Quantitative Analysis of Florigen for the Variability of Floral Induction in Cabbage/Radish Inter-generic Grafting

Abstract

Grafting-induced flowering is a key phenomenon to understand systemic floral induction caused by florigen. It can also be used as a breeding technique enabling rapid seed production of crops with long generation times. However, the degree of floral induction in grafted plants is often variable. Moreover, it is difficult in some crop species. Here, we explored the factors promoting variability in the grafting-induced flowering of cabbage (Brassica oleracea L. var. capitata), an important vegetable crop with a long generation time, via the quantitative analysis of florigen accumulation. Significant variability in the flowering response of grafted cabbage was observed when rootstocks of different genotypes were used. As reported previously, B. oleracea rootstocks did not induce the flowering of grafted cabbage plants, but radish (Raphanus sativus L.) rootstocks unstably did, depending on the accessions used. Immunoblotting analysis of the FLOWERING LOCUS T (FT) protein, a main component of florigen, revealed that floral induction was quantitatively correlated with the level of accumulated FT protein in the grafted scion. To identify rootstock factors that cause variability in the floral induction of the grafted scion, we investigated FT protein accumulation and flowering response in grafted scions when the transcription levels of FT and the leaf area of rootstocks were altered by vernalization, daylength and leaf trimming treatments. We concluded that increasing the total amount of FT protein produced in the rootstock is important for the stable floral induction of the grafted cabbage, and this can be accomplished by increasing FT transcription and the leaf area of the rootstock.

Introduction

Grafting-induced flowering has been intensively studied in many plant species as a key phenomenon to understand systemic floral regulation in plants (Chailakhyan 1937, Zeevaart 1976). This process involves plants under non-inductive conditions being induced to flower by grafting to florally induced donor plants. Historically, this phenomenon evoked the idea that ‘florigen’—a transmissible floral inducer common to a wide range of plant species—was synthesized in the leaves under specific inductive conditions and transported to the shoot apex to induce flowering (Chailakhyan 1937). The identity of florigen was unknown for many years, but the small globular protein FLOWERING LOCUS T (FT) and its orthologs have recently been identified as major florigen components (Kobayashi et al. 1999, Abe et al. 2005, Lifschitz et al. 2006, Corbesier et al. 2007, Lin et al. 2007, Tamaki et al. 2007, Notaguchi et al. 2008). Subsequent grafting experiments have confirmed that FT is translocated from induced donor leaves to the recipient shoot apex via the graft junction and is a causal factor of grafting-induced flowering (Corbesier et al. 2007, Lin et al. 2007, Notaguchi et al. 2008).

Grafting-induced flowering also has the potential to be used as a breeding technique, since rapid induction of flowering reduces the time for crossing and seed production, thereby permitting shorter breeding cycles. Grafting-induced flowering is currently used to induce flowering in several crop species that have long generation times or show difficulty in flowering under natural conditions. These species include sweet potato (Ipomoea batatas; Sunakawa 1966), citrus (Citrus sinensis; Soares et al. 2020) and jatropha (Jatropha spp.; Ye et al. 2014, Tang et al. 2022). Our research group is developing techniques to shorten the time to flowering of cabbage (Brassica oleracea L. var. capitata) by grafting (Motoki et al. 2019). Cabbage is an important leaf vegetable cultivated throughout the world and requires long generation times for breeding. Due to the existence of a juvenile phase and a strong vernalization requirement for flowering (Miller 1929), cabbage can take 6 months or more for each generation, which is much longer than other cruciferous crops. In an extreme cabbage accession, flowering may not occur even after overwintering (Kinoshita et al. 2021). Since grafting-induced flowering permits skipping the vernalization treatment altogether by directly inducing cabbage flowering (Kagawa 1957, Motoki et al. 2019), it is relatively faster than conventional methods of promoting flowering using climate chambers.

Despite this advantage, the variability and low reproducibility of floral induction in grafted cabbage have made this technique impractical for reliable breeding (Hamamoto and Yoshida 2012). It has been observed that the flowering response of grafted cabbage varies greatly depending on the rootstock genotype. Cabbage is known not to be induced to flower when grafted onto florally induced B. oleracea rootstocks but can flower only when grafted onto radish (Raphanus sativus L., Brassicaceae) rootstocks (Kagawa 1957, Motoki et al. 2019). In addition, the flowering response of grafted cabbage can differ even among radish accessions (Motoki et al. 2019). The rootstock factor causing this variability remains unknown but must be identified if grafting-induced flowering is to be used for practical cabbage breeding schemes. Similar variability or difficulty in producing grafting-induced flowering has also been reported in other crop species. Studies have reported failures in floral induction in recipient plants even when an FT-overexpressing transformant was used as a donor rootstock (Tränkner et al. 2010, Zhang et al. 2010, Wenzel et al. 2013, Bull et al. 2017, Odipio et al. 2020, Wu et al. 2022). In these cases, rapid floral induction in the FT-overexpressing transformant itself was observed. Therefore, floral induction of the donor rootstock may not be sufficient to cause successful grafting-induced flowering in general.

