The germinated rice seeds were grown in a plant greenhouse at 30°C during 14 h daytime and 25°C with 10 h night under 70% humidity. All seeds were harvested at maturity stage, and stored at room temperature after drying at 42°C for 7 days.

The konck-out vector Cas9-OsLTPL23-gRNA was generated as previously reported (Xu et al., 2014). The specific sgRNA sequences targeted for OsLTPL23 were inserted into the pHUN4c12 vector digested with Bsa I-HF using T4 DNA ligase reaction (2011A; Takara, Japan). The resultant construct Cas9-OsLTPL23-gRNA was transformed into the japonica variety NIP through the Agrobacterium tumefaciens strain EH105-mediated stable transformation (Nishimura et al., 2006). Resistant rice calli were vigorously grown in hygromycin-containing medium, and finally transferred to regeneration medium to obtain transgenic plants.

To identify the osltpl23 mutants, the genomic DNA was extracted from the leaves of hygromycin-resistant plants using CTAB method, and then used as template to perform PCR amplification with OsLTPL23-Cas9-detection primers (Supplementary Table S4). All the DNA sequences of PCR products from the above plants were directly determinate to identify Cas9-editing events using Sanger sequencing techniques. Homozygous mutants and heterozygotes were indicated as normal sequencing chromatograms carrying simple indels and superimposed sequencing chromatograms, respectively.

To determine transgenic-free mutants, the T1 generation plants of Cas9-cutting lines were analyzed by PCR amplification with HPT-specific primers (Supplementary Table S4). The Cas9-OsLTPL23-gRNA vector and genomic DNA of NIP were chosen as positive control and negative control, repectively. The HPT-negative mutation lines were marked as transgenic-free mutants.

Fifty seeds per replicate of each osltpl23 mutants and NIP were germinated on Petri dishes containing moistened paper towels at 26°C for 6 days. The seed germination criterion and germination rate determination were as stated in Wang et al. (2021). The seeds germination assay was conducted in triplicate.

Total RNA was isolated from the germinating seeds (12, 24, 48, and 72 h post imbibition (hpi)) and various tissues at two developmental stages (root, leaf and sheath of five-leaf stage seedlings; stem, flag leaf and spike of booting stage plants) of NIP, and two-week-old seedlings of NIP, osltpl23-1 and osltpl23-2 mutants using the TRIzol™ reagent (15596-026; Invitrogen, Carlsbad, California, United States). 1 μg total RNA was used to first-strand cDNA synthesis by the ReverTra Ace qPCR RT Master Mix with gDNA Remover (FSQ-301; TOYOBO, Osaka, Japan). The qRT-PCR was performed using TransStart^® Tip Green qPCR SuperMix (+Dye I) (AQ142-11; TransGen, Beijing, China) with corresponding primers on CFX96^® Real-Time PCR system (Bio-Rad, Hercules, California, USA), and OsACTIN genes was used as the internal control for normalization. A complete list of primers is included in Supplementary Table S4. The relative transcript levels of indicated genes were quantified by 2^−ΔΔCT method from the qRT-PCR data of three biological replicate experiments with three independent repeat (Schmittgen and Livak, 2008).

The protein-coding sequences of OsLTPL23 (without the termination codon) were cloned from the cDNA of 2-weeks old NIP seedlings, and inserted into the PMDC43 and the pSAT6-eYFP-N1 vector, respectively. Then, the plasma membrane marker AtPIP2A-mCherry (Cutler et al., 2000) or nucleus marker bZIP73-mCherry (Liu et al., 2019) were separately introduced into N. benthamiana epidermal cells with PMDC43-OsLTPL23 construct via Agrobacterium tumefaciens strain EH105-mediated transient transformation or infiltrated into rice protoplasts with pSAT6-OsLTPL23-eYFP-N1 construct using polyethylene glycol (PEG)-mediated transformation method, respectively (Chen et al., 2008). The fluorescence intensity of recombinant proteins was photographed at ZEISS LSM 710 NLO (Carl Zeiss, Oberkochen, Germany).

The primary sequences of OsLTPL23 were blasted and aligned to the plant homologues in NCBI and Phytozome database. A neighbour-joining phylogenetic tree between OsLTPL23 and its homologues was constructed by MEGA X with the Poisson correction model, pairwise deletion for gaps/missing data treatment and 1,000 of bootstrap replicates (1,000 replicates) (Kumar et al., 2016).

