The Indica rice cultivar 9311, which has a partial loss of function NAL1 allele, and the near-isogenic line (9311-NIL), which is derived from the Japonica rice cultivar Nipponbare (Zhang et al., 2014) and has the fully functional NAL1 allele, were used to generate the NAL1 hybrid line. The thermo-sensitive genic male sterile line PA64S had the fully functional NAL1 allele, which was the same as 9311-NIL, and Y58S had the partially functional allele, which was the same as 9311. The near-isogenic sister line of the super hybrid rice variety (LYP9 or YLY1) was generated by crossing the corresponding male sterile line maternal parent (PA64S or Y58S) with 9311-NIL, respectively.

The field experiments were executed in two consecutive summer growth seasons (2022 and 2023) at the experimental station of the China National Hybrid Rice R&D Center (HHRRC) located in Changsha (28.2°N, 113.2°E). Seeds of each line were sown in the seedling nursery and 25-day-old seedlings were transplanted into the field with a density of one plant per 20 cm × 20 cm. The fields were managed according to local agricultural practices for growing rice: 225 kg N ha^–1, 135 kg P₂O₅ ha^–1, and 250 kg K₂O ha^–1.

Flag leaf length, width, and area at the early grain filling stage were measured using a handheld laser leaf area meter (Ci-203, CID, USA). Nine flag leaves from nine plants were measured for each line. For total N content analysis, flag leaves at the early grain filling stage and straws and grains at the maturity stage from the Changsha paddy field were collected separately. Ten replicates, each containing three plants, were measured for each line. The dried samples were milled and analyzed with an NC Analyzer (FlashSmart NC; Thermo Fisher, USA).

The net leaf photosynthetic CO₂ assimilation rate (A) under different photosynthetic photon flux density (PPFD) (A–Q curve) was measured according to Chang et al. (2020). Measurements were made on the flag leaves at the heading stage between 09:00 am and 16:00 pm on clear days with a portable photosynthesis system (LI-6800, Li-Cor, USA) in the 2023 summer season at Changsha. For each line, five flag leaves from five different plants were used.

Canopy photosynthesis was measured using a canopy photosynthesis and transpiration system (CAPTS-100, MilletHill, China). For each measurement, the chamber (with a size of 100 cm * 100 cm) contained five plants in each line. A total of 25 plants in each line were used. A detailed description of the protocol used for data acquisition and analysis is described by Song and Zhu (2018).

The plant canopy architecture was analyzed at the grain-filling stage using a plant 3D imaging and modeling system (a-CTP, MilletHill, China). Five plants without any broken leaves in each line were selected. Each plant was photographed at omnidirectional angles individually, and then the 3D plant architecture information was integrated into the point cloud data using Metashape Agisoft software. The point cloud data were reconstructed by the CERS suite software using default values for the analysis parameters.

Proteins of the axillary buds at the tillering stage grown in the paddy field were powdered using liquid nitrogen and extracted with SDT lysis buffer (2% SDS, 100 mM NaCl, and 1/100 volume of DTT). Protein sample concentration was quantified by using a Bradford protein assay kit with BSA. Then each protein sample was hydrolyzed with trypsin (100 ng) and the volume was made up to 100 μL with DB lysis (8 M Urea, 100 mM TEAB, pH 8.5). The samples were subjected to UHPLC-MS/MS analysis. The data analysis and visualization were conducted by Novogene Co., Ltd. (Beijing, China) using the Proteome Discoverer 2.2 (PD 2.2, Thermo Fisher Scientific) platform, Biognosys Spectronaut v. 9.0, and R statistical framework. MS2-based label-free quantification was carried out by analyzing data-independent acquisition (DIA) raw data using Biognosys Spectronaut v.9. Differentially expressed proteins (DEPs) were identified according to a fold change >1.5 and a corrected P-value < 0.05. Functional and pathway analysis of the DEPs was conducted using Gene Ontology (GO) annotation (http://www.geneontology.org/).