In this study, we explored the factors that contribute to the variable flowering response caused by the use of different rootstocks in grafting-induced flowering of cabbage. We hypothesized that differences in the accumulation of florigen in grafted cabbage plants cause variability in the flowering response. This kind of quantitative action of florigen has been previously observed in grafted plants (Tang et al. 2022) as well as in intact plants (Liu et al. 2012, S. C. Yoo et al. 2013, Endo et al. 2018, Susila et al. 2021). To investigate this hypothesis, we developed an antibody for the FT proteins of cabbage and radish and quantitatively measured their accumulation in cabbage plants grafted to different rootstocks. After confirming the association between the levels of FT protein accumulation and the flowering response, we further explored rootstock factors affecting the levels of FT accumulation in grafted cabbage plants. Two candidate factors, namely the level of FT transcription and leaf area, differed among the rootstock genotypes and could be artificially manipulated by vernalization, daylength and leaf trimming treatments. We then performed a grafting experiment with these treatments. The results of this experiment confirmed that both FT expression and leaf area in the rootstock needed to meet some threshold to cause FT protein accumulation and induce flowering in the grafted cabbage plants. This study therefore provides insight into the mechanisms affecting the variability of grafting-induced flowering and provides perspective on how stable floral induction by grafting for the rapid breeding of cabbage can be achieved.

Results

Differences in the flowering responses of cabbage scions grafted onto various early-flowering B. oleracea and R. sativus rootstocks

To reconfirm that the variability of the flowering response of grafted scions depends on the genotype of the rootstock (as observed in a previous study), young cabbage seedlings (3–4 weeks after sowing) were grafted under non-vernalized conditions to rootstocks of different genotypes (experiment 1). In a previous study, we observed a higher frequency of flower bud differentiation in cabbage scions when grafted onto early-flowering radish accessions (Motoki et al. 2019). Therefore, we used two early-flowering accessions from B. oleracea and five accessions from R. sativus as rootstocks (Supplementary Table S1). For all radish accessions, the appearance of flower buds and the bolting of the rootstocks were observed within 30 days after sowing (DAS) in average without vernalization treatment (Supplementary Table S2). For B. oleracea accessions, bolting cannot be used as an indicator of floral induction because stem elongation occurs regardless of floral induction in this species. Flower bud appearance is usually observed at 25–30 DAS in the TO1000 accession and at 40–50 DAS in the Kairan accession under the same conditions (data not shown). Grafting was performed at the day of bolting in R. sativus accessions and at 30 DAS in B. oleracea accessions. Grafting success rates ranged from 78% to 100% (Supplementary Table S2). The flowering response of grafted cabbage plants varied greatly depending on rootstock genotype, with the values of the percentage of scions with differentiated flower buds ranging from 0% to 100% by the end of the grafting experiment (Fig. 1A, B). Neither of the two B. oleracea accessions induced the flowering of grafted scions (Fig. 1A, B), which was consistent with the results of previous studies showing that B. oleracea rootstocks cannot induce flowering in grafted cabbage plants (Kagawa 1957, Motoki et al. 2019). On the other hand, we also observed substantial differences in the percentage of scions with differentiated flower buds among the five radish accessions (Fig. 1A, B). The number of days to flower bud appearance and the number of flowers opened by 63 days after grafting (DAG) in the scion also differed among radish accessions (34.8–60.6 d and 0–95.4 flowers, respectively, Supplementary Table S2). Taken together, these results suggest that there are quantitative differences among radish accessions with respect to the ability to induce flowering in grafted cabbage.

In addition to reproductive growth, we also examined the vegetative growth of the grafted cabbage plants. In cabbage, the maximum stem diameter is used as a proxy for vegetative growth and physiological age. This is because there is a correlation between the maximum stem diameter and vegetative growth parameters such as above-ground fresh weight and the number of expanded leaves and because it can be measured non-destructively (Ito and Saito 1961; Supplementary Fig. S1). Thus, the maximum stem diameter at 35 DAG was used as the parameter characterizing the vegetative growth of grafted scions in this study. We found significant differences in the stem diameter of grafted cabbage plants among rootstock accessions (Supplementary Table S2). Moreover, the stem diameter of the grafted scion strongly correlated with the total area of leaves left on the rootstock on the day of grafting (Fig. 1C). This indicated that scion growth was dependent on the translocation of assimilates from the rootstock. This is probably caused by the removal of mature leaves of the scion (see the section ‘Materials and Methods’), which we performed to maintain the sink activity of the scion, as used in previous studies (Hamner and Bonner 1938, Kagawa 1957).

Quantification of FT protein accumulation in grafted cabbage scions

To examine the potential causes of the differences in the flowering response of the grafted cabbage, we investigated the relationship between the flowering response of the grafted scion and FT protein accumulation, which is known to be involved in the grafting-induced flowering response in several plant species (Corbesier et al. 2007, Lin et al. 2007, Notaguchi et al. 2008). For this purpose, we developed an anti-peptide antibody for the FT proteins found in radish and cabbage (details are described in Supplementary Methods). A validation experiment involving immunoblotting and immunoprecipitation mass spectrometry analysis showed that the anti-FT antibody developed here could detect R. sativus FT, B. oleracea FT and B. oleracea TWIN SISTER OF FT (TSF) protein and could weakly interact with R. sativus TSF protein (Supplementary Figs. S2–S7; Supplementary Table S3). We also confirmed the quantitativity of the immunoblotting analysis using this antibody (Supplementary Fig. S8).

