CRISPR/Cas9-Mediated Dual Editing of BcSGR1 and BcSGR2 Exerts Additive Enhancement on Delaying Leaf Senescence in Non-Heading Chinese Cabbage

Highlights

What are the main findings?

• Dual CRISPR/Cas9 knockout of BcSGR1 and BcSGR2 additively delays leaf senescence, boosting chlorophyll 3.15-fold, cutting MDA to one-quarter, and extending shelf life to 11 days.
• Optimized Agrobacterium transformation (petiolate cotyledons + AgNO₃) achieved 26% efficiency without agronomic penalties.

What are the implications of the main findings?

• Provides a heritable breeding strategy for stay-green, longer-shelf-life non-heading Chinese cabbage.
• Reveals SGR paralog functional divergence, transferable to other Brassica crops.

Abstract

Leaf senescence negatively impacts yield and postharvest quality in non-heading Chinese cabbage (Brassica rapa ssp. chinensis), and the STAY-GREEN (SGR) homologs BcSGR1 and BcSGR2 function as core regulators of chlorophyll degradation. In this study, we constructed CRISPR/Cas9 single- and dual-gene editing vectors using four high-efficiency, high-specificity single-guide RNAs (sgRNAs). By optimizing the transformation system using petiolate cotyledon explants supplemented with 6 mg·L⁻¹ AgNO₃, we obtained a PCR-positive transformation efficiency of 26% (this value reflects transformation efficiency; the actual gene editing efficiency for each mutant line is presented in the target sequencing results) and generated BcSGR1 (YB1) and BcSGR2 (YB2) single mutants and a BcSGR1/BcSGR2 double mutant (DKO). All mutant lines exhibited a pronounced stay-green phenotype and delayed leaf senescence. The DKO line produced 3.15-fold higher chlorophyll and only one-quarter of the wild-type MDA level, indicating an additive enhancement effect of the double knockout. Compared with the wild type (3-day shelf life), YB1, YB2 and DKO extended postharvest longevity to 7, 8 and 11 days, with senescence delayed by 15, 18 and 22 days, respectively. Notably, whereas the YB1 and YB2 displayed mild leaf shriveling at later developmental stages, the DKO maintained normal leaf morphology. Collectively, our findings demonstrate that dual editing of BcSGR1 and BcSGR2 exerts an additive effect on delaying leaf senescence and enhances postharvest shelf life without incurring significant yield penalties under the experimental conditions. This study provides a feasible and effective strategy for breeding stay-green cultivars of non-heading Chinese cabbage.

1. Introduction

Non-heading Chinese cabbage (Brassica rapa ssp. chinensis) is one of the most widely cultivated leafy vegetables in China and plays a pivotal role in ensuring year-round vegetable supply; however, leaf senescence-induced quality deterioration severely compromises its commercial value [1]. This crop is highly susceptible to postharvest water loss and wilting, with leaf yellowing—the primary cause of quality decline—rendering it unmarketable after only four days of ambient storage, resulting in substantial economic losses [2]. Current postharvest preservation strategies rely predominantly on physical and chemical treatments, including low-temperature storage and 1-methylcyclopropene (1-MCP) fumigation [3]. Previous studies have demonstrated that 1-MCP alleviates postharvest yellowing by modulating transcription factors such as BcERF109 and BcNAC083 to inhibit the expression of chlorophyll catabolic genes [4,5]. Although these treatments can moderately extend shelf life, their application is constrained by high operational costs, inconsistent efficacy across different cultivars and handling conditions, and, most fundamentally, their failure to address the root cause of leaf yellowing from a genetic perspective [6]. Consequently, identifying core senescence-regulatory genes and actively manipulating the senescence process through genetic improvement has emerged as a pivotal research direction to overcome existing preservation bottlenecks. The present work is grounded precisely in this need, aiming to establish genetic approaches for mitigating leaf senescence and to provide a practical example for the development of durable, low-cost, and environmentally friendly genetic improvement strategies, thereby holding significant practical value for ensuring the sustainable development of the non-heading Chinese cabbage industry.

