The Tibetan turnip KTRG-B49 response to freezing tolerance
We used the turnip KTRG-B49 variety from the Qinghai-Tibet Plateau to assess the freezing tolerance of turnip. As expected, compared with the control seedlings, most of the treated seedlings survived (Fig. 1A). With the extension of freezing stress time, the electrolyte leakage of the turnip cells increased gradually (Fig. 1B). When the plants were grown at -4 ℃ for 2 h, they displayed more serious damage than plants grown under standard conditions. After recovery growth for 12 h in an ambient environment, the plants were restored and had an electrolyte leakage of 23.9 %, but there was still a high survival rate of 78.9 % (Fig. 1C). These data indicated that KTRG-B49 had the reproductive characteristics of turnip on the Qinghai-Tibet Plateau and can respond to low temperatures.
Transcriptome changes in KTRG-B49 during freezing treatment
KTRG-B49 plants that underwent freezing treatment were used for RNA-seq to determine the genes responsible for tolerance. Totals of 144.16 Gb clean data were obtained for 21 KTRG-B49 individuals, with more than 6.18 Gb for each sample. The sequencing quality scores of 30 (Q30) of the samples were greater than 89.03 %, indicating the high quality of the reads (Table S2). Here, 83.27 %-89.72 % of the clean reads were mapped to the turnip reference genome using TopHat software with an average mapping rate of 86.56 % (Table S3). Therefore, more than half of the turnip reference genes were expressed in each sample, which was sufficient for further differential gene expression analysis. A correlation analysis of the expression levels among the samples (Fig. S2) and the FPKM distribution for all samples (Fig. S3) showed that the repeatability of the expression level of each sample was good. In addition, the results of principal component analysis (PCA) based on the 21 samples showed that the first component explained 38.9 % of the variation, and the second component explained 36 % of the variation (Fig. 2A). KTRG-B49 could be clearly separated five distinct clusters along the first component axis. This finding indicated a close relationship between CK and C1 and between F2 and F4, which may have similar expression patterns. The differential expression analysis among different treatments can be inferred based on PCA.
The polyamine synthesis pathway is associated with the freezing responses of KTRG-B49
To investigate the gene expression levels, we calculated the FPKM values using the reads from RNA-seq. The number of differentially expressed genes (DEGs) and functional annotation information are shown in Fig. S4 and Table S4, respectively. In the analysis, the DEGs were statistically grouped into six subclusters based on their expression patterns in turnip at different treatment stages (Fig. S5). Interestingly, 347 genes in subcluster_1 demonstrated a similar trend in the PCA of the transcriptomic landscape, which deserves further study (Fig. 2B). Accordingly, KEGG enrichment analysis of the DEGs in subcluster_1 was conducted to identify pathways that played important roles in the freezing process. The top five pathways for the upregulated DEGs are displayed, and major pathways related to circadian rhythm-plant, plant-pathogen interaction, polyamine biosynthesis, and arginine (Fig. 2C), which could indicate their involvement in turnip freezing tolerance. Here, based on these results and the role of polyamine in freezing stress, we focused on the pathway of polyamine synthesis (Fig. 2D). The expression levels of the DEGs in the polyamine pathway were verified by qRT-PCR (Fig. 2E). Of these genes, BrrADC1, BrrADC2.1, BrrADC2.2, and BrrAIH1.1 accumulated continuously under freezing stress, showing a similar trend to subcluster_1. These DEGs related to polyamine synthesis may play a special role in enhancing the freezing resistance in turnip.