Next, we checked whether the FT protein could be detected in grafted cabbage scions using the developed anti-FT antibody. Since previous studies have shown that the FT protein also accumulates in sink tissues other than the shoot apex (Navarro et al. 2011, Endo et al. 2018), we investigated the FT protein accumulation level not only in the shoot apex but also in the leaf lamina and the midrib of the young leaves of the grafted scion plants (Fig. 2A, B). Immunoblotting analysis showed that a band of FT protein was detected only in the protein extracted from cabbage scions grafted onto G2 × CH F₁ radish but not in the protein extracted from non-grafted cabbages and cabbage scions grafted onto G2-IL1 radish (Fig. 2B). Thus, we confirmed that the FT protein was transmitted from florally inductive radish rootstocks to the grafted cabbage and that this FT protein could be detected by the developed antibody. Furthermore, the FT protein accumulation level in the tissues of the young leaf was almost comparable to that in the shoot apex of the same plant (Fig. 2B). Therefore, in the following experiments, we extracted protein using the midrib of the young leaf of the scion to avoid destructive effects on the growth of the scion. Using this method, we then measured the FT protein accumulated in grafted scion plants by determining the relative band intensity of the FT protein normalized against the total protein signal. As a result, we observed significant differences in the amount of FT protein accumulated in grafted scions among the rootstock accessions (Fig. 2C). When we examined this data along with the flowering response patterns of the grafted scions, we found that the FT protein accumulation level was significantly higher in scions that had differentiated flower buds by the end of the experiment (Fig. 2D). In addition, there was a significant negative correlation between the number of days to flower bud appearance of the scion and the level of FT protein accumulation (Fig. 2E, r = − 0.59, P < 0.01). These results suggested that differences in the flowering response of scion plants grafted onto different rootstock accessions were caused by differences in the level of FT protein accumulation in the scion.

To clarify the factors that cause differences in FT protein accumulation in scions, we measured FT transcript expression in the leaves of the rootstock. As mentioned above, we could not detect FT protein in scion plants grafted onto B. oleracea accessions; however, TO1000 showed significantly higher FT expression than Kairan (Fig. 2F). In the R. sativus accessions, we observed significant differences in FT expression, with a mean difference of up to 18.9-fold (Fig. 2G, between G2-IL1 and G2 × CH F₁). There was also a significant but weak positive correlation between the level of FT transcription in the radish rootstock and the level of FT protein accumulation in the grafted cabbage (Fig. 2H, r = 0.47, P < 0.01). When the FT transcript expression level was compared between the two B. oleracea accessions and the two representative R. sativus accessions (G2-IL1 and G2 × CH F₁) by absolute quantification, G2-IL1 showed comparable and G2 × CH F₁ radish showed a much higher level of FT expression than B. oleracea accessions (Supplementary Fig. S9A, B). In agreement with this, the FT protein could be detected only in G2 × CH F₁ plants (Supplementary Fig. S9C). Therefore, we hypothesized that the variability in FT protein accumulation and floral induction in cabbage plants grafted onto different rootstocks was caused by differences in the level of transcription of FT in the rootstock. To examine this hypothesis, we next performed the following grafting experiment using radish rootstocks with altered FT transcription levels by vernalization treatment.

Effect of seed vernalization treatments of the rootstock

In radish, the low-temperature treatment promotes flowering at the imbibed seed stage (Hagiya 1951). Since it was reported in radish that FT transcription was upregulated according to floral acceleration induced by vernalization treatment (Jung et al. 2020), we first investigated the effect of seed vernalization treatments of the rootstock on the flowering response of the grafted cabbage (experiment 2). The radish accessions G2-IL1 and G2 × CH F₁, whose grafted scions showed different flowering responses in experiment 1, were used for this experiment. 0–14-day long seed vernalization treatments were applied to rootstocks, and then grafting was performed on bolted rootstocks as in experiment 1 (experiment 2-1–2-2 in Supplementary Table S2). While both radish accessions do not require low-temperature exposure for flowering, the vernalization treatment significantly accelerated the flowering of these radish rootstock accessions (Fig. 3A). We confirmed that FT expression in the leaf of the rootstock had significantly increased in response to the vernalization treatment (Fig. 3B). However, contrary to our expectation, the flowering response of the grafted scion was not promoted by the vernalization treatment in either G2-IL1 or G2 × CH F₁, despite increased FT expression in their rootstocks (Fig. 3C; Supplementary Table S2). In particular, G2-IL1 did not induce the flowering of grafted scions at all under any vernalization treatment (Fig. 3C). For G2 × CH F₁, the percentage of flower-bud-differentiated scions was the highest in the non-vernalized treatment (Fig. 3C) and the number of days to flower bud appearance was also the shortest on average in the non-vernalized treatment (Supplementary Table S2).