The STAY-GREEN (SGR) gene family has been well documented as a core regulator modulating chlorophyll degradation and leaf senescence progression, and SGR proteins are localized to chloroplasts. Studies in Arabidopsis thaliana and rice have firmly established SGR as a master positive regulator of leaf senescence [7,8]. Recent genome-wide characterization of the SGR gene family in horticultural crops such as tea has revealed functional divergence among paralogs: SGR1 is exclusively localized to chloroplasts, whereas SGR2 and SGRL display a wider subcellular distribution pattern [9]. In Arabidopsis, SGR1 accelerates chlorophyll degradation, and loss-of-function mutations of SGR1 markedly delay leaf yellowing; conversely, SGR2 heterodimerizes with SGR1 to repress its activity, functioning as a negative regulator [10]. Although this regulatory model, derived primarily from Arabidopsis, suggests that simultaneous knockout of SGR1 and SGR2 would offer limited practical benefit and could perturb the overall chlorophyll metabolic network [11], whether this paradigm holds in non-heading Chinese cabbage remains unclear and requires systematic investigation. In recent years, CRISPR/Cas9 genome editing, valued for its high precision and editing efficiency, has been deployed successfully in Brassica crops to accelerate functional gene characterization and trait improvement [12,13]. Previous studies have reported SGR1-associated stay-green mutants in Chinese cabbage and pak choi [14]. However, the biological functional role of BcSGR2 and its potential interaction with BcSGR1 remain largely elusive. Based on the foregoing evidence and the absence of direct functional studies on BcSGR2 in this crop, we hypothesize that BcSGR2 in non-heading Chinese cabbage may have functionally diverged from its Arabidopsis ortholog. Furthermore, simultaneous disruption of both BcSGR1 and BcSGR2 could yield additive enhancement on senescence regulation. To test these predictions, we generated single and double CRISPR/Cas9 knockout mutants for systematic phenotypic and physiological evaluation.

In this study, the CRISPR/Cas9 genome editing system was employed to systematically elucidate the regulatory roles of BcSGR1 and BcSGR2 in leaf senescence of non-heading Chinese cabbage, with particular emphasis on the additive effects conferred by simultaneous mutagenesis of both genes. In this study, we have performed the following: (i) design high-specificity sgRNAs targeting BcSGR1 and BcSGR2, and construct high-efficiency CRISPR/Cas9 vectors for single-gene and dual-gene editing; (ii) generate BcSGR1 single-knockout, BcSGR2 single-knockout and double-knockout mutants via an optimized Agrobacterium-mediated genetic transformation system, and conduct systematic phenotypic characterization of the mutant materials; (iii) determine physiological parameters associated with leaf senescence and oxidative stress, and perform comparative analyses of parameter differences among single mutants, double mutants and the wild type. This study provides experimental evidence for elucidating the functional divergence pattern of BcSGR1 and BcSGR2 in non-heading Chinese cabbage, and offers germplasm resources and technical references for breeding stay-green vegetable cultivars with excellent postharvest quality.

2. Materials and Methods

2.1. Plant Materials and Reagents

The elite cultivar of non-heading Chinese cabbage (Brassica rapa ssp. chinensis), ‘Suzhou Qing’, was used as the recipient material for gene editing; aseptic seedlings were prepared by the Vegetable Genetics and Breeding Laboratory of the Sanya Research Institute of Nanjing Agricultural University. Agrobacterium rhizogenes strain K599, Escherichia coli strain DH5α, and the CRISPR/Cas9 vector backbone pYLCRISPR/Cas9P35S-N were maintained in our laboratory. Restriction endonucleases, T4 DNA ligase, and other reagents were purchased from TaKaRa (Bio Inc., Shiga, Japan); PCR primers were synthesized by Sangon Biotech (Shanghai) Co., Ltd. (Shanghai, China).

2.2. sgRNA Design and Screening

Based on the coding region sequences of the non-heading Chinese cabbage SGR1 (Bra020829) and SGR2 (Bra000755) genes retrieved from the NCBI database, eight candidate sgRNAs targeting the conserved functional domain of each gene using the online tool CRISPR-P 2.0. Genome-wide off-target site prediction was performed with Cas-OFFinder, with the maximum number of allowed mismatches set to ≤3, and candidate sequences with high off-target risk were excluded. Using non-heading Chinese cabbage genomic DNA as the template, ~500 bp fragments flanking the target sites were amplified by PCR; Cas9 mRNA and candidate sgRNAs were synthesized via in vitro transcription, and the in vitro cleavage efficiency was determined to screen for sgRNAs with an in vitro cleavage efficiency > 85%. Finally, BcSGR1 sgRNA1 (CCAACGCTCCCTAGAACTTA) and sgRNA2 (TGATTCGAAGATCGCTGGTC), together with BcSGR2 sgRNA1 (TCACAGTGACATAACCGCTA) and sgRNA2 (TGTTCCGGGACGAAGTAGTG), were selected for subsequent vector construction.

2.3. Construction of CRISPR/Cas9 Editing Vectors

Single-gene and dual-gene editing vectors were constructed using the Golden Gate assembly method. The selected sgRNA coding sequences were synthesized as complementary oligonucleotide pairs and individually cloned into pYL-U6 series vectors to obtain sgRNA expression cassettes. The BcSGR1 and BcSGR2 sgRNA expression modules were separately inserted into the linearized pYLCRISPR/Cas9P35S-N vector yielding the single-gene editing vectors pYLCRISPR/Cas9-BcSGR1 and pYLCRISPR/Cas9-BcSGR2. For the dual-gene editing vector (designated pYL-Cas9-S1S2), all four sgRNA expression cassettes were assembled into the pYLCRISPR/Cas9P35S-N backbone via one-step digestion-ligation. The ligation products were transformed into E. coli DH5α competent cells, and after verification by restriction enzyme digestion and Sanger sequencing, the positive plasmids were extracted and transformed into A. rhizogenes K599 [15].