For Western blot analysis, a fragment of a 513 bp coding sequence of NAL1 was amplified by a primer pair of NAL1-Anti-For1/Rev1 ( Supplementary Table S1 ) and was fused into a PET-28a-SuMO vector. The purified fusion protein was injected into rabbits to produce polyclonal antibodies against NAL1. Total proteins were extracted from the axillary bud of NAL1 plants by using a Plant Total Protein Extraction Kit (Solarbio, Beijing, China). Immunoblots were performed using primary antibodies against NAL1 and ACTIN (Cat# AT0001, Engibody, 1:2000 dilution). After incubating with secondary antibodies HRP-labeled Goat Anti-rabbit IgG (H + L) (Cat# AT0097, Engibody; 1:2000 dilution) and HRP-labeled Goat Anti-mouse IgG (H + L) (AT0098, Engibody; 1:2000 dilution), respectively, the immunoblot signal was visualized using the Immobilon Western HRP substrate (Cat# WBKLS0100, Millipore).

Total RNA was extracted from the axillary bud using TRIzol reagent (Invitrogen). Reverse transcription was performed with the ReverTra Ace qPCR RT Master Mix with gDNA Remover (Toyobo) using 1 μ g total RNA. Quantitative RT-PCR analysis was carried out with a LightCycler 480 engine (Roche) using the LightCycler480 SYBR Green I Master Mix (Roche). The relative quantification method (2^−ΔΔCT) was used to evaluate the quantitative variation of expression after normalization to Ubiquitin. Five replicates of each line were analyzed to produce the mean values for NAL1 expression level. The primers used for RT-qPCR analysis are listed in Supplementary Table S1 .

The plant height, biomass, and yield-related data were collected from plants in each plot, excluding the two marginal columns. The yield-related heterosis contribution rate of NAL1 was calculated according to Wang et al. (2019). For biomass and yield evaluation of the hybrid rice from LYP9 and YLY1 with the corresponding NIL lines, plants from 2 rows x 4 columns were taken from the inside of the plots to eliminate boundary effects. Four replicates were measured for each line.

In a previous study, we elucidated that the fully (9311-NIL) and the partially functional (9311) NAL1 alleles have different specific advantageous yield-related traits (Ouyang et al., 2022). As the advantageous traits of the two homozygous NAL1 alleles showed great complementarity, we speculated that heterozygous plants may have a better yield performance than the homozygous plants. To verify this hypothesis, the two elite lines and their crossed offspring were used for further analysis. In the two consecutive years of field trials at Changsha, we found there were no significant difference between 9311 and 9311-NIL in biomass and yield. However, the NAL1 hybrid plants exhibited significantly higher biomass and yield than both homozygous parents ( Figures 1A–D ). The yield performance of the NAL1 hybrid plants that exceeded the better parent (named better-parent heterosis, BPH) was up to 27.15% ( Figure 1E ). The solo NAL1 locus could provide an 8.64% heterosis effect contribution rate ( Figure 1F ). These results suggest the heterozygous NAL1 variety has improved growth performance and yield production.

We further analyzed the three yield components and found that contrary to TGW, both the panicle number and the spikelet number per panicle were greatly increased in the NAL1 hybrid plants ( Figure 2 ; Supplementary Figure S1 ). When compared with the two homozygous parents, the hybrid plants had significantly higher panicle numbers ( Figures 2A, B ). Even though the number of spikelets per panicle was between that of the two homozygous parents ( Figures 2C, D ), the total filled grain number per plant in the hybrids also showed a significant increase ( Figures 2E, F ). These suggest that the heterozygous NAL1 allele has BPH for panicle number and the middle-parent heterosis (MPH) for number of spikelets per panicle compared to the homozygous plants.

Leaf type analysis showed that the heterozygous NAL1 plants had decreased flag leaf length, width and area than 9311-NIL, but the three traits were significantly increased when compared with 9311 ( Figure 3 ). Furthermore, A-Q curve analysis showed that the NAL1 hybrid plants had similar leaf photosynthetic CO₂ uptake rate with 9311 under different PPFD, which both were higher than 9311-NIL ( Figure 4 ). Although we found no significant differences between the hybrids and two homozygous parents in flag leaf N content, the leaf N accumulation was gradually increased from the 9311, hybrids to 9311-NIL ( Supplementary Figure S2 ). These suggest that the heterozygous NAL1 plants not only have the MPH in leaf type but also have the comparative leaf photosynthetic capacity with parent 9311.