The level of accumulated FT protein in scion plants tended to increase with increasing duration of the vernalization treatment in G2-IL1, but there was considerable variation in this response (Fig. 3D). On the other hand, G2 × CH F₁ showed the highest accumulation of FT protein in scions in the non-vernalized treatment (Fig. 3D). The FT protein accumulation level was significantly higher in scions that had differentiated flower buds by the end of the experiment (Fig. 3E), although the FT protein was clearly detected in some individuals where flower bud differentiation did not occur (Fig. 3E). In summary, the vernalization treatment increased the expression of FT transcripts in the rootstock but had almost no or a negative effect on FT protein accumulation and flowering response in the scion. In this experiment, we found no correlation between the expression levels of FT transcripts in the rootstock and the level of accumulated FT protein in the scion (Fig. 3F). This indicated that there are factors other than FT expression levels in the rootstock that affect the amount of FT protein accumulated in the scion.

In this experiment, we observed that the vernalization treatment also significantly reduced the leaf area of rootstocks at the day of grafting in both of the accessions (Fig. 3G). This was probably due to the fact that bolting occurred earlier. As we also observed differences in the leaf area among the rootstocks in experiment1 (Fig. 1C), we speculated that the leaf area of the rootstock has an effect on FT protein accumulation and hence on flowering response in the grafted scion. Thus, we separately investigated the effect of the FT transcription level and rootstock leaf area in the following experiments.

Effect of daylength before and after grafting

Taking advantage of the flowering characteristics of radish, which is a long-day (LD) plant (Hagiya 1951), we performed a grafting experiment using rootstocks with the same leaf area but with different levels of FT transcription by applying daylength treatments (experiment 3). The following three conditions were applied to G2 × CH F₁ rootstocks and their grafts: LD conditions of 16-h light/8-h dark throughout the growing period (LD/LD), LD conditions up to grafting and middle-day (MD) conditions of 12-h light/12-h dark after grafting (LD/MD) or MD conditions up to 21 DAS and LD conditions afterward (Fig. 4A). The number of days to bolting, the leaf area of the rootstock and the vegetative growth of the grafted scion were comparable between the LD/LD and LD/MD treatments (Fig. 4B, C, D). On the other hand, in the MD/LD condition, the bolting of the rootstock was significantly delayed, the leaf area of the rootstock was significantly greater and the scion growth was significantly higher than in the other two treatments (Fig. 4B, C, D). This confirmed the repressive effect of shorter daylength on the flowering of radish plants. As expected, the FT transcription levels of the rootstock at 35 DAG was significantly lower in LD/MD plants and did not differ between LD/LD and MD/LD plants (Fig. 4E). The flowering response of scions grafted onto these rootstocks showed that almost all the LD/LD and MD/LD scions reached the flower opening stage, whereas more than half of the LD/MD scions continued growing only vegetatively until the end of the experiment (Fig. 4F). Moreover, the level of accumulated FT protein in the scion was significantly lower in LD/MD plants and did not differ between LD/LD and MD/LD plants (Fig. 4G). Taken together, these results suggest that the transcription level of FT in the rootstock determines the level of FT protein accumulation and the flowering response of the scion if the leaf area of the rootstock is the same.

Effect of leaf trimming of rootstock plants

We then performed another grafting experiment using rootstocks with the same FT transcription level but with different leaf areas (experiment 4). On the day of grafting, zero, one or two leaves of G2 × CH F₁ radish were cut from the base of the petiole starting from the largest leaf, and cabbage scions were then grafted onto the rootstock (Fig. 5A). The leaf area of the rootstock decreased accordingly with the number of trimmed leaves (Fig. 5B). The stem diameter of the grafted cabbage also decreased significantly in response to leaf trimming (Fig. 5C; Supplementary Fig. S10), which confirmed that the rootstocks experienced a decrease in source capacity. In addition, the flowering response of the grafted cabbage plants was repressed as the number of trimmed leaves increased (Fig. 5D), and the level of accumulation of FT protein in the scion also decreased in response to trimming, although not significantly (Fig. 5E). When the relationship between the leaf area of the rootstock and the accumulation of the FT protein in the scion was examined in individual plants, we found a significant positive correlation (Fig. 5F). Taken together, these results showed that the leaf area of the rootstock determines the level of FT protein accumulation and the flowering response in the grafted cabbage when the level of FT transcription in the rootstock is the same.

Discussion

Variation in the flowering response of the grafted cabbage is quantitatively correlated with the level of accumulation of FT protein