2.4. Agrobacterium-Mediated Transformation and Optimization of the Regeneration System

Explant preparation: Plump ‘Suzhou Qing’ seeds were surface-sterilized with 75% ethanol for 2 min and 25% sodium hypochlorite solution for 10 min, then the sterilized seeds were sown on 1/2 MS medium and cultured at 25 °C under a 16 h photoperiod for 8 d to obtain aseptic seedlings.

Agrobacterium preparation: Agrobacterium (K599) harboring the target editing vector was inoculated into liquid LB medium supplemented with 50 mg·L⁻¹ kanamycin (Kan) and 50 mg·L⁻¹ streptomycin (Sm) and cultured overnight at 28 °C with shaking at 250 rpm. After centrifugation, the supernatant was discarded and the pellet was resuspended in liquid MS medium; the OD₆₀₀ was adjusted to 0.6, 150 μmol·L⁻¹ acetosyringone (AS) was added, and the suspension was incubated at room temperature in the dark for 3 h (or with gentle shaking for 1.5 h) prior to use.

Infection and culture: Petiolate cotyledons and hypocotyls excised from aseptic seedlings were used as explants. After 2 days of pre-culture, the explants were immersed in the Agrobacterium suspension and inoculated with gentle shaking for 10 min; the suspension was blotted off and the explants were subjected to 3 days of co-culture. The explants were then transferred to selective regeneration medium supplemented with 6 mg·L⁻¹ AgNO₃ for adventitious-shoot induction; after 30 days of culture, regenerated adventitious shoots were transferred to rooting medium for root induction, and intact regenerated plants were finally obtained (Figure 1) [16].

Four comparative treatments were established, combining explant type (petiolate cotyledon/hypocotyl) with the presence or absence of AgNO₃. The explant browning rate, adventitious-shoot induction rate, and PCR-positive transformation efficiency were recorded to optimize the regeneration system. As reported in previous studies, AgNO₃ functions as an ethylene inhibitor and can effectively alleviate explant browning in Chinese cabbage, while an optimized hormone combination is critical for adventitious shoot differentiation. For each treatment, no fewer than 120 explants were used, and three independent transformation experiments were performed.

The co-culture medium was formulated with MS basal salts supplemented with 30 g·L⁻¹ sucrose and 7 g·L⁻¹ agar, adjusted to pH 5.8. For adventitious shoot induction, explants were cultured on regeneration medium containing MS basal salts supplemented with 30 g·L⁻¹ sucrose, 7 g·L⁻¹ agar, 200 mg·L⁻¹ carbenicillin, 200 mg·L⁻¹ timentin, and 6 mg·L⁻¹ AgNO₃. For rooting, shoots were transferred to 1/2 MS medium supplemented with 200 mg·L⁻¹ carbenicillin, 200 mg·L⁻¹ timentin, and 0.25 mg·L⁻¹ α-naphthaleneacetic acid (NAA).

2.5. Screening, Identification, and Phenotypic Evaluation of Gene-Edited Plants

Genomic DNA was extracted from the regenerated plants and initial screening was performed via PCR amplification of the Cas9 gene, using the primers Cas9-F (5′-GCCATCCTGTCGCTGTTG-3′) and Cas9-R (5′-GTCGATGTTGTCGGTGGT-3′). For Cas9-positive plants, the BcSGR1/BcSGR2 target regions were amplified by PCR, and the amplicons were subjected to Sanger sequencing at Sangon Biotech (Shanghai) Co., Ltd. for sequencing. Mutation types were analyzed by sequence alignment, and BcSGR1 single-mutant (YB1), BcSGR2 single-mutant (YB2), and double-mutant (DKO) plants were identified for subsequent analysis. All phenotypic assays were performed on T₀ regenerants. For detailed phenotypic characterization, three representative lines were selected for each genotype (YB1, YB2 and DKO), and all available T₀ plants of each selected line (nine for YB1, nine for YB2, and three for DKO) were evaluated. The individual T₀ plants within each selected line were regenerated from the same editing event; therefore, they constitute biological replicates within the line rather than independent editing events.

2.6. Phenotypic and Physiological Measurements

Wild-type (WT) and mutant plants were grown in an experimental field under conventional field management arranged in a randomized complete block design with three blocks (replicates). The initiation of leaf senescence was recorded starting from 60 days after sowing (DAS). After harvest, functional leaves of uniform size were selected and stored at room temperature (25 °C) under controlled relative humidity (60–70%); leaf yellowing was observed daily and the shelf life was recorded. Shelf life was defined as the duration until 50% of the leaf area turned yellow, at which point the leaves were considered to have lost commercial value. Chlorophyll content was determined by the 95% ethanol extraction method to evaluate the extent of chlorophyll degradation, whereas malondialdehyde (MDA) content was determined by the thiobarbituric acid method according to established protocols [17]. Plant height, maximum leaf area, and yield per unit area (kg·m⁻²) were measured to assess the effects of gene editing on major agronomic traits. The experimental unit for field observations and agronomic measurements was the individual plant. In each block, three plants per genotype were planted for WT, YB1, and YB2, together with one DKO plant; thus, nine plants each of WT, YB1, and YB2 and three DKO plants were evaluated in total. For the physiological assays, one biological replicate per genotype was derived from each block: leaf tissue was pooled from the three plants of that genotype within the block for WT, YB1, and YB2, whereas the single DKO plant within the block constituted one biological replicate, yielding three biological replicates per genotype.