Canopy architecture is a major yield determinant of crops because of its effect on light distribution, light interception, and canopy photosynthesis (Zhu et al., 2010; Long et al., 2015). We investigated the canopy architectures of the hybrid plants and the two homozygous parents. The results showed that there were no differences in plant height between the hybrids and the homozygous parents ( Supplementary Figure S3 ). The canopy occupation volume (COV), which is a key architectural trait and can reflect the synergistic effect of leaf area index (LAI) and leaf angle on canopy photosynthetic capacity (Liu et al., 2021), showed great variance among three NAL1 lines. The results showed that 9311 had a significantly higher COV value than 9311-NIL. However, the COV of 9311 was still significantly lower than the hybrids ( Figure 5A ). This suggests that the NAL1 hybrid plants had greater canopy photosynthetic capacity than 9311 and 9311-NIL. To confirm this, we further investigated the canopy photosynthetic rate of the heterozygous NAL1 plant and the homozygous parents during the grain-filling stage. The results showed that the hybrid plants had the highest total canopy photosynthetic CO₂ uptake rate under both high and low light conditions compared with 9311 and 9311-NIL ( Figure 5B ). These results suggest that the heterozygous NAL1 allele improved canopy architectures and contributed to higher canopy photosynthetic efficiency.

A previous study confirmed that NAL1 acts as a serine protease (Li et al., 2023), thus, in order to further illustrate the difference between the parents and hybrid regarding translation levels, we performed a proteomic analysis using axillary buds from 9311, 9311-NIL, and the hybrids. Principal component analysis (PCA) revealed that the replicates of each NAL1 genotype were clustered into the same group ( Supplementary Figure S4A ). A total of 708 differentially expressed proteins (DEPs) were found to be significantly differentially expressed (corrected P-value <0.05, log2-fold change >1.5) in two compared groups, including 337 DEPs specifically in the hybrid and 9311 compare group (Hy-9311), 295 DEPs specifically in the hybrid and 9311-NIL compare group (Hy-NIL), and 76 DEPs in both groups ( Supplementary Figure S4B ; Supplementary Data Sheet 1 ). There were 177 DEPs upregulated and 236 DEPs downregulated in the Hy-9311 group, and 114 DEPs upregulated and 257 DEPs downregulated in the Hy-NIL group ( Supplementary Data Sheets 2 , 3 ). This suggests the total protein expression levels of DEPs in the hybrid plants were downregulated.

Gene ontology (GO) enrichment analysis showed that most of the downregulated DEPs in the NAL1 hybrid plants were enriched in oxidation-reduction process, single-organism metabolic process, fatty acid biosynthetic process, and defense response when compared with 9311 or 9311-NIL. In contrast, the upregulated DEPs in the hybrids were mainly involved in DNA binding, nucleosome assembly, and embryo development ( Supplementary Figure S5 ). These suggest that the organic components metabolic process might be weakened, but the transcription activity was strengthened in the hybrids. There were several critical proteins were significantly upregulated in the NAL1 hybrid plants when compared with the parents, such as Squamosa promoter-binding-like proteins (OsSPL4, OsSPL5, OsSPL14, and OsSPL18), OsALS3, and OsAPO1 ( Figure 6 ). All of them have been identified to play key roles in regulating the plant architecture and grain yield previously. These suggest that the heterozygous NAL1 alleles might cooperate with OsSPLs, OsALS3, and OsAPO1 to regulate tiller and panicle development. It is interesting that the protein levels of NAL1 gradually decreased from 9311 to the hybrids and to 9311-NIL according to the proteomic data ( Figure 7A ). The Western blot assay further confirmed that 9311 had the highest NAL1 protein level, and 9311-NIL had the lowest protein level by using the anti-body of NAL1 ( Figure 7B ). Furthermore, a similar change pattern for NAL1 among the three lines was also found in the transcription level ( Supplementary Figure S6 ). These results suggest that the NAL1 hybrids have middle abundant NAL1 protein and transcription levels.