In this study, we observed that the flowering response of the grafted cabbage varied greatly depending on the genotype of the rootstock; this finding agreed with those of previous studies (Kagawa 1957, Motoki et al. 2019). Moreover, even when grafted onto early-flowering radish accessions, whose difference in the average time to bolting was only ∼6 d, the flowering response of the cabbage scions varied considerably, ranging from complete vegetative growth to complete flower bud differentiation (Fig. 1A, B). This finding showed that the donor plant itself being induced to flower is not a sufficient condition for floral induction of the grafted cabbage. To clarify the cause of this variability in flowering response, we developed an antibody to detect the native FT protein of R. sativus and B. oleracea and examined the relationship between the flowering response of the scion and its level of accumulated FT protein. The results showed that the variable flowering response of the scion was consistent with the level of accumulated FT protein (Fig. 2D). We also confirmed this tendency in other grafting conditions where vernalization, daylength and leaf trimming treatments were applied to the rootstock (Figs. 3E, 4G, 5F). This indicated that the transmission of FT protein from the donor rootstock causes floral induction in the recipient cabbage, which is similar to the pattern found in previous studies of other plant species (Lin et al. 2007, Notaguchi et al. 2008, Tang et al. 2022). Furthermore, we observed that the level of FT protein accumulation in the scion was correlated not only with the presence or absence of flower bud differentiation but also with quantitative indices of floral induction, such as days to flower bud appearance and the number of opened flowers (Fig. 2E; Supplementary Fig. S11). In previous studies investigating FT transmission, the accumulation of the FT protein at the shoot apex was found to be quantitatively related to the flowering response (Liu et al. 2012, S. C. Yoo et al. 2013, Endo et al. 2018, Susila et al. 2021). Therefore, it may be that variation in the grafting-induced flowering of cabbage is caused by quantitative differences in the amount of FT protein translocated to the scion.

In contrast, in this study the threshold level of FT protein accumulation required for the floral induction of cabbage was unclear in some cases. For example, during the seed vernalization experiment, when vernalized G2-IL radish plants were used as rootstocks, some individuals were not induced to flower even though FT was clearly detected in the scion (Fig. 3D). Functional mutations in the FT protein are probably not the cause of this phenomenon since we found no non-synonymous mutations in the coding region of FT in G2-IL1 or any other radish accessions (data not shown). One possible reason for differences in floral induction may be that the threshold level of the FT protein required for successful induction may vary depending on the state of the scion (e.g. its physiological age). Cabbage is a plant-vernalization-type plant, which becomes sensitive to low temperatures only after developing to a size threshold characterized by the expansion of many true leaves (Miller 1929); this property may also affect grafting-induced flowering. In addition, we observed less vegetative growth in scions grafted onto G2-IL1 plants than in those grafted onto G2 × CH F₁ plants (Supplementary Fig. S12A). This difference seems to be related to the fact that G2-IL1 plants show earlier leaf senescence (Supplementary Fig. S12B, C). Thus, it is possible that the physiological age of the scions grafted onto G2-IL1 did not advance sufficiently, thereby inhibiting floral induction. Since the molecular mechanisms involved in the plant-vernalization-type response of cabbage plants have not yet been elucidated, it cannot be discussed further in this study. However, we consider it to be of interest for future studies.

Possible reason why B. oleracea rootstocks lack ability to induce flowering in grafted cabbages

Although the anti-FT antibody used in this study can detect the FT protein of B. oleracea (Supplementary Fig. S6A), the FT protein was not detected from cabbage scions grafted onto B. oleracea rootstocks (Fig. 2C). This seemed to be caused by the lower transcription and protein expression of FT in these B. oleracea accessions (Supplementary Fig. S9). Considering the fact that the FT homolog of B. oleracea has been shown to promote flowering in Arabidopsis transformants (Itabashi et al. 2019), the failure of B. oleracea rootstocks to induce flowering in grafted cabbages in this study was possibly due to the lower ability to express the FT protein.

However, we cannot exclude the possible involvement of the difference in the function, mobility or stability of the FT protein to the difference in the flowering-inducing ability between B. oleracea and R. sativus rootstocks. This is because there are several amino acid polymorphisms between RsFT and BoFT (Supplementary Fig. S4), including the one that was reported to cause functional mutation in Arabidopsis (Gln-140, Ho and Weigel 2014). A previous study, which observed significant difference in the flowering-promoting activity among the FT homologs from several crop species (Yamagishi et al. 2014), also indicates the possible difference in the flowering-promoting activity between RsFT and BoFT. To clarify these possibilities, it will be firstly necessary to find B. oleracea accessions that express higher levels of the BoFT protein (or develop transformants that do) and to investigate if they can induce the flowering of the grafted cabbage as well as R. sativus rootstocks. Another approach would be to directly test the interaction between these FT homologs and the cabbage homologs of the known partners of FT such as FD and FD PARALOG (FDP) (Abe et al. 2005, Taoka et al. 2011), FT INTERACTING PROTEIN 1 (FTIP1) (Liu et al. 2012) and SODIUM POTASSIUM ROOT DEFECTIVE1 (NaKR1) (Zhu et al. 2016). We found three FD homologs (XM_013733162, XM_013769361 and XM_013741136), two FDP homologs (XM_013751187 and XM_013740745), two FTIP1 homologs (XM_013774682 and XM_013736140) and two NaKR1 homologs (XM_013763098 and XM_013757038) from the NCBI Reference Sequence Database of B. oleracea, which could be the potential targets of the interaction analysis in the future.

