Metabolome analysis of different colored bell peppers
Three pepper inbred lines were used in the study, IL-Y, IL-O and IL-R (Fig. 1A), and fruit color of the three lines was green at 50 days after planting (50 DAP), the corresponding samples named YG, OG and RG. The color transformed to yellow, orange and red at 65 DAP, and the samples named YM, OM and RM, respectively. The results from metabolome analysis on samples YG, OG, RG, YM, OM and RM showed that 68 carotenoid compounds were detected, including 61 lutein (yellow pigment) and 7 carotenes (orange pigment), and among the examined carotenoids, 54 were identified in at least one sample (Table S1).
At 50 DAP, the carotenoids accumulated in the green fruits of IL-Y (YG), IL-O (OG) and IL-R (RG) were mainly lutein, β-carotene and neoxanthin, but other compounds were extremely low (Fig. 1B, C and D). With the ripening of fruits, total carotenoid contents in mature fruits of IL-Y (YM), IL-O (OM) and IL-R (RM) were 571.29 µg/g, 2098.27 µg/g and 575.62 µg/g, and it was 6.06, 1.38 and 1.73 times higher than that in YG, OG and RG, respectively. The species were also increased from 20 (YG), 18 (OG) and 17 (RG) to 39 (YM), 49 (OM) and 27 (RM) (Table S2). In YM, the main carotenoids were lutein (31.36%), (E/Z) phytoene (27.89%), violaxanthin (10.96%), violaxanthin myristate (4.55%) and zeaxanthin (3.42%) (Fig. 1E). While in OM, zeaxanthin (37.93%) was the highest, followed by (E/Z) phytoene (12.69%), zeaxanthin dimyristate (11.97%), lutein (8.02%) and lutein dimyristate (4.42%) (Fig. 1F). The accumulation of (E/Z) phytoene were both high in YM and OM, but in RM, red capsanthin was the main component of carotenoids, reaching 71.02%, which is different from YM and OM (Fig. 1G). Other carotenoids compounds varied from 0 to 5.9%. In addition, the content of lutein in RM was extremely low (0.43 µg/g), which was also different from the high content of lutein in YM and OM (Fig. 1E, F and G, Table S2).
The metabolites heatmap showed significant differences among green fruits samples YG, OG and RG, and mature fruits of YM, OM and RM (Fig. 2A). Under the selection of Fold_Change (FC) ≥ 2 or ≤ 0.5, differential metabolites of three bell pepper varieties at the green ripening stage were less, ranging from 5 to 8 (Fig. 2B). At the mature stage, OM showed a higher accumulation of carotenoids compared to YM and RM. Then the differentially accumulated carotenoids (DACs) among samples were also screened (Table S3). In YM_vs_OM group, 40 DACs were identified, including 33 up-regulated and 7 down-regulated. Among the DACs, zeaxanthin and zeaxanthin-related compounds were significantly up-regulated, and the same as to orange carotenes, α-carotene and β-carotene, while violaxanthin and violaxanthin-related compounds were significantly down-regulated. Similarly in RM_vs_OM group (44 up-regulated and 3 down-regulated), zeaxanthin, zeaxanthin-related compounds, lutein-related compounds, and carotenes were also highly up-regulated, while the red pigments capsanthin and capsorubin were down-regulated. In YM_vs_RM comparison group, 41 DACs were identified, among which capsanthin and capsorubin were significantly up-regulated, while lutein and related compounds were conversely down-regulated. A total of 47, 29 and 40 DACs were identified between green and mature stages for groups OG_vs_OM, RG_vs_RM and YG_vs_YM, respectively (Fig. 2B, Table S3). Venn diagram showed 24 DACs common to the three comparison groups (Fig. 2C), detail information of DACs was showed in Table S3. The above results indicated that the unique accumulation pattern of carotenoid-related metabolites was the main reason for the color difference in bell pepper fruits.
RNA-seq detection and analysis of different colored bell peppers
To understand the molecular regulatory network on fruit color formation of yellow, orange and red bell pepper, we performed transcriptome sequencing analysis on YG, YM, OG, OM, RG and RM with three biological replicates per sample (-1, -2, and − 3). Data analysis showed 124.82 Gb of total clean data was obtained, and clean reads per sample ranged from 41358224 to 52919370 (Table 1). The Q20 of all samples was above 97.3%, and Q30 was above 92.37%, and mean GC content was 42.53% (Table 1). The principal component analysis (PCA) score plot showed PC1, PC2 and PC3 accounted for 40.38%, 11.96% and 9.18% of the variance, respectively. Samples YG, YM, OG, OM, RG and RM were clearly separated in the analysis chart, and their three replicates were closely clustered (Fig. 3A).