2.7. Statistical Analysis

Experimental data were subjected to analysis of variance using SPSS 26.0 software. Normality (Shapiro–Wilk test) and homogeneity of variance (Levene’s test) were verified before one-way ANOVA. Multiple comparisons were conducted using Tukey’s honestly significant difference (HSD) test, with a significance threshold set at p < 0.05. Agronomic traits are presented as mean ± standard deviation (SD) of individual plants (n = 9 for WT, YB1, and YB2; n = 3 for DKO), and physiological indices as mean ± SD of three biological replicates (blocks) per genotype, as defined in Section 2.6. To evaluate whether the DKO phenotype exceeded the additive expectation of the two single mutant (YB1, YB2) effects, the expected additive value was calculated as E_add = YB1 + YB2 − Y_WT, where Y_WT, YB1, and YB2 denote the mean phenotypic values of the wild-type, single mutant YB1, and single mutant YB2, respectively, following the additive model described by Mani et al. (2008) [18]. The deviation from additivity (Δ = Y_DKO − E_add) was then tested for statistical significance using a two-tailed Student’s t-test, with the standard error of Δ estimated by propagating the standard errors of the four genotype group means and the degrees of freedom approximated by the Welch–Satterthwaite equation. A p value < 0.05 was considered significant; a non-significant Δ indicated an additive effect, a significant positive Δ indicated the additive enhancement, and a significant negative Δ indicated an antagonistic interaction. Because the plants were arranged in a randomized complete block design, block effects were first evaluated within a general linear model and were non-significant for all traits (p > 0.05); one-way ANOVA was therefore used for comparisons among genotype means. In addition, the four genotypes (WT, YB1, YB2, and DKO) were analyzed within a 2 × 2 factorial framework (BcSGR1: intact vs. knockout; BcSGR2: intact vs. knockout) using two-way ANOVA to partition the main effects of BcSGR1 and BcSGR2 and their interaction; of note, the deviation statistic Δ is algebraically identical to the BcSGR1 × BcSGR2 interaction contrast (Y_DKO − Y_YB1 − Y_YB2 + Y_WT).

3. Results

3.1. Screening of sgRNAs and Construction of Editing Vectors

In vitro cleavage activity assays confirmed that all four candidate sgRNAs exhibited cleavage efficiencies exceeding 85%, among which BcSGR2-sgRNA1 achieved the highest activity at 93%. Comprehensive genome-wide off-target prediction further revealed that each sgRNA harboured a maximum of two putative off-target loci, all situated within intergenic non-coding regions, demonstrating robust targeting specificity (

Table 1. Sequences of sgRNAs designed for BcSGR1 and BcSGR2 and their corresponding in vitro cleavage efficiencies.

Target Gene	sgRNA Name	Sequence (5′–3′)	PAM	In Vitro Cleavage Efficiency (%)
BcSGR1	S1-g1	CCAACGCTCCCTAGAACTTA	CGG	92
BcSGR1	S1-g2	TGATTCGAAGATCGCTGGTC	CGG	88
BcSGR2	S2-g1	TCACAGTGACATAACCGCTA	AGG	93
BcSGR2	S2-g2	TGTTCCGGGACGAAGTAGTG	GGG	89

). Subsequent restriction endonuclease digestion and Sanger sequencing verification validated the successful assembly of three CRISPR/Cas9 constructs: the single-target editing plasmids pYLCRISPR/Cas9-BcSGR1 and pYLCRISPR/Cas9-BcSGR2, and the dual-gene editing vector pYL-Cas9-S1S2. Sequencing chromatograms verified accurate insertion of each sgRNA cassette into the multiple cloning site of the vector backbone.

3.2. Optimization of the Regeneration System

Significant differences in stable transformation efficiency of ‘Suzhou Qing’ were detected among the distinct regeneration regimes (

Table 2. Effects of different regeneration conditions on the genetic transformation efficiency in non-heading Chinese cabbage.

Explant	Medium (AgNO3)	Explant Browning Rate (%)	Adventitious-Shoot Rate (%)	PCR-Positive Transformation Efficiency (%)
Petiolate cotyledon	+(6 mg·L−1)	15 ± 5	38 ± 6.5	26 ± 2.9 a
Petiolate cotyledon	−	44 ± 10	20 ± 5.6	12.0 ± 2.1 b
Hypocotyl	+(6 mg·L−1)	30 ± 8	16 ± 4	10.5 ± 1.6 b
Hypocotyl	−	65 ± 12	6 ± 2	3.1 ± 1.5 c

Values are means ± SD; the same lowercase letter within a column indicates no significant difference (p < 0.05, Tukey’s HSD test).