High expression of FT transcripts in the rootstock is necessary but not sufficient to cause the accumulation of FT protein and induce flowering in grafted cabbage

Previous studies have reported that the flowering responses of recipient plants were associated with FT transcription levels in donor plants (Notaguchi et al. 2008, S. J. Yoo et al. 2013, Tang et al. 2022). However, in the current study, we found that the FT transcription levels of the donor rootstock alone could not explain the flowering response of the receptor scion. This was especially true of the vernalization-treatment experiment (Fig. 3). Here, the FT transcription level of the rootstock did not coincide with the level of FT protein accumulation in the scion (Fig. 3F), suggesting that the level of FT protein accumulation and floral induction in the scion may be affected by other factors. In consideration of the observed differences in both FT transcription levels and leaf areas among radish genotypes (Fig. 1C; Fig. 2G), we separately investigated the effects of these two factors on FT protein accumulation in grafted scions. First, we confirmed by a daylength-manipulation experiment that the FT transcription level of the rootstock had a clear effect on FT protein accumulation in the grafted scion if the leaf area was the same among rootstocks (Fig. 4). Next, we confirmed by a leaf trimming experiment that the leaf area determined the level of FT protein accumulation in the scion if the transcription level of FT was comparable in all rootstocks (Fig. 5). This was consistent with the observations of previous grafting studies, which showed that reducing the leaf area of donor plants used for grafting suppressed the floral induction of the receptor plant (Zeevaart 1958, Suge 1986). Therefore, we suggest that both the level of FT transcription and the leaf area of the rootstock contribute to the level of FT protein accumulation in the grafted scion, and both FT transcription and leaf area need to be high to cause floral induction in grafted cabbage.

Based on these results, the discrepancy observed in the vernalization-treatment experiment between the FT transcription level of the rootstock and the level of FT protein accumulation in the scion seems to be primary related to the reduced leaf area in rootstocks caused by floral acceleration (Fig. 3A, G). It is possible that a smaller leaf area may have counteracted the effect of the increased transcription level of FT brought about by the vernalization treatment and thereby reduced the total amount of FT protein produced in the whole rootstock plant. Our results indicated that the leaf area of the rootstock, rather than the FT transcription level, may be a restricting factor for grafting-induced flowering in some cases. We suppose that previous studies where failures in grafting-induced flowering were observed despite the use of FT-overexpressing transformants (Tränkner et al. 2010, Zhang et al. 2010, Wenzel et al. 2013, Bull et al. 2017, Odipio et al. 2020, Wu et al. 2022) may be related to the deficiency in the amount of leaves producing FT. Another potential cause for the discrepancy between the FT transcription level in the rootstock and the FT protein accumulation level in the scion might be the possible reduction in the transcription level of FT after the end of vernalization treatment. It is possible that the FT transcription level of the radish rootstocks decreased in the warm condition after the vernalization exposure, as observed in some perennial cruciferous species (Satake et al. 2013, Irwin et al. 2016). Because sampling was conducted only once at 14 d after the vernalization treatment in the current experiment, the FT transcription level of the radish rootstocks may have been overestimated as the entire growing period. The time-course observation of the FT transcription level will be needed for further elucidation of the relationship between the FT transcription level in the rootstock and FT protein accumulation in the grafted scion.

Perspective on stable floral induction by grafting

For breeding and seed production, induction of flowering by grafting is useful for crops that have long generation times or have difficulty in flowering under natural conditions. However, as we observed in this study, grafting-induced flowering is often variable even when florally induced donor rootstocks are used. In this study, we showed that variability in floral induction in cabbage/radish grafting was associated with two rootstock traits: the level of FT transcription and leaf area. Both of these traits contribute to the level of FT protein accumulation in the grafted cabbage. Here, we discuss the factors affecting these traits and the possibility of manipulating them to permit stable floral induction by grafting.

We found large differences in the FT transcription levels of the radish accessions grown under the same conditions (Fig. 2G). This indicated that genetic factors determine the transcription level of FT in radish, and it is possible that genetic selection may be performed for this trait. Previous studies observed that FT transcription levels tended to be higher in early-flowering radish accessions than in radish accessions that require vernalization to flower (Jung et al. 2020, Han et al. 2021). In this study, we found a correlation between the transcription level of FT and the days to bolting among radish accessions (Supplementary Fig. S13A). Therefore, early-flowering radish accessions seem to be suitable for preparing the rootstock to ensure high levels of FT transcription. In addition, we also found that vernalization treatments significantly promoted both floral induction and the transcription level of FT even in early-flowering radish accessions that do not require vernalization for floral induction (Fig. 3A, B). Although the difference in the days to bolting was about 1 week between non-vernalized and 7-day-vernalized G2-IL1 plants, the difference in FT transcription levels between these conditions was approximately 200-fold (Fig. 3A, B). This shows that vernalization treatments exert significant effects on the transcriptional activation of FT even in early-flowering radish accessions. We also confirmed that LD conditions were necessary to induce high FT expression in radish (Fig. 4), which is in agreement with the fact that radish is an LD plant (Suge and Rappaport 1968).