The dry rice seeds were successively ground into fine powder filtered with 100-, 200-, and 400-mesh sieves. The amount of starch and soluble sugar was separately quantified with 0.03 g and 0.2 g samples according to the manufacturer’s instructions (BC0705, BC0035; Solarbio, Beijing, China).

All experimental data were performed with GraphPad Pism 8 software and statistical analyses among samples were compared using Student’s t-test at the 5% and 1% levels of probability.

To reach a better understanding of nsLTP-based regulation of seed vigor and identify the nsLTPs responsible for seed germination in rice, we functionally analyzed the reported rice nsLTP members with gene expression in seed or embryo (Wang et al., 2012). Through phylogenetic relationship, we found that a total of 15 nsLTP-encoding proteins can be classified into five clade (Figure 1A), which agrees with previous investigation (Wang et al., 2012). Followed the tissue expression profiles from the Plant Regulomics database, the transcripts of the nsLTP genes, OsLTPL8, OsLTPL9, OsLTPL22 and OsLTPL23, in clade II were specifically enriched in the developing seed, with high expression in 5 days after pollination (DAP)-seed and -embryo and low accumulation in 5 DAP-endosperm, suggesting that this clade may be involved in carbohydrates accumulation during rice seed maturation (Figure 1B; Supplementary Table S1). The phytohormones ABA negatively regulates seed germination and maintains seed dormancy (Finkelstein et al., 2008; Rajjou et al., 2012; Penfield, 2017). To determine the candidate genes controlling seed germination, the responsive profiles of these nsLTP genes to ABA were downloaded from TENOR (https://tenor.dna.affrc.go.jp/) and investigated. The transcript abundance of OsLTPL5, OsLTPL8, OsLTPL9, OsLTPL11, OsLTPL22, OsLTPL23, OsLTPL26 and OsLTPL28 genes gradually decreased along ABA treatment, providing the possibility of OsLTPL8, OsLTPL9, OsLTPL22 and OsLTPL23 in regulating seed germination (Figure 1C; Supplementary Table S2). Further, we found that the transcription expression of three genes, OsLTPL16, OsLTPL23 and OsLTPL26, are induced in imbibed seeds, which reinforces the reliability of OsLTPL23 participating in seed germination (Figure 1D; Supplementary Table S3; Galland et al., 2014). Collectively, these data pointed to the idea that OsLTPL23 is the most candidate related to seed vigor and seed germination in these rice nsLTP members.

To ascertain the potential relationship between OsLTPL23 and seed germination, a phylogram of OsLTPL23 and its plant homologues was firstly generated through the amino acid (aa) sequence based neighbor-joining algorithm (Figures 2A, B). Homology analysis in 7 model plants displayed that the proteins high identity to OsLTPL23 are presented in gramineous C3 plants, including Triticum aestivum, Hordeum vulgare, Brachypodium distachyon. Unfortunately, there were no reports on the function of those proteins in seed germination. By the protein structure analysis, OsLTPL23 encodes a typical plant non-specific lipid transfer protein with 120 amino acids, including one 26 aa N-terminal hydrophobic signal peptide and the 8 CM in bifunctional inhibitor/plant lipid transfer protein/seed storage helical domain (Figure 2A, C). Subsequently, we isolated total RNA from six tissue samples of NIP at vegetative growth and reproductive growth stages to verify the tissue specific expression pattern of OsLTPL23. The qRT-PCR analysis indicated that the highest gene expression level of OsLTPL23 is presented in rice spikes, followed by that in roots and leaves at five-leaf stage, and decreased in flag leaves, stem and sheathes at booting stage (Figure 3A). These results basically agreed with the data from the Plant Regulomics database, supporting the function of OsLTPL23 in seed development.