Table 1
Transcriptome sequencing data quality index.
Sample | Raw Reads | Clean Reads | Clean Base(G) | Q20(%) | Q30(%) | GC Content(%) |
OG-1 | 46530998 | 44655450 | 6.7 | 97.62 | 93.15 | 42.65 |
OG-2 | 43121018 | 41358224 | 6.2 | 97.59 | 93.08 | 42.71 |
OG-3 | 47223748 | 45014448 | 6.75 | 97.53 | 92.97 | 42.55 |
OM-1 | 49292006 | 46178988 | 6.93 | 97.72 | 93.43 | 42.57 |
OM-2 | 48808226 | 45865292 | 6.88 | 97.61 | 93.2 | 42.47 |
OM-3 | 53698840 | 52919370 | 7.94 | 97.86 | 93.54 | 42.56 |
RG-1 | 48982534 | 46584348 | 6.99 | 97.3 | 92.37 | 42.73 |
RG-2 | 47078894 | 44053318 | 6.61 | 97.76 | 93.5 | 42.57 |
RG-3 | 47272994 | 45078648 | 6.76 | 97.64 | 93.22 | 42.9 |
RM-1 | 47140830 | 45427904 | 6.81 | 97.32 | 92.43 | 42.28 |
RM-2 | 49236310 | 46708260 | 7.01 | 97.54 | 92.96 | 42.46 |
RM-3 | 48691662 | 46481656 | 6.97 | 97.4 | 92.6 | 42.35 |
YG-1 | 51603776 | 49636440 | 7.45 | 97.41 | 92.65 | 42.66 |
YG-2 | 47528914 | 45475804 | 6.82 | 97.44 | 92.76 | 42.71 |
YG-3 | 52411018 | 50004026 | 7.5 | 97.58 | 93.05 | 42.59 |
YM-1 | 47862660 | 45657656 | 6.85 | 97.53 | 92.99 | 42.33 |
YM-2 | 45830776 | 43821608 | 6.57 | 97.39 | 92.65 | 42.23 |
YM-3 | 49055178 | 47171162 | 7.08 | 97.44 | 92.8 | 42.15 |
Under the thresholds of log2 (FC) ≥ 1 and FDR ≤ 0.05, 14961 differentially expressed genes (DEGs) were screened out in three bell pepper lines between green and mature stages. DEGs distribution showed in volcano plots suggested significant differences in samples YG, YM, OG, OM, RG and RM (Fig. S1). At the green ripening stage, we identified 1854, 2467 and 3059 DEGs in YG_vs_OG, YG_vs_RG and RG_vs_OG, respectively. Among them, 1099, 1452 and 1435 genes were up-regulated, while 755, 1015 and 1624 genes were conversely down-regulated. (Fig. 3B). In total, 240 common DEGs were identified at green ripening stages of yellow, orange and red fruits (Fig. 3C). At the mature stage, 2468 (1034 up- and 1434 down-regulated), 3091 (1397 up- and 1694 down-regulated) and 2930 (1500 up- and 1430 down-regulated) DEGs were screened in comparison groups RM_vs_OM, YM_vs_OM and YM_vs_RM, respectively. Venn diagrams showed that 273 DEGs were common, indicating that these DEGs may be responsible for the key differences in fruit color among the three bell peppers at mature stage(Fig. 3B and C). For two different stages of the same cultivar, we detected 3205 up- and 5923 down-regulated DEGs in OG_vs_OM, 3479 up- and 6474 down- in RG_vs_RM, and 3166 up- and 5998 down- in YG_vs_YM. Venn diagram showed 5735 common DEGs detected in all comparison groups, indicating that these DEGs might be important reasons for the color change of three bell peppers lines (Fig. 3B and C).