When considering rootstock size, plants with larger leaf areas can induce flowering to a greater degree in grafted cabbage if rootstocks with comparable FT transcription levels are used (Fig. 5). However, because floral acceleration reduces leaf area by shortening the vegetative growth period (Fig. 3A, G), there is a trade-off between leaf area and FT expression (Supplementary Fig. S13B). One possible way to avoid this trade-off would be to delay the floral induction of rootstocks transiently. In the daylength-manipulation experiment, we observed that in the MD/LD treatment, the floral induction of the rootstock was delayed, the leaf area of the rootstock had increased and the growth of the grafted scion was also greater than in the LD/LD treatment (Fig. 4B–D). On the other hand, FT expression in rootstocks after grafting was similar in both the MD/LD and the LD/LD treatments (Fig. 4E), suggesting that reducing daylength before grafting can increase the leaf area of the rootstock without hampering FT expression after grafting. Another approach would be to select rootstocks with the genetic potential for a larger leaf area and less leaf senescence. This may be possible given the observed differences in leaf area and leaf longevity among radish accessions (Fig. 1C; Supplementary Fig. S12) as well as morphological diversity in shape and size in both roots and leaves within extant radish genetic resources (Yamagishi 2017).

In conclusion, in this paper we showed that in inter-generic cabbage/radish grafts, the level of FT transcription and the leaf area of the rootstock both contribute to variation in the flowering response of the grafted cabbage by altering the levels of FT protein accumulation. The results of this study will not only accelerate cabbage breeding but may also provide fundamental knowledge on the quantitative action of florigen in grafting. Moreover, it may also contribute to the development of new grafting techniques for rapid floral induction in other crop species.

Materials and Methods

Plant materials and growth conditions

Plants were grown in a growth room maintained at 22°C ± 2°C for LD (16 h/8 h, light/dark) conditions or at 23°C ± 2°C for MD (12 h/12 h, light/dark) conditions. Light was provided by fluorescent lamps (photosynthetic photon flux density of 120 μmol m⁻² s⁻¹, NEC Lighting, Ltd., Japan). We used the following accessions as rootstock plants: B. oleracea TO1000, B. oleracea var. alboglabra Kairan, R. sativus var. caudatus G2-IL1, G3-IL1, CH-IL1 and G2 × CH F₁ (F₁ progeny derived by crossing G2-IL1 and CH-IL1) and G2 × G3 F₁ (F₁ progeny derived by crossing G2-IL1 and G3-IL1). The origin of each accession is listed in Supplementary Table S1. Seeds were sown on wet filter paper and germinated in the dark at 22°C for 1–2 d. Germinated seeds were then vernalized at 2°C in the dark for 0–28 d depending on the experiment (Supplementary Table S2). Vernalized seeds were then transplanted into 7.5 cm diameter plastic pots filled with granular rockwool (Nippon Rockwool Corp., Tokyo, Japan) and cultivated further in the growth room. We used seeds of ‘Matsunami’ (Ishii Seed Growers Co., Ltd., Shizuoka, Japan) cabbage as scion plants. These were grown in Jiffy-7^® 42-mm peat pellets (Jiffy Products of America, Inc. Batavia, IL, USA) in a growth room under the LD conditions described above for 3–4 weeks. All plants were irrigated and fertilized from below using a half-strength nutrient solution (Enshi-shoho, formulated by the National Horticultural Research Station, Japan).

Grafting of cabbage scions to rootstock plants

Grafting was conducted according to a previously published protocol (Motoki et al. 2019) with some modification. For radish rootstocks, seedlings that had bolted to a height of 7 cm from the top of the hypocotyl were used as rootstocks. For two accessions of B. oleracea, seedlings at 30 DAS were used as rootstocks. The stem of the rootstock was cut at a height of 4–5 cm from the top of the hypocotyl. A cabbage scion was then cut to be wedge-shaped at the stem with two to three expanded leaves remaining and was grafted onto the stem of the rootstock by cleft grafting. After grafting, the scion was covered with a clear polyethylene bag to maintain high humidity. Two different curing methods described in Supplementary Fig. S14 were used depending on the experiment (Supplementary Table S2). In most cases, the scion and the rootstock were fully connected 1–2 weeks after grafting. To promote the translocation of assimilates from the rootstock to the scion, all lateral shoots of the rootstock were removed. New leaves on the scion were also continuously removed until only five newly expanded leaves (2 cm in length or more) remained; this was to ensure that the scion could preserve sink activity. After the opening of the first flower, all leaves longer than 2 cm in length were continuously removed. Plant growth continued until 63 DAG.

Measurements of the vegetative and reproductive growth of the rootstock and scion

The total leaf area of the rootstock was estimated from the length and width of the largest leaf and the number of leaves left on the rootstock using a regression equation as described by a previous study (Motoki et al. 2019). Grafting was regarded as failed if the scion plant wilted without the polyethylene bag at 21 DAG. The appearance of flower buds was evaluated by daily visual inspection. On 35 DAG, the maximum stem diameter of the scion was measured using a caliper. On 63 DAG, the total leaf number of the scion to flower bud and the total number of opened flowers were counted, and scion flower bud differentiation was evaluated under a microscope if an obvious flower bud was not visible.