). For petiolate cotyledon explants cultured on medium supplemented with 6 mg L⁻¹ AgNO₃, the browning incidence was 15 ± 5%, the adventitious shoot induction rate was 38 ± 6.5%, and the PCR-verified transgenic efficiency was 26 ± 2.9%. For hypocotyl explants cultivated without AgNO₃, the browning incidence was 65 ± 12%, and the transformation efficiency was 3.1 ± 1.5%.

3.3. Molecular Identification of Positive Mutant Plants

Twenty-one independent regenerants were recovered from tissue culture, 18 of which yielded amplicons diagnostic for the Cas9 transgene via PCR screening, corresponding to an overall transgenic recovery rate of 85.71%. Sanger sequencing across target loci identified nine independent lines harbouring monoallelic edits at the BcSGR1 locus. Hi-TOM amplicon sequencing of embryogenic callus derived from these BcSGR1-edited lines uncovered no sequence variation at the sgRNA1 target, yet detected a single-base insertion immediately downstream of the sgRNA2 cleavage site; wild-type reads constituted 62.90% of the sequencing pool, with the 1 bp insertion variant comprising the remaining 37.10% (Figure 2;

Table 3. Hi-TOM sequencing analysis of YB1 embryogenic callus mutants.

Sort	Reads	Ratio	Left Variation	Right Variation	Left Variation	Right Variation
1	4821	62.90%	WT	WT	-	-
2	2843	37.10%	WT	1I	-	C

). This sgRNA2 binding motif resides within a highly conserved coding domain of BcSGR1 (Figure 3). The predominant lesion corresponded to a 1 bp deletion at the sgRNA2 target, alongside larger 4 bp and 8 bp truncations at the sgRNA1 site and multi-site deletions spanning both protospacers, each predicted to perturb conserved polypeptide architecture within the coding frame. Sequence alignment spanning the BcSGR2 sgRNA1 target further corroborated precise on-target mutagenesis within the intended genomic interval (Figure 3). Distinct mutational signatures were recovered from lines targeted at BcSGR2 (

Table 4. Hi-TOM sequencing analysis of the mutation profiles in YB2 embryogenic callus mutants.

Sort	Reads	Ratio	Left Variation	Right Variation	Left Variation	Right Variation
1	2175	27.38%	WT	WT	-	-
2	1620	20.40%	WT	1D	-	T
3	873	10.99%	8D	WT	CATAAACCG	-
4	751	9.45%	4D	WT	ACCG	-
5	574	7.23%	8D	1D	CATAAACCG	T

), among which nine isolates represented BcSGR2 single mutants. Three regenerants carried combinatorial lesions characteristic of dual-gene knockout, harbouring the identical 1 bp insertion within BcSGR1 and 4 bp deletion within BcSGR2 recovered in the respective single mutant backgrounds YB1 and YB2. All identified allelic variants were predicted to generate frameshift mutations in the coding regions of BcSGR1 or BcSGR2, resulting in truncated and potentially non-functional proteins.

3.4. Leaf Senescence and Shelf-Life Phenotypes of Edited Plants

Field phenotyping revealed that both wild-type plants and mutants maintained vibrant green foliage under field conditions (Figure 4). Notably, the results demonstrate that the YB1 and YB2 single mutants exhibited some degree of leaf shriveling at later developmental stages, whereas the DKO double mutant displayed normal leaf morphology without such abnormalities. Postharvest shelf-life assays under ambient storage conditions further quantified the divergence in leaf yellowing progression. Wild-type leaves reached 50% yellow leaf coverage at 3 days postharvest, a threshold corresponding to complete loss of marketable quality, while the functional leaves of YB1, YB2, and DKO still exhibited less than 20% yellowing after 3 days, indicating a delayed onset of yellowing (Figure 5). This senescence-delaying effect was most pronounced in the dual mutant genotype: at 20 days, the functional leaves of YB1 and YB2 exhibited approximately 80% yellowing, whereas those of DKO showed only about 60% yellowing. No statistically significant differences in storage longevity were detected between the two single mutant backgrounds (Figure 5). These results indicate that the double mutant decelerates chlorophyll degradation to a greater extent than the single mutants.