DEGs functional prediction
DEGs functional prediction of GO and KEGG enrichment analysis was performed to elucidate their roles in fruit color change of bell pepper (Fig. S2 and S3). All the DEGs were enriched in molecular functions, cellular components and biological processes. KEGG enrichment analysis showed the highest abundance pathway was metabolic (ko01100) and secondary metabolites biosynthesis (ko01110). Most importantly, pathway of carotenoid biosynthesis (ko00906) was significantly enriched in groups YG_vs_YM, OG_vs_OM, RG_vs_RM, RG_vs_OG, YG_vs_RG, RM_vs_OM and YM_vs_RM (Fig. S3), providing a powerful clue on genes expression in fruit coloration of bell pepper. The DEGs of group YG_vs_OG was enriched in biosynthesis pathways of phenylpropanoid (ko00940), anthocyanin (ko00942) and flavonoid (ko00941). For group RG_vs_OG, DEGs was enriched in biosynthesis of phenylpropanoid (ko00940) and anthocyanin (ko00942), and DEGs of YG_vs_RG was also significantly enriched in anthocyanin biosynthesis pathway (ko00942). The results are consistent with previous study that flavonoids are mainly synthesized in the early stage of fruit44, suggesting DEGs involved in these three pathways may be the reason for the early discoloration in bell pepper.
DEGs related to carotenoid biosynthesis
A total of 48 DEGs were screened to be related to carotenoid metabolism and assigned to pathway ko00906 (Table S4). After excluding 25 uncertain genes and low-expression genes, 23 DEGs were detected in at least one comparison group. Based on gene functional annotation, homology of these 23 genes was identified, which included 2 PSY (gene-LOC107859651, gene-LOC107868281), 1 PDS (gene-LOC107861625), 1 Z-ISO (gene-LOC107850257), 1 ZDS (gene-LOC107839468), 2 LCY-E (gene-LOC107840923, gene-LOC107852092), 1 LCYB (gene-LOC107869983), 2 BCH (gene-LOC107873401, gene-LOC107863219), 1 ZEP (gene-LOC107872926), 1 CRTISO (gene-LOC107854534), 1 CYP97 (gene-LOC107850957), 1 CCS (gene-LOC107875664), 1 NCEDs (gene-LOC107852027), 1 CCDs (gene-LOC107870081), 3 SDRS (gene-LOC107843702, gene-LOC107862477, gene-LOC107855067), 2 AAO (gene-LOC107847514, gene-LOC107847367) and 2 D27s (gene-LOC107863814, gene-LOC107875230).
In plants, phytoene is the first product of carotenoids biosynthesis pathway, which is formed by PSY catalyzing condensation of two GGPP molecules25. In our study, compared with the green ripening stage, expression of PSY1 (gene-loc107868281) was significantly increased in three bell pepper lines at coloring stage, which was corresponding to the higher content of phytoene in OM, RM and YM. The CCS gene (gene-LOC107875664) was only expressed in RM, consistent with capsaicin and capsorubin were detected only in RM. The expression level of BCH1 gene (gene-LOC107863219) increased most significantly in OG_vs_OM, RG_vs_RM, YG_vs_YM comparison groups, while the expression of BCH2 gene (gene-LOC107873401) decreased significantly, showing different expression patterns. Compared with the green ripening stage, other key genes including PDS (gene-LOC107861625), ZDS (gene-LOC107839468), Z-ISO (gene-LOC107850257), LCYB (gene-LOC107869983) and ZEP (gene-LOC107872926) also increased significantly. However, cyclase genes LCYE (gene-Loc107852092, gene-LOC107840923) and CYP97 (gene-Loc107850957) involved in α-branch synthesis decreased during maturation. In addition, downstream products of carotenoids produces precursors for abscisic acid (ABA) and strigolactone (SL)14–15. One NCED, three SDR, two AAO (related to ABA synthesis) and two D27 (related to SLs synthesis) were significantly differentially expressed in bell pepper lines at different periods (Table S4).
The direct carotenoids precursor, GGPP, is synthesized under the catalysis of GGPPS22–24. Two GGPPS (gene-LOC107843186, gene-LOC107867046) with high expression in YM, OM and RM were identified in our study, providing sufficient precursor compounds for carotenoid synthesis, which may be one of the reasons for the high carotenoid content of the three bell pepper lines at mature stage (Table S4).
DEGs analysis related to carbon and lipid metabolism
Previous studies showed positive correlation between sugar and pigment accumulation45–47. We found DEGs in groups YG_vs_YM, OG_vs_OM and RG_vs_RM were significantly enriched in carbon metabolism (ko01200) (Fig. S3). At least 80 DEGs were detected in one comparison group (Table S5). In groups YG_vs_YM, OG_vs_OM and RG_vs_RM, 32, 54 and 45 DEGs were up-regulated, while 17, 12 and 11 DEGs were down-regulated, respectively (Table S5).