Putrescine was specifically enriched in KTRG-B49 upon freezing treatment
To confirm that the polyamine pathway was associated with turnip freezing tolerance, we analyzed polyamine accumulation in KTRG-B49. The polyamine profiles were assessed with a focus on agmatine, putrescine, cadaverine, and spermidine. As shown in Fig. 3A, the contents of agmatine and spermidine remained almost unchanged during the freezing treatment when compared with those of the CK without freezing stress group. However, freezing treatment significantly increased the levels of putrescine, and its contents were especially high (25.34-32.27 nmol g-1 FW) compared with those of the CK group (24.98 nmol g-1 FW). Furthermore, the cadaverine levels were also significantly different from those of the CK group in the 12 h recovery stage after freezing treatment, although the contents were relatively low. Therefore, these specifically increased metabolites upon freezing treatment, especially putrescine, may be related to the freezing tolerance of KTRG-B49.
Gene correlation network analysis
First, the expression patterns of 3886 DEGs (FPKM ≥ 1) obtained from transcriptome sequencing were analyzed by WGCNA, and they were divided into three modules according to the similarity of their expression patterns (Fig. 3B). The numbers of unique genes in each module were 2378, 571, and 937 for the black, green, and turquoise colors, respectively. The gene expression profile of the entire module was obtained (Fig. S1). Then, the correlations between the modules and the traits were analyzed, and the data were shown in Fig. 3C. One module of interest was the turquoise module; the correlation coefficient of that module with putrescine was the highest (0.78, P-value = 0.00003), indicating that the module ‘Turquoise’ was closely related to putrescine accumulation in turnip. To further identify ‘Turquoise’ modular features, we used KEGG pathway analysis for deep analysis (Fig. S6). Six genes were included in the METurqoise module and were found to be involved in the ‘arginine and proline metabolism’ pathway (ko0330, M00133). Among these genes, BrrADC2.2 (GenBank number: MN630676) was found to participate in polyamine synthesis, and the increase in BrrADC2.2 expression was consistent with the accumulation of putrescine under freezing treatment. Taken together, these results suggested that BrrADC2.2 may play a key role in polyamine synthesis under freezing stress in turnip, and the upregulation of BrrADC2.2 expression and the increase in putrescine could be essential for freezing tolerance of turnip.
Identification of upstream regulatory transcription factors of BrrADC2.2
We performed yeast one-hybrid assays to screen the transcription factors interacting with BrrADC2.2 promoter as bait and the turnip cDNA library as prey, and positive colonies were partially sequenced and identified by BLAST analysis. Finally, we identified a cDNA of a B. rapa ICE1-like transcription factor and named it BrrICE1.1 (GenBank number: MN630673) in our turnip. BrrICE1.1 has the entire set of signature motifs required for defining a typical bHLH transcription factor (Fig. S7). In addition, ICE1 was reported to be able to bind specifically to the MYC recognition sequence (CANNTG). Thus, we further analyzed the promoter region of BrrADC2.2 using the PlantCARE database (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/). As expected, there was a potential MYC recognition element (CATTTG) (958-963) within the 1-kb region upstream of the ATG (Fig. 4A, Table S5). Therefore, we speculated that BrrICE1.1 combined with the CATTTG element in the promoter region of BrrADC2.2 plays a role in freezing tolerance. Moreover, we analyzed the promoter sequence of other DEGs in the polyamine pathway, and an additional six DEGs (BrrADC1, GenBank number: MN630674; BrrADC2.1, GenBank number: MN630675; BrrAIH1.1, GenBank number: MN630677; BrrAIH1.2, GenBank number: MN630678；BrrSAMDC1.1, GenBank number: MN630679；BrrSPD2.1, GenBank number: MN630681) were found to contain the MYC recognition element (Table S5).
BrrICE1.1 bound to the promoters of polyamine pathway DEGs in vitro and vivo
Possible interactions of BrrICE1.1 with the MYC elements of the above gene promoters were tested using a yeast one-hybrid system. We found that BrrICE1.1 could activate BrrADC2.1pro::pAbAi, BrrADC2.2pro::pAbAi, BrrAIH1.1pro::pAbAi, and BrrAIH1.2pro::pAbAi, but not their mutant, suggesting that the MYC elements, CATTTG sites for BrrADC2.1 and BrrADC2.2, and CAATTG sites for the BrrAIH1.1 and BrrAIH1.2 promoters were necessary and required for BrrICE1.1 binding in yeast (Fig. 4B).
