Development of genome-edited brsoc1 parental lines to breed cultivars with delayed bolting
We explored effective methods for introducing the late-bolting trait into various Chinese cabbage inbred lines, given the substantial time spent in our previous study on obtaining Cas9-free plants to conduct backcross breeding. To efficiently introduce mutated brsoc1 genes, we developed several brsoc1 lines for crosses (Fig. S1A). From 85 T2 plants derived from backcrossing (BC) in our previous work [18], we successfully identified several plants with late bolting (Fig. S1B). Regrettably, we only obtained seeds for two lines: BC T2-16, heterozygous at all three mutated BrSOC1 loci, with the genotype brsoc1-1/BrSOC1-1 brsoc1-2/BrSOC1-2 brsoc1-3/BrSOC1-3; and BC T2-31, homozygous for brsoc1-1 and brsoc1-3 and heterozygous for brsoc1-2, with the genotype brsoc1-1 brsoc1-2/BrSOC1-2 brsoc1-3 (Fig. S1C). We obtained brsoc1-1 brsoc1-2 plants from the progeny of BC T2-16, and brsoc1-1 brsoc1-3 and brsoc1-1 brsoc1-2 brsoc1-3 plants from the progeny of BC T2-31 to serve as parental lines in the next generation (Fig. S1D and E).
An early-bolting inbred line is minimally effective for introducing the delayed bolting conferred by brsoc1 genes to F1 hybrids
To investigate the potential for developing late-bolting varieties by introducing genome-edited brsoc1 genes, we examined backcrossed progeny from our previous cross between lines carrying genome-edited brsoc1s and the inbred line 20, an early-bolting Chinese cabbage cultivar [18] (Fig. 1A). We selected F1 hybrids heterozygous at two or more BrSOC1 loci, specifically 20 × brsoc1-1 brsoc1-2, 20 × brsoc1-1 brsoc1-3, and 20 × brsoc1-1 brsoc1-2 brsoc1-3 (Table S1). All selected hybrids showed an insertion/deletion (InDel) frequency of 50% for those BrSOC1s at which they harbored a mutant allele, confirming their genotypes (Fig. 1B, Table S1).
However, the F1 hybrids obtained from this early-bolting inbred line only exhibited bolting delayed by 1 − 5 days. Unlike homozygous brsoc1s, which significantly delayed bolting, the F1 hybrid plants carrying one mutated copy at two BrSOC1 loci, 20 × brsoc1-1 brsoc1-2 and 20 × brsoc1-1 brsoc1-3, did not show a significantly later bolting time compared with the 20 maternal line after 35 days of vernalization treatment. However, F1 hybrid plants harboring one mutated copy at all three BrSOC1 loci, 20 × brsoc1-1 brsoc1-2 brsoc1-3, did bolt later than the 20 inbred line, with a median difference of about 5 days with a high range of variation (Fig. 1C and D). All F1 hybrid plants produced more rosette leaves until bolting relative to the 20 inbred line, with 14 rosette leaves in the 20 line, 18 in the F1 hybrids 20 × brsoc1-1 brsoc1-2 and 20 × brsoc1-1 brsoc1-3, and 20 in the F1 hybrid 20 × brsoc1-1 brsoc1-2 brsoc1-3. Based on the above bolting times, we conclude that the early-bolting trait is likely dominant over the brsoc1 mutations, although we do note a dose-dependent effect in the delay of bolting as more BrSOC1 loci are mutated. Nevertheless, an early-bolting inbred line such as 20 is not suitable for crossing with plants genome-edited at BrSOC1s to significantly delay bolting.
Selection of late-bolting inbred lines for crossing to brsoc1 lines
To test the effectiveness of crossing brsoc1 lines to late-bolting inbred lines, we looked for inbred lines with delayed bolting times among a set of lines developed through traditional breeding. Accordingly, we obtained ten inbred lines (JN01 to JN10) and measured their bolting time after 35 days of vernalization treatment. After 25 days of growth following release from vernalization, three inbred lines (JN05, JN07, and JN10) had completed flowering, and another inbred line had started to bolt (Fig. 2A). By contrast, the inbred lines JN03, JN06, and JN08 had not started to bolt at that time. Indeed, the average bolting time of JN06 was 30 days, while that of JN08 was 33 days. In agreement with their later bolting time, the inbred lines JN06 and JN08 produced more leaves by the time of bolting than the other inbred lines (Fig. 2B).
To assess the function of BrSOC1s in these inbred lines, we conducted reverse transcription quantitative PCR (RT-qPCR) analysis to examine the expression levels of the three BrSOC1 genes in plants exposed to 35 days of vernalization treatment or maintained at 22°C. The transcript levels of BrSOC1-1 increased after vernalization in all inbred lines, although reaching lower levels in vernalized JN03 and JN08 plants than in the other inbred lines. BrSOC1-2 was strongly induced following vernalization treatment in JN02, JN05, and JN07, but showed a more modest increase in JN03, JN06, and JN08 after vernalization treatment, with lower transcript levels than in the other inbred lines. Similarly, the transcript levels of BrSOC1-3 were the lowest in vernalized JN03, JN06, and JN08 plants (Fig. 2C). Although JN03 displayed late bolting and low expression of all BrSOC1 genes, this inbred line showed abnormal growth after vernalization, making it unsuitable for crosses (Fig. 2A, magnified field of JN03). Based on their bolting time, the number of rosette leaves produced at the time of bolting, and their expression levels of BrSOC1 genes, we chose JN06 and JN08 for crossing with brsoc1 lines.