Total RNA extraction, cDNA synthesis and quantitative reverse transcription-PCR

Leaf disks were collected from the tip of the largest leaf of the rootstock on 14 DAS, 21 DAS or 35 DAG at 0–0.5 h before the end of the light period. Total RNA was extracted from the leaf samples using Sepasol RNA I Super G (Nacalai Tesque, Inc., Kyoto, Japan), purified using an Econospin™ for RNA (Epoch Life Sciences, Missouri, TX, USA) and reverse transcribed using ReverTra Ace^® qPCR RT Master Mix with gDNA Remover (Toyobo Co., Ltd., Osaka, Japan) according to the manufacturer’s instructions. Subsequently, 1 μL of 20-fold diluted reverse transcription (RT) product was used as a template for quantitative RT-PCR (RT-qPCR). RT-qPCR was performed using a THUNDERBIRD^® SYBR^® qPCR Mix kit (Toyobo Co., Ltd.) according to the manufacturer’s instructions. The reaction was performed using a LightCycler® 480 system (Roche Diagnostics K.K., Tokyo, Japan). RT-qPCR cycling was performed as follows: 95°C for 5 min, followed by 40 cycles at 95°C for 10 s, 60°C for 30 s and 72°C for 30 s. Single-target product amplification was evaluated using a melting curve. For the absolute quantification, dilution series of the linearized plasmid DNA containing coding sequence of each FT gene were used as a standard. All RT-qPCR primers used in this study are listed in Supplementary Table S4.

Protein extraction and quantification

A sample of the midrib of a young leaf (1–2.5 cm in length) was collected from a grafted scion on 35 DAG at 0–0.5 h before the end of the light period. Total protein was extracted according to the method of Wang et al. (2010) with minor modification. In brief, tissue was frozen in liquid nitrogen in a microcentrifuge tube, ground by bead beating using TissueLyser II (Qiagen, Valencia, CA, USA) and homogenized in approximately 5× volume of extraction buffer [50 mM Tris–HCl, pH 9.0, 2% (w/v) sodium dodecyl sulfate (SDS), 5 mM ascorbic acid, 0.1% (v/v) 2-mercaptoethanol and 1× protease inhibitor cocktail (P9599; Sigma-Aldrich Co. LLC, St. Louis, MO, USA)]. The homogenate was subsequently centrifuged twice for 10 min at 16,100×g. The supernatant was used for further analysis. The total soluble protein was quantified using an XL Bradford assay (Intégrale Co., Ltd, Tokushima, Japan).

Immunoblotting analysis

The extracted protein was diluted with an equal volume of 2× SDS sample buffer [125 mM Tris–HCl, pH 6.8, 4% (w/v) SDS, 30% (w/v) glycerol, bromophenol blue, 10% (v/v) 2-mercaptoethanol] and incubated at 95°C for 3 min. In total, 30 µg of protein was then loaded on a 16% (w/v) SDS–polyacrylamide gel, separated by electrophoresis and transferred onto a polyvinylidene fluoride membrane (Immobilon-P; Cytiva, MA, USA). After transferring, the membrane was stained with Ponceau S staining solution (Beacle, Inc., Kyoto, Japan), and a picture of the membrane was taken using a scanner. After blocking with PBST buffer (phosphate-buffered saline containing 0.1% Tween 20) containing 1% skim milk at a room temperature of 20–25°C, the membrane was incubated with an anti-FT antibody (Pos2, developed in this study) diluted to 300 ng/ml in Can Get Signal^® Immunoreaction Enhancer Solution 1 (Toyobo Co., Ltd.) at room temperature. Next, the membrane was incubated with a 1:20,000 diluted anti-rabbit IgG, Horseradish peroxidase-linked antibody (Cell Signaling Technology, Inc., Danvers, MA, USA) with Can Get Signal^® Immunoreaction Enhancer Solution 2 (Toyobo Co., Ltd.) at room temperature. Detection was performed in TMB solution for Western blotting (Nacalai Tesque, Inc.) for 15 min at 23°C. After washing and drying, a picture of the membrane was taken using a scanner, and the intensity of the target band was quantified with Image Lab Software (Version 6.1; BioRad, Hercules, CA, USA). The FT protein band intensity was normalized for variation in total protein loading in the corresponding lane, as quantified by the Ponceau S-stained membranes (Romero-Calvo et al. 2010), and was then normalized using a positive control (a mixture of protein extracted from the leaf petiole of a 14-day-vernalized G2-IL1 and a 0-day-vernalized Haya radish) on the same blot.

Supplementary Data

Supplementary Data are available at PCP online.

Data Availability

The data underlying this article are available in the article and in its online supplementary material.

Funding

A Grant for Scientific Research from the Faculty of Agriculture, Kindai University, Japan Society for the Promotion of Science (JSPS) [KAKENHI Grant-in-Aid for JSPS Fellows 19J15038 and grant-in-aid for Young Scientists (B) 20K15518] and Kyoto University start-up grant.

Acknowledgements

We would like to say a special thanks to Prof. Susumu Yazawa who gave us an abundance of valuable information, resulting from his long-term investigations into cabbage flowering. Original seeds of G2-IL1 and G3-IL1 were kindly provided by the Genebank Project, NARO, Japan (JP No. 76703 and JP No. 76704, respectively). Mass spectrometry analysis was performed with the kind support of the Medical Research Support Center, Graduate School of Medicine, Kyoto University. We thank Dr. Sho Ohno (Kyoto University, Japan) for kindly providing us with the protocol for protein extraction and immunoblotting analysis. We are grateful to Mrs. Saori Kawaguchi for technical assistance with laboratory experiments.

Disclosures

The authors have no conflicts of interest to declare.

References