3.5. Changes in Physiological Indices of Edited Plants

Quantification of total chlorophyll concentrations revealed substantially elevated pigment levels across all mutant genotypes relative to wild-type controls (Figure 6A). Chlorophyll concentrations in YB1 and YB2 reached 1.4- and 1.5-fold wild-type levels, respectively, while the DKO line accumulated chlorophyll at 3.15-fold the wild-type abundance. This value far exceeded the additive expectation of the two single-gene mutants, reflecting a prominent additive enhancement effect following simultaneous dual-gene disruption, which was statistically confirmed by Student’s t-test (Figure 6A). Comparable additive phenotypic shifts have been documented in multi-target CRISPR editing studies of cereal crops including rice and wheat. Malondialdehyde (MDA) quantification yielded contrasting results: wild-type plants accumulated markedly higher MDA concentrations than all mutant lines, with wild-type MDA levels approaching fourfold those detected in DKO foliage (Figure 6B). Equivalent MDA pools were measured between YB1 and YB2. To further evaluate the effect of gene editing on oxidative stress, the superoxide anion radical (${\mathrm{O}}_2^{-}$) production rate and hydrogen peroxide (H₂O₂) content were measured in each line. Collectively, these observations indicate that dual-gene mutagenesis effectively mitigates membrane lipid peroxidation and preserves cellular membrane integrity. In contrast, YB1 and YB2 retained marketable quality until 7 days, and the DKO line sustained acceptable commercial performance through 10 days. Both ${\mathrm{O}}_2^{-}$ production rate and H₂O₂ content in all mutant lines were significantly lower than those in wild-type. The ${\mathrm{O}}_2^{-}$ production rates of YB1 and YB2 were 73.4% and 66.0% of the wild-type level, respectively, while their H₂O₂ contents were 71.4% and 63.8% of the wild-type level, respectively. The DKO line exhibited the lowest ${\mathrm{O}}_2^{-}$ production rate and H₂O₂ content, which were only 33.7% and 33.0% of the wild-type level, respectively (Figure 6C,D). The consistent trends between ${\mathrm{O}}_2^{-}$ production rate and H₂O₂ content further confirmed that dual-gene knockout effectively reduced ROS accumulation and alleviated oxidative damage. Two-way ANOVA within the 2 × 2 factorial framework further revealed significant main effects of BcSGR1 and BcSGR2 and a significant BcSGR1 × BcSGR2 interaction for chlorophyll content, fully consistent with the additive-expectation analysis.

Measurement of the weight loss rate revealed an upward trend over time in all plant materials (WT, YB1, YB2, and DKO). The weight loss rate of the wild-type (WT) increased sharply after 5 days of storage, reaching a maximum of 31% by 10 days, indicating rapid deterioration of postharvest quality. In contrast, the gene-edited lines exhibited considerably lower weight loss rates, with the DKO double mutant showing the smallest reduction. At 10 days, the weight loss rate of the DKO line was only approximately 14%, which was significantly lower than that of the WT and slightly lower than those of YB1 (18%) and YB2 (17%) (Figure 7).

The significant reduction in oxidative stress markers and the effective maintenance of postharvest weight indicate that cells of the double mutant line experienced reduced oxidative damage and decreased water transpiration.

3.6. Agronomic Trait Analysis of Edited Plants

Quantification of core agronomic traits revealed no statistically significant divergence in plant height, maximum leaf area and plant weight among YB1, YB2, DKO mutants, and wild-type controls. The DKO genotype exhibited a plant height of 24.2 ± 2.3 cm, maximum leaf area of 141.2 ± 10.2 cm², and yield of 131.95 ± 6.37 g. These phenotypic observations suggest that, under the conditions of this experiment, targeted mutagenesis of BcSGR1 and BcSGR2 predominantly modulates leaf stay-green characteristics without imposing detectable detrimental impacts on the elite agronomic attributes of non-heading Chinese cabbage (

Table 5. Comparison of agronomic traits between gene-edited and wild-type plants.

Line	Plant Height (cm)	Maximum Leaf Area (cm2)	Plant Weight (g)
WT	22.5 ± 1.8 a	125.1 ± 8 a	130.78 ± 11.35 a
YB1	23.2 ± 2.3 a	129.5 ± 10 a	133.27 ± 8.08 a
YB2	23.1 ± 2.1 a	137.6 ± 9 a	129.31 ± 3.26 a
DKO	24.2 ± 2.3 a	141.2 ± 10.2 a	131.95 ± 6.37 a

Values are means ± SD; lowercase letter “a” indicates no significant difference (p < 0.05, Tukey’s HSD test). Data are means of three individual plants.

).