Crossing genome-edited brsoc1 lines with late-bolting inbred lines effectively delays bolting time
We generated F1 hybrids by crossing JN06 or JN08 as the maternal line, since these inbred lines are male sterile, with genome-edited brsoc1 lines as the pollen donor. We collected mature F1 hybrid seeds from all crosses (Fig. S2).
We confirmed that all F1 hybrids exhibited an InDel frequency of 50% at the BrSOC1 loci at which they harbored one mutated copy (Fig. 3A). We assessed the bolting phenotype of all parental lines and their F1 hybrids following 35 days of vernalization treatment (Fig. 3B). None of the F1 hybrid plants had developed an inflorescence at 33 days after vernalization treatment, in contrast to their maternal lines, JN06 and JN08. Indeed, the F1 hybrid plants JN06 × brsoc1-1/BrSOC1-1 brsoc1-2/BrSOC1-2 brsoc1-3/BrSOC1-3 and JN08 × brsoc1-1/BrSOC1-1 brsoc1-2/BrSOC1-2 brsoc1-3/BrSOC1-3 only started to bolt after 40 days (JN06 F1 hybrids) and 35 days (JN08 F1 hybrids) after vernalization.
In more detail, F1 hybrid plants carrying one mutated brsoc1 gene at each of two loci (JN06 × brsoc1-1 brsoc1-3, JN08 × brsoc1-1 brsoc1-2, and JN08 × brsoc1-1 brsoc1-3) bolted at 33 to 35 days after vernalization treatment, approximately 3–5 days later than their respective maternal line, JN06 or JN08 (Fig. 3C). These hybrids also had at least three more rosette leaves than plants from the maternal lines. The two F1 hybrids heterozygous at all three BrSOC1 loci (JN06 × brsoc1-1 brsoc1-2 brsoc1-3 and JN08 × brsoc1-1 brsoc1-2 brsoc1-3) exhibited further delayed bolting, by 12 days for F1 hybrids derived from JN06 and 6 days for F1 hybrids derived from JN08. Moreover, these hybrids produced at least seven more rosette leaves than their respective maternal line. Importantly, the number of leaves in the two F1 hybrids heterozygous at all three BrSOC1 loci was similar to that of the brsoc1-1 brsoc1-3 double mutant. These results suggest that F1 hybrids produced by breeding can be effective in delaying bolting time without having to generate homozygous mutants through CRISPR/Cas9, thus reducing the time-consuming and labor-intensive aspects of the process.
Genetic inactivation of BrSOC1s represses the expression of downstream floral homeotic genes
As the F1 hybrid plants heterozygous for mutated brsoc1s at two or more loci showed delayed bolting, we wished to explore the consequences of these mutations on the expression of flowering-related genes downstream of SOC1. In Arabidopsis, SOC1 expression is activated by the FLOWERING LOCUS T (FT)–FD complex [23]; SOC1 collaborates with the floral homeotic genes APETALA1 (AP1) [24], LEAFY (LFY) [25], and FRUITFULL (FUL) [26] for meristem specification. We looked for the Chinese cabbage homologs of these floral homeotic genes using the Brassica database (BRAD; http://brassicadb.cn), yielding BrAP1 (Bra038326), BrLFY (Bra029305), and two genes for BrFUL (BrFUL1 [Bra029347] and BrFUL2 [Bra035952]). The transcript levels of all of these genes increased significantly in vernalized JN06 and JN08 plants relative to plants maintained at 22ºC. Notably, we noticed less pronounced increases in expression levels in many of the F1 hybrids compared with JN06 and JN08, with a greater effect observed as more BrSOC1 copies were inactivated (Fig. 4). Expression levels in the two F1 hybrids JN06 × brsoc1-1 brsoc1-2 brsoc1-3 and JN08 × brsoc1-1 brsoc1-2 brsoc1-3 were similar to those in the brsoc1-1 brsoc1-3 double mutant after vernalization. This finding is consistent with the bolting phenotype of these genotypes under vernalization.
F1 hybrids are more resistant to bolting than commercial varieties
To ensure practical applicability, we compared the late-bolting phenotype of the F1 hybrids to that of five commercial varieties known for their late-bolting characteristics. We exposed all seedlings to 35 days of vernalization treatment and transferred the plants to small pots to assess bolting time effectively. Remarkably, all five commercial varieties had finished bolting at 22 days after release from vernalization treatment, whereas the first F1 hybrid plants from the crosses JN06 × brsoc1-1 brsoc1-3 and JN08 × brsoc1-1 brsoc1-3 only started to bolt at 24 days after vernalization (Fig. 5A). By 27 days after release from vernalization, the two sets of F1 hybrids either displayed a short flowering stalk or had not developed inflorescences, surpassing both their maternal lines and the commercial lines. The F1 hybrids bolted on average at 25 days after vernalization treatment and produced an average of 23 leaves, representing a significant difference of 5 days and 5 leaves compared with the commercial lines (Fig. 5B). This result indicates that the development of F1 hybrids holds promise for practical commercial applications.