Several investigations showed that the rice nsLTP family members specifically localized to the plasma membrane, cytoplasm and nucleus to regulate the cold tolerance, fertility, seed development and low-temperature germinability, respectively (Fujino and Sekiguchi, 2011; Li et al., 2020; Zhao et al., 2020; Chen et al., 2022). To detect the functional compartment of OsLTPL23 at subcellular level, we monitored the fluorescence intensity of OsLTPL23 recombinant protein in different transient expression systems. The fluorescence signal in Nicotiana benthamiana exhibited that the fusion protein GFP-OsLTPL23 clearly localizes to the nucleus and colocalizes with the plasma membrane marker AtPIP2A-mCherry (Figure 3B). To further determine the subcellular localization of OsLTPL23, pSAT6-eYFP, OsbZIP73-mCherry, AtPIP2A-mCherry and pSAT6-OsLTPL23-eYFP constructs were also introduced into the rice protoplast. Laser scanning confocal microscope showed that the recombinant protein OsLTPL23-eYFP separately colocalizes with the plasma membrane marker AtPIP2A-mCherry and nucleus marker OsbZIP73-mCherry in rice (Figure 3C), implying that OsLTPL23 may be involved in transcriptional regulation, as well as plasma membrane biological function.

To explore the involvement of OsLTPL23 gene in rice seed germination, the generation of osltpl23 mutants in the NIP background was achieved through Cas9-induced gene editing (Figure 4A). Specifically, the target sequences were located at the position of 21–40 bp in the first exon of OsLTPL23 (Figure 4B). After the sequencing of site-specific PCR products, 12 osltpl23 mutants, including four homozygous and eight heterozygous mutations, were recovered from 21 T₀ hygromycin-resistant transgenic plants (57.1%). To obtain more allele mutation types, the zygosity analysis of eight heterozygous mutants were carried out through the T vector sequencing of PCR products. Based on the results of zygosity analysis, 33.3%, 19.0% and 33.3% of the mutations were separately nucleotide insertions, deletions and substitutions, and others were characterized by the chimeric mutation (Figure 4C; Supplementary Figure S1).

To obtain the transgene-free plants, all the T₀ mutants were self-pollinated, and we isolated two T-DNA-free homozygous lines from T₁ generation of osltpl23-1 and osltpl23-2 mutants. As shown in Figures 4C, D, the osltpl23-1 and osltpl23-2 mutants separately harbored a 1 bp deletion and insertion between nucleotide 37 and 39 of the OsLTPL23 coding region, which result in a truncated and recombinant variant of the OsLTPL23 respectively. Besides, we noticed that osltpl23-3 and osltpl23-4 mutants produce nearly identical protein variants with osltpl23-2 mutant. Given no differentially expressed OsLTPL23 in all the osltpl23-2, osltpl23-3, and osltpl23-4 mutants (Supplementary Figure S2), the osltpl23-1 and osltpl23-2 mutants were subjected to further investigation.

To validate the germination-responsiveness of OsLTPL23, seeds from wild type NIP, osltpl23-1 and osltpl23-2 mutants were harvested and stored. The seeds stored for 18 months post-harvest were soaked in distilled water to calculate the germination rate. The osltpl23 mutants supported distinctly slower germination before 24 hpi and lower germination percentage at 24 hpi to 96 hpi than NIP (Figures 5A–C). At 24 hpi, the germination percentage of NIP (11.83%) was clearly higher than that of osltpl23 (0%). Subsequently, there were a sharply augmented difference between NIP and osltpl23 mutants at 48 hpi and 72 hpi, which the germination percentages of NIP were 2.27- and 2.76-fold, and 1.73- and 1.86-fold that of osltpl23-1 and osltpl23-2 mutant, respectively. Eventually, the germination percentages of NIP and osltpl23 mutants at 96 h reached 98.20%、76.00% and 69.33%, respectively (Figure 5C). These data indicated that the dysfunction of OsLTPL23 delays and inhibits seed germination.

In previous protein structure analysis, OsLTPL23 has a bifunctional inhibitor/plant lipid transfer protein/seed storage helical domain. Further functional domain analysis showed that it may act as trypsin-alpha amylase inhibitor (https://www.ebi.ac.uk/interpro/entry/pfam/PF00234/). In rice, alpha-amylas isozymes are critical to convert the stored starch into soluble sugar for nourishing the seedling establishment (Damaris et al., 2019). To validate the role of OsLTPL23 in the conversion between starch and soluble sugar, we measured the starch and soluble sugar contents in the dry seeds of NIP, osltpl23-1 and osltpl23-2 mutants. The results uncovered that the starch contents of osltpl23-1 and osltpl23-2 mutants, with separately 268.90 mg/g and 236.17 mg/g, were significantly reduced compared with NIP (332.73 mg/g), whilst the soluble sugar contents were increased 34.11% and 48.19% relative to 14.98 mg/g in NIP (Figures 5D, E). There are reports that soluble sugar delays seed germination (Zhu et al., 2009), we speculated that the high soluble sugar contents may be a contributor responsible for the slower germination in osltpl23 mutants.