Next, using the well-established transient expression assay of N. benthamiana leaves, we further verified the transcriptional activation activity of BrrICE1 with above four genes (Fig. 4C). Coexpression of 35S::BrrICE1.1 with BrrADC2.1pro::LUC,BrrADC2.2pro::LUC, BrrAIH1.1pro::LUC, and BrrAIH1.2pro::LUC could detect LUC activity, indicating that BrrICE1.1 activated their expression, respectively. Furthermore, ChIP experiments using BrrICE1.1-6Flag transgenic hair roots and an anti-Flag antibody were employed to confirm the binding of BrrICE1.1 to these four gene promoters in vivo. qPCR revealed that only the BrrICE1.1-6Flag protein could immunoprecipitate the BrrADC2.2 promoter region containing an element of CATTTG (compared to the IgG and WT) (Fig. 4D). However, BrrADC2.1, BrrADC2.2, BrrAIH1.1 and BrrAIH1.2 were not detected. Together, these data suggested that BrrICE1.1 could directly bind to the promoter of BrrADC2.2 and that the CATTTG elements presented in the BrrADC2.2 promoter were the binding sites for BrrICE1.1.
Transgenic and RNAi hair roots showed altered levels of BrrADC2.2 transcripts and polyamines
The effect of the BrrADC2.2 gene on putrescine synthesis was further verified by using A. rhizogenes-mediated infection of the cotyledons of turnip to obtain BrrADC2.2 transgenic and RNAi hairy roots (Fig. 5A). Fluorescence of the hairy roots overexpressing BrrADC2.2-OE was obtained by confocal laser microscopy (Fig. 5B). Moreover, we tested the expression levels of BrrADC2.2 in BrrADC2.2-OE and BrrADC2.2-RNAi using qRT-PCR analysis (Fig. 5C). The expression of BrrADC2.2 was significantly up- and down-regulated in BrrADC2.2-OE and BrrADC2.2-RNAi hair roots, respectively, compared with the control, indicating that the BrrADC2.2 gene was successfully expressed in the roots. Additionally, the putrescine content of transgenic BrrADC2.2-OE measured by LC-MS was significantly higher than that of the control. As expected, BrrADC2.2-RNAi reduced the putrescine content (Fig. 5D). Both the BrrADC2.2 expression levels and putrescine contents indicated that BrrADC2.2 displayed a crucial role in putrescine synthesis.
BrrICE1.1 directly regulated BrrADC2.2 in putrescine synthesis
To identify the role of BrrICE1.1 in putrescine synthesis, we obtained transgenic and RNAi hairy roots (Fig. 5A). The BrrICE1.1-GFP fusion protein emitted a green fluorescent signal in the membranes and nuclei of the hair roots (Fig. 5B). Interestingly, cell membrane localization of BrrICE1.1 was present in turnip but absent in Arabidopsis (Fig. 5E, left). To explore the membrane localization of turnip BrrICE1.1, we analyzed the amino acid sequence characteristics of BrrICE1.1 and AtICE1 (Fig. 5F). We found that the C-terminus was conserved, while at the N-terminus, BrrICE1.1 had an extra amino acid sequence (GGGGERV). We speculated that this amino acid sequence affected the location of BrrICE1.1. Hence, we deleted it to test the location of BrrICE1.1 and we were surprised to find that its membrane localization disappeared and only nuclear localization was observed in N. benthamiana leaves (Fig. 5E, right). Therefore, this extra amino acid sequence was indeed the reason for the membrane localization of BrrICE1.1.
