang et al., 2023). These findings suggest these regulatory elements may contribute to heterosis by affecting the translation of yield-related genes, revealing important differences in transcriptional and translational activity between hybrids and their parental lines.

In summary, the authors demonstrated that certain heterosis hypotheses hold true not only at the transcriptional level but also at the translational level. Their next objective is to identify genes in the hybrid that are significantly more highly expressed at the translational level compared to the parental lines, and among these, to pinpoint those that positively influence rice yield. The findings suggest that different genes may adopt distinct strategies to regulate their translation. Since translational regulation is faster and more direct than transcriptional control, the authors propose that future research in hybrid rice breeding should place greater emphasis on the role of translational regulatory mechanisms in gene expression.