4. Discussion

4.1. Functional Regulation of BcSGR1 and BcSGR2 Mutagenesis on the Stay-Green Phenotype in Non-Heading Chinese Cabbage

SGR proteins function as master regulators governing chlorophyll catabolism. Similar stay-green phenotypes caused by disruption of chlorophyll degradation-related genes have been reported in Brassica crops, supporting the expected functional consequences of BcSGR1/BcSGR2 knockout [11,12]. In the present study, a single-nucleotide insertion in BcSGR1 induced translational frameshift, whereas diverse deletion alleles within BcSGR2 disrupted its conserved functional domain, collectively abolishing native gene activity (Table 3 and Table 4). Mutant lines exhibited robust suppression of chlorophyll degradation and delayed leaf senescence, consistent with previous findings that impaired SGR1-related chlorophyll degradation pathways contribute to stay-green phenotypes in Brassica species [11,12]. Contrary to the canonical negative regulatory role of AtSGR2 in Arabidopsis, targeted disruption of BcSGR2 alone substantially elevated leaf chlorophyll abundance and prolonged postharvest storability in non-heading Chinese cabbage. This observation indicates substantial functional divergence of BcSGR2 orthologs across Brassicaceae species; rather than merely a negative effector, BcSGR2 may serve as a core mediator of chlorophyll turnover. Beyond chlorophyll catabolism, SGR is also involved in plastid remodeling during leaf senescence [19]. These findings challenge conventional functional inferences drawn solely from model plant systems, expand the functional framework of the SGR gene family, and underscore the necessity of independent phenotypic validation when translating gene functional data from Arabidopsis to cultivated crops. In addition, we observed leaf shriveling in the YB1 and YB2 single mutants (Figure 4A,B), whereas the DKO double mutant displayed normal leaf morphology (Figure 4C). This phenomenon may be associated with chloroplast accumulation-induced cellular structural stress when chlorophyll degradation is partially blocked; complete disruption of the pathway in the DKO line may permit cellular rebalancing and thereby avoid morphological abnormalities. However, the underlying mechanism requires further investigation. Future work could further increase the editing frequency of BcSGR-targeted lines by adopting high-efficiency transformation strategies optimized for Brassica crops, such as the application of tenoxicam, optimized phytohormone ratios and visual reporter-based screening systems [20].

4.2. Possible Mechanisms Underlying the Additive Enhancement Effect of Dual BcSGR Editing

Dual knockout genotypes displayed far stronger stay-green performance than either single mutant allele, with chlorophyll concentrations reaching 3.15-fold wild-type levels -an accumulation exceeding the additive phenotypic output of the two individual single-gene disruptions, indicative of a robust additive enhancement effect, which was statistically confirmed by Student’s t-test (p < 0.05). These two results—markedly decreased oxidative stress markers (MDA, ${\mathrm{O}}_2^{-}$, H₂O₂) (Figure 6B–D) and reduced postharvest weight loss (Figure 7)—further indicate that simultaneous knockout of BcSGR1 and BcSGR2 significantly delays postharvest senescence. The phenotypic discrepancy between single and double mutants is primarily attributed to the functional divergence and complementary roles of duplicated SGR paralogs in chlorophyll catabolism, which can be systematically predicted through genome-wide evolutionary and subcellular functional analysis of gene families [21,22]. In studies of other plant species, subcellular localization assays in tea plants have revealed distinct compartmentalization patterns for two SGR paralogs: one localizes exclusively to chloroplasts, while the other distributes across chloroplasts, plasma membranes and nuclei, implying that SGR2 orthologs may have expanded regulatory repertoires relative to SGR1 [9]. This pattern is consistent with the functional divergence observed for BcSGR1 and BcSGR2 in the present study. Such functional differentiation among paralogous copies is widely observed within Brassica vegetable species [23,24]. Sophisticated multi-gene cloning and editing systems have been established for model plants and crops, enabling efficient simultaneous mutagenesis of paralogous loci and facilitating the exploration of gene functional redundancy [25]. Consistently, practical multiplex editing combined with haploid breeding strategies in cereal crops has successfully generated superior mutant materials with pyramided agronomic traits, verifying the broad applicability of multi-locus editing for phenotypic enhancement [26].

Transcriptional regulation of senescence-associated genes represents a core regulatory module governing leaf aging and postharvest quality deterioration in vegetable crops, and CRISPR-mediated gene modulation of senescence-related pathways has become a mature strategy for horticultural crop improvement [27]. For example, CoBBX24 overexpression downregulates SAG12, SAG29, NYC1, NYE1, and NYE2 to delay leaf senescence in Arabidopsis, suggesting that SGR family members may themselves be subject to transcriptional coordination [28]. We hypothesize that BcSGR1 and BcSGR2 may assemble into functional homodimeric or heterodimeric complexes to co-govern chlorophyll degradation in non-heading Chinese cabbage, although this hypothesis requires experimental validation through protein–protein interaction assays. Dual mutagenesis would be expected to fully abrogate formation of this putative protein complex and thereby block chlorophyll catabolism, whereas single-gene knockout only partially dampens complex activity and yields milder stay-green traits. This mechanistic model suggests potential functional interplay within SGR orthologs across divergent crop taxa and provides a candidate theoretical framework for multiplex editing strategies to accelerate horticultural trait innovation, pending further experimental confirmation.

4.3. Optimized Regeneration and Transformation Pipeline for CRISPR Mutagenesis in Non-Heading Chinese Cabbage

Agrobacterium-mediated transformation of non-heading Chinese cabbage is inherently constrained by stringent genotype dependency, severe explant browning, and low stable transformation rates. Conventional transformation protocols for this crop suffer from poor efficiency, limiting broad deployment of CRISPR genome-editing platforms, especially for multiplex gene-knockout experiments [20,29]. Conventional SpCas9 recognizes NGG PAMs, restricting targetable sites to approximately 11.97% of the Brassica rapa genome. Engineered variants—Cas9-NG/SpG (relaxed NGN PAMs) and SpRY (near-PAM-less)—have been validated in Brassica protoplasts and cabbage T₀ plants, though they raise off-target and self-editing concerns [30]. Through systematic screening of optimal explant materials and exogenous supplementation with silver nitrate, the present pipeline elevated the PCR-positive transformation efficiency to 26% (Table 2), outperforming previously optimized cabbage transformation systems (19.33–20.00%) [20].