Efforts were further made to compare the expression levels and putrescine contents among transgenic, RNAi and WT hairy roots. The expression level of BrrICE1.1 was slightly upregulated and BrrADC2.2 was significantly upregulated in transgenic BrrICE1.1-OE hairy roots compared to WT roots, while the expression levels in BrrICE1.1-RNAi hairy roots were lower than those in the WT (Fig. 5G). With respect to putrescine contents, LC-MS measurement showed that transgenic lines increased the levels of putrescine, in contrast to a reduction in the RNAi hairy roots (Fig. 5H). The expression level analysis together with the putrescine content analysis presented here may indicate that BrrICE1.1 positively regulates the expression of BrrADC2.2 in freezing stress of turnip.
Effect of exogenous application of putrescine and its inhibitor on the freezing tolerance of turnip
To confirm the role of putrescine in freezing tolerance in turnip, exogenous putrescine and inhibitor (DFMA) were applied to the KTRG-B49 under freezing conditions, respectively (Fig. 6). There were no obvious changes in plant morphology between putrescine-treated, inhibitor-treated and control plants before application of the freezing treatments. When submitted to freezing treatment, the putrescine-treated plants suffered significant less injury compared with the control, whereas the inhibitor-treated plants the most severely injured (Fig. 6A). To further confirm the results obtained above, electrolyte leakage was used to measure the plants after freezing treatment. Consistently, electrolyte leakage of putrescine-treated leaves exhibited a significantly lower electrolyte leakage after freezing treatment (Fig. 6B). These results suggested that putrescine could play an important role in turnip freezing tolerance.
The CBF signaling pathway was enhanced by freezing stress in KTRG-B49
CBF represents a well-documented pathway involved in plant cold responses. To clarify the possible signal transduction, we monitored the expression levels of BrrICE1.1, BrrCBF3 and the downstream BrrCOR15A genes in turnip under freezing stress. We observed an elevation of all three genes, although significance was not observed in all periods of the freezing treatment (Fig. 7). Next, we asked whether BrrICE1.1 regulated BrrCBF3 using in vivo and in vitro methods. We first analyzed the promoter region of BrrCBF3 and found that it contained the MYC recognition element (CATTTG) in a 1623 region (Fig. 8A), and thus, we confirmed that BrrICE1.1 could bind to BrrCBF3. Indeed, Y1H assays revealed that BrrICE1.1 could activate BrrCBF3pro::pAbAi, suggesting that the BrrICE1.1 binding sites may be CATTTG. The activation of BrrICE1.1 on the reporters was completely abolished when CATTTG was deleted (mBrrCBF3pro), suggesting that the CATTTG element in the BrrCBF3 promoter was necessary and required for BrrICE1.1 binding in yeast (Fig. 8B). We thus performed LUC assays to determine the effect of BrrICE1.1 on the expression of a reporter containing the BrrCBF3 promoter fused with the LUC reporter gene. We found that coexpression of 35S::BrrICE1.1 with the BrrCBF3pro::LUC reporter led to an obvious increase in LUC activity, indicating that BrrICE1.1 activated the expression of BrrCBF3 (Fig. 8C). To confirm this effect, we used ChIP-qPCR analysis. An IgG antibody and WT were used as controls, and two different sets of primers, CBF3 promoter primers containing CATTTG elements (BrrCBF3) and CBF3 gene background primers (BrrCBF3-GD), were used to test the BrrICE1.1-6Flag immunoprecipitated complexes. Notably, BrrCBF3 promoter primer enrichment was significantly increased in BrrICE1.1-6Flag immunoprecipitated complexes (Fig. 8D). This finding implied that BrrICE1.1 directly regulated BrrCBF3. To examine the function of BrrCBF3 and BrrICE1.1, we generated BrrICE1.1 overexpression and RNAi hair roots. Overexpression of BrrICE1.1 led to enhanced induction of the BrrCBF3 gene, whereas the expression level of BrrCBF3 was downregulated in the BrrICE1.1-RNAi hair roots (Fig. 8E). These investigations suggested that the BrrICE1.1 signal pathway could be associated with CBF-mediated freezing tolerance in turnip.