ABA is a central player in controlling seed dormancy and germination (Finkelstein et al., 2008; Rajjou et al., 2012; Penfield, 2017). To investigate whether ABA is involved OsLTPL23-trigger seed germination reduction, we firstly checked the expression level of ABA-responsive genes OsABI3, OsABI4 and OsABI5, which are major downstream components of ABA signalling pathway in seed dormancy and seed germination (Lopez-Molina et al., 2001; Liu et al., 2013; Luo et al., 2021; He et al., 2022). The transcripts of OsABI3 and OsABI5 in osltpl23 mutants obviously accumulated compared with those of NIP within 72 hpi, whereas the relative OsABI4 expression in osltpl23-1 mutant were only higher than NIP at 24 hpi (Figures 6A–C). Pioneer report has demonstrated that the ABA content of abi4 seeds was comparable with wild type after stratification (Shu et al., 2013). Therefore, we speculated that the amounts of ABA in osltpl23 mutants may be higher than that in NIP.

To investigate whether endogenous ABA is increased in osltpl23 mutants, the transcription levels of crucial ABA synthesis and metabolism genes (Figures 6D–J), including OsNCEDs (ABA biosynthesis) and OsABA8oxs (ABA degradation), were quantified among the imbibed seeds of NIP, osltpl23-1 and osltpl23-2 mutants. The gene transcription amounts of OsNCED1, OsNCED2, OsNCED3, and OsNCED4 were clearly upregulated in osltpl23 lines, compared with NIP within 12–48 hpi, whilst the increase trend of all the OsNCEDs were blocked at 72 hpi (Figures 6D–G). Concurrently, the mRNA amounts of OsABA8ox2 and OsABA8ox3 were markedly abolished in osltpl23 mutants relative to those in NIP at 12–72 hpi, whereas the transcription expression of OsABA8ox1 was gradually decreased within 3 days post imbibition (Figures 6H–J). Taken together, these data exposed that the disruption of OsLTPL23 may promote the endogenous ABA accumulation, thus leading to lower germination rate of osltpl23 mutants.

The transcript enrichment of OsABI3, OsABI5, OsNCEDs, and accumulative soluble sugar observed in osltpl23 mutants during seed germination provide a indicative cue that OsLTPL23 may regulate post-germination growth of rice. To verify the possibility, we observed the growth phenotype of 2-week old seedlings grown in greenhouse. Phenotypic investigation showed that the plants from osltpl23-1 and osltpl23-2 lines have significantly lower seedling height, with 30.74% less and 37.35% less, than the control plants, respectively (Figures 7A, B). To confirm whether the post-germination seedling growth inhibition of osltpl23 lines is ABA-dependent, we detected the mRNA levels of plant stress hormone-related genes in 2-week old seedlings between mutants and NIP. The qRT-PCR data displayed that all three genes, OsABA8ox1, OsABA8ox2 and OsABA8ox3 in ABA metabolic pathway are evidently increased in osltpl23 mutants relative to those of NIP, whereas ethylene signal transduction gene OsEiN3, salicylic acid signal pathway gene OsNPR1, and jasmonic acid-responsive transcription factor OsMYC2 have abortive or minor transcript enrichment in osltpl23 lines compared with those of NIP (Figure 7C). These data jointly pinpoint the idea that the ABA is the major limiting factor for early seedling growth in osltpl23 mutants. Meanwhile, we also checked the transcription levels of defence-related marker genes OsPR1a, OsPR1b, OsWRKY45 and OsPAD4 in all plants. All these biotic stress-responsive genes in the osltpl23 lines exhibited lower gene expression level compared with those in NIP (Figure 7D). All these results indicated that OsLTPL23 may positively regulate the rice early seedling growth.