Petiolate cotyledons serve as ideal explants owing to their high regenerative competence and susceptibility to Agrobacterium infection [12]. Silver nitrate functions as an ethylene biosynthesis inhibitor, mitigating oxidative browning and stimulating adventitious shoot regeneration. Notably, the efficiency-boosting strategy developed here operates via a distinct biochemical pathway compared with the protocol reported by Wang et al., who improved regeneration by exogenous tenoxicam application to repress salicylic acid signaling [20]. Despite divergent mechanistic routes, both approaches converge on overcoming genotype-dependent regeneration barriers in Brassica rapa. The refined transformation workflow established herein delivers actionable technical references for CRISPR mutagenesis research in non-heading Chinese cabbage and additional Brassica ceae vegetable species [31,32].

4.4. Limitations of This Study

Several limitations should be acknowledged. First, off-target assessment was based solely on computational prediction and lacks experimental verification (e.g., by Sanger sequencing or targeted deep sequencing); therefore, conclusions regarding editing specificity should be interpreted with caution (Table 1). Second, three representative lines per genotype were subjected to detailed phenotypic analysis; the potential variation among independent transformation events was not fully assessed. Future studies with additional independent lines will strengthen the generalizability of the conclusions. Moreover, all phenotypic data were obtained from T₀ regenerants; therefore, the heritability and genetic stability of the edited alleles in subsequent generations remain to be demonstrated (Figure 4). Third, we did not perform transcript-level (RT-qPCR) or protein-level validation to directly confirm the loss of mRNA or protein, nor did we examine the expression changes in downstream chlorophyll catabolic genes, leaving the mechanistic interpretation incomplete. Fourth, agronomic evaluation was conducted under a single environmental condition; the adaptability of the edited lines to different soils, climates, and management practices remains to be tested (Figure 6 and Figure 7). Fifth, the hypothesis regarding SGR1–SGR2 complex formation lacks experimental support from protein interaction assays. These limitations suggest that the conclusions of this study should be considered preliminary, and confirmation of the relevant molecular mechanisms awaits further experimental investigation.

4.5. Translational Potential of CRISPR Genome Editing for Genetic Improvement of Non-Heading Chinese Cabbage

Historical research dissecting stay-green physiology in non-heading Chinese cabbage has relied predominantly on spontaneous natural variants, which carry inherent drawbacks including lengthy breeding cycles and limited allelic diversity. CRISPR/Cas9-mediated genome editing enables precise, user-defined mutagenesis of target loci. Multiple streamlined multiplex editing assembly systems have been developed in recent years, among which Golden Gate cloning represents a prevailing toolkit for constructing multi-target CRISPR/Cas9 constructs. This system enables the design of highly efficient vectors for dual mutagenesis [29]. Stable heritable stay-green germplasm can be generated within merely two to three generations using this platform, providing invaluable mutant resources for deep functional dissection of BcSGR1 and BcSGR2 [33].

Globally, CRISPR mutagenesis has been widely deployed to engineer stress resilience, fruit and leaf quality, and postharvest shelf-life traits across vegetable crops [27]. Complementary to transcription factor engineering—such as leveraging the light-responsive BASIC PENTACYSTEINE (BPC) family to enhance environmental adaptability—precision genome editing of core chlorophyll catabolic genes offers a direct route to manipulate stay-green phenotypes [29]. Alternatively, translational modulation via upstream open-reading-frame (uORF) targeting—either through genome editing or antisense oligonucleotide intervention—represents another promising strategy to fine-tune gene activity without complete gene knockout [34,35]. For instance, Lee et al. disrupted HY5 in Chinese cabbage to reinforce endoplasmic reticulum stress tolerance [13]. Our dual-edited germplasm exhibits retarded leaf senescence and extended postharvest storability without compromising elite agronomic performance, under the conditions tested, furnishing critical germplasm materials and mechanistic frameworks for breeding senescence-delayed non-heading Chinese cabbage varieties. Beyond the focal crop, the multiplex editing pipeline described here is transferable to other Brassica ceae crops including rapeseed and head cabbage, offering a replicable research paradigm for dissecting leaf senescence regulatory networks across horticultural Brassica species [27].

5. Conclusions

This study established an efficient regeneration system (26% PCR-positive transformation efficiency) and achieved dual CRISPR/Cas9 editing of BcSGR1 and BcSGR2 in non-heading Chinese cabbage, exerting a prominent additive enhancement effect that increased chlorophyll content by 3.15-fold, delayed leaf yellowing and mitigated agronomic trait penalties. These findings provide evidence for the interplay between BcSGR1 and BcSGR2 in regulating leaf senescence and offer a feasible strategy for breeding stay-green cultivars.