In plant cells, carotenoids usually accumulate in chromoplasts, a kind of specialized storage organelles (plastids)48. In chromoplast differentiation, internal membrane remodeling is a prominent feature, and lipid substances synthesis plays an important role during the process49. In the pathway of fatty acid metabolism (ko01212), 44, 52 and 56 DEGs were detected in groups YG_vs_YM, OG_vs_OM and RG_vs_RM, respectively. Among them, 17, 22 and 25 DEGs were up-regulated, and 27, 30 and 31 were down-regulated, respectively (Table S6). In the fatty acid degradation pathway (ko00071), 17, 15, 21 DEGs were up-regulated, while 19, 15 and 22 were down-regulated in YG_vs_YM, OG_vs_OM and RG_vs_RM, respectively (Table S6). Besides, in the fatty acid biosynthesis pathway (ko00061), 7, 17 and 13 genes of YG_vs_YM, OG_vs_OM and RG_vs_RM were up-regulated, while 12, 12 and 13 genes were down-regulated (Table S6).
DEGs co-expression analysis
We performed weighted gene co-expression network analysis (WGCNA) using DEGs identified, and obtained 13 main branches (Fig. 4A, Table S7). Modules blue, red and magenta had a positive correlation with carotenoid metabolites, while turquoise had a negative correlation. The greenyellow module had the highest correlation with capsaicin and capsorubin, which was characteristic in RM, indicating genes in this module play more important roles in red pigment formation (Fig. 4B). Expression heat map of co-expressed genes of blue module in OM, YM and RM were much higher than OG, YG and RG, while genes of turquoise module was just the opposite, indicating that genes of the two modules played opposite roles in carotenoid synthesis (Fig. 4C). DEGs in yellow module was highly correlated with YM, while genes of magenta and greenyellow module were only highly expressed in OM and RM, showing obvious variety specificity (Fig. 4C and D). These results indicated DEGs in different modules may play different roles in carotenoid metabolites synthesis, and subsequent analysis also focuses on these six modules.
Transcription factors (TFs) prediction for carotenoid metabolism regulation
Transcription factors (TFs) are also key factors in plant carotenoid biosynthesis. MYB, MAD-box, bHLH, WRKY, NAC and NY-F play important roles in regulating transcription of carotenoid metabolism related genes40–42. In blue module, 210 TFs were identified, including 16 MYB, 10 NAC, 4 WRKY, 8 MAD, 2 BHLH, 11 AP2/ERF, 9 NF-Y, and 3 HD-ZIP (Table S8). Expression abundance of 4 MYB (gene-LOC107850892, gene-LOC107867188, gene-LOC107868015 and gene-LOC107871103), 3 HD-ZIP (gene-LOC107843791, gene-LOC107862495 and gene-LOC107863056), 4 NAC (gene-LOC107842449, gene-LOC107845296, gene-LOC107847606 and gene-LOC107867776), 4 AP2/ERF (gene-LOC107855040, gene-LOC107862115, gene-LOC107867626 and gene-LOC107870958), 3 MAD (gene-LOC107847473, gene-LOC107855404 and gene-LOC107878477), 3 WRKY (gene-LOC107840352, gene-LOC107859607 and gene-LOC107872867) and 1 NF-Y (gene-LOC107861703) was higher in the colored mature fruit, indicating they may contribute more in fruit color transformation of bell pepper (Table S8). Notably, one MAD family member (gene-LOC107847473) was almost not expressed at the green ripening stage, but showed high expression at the mature stage with the fold change ranging from 9.18 to 15.38, indicating its pivotal role in regulating carotenoids biosynthesis in bell pepper. In the turquoise module, 269 TFs genes were identified including 25 AP2/ERF, 20 bHLH, 8 MAD, 12 MYB, 4 NAC, 8 HD-ZIP, 6 WRKY and 4 NF-Y, and detected HD-ZIP and WRKY family members were identified as negative regulators associated with carotenoid metabolism (Table S8). In the yellow module, 56 TFs genes were identified, and expression of 3 bHLH (gene-LOC107838913, gene-LOC107840824 and gene-LOC107862125) and 1 AP2/ERF (gene-LOC107864060) in YM were much higher than other samples (Table S8). Besides, 15, 4 and 4 TFs were predicted in red, magenta and greenyellow modules, respectively. Interestingly, one GRAS (gene-LOC107853465) was highly expressed only in greenyellow module, but no GRAS family members have been reported in carotenoids synthesis regulation, so their functions need to be further studied (Table S8).