The above agronomic trait data suggest that NAL1 would have great value in hybrid rice breeding. To further confirm this, we chose two super hybrid rice varieties, Liang-you-pei-9 (LYP9) and Y-liang-you-1 (YLY1), which shared the same male parent, 9311, but had the different female parents, i.e., PA64S and Y58S, respectively. According to our hypothesis, if we cross 9311-NIL with the female parents PA64S and Y58S separately, the yield of the resulting hybrid lines (LYP9-NIL and YLY1-NIL) will change accordingly. The paddy field experiment results showed a significant decrease in biomass and grain yield in LYP9-NIL when compared with LYP9 ( Figures 8A, C ). The yield reduction was 9.25%. The opposite was also true after the homozygous NAL1 locus in YLY1 was substituted with heterozygous NAL1. When compared with YLY1, YLY1-NIL had an increased grain yield of 10.1% ( Figures 8B, D ). These suggest that the hybrid rice yield can be further increased by using heterozygous NAL1.

Crop yield is a complex trait determined by the balance of photosynthetic assimilates (source) and photosynthate utilization (sink) (Li et al., 2018). A large number of genes have been identified to play important roles in regulating the source and or the sink (Xing and Zhang, 2010; Zuo and Li, 2013; Li et al., 2018; Wang et al., 2018; Ren et al., 2023). NAL1 has been confirmed to pleiotropically affect plant height, leaf morphology, photosynthesis efficiency, spikelet number, panicle number, and so on. This greatly constrains the application potential of NAL1 as a result of the trade-off traits related to the source and sink. In this study, we tested an effective method for comprehensively utilizing the advantageous traits of two NAL1 functional haplotypes. Our results showed the NAL1 hybrid plants had a larger leaf size than 9311 ( Figure 3 ) and had a higher leaf photosynthetic rate than 9311-NIL ( Figure 4 ). These results indicate that the combined NAL1 alleles had synergetic effects on plant architecture and leaf photosynthesis. Furthermore, the canopy photosynthesis capacity of the hybrid plants was also enhanced with the highest panicle number ( Figures 2A, B ), the largest canopy occupation volume ( Figure 5A ), and the highest canopy photosynthetic rate ( Figure 5B ) when compared with 9311 and 9311-NIL. All these results together contribute to a strengthening of the plant source by the combined NAL1 alleles. Similarly, our results also showed the NAL1 hybrid plants had the highest panicle number and total filled grain number per plant compared with 9311 and 9311-NIL ( Figures 2A, B, E, F ). This suggests the hybrid NAL1 allele had over-dominance effects in panicle number and total filled grain number per plant. Moreover, the hybrids had an MPH performance for spikelet number per panicle ( Figures 2C, D ). As a result, the final sink size of the NAL1 hybrids was increased by the synergistic effect of yield component traits.

The role of NAL1 in rice yield has been identified in many previous studies (Takai et al., 2013; Zhang et al., 2014; Xu et al., 2015; Hirotsu et al., 2017; Ouyang et al., 2022; Li et al., 2023; Zhai et al., 2023; Wang et al., 2024). The ambiguous effect on yield may be due to the concurrent impacts of NAL1 on source and sink-related traits (Ouyang et al., 2022). As the trade-off effect among yield component traits may be the main reason for a limited rice yield, a balanced or coordinated relationship between the panicle, grain numbers, and grain weight seems to make it easier to produce a high yield. However, it is still important to explore effective ways to apply the genes for higher yield production. One of the effective methods to relieve the trade-off effect might be genome editing of the promoter of the target gene using CRISPR/Cas9 to generate beneficial alleles. This method has the most successful examples in rice (Song et al., 2022) and tomato (Rodriguez-Leal et al., 2017). In this study, we tested another effective way to apply NAL1 for higher rice production yield. Our result showed the NAL1 hybrid plants had increased source size and activity which together contribute to significantly higher biomass and yield than both the homozygous parents ( Figures 1A–D ). The calculated heterosis effect contribution rate offered by the solo NAL1 locus reached 8.64% ( Figure 1F ). Furthermore, when the heterozygous NAL1 alleles were substituted with homozygous NAL1 in the super hybrid rice LYP9, the yield reduction was up to 9.25% ( Figure 8C ). In contrast, when homozygous NAL1 in YLY1 was substituted with the heterozygous alleles, YLY1-NIL had an increased grain yield of 10.1% ( Figure 8D ). These results further confirm the positive over-dominant effect of the NAL1 locus and the heterozygous NAL1 allele has great potential to improve rice production yield.