Candidate hub genes for carotenoid synthesis in different colored bell pepper
Correlation networks were constructed between genes within six key modules. In blue module, a total of 333 DEGs showed high co-expression with edge weight over 0.5, including 8 TFs, 6 carotenoid synthesis genes, 13 carbon metabolism pathway genes, and 13 genes involved in fatty acid synthesis and metabolism (Table S9), and mitochondrial carrier protein (gene-LOC107839282) and Zeta-carotene desaturase (ZDS, gene-LOC107839468) were hub genes (highly connected genes) in this module (Fig. 5A), and mitochondrial carrier protein (gene-LOC107839282) had high correlation with carotenoid synthesis genes ZDS (gene-LOC107839468), PDS (gene-LOC107861625), BCH (gene-LOC107863219), and GGPS (gene-LOC107843186), and TFs MAD (gene-LOC107879769), NAC (gene-LOC107842449 and gene-LOC107850240), indicating its key function in carotenoids synthesis. The turquoise module contained 284 co-expressed genes, including 22 TFs, 1 carotenoid decomposition gene, 3 carbon metabolism pathway genes, and 2 genes related to fatty acid synthesis and metabolism (Table S9). Hypothetical protein (gene-LOC107840002) and Ras-related protein (gene-LOC107839339) had the highest connectivity (Fig. 5B), and were both highly correlated with the carotenoid-breakdown gene CCD (gene-LOC107870081), which was negatively related to carotenoid metabolism. The yellow module contained 79 co-expressed genes, and gene-LOC107839932, putative L-ascorbate peroxidase (gene-LOC107838875) and excision repair protein (gene-LOC107870806) had the highest connectivity (Fig. 5C, Table S9). In magenta module, genes novel.3368, Chaperonin-like RBCX protein (gene-LOC107861269) and isocitrate dehydrogenase (gene-LOC107854067) had the highest connectivity (Fig. 5D, Table S9). In the red module envoloved of 63 genes, of which obg-like ATPase 1 (gene-LOC107853317) and methionine-tRNA ligase (gene-LOC107873333) showed the highest connectivity (Fig. 5E, Table S9). There were 52 genes in greenyellow module, and novel.3256 and gene-LOC107869054 were the hub genes (Fig. 5F, Table S9).
Metabolite and gene associations underlying color formation in bell pepper fruit
To further analyze the relationship between metabolites and related genes in carotenoid biosynthesis pathway during yellow, orange and red pepper coloration, we have established a network based on key metabolites and related genes (Fig. 6). In different pepper fruit samples undergoing color changes (YG, YM, OG, OM, RG and RM), 7 key carotenoid metabolites and 14 key biosynthetic genes were involved in the regulation. The 14 key genes are: two GGPPS genes (gene-LOC107843186 and gene-LOC107867046), PSY1 (gene-LOC107868281) and PSY2 (gene-LOC107859651), PDS (gene-LOC107861625), Z-ISO (gene-LOC107850257), ZDS (gene-LOC107839468), CRTISO (gene-LOC107854534), LCYB (gene-LOC107869983), BCH1 (gene-LOC107863219) and BCH2 (gene-LOC107873401), LCYE (gene-LOC107840923 and gene-LOC107852092), CCS (gene-LOC107875664). Among them, GGPPS, PSY1, PDS, Z-ISO, ZDS, LCYB, BCH1, and LCYE were highly expressed at mature stage (65 DAP), positively regulating pigment accumulation. In contrast, PSY2, CRTISO, and BCH2 showed higher expression in the green ripening fruit stage (50 DAP), which may be related to pigment accumulation in the green fruit.
The 7 key carotenoid metabolites are: α-carotene, β-carotene, β-cryptoxanthin, zeaxanthin, lutein, capsanthin, and capsorubin. In YM and OM samples, α-carotene content was higher, related to high expression of LCYB gene. Genes BCH2 and BCH1 promoted lutein accumulation at green ripening and mature stages, respectively, laying the foundation for pepper fruit color changes. The significantly higher accumulation of zeaxanthin in OM samples was a key factor in color transition from green to orange. Furthermore, the CCS gene was only expressed in RM samples, and it determined capsanthin and capsorubin accumulatio, leading to red color in mature RM. The distinct accumulation patterns of different metabolites are closely related to specific DEGs during fruit development, which collectively regulate the color changes in pepper fruits.
qRT-PCR analysis
In order to confirm the accuracy of RNA-Seq data, we performed qRT-PCR experiments on the carotenoid biosynthesis pathway genes PSY1, PSY2, PDS, Z-ISO, ZDS, CRTISO, LCYB, LCY-E, CYP97, BCH, and CCS (Fig. 7). The expression levels of 12 genes were basically consistent with FPKM results, which supports the reliability of RNA-seq data in our study.