Transcriptomic profiling of purple broccoli reveals light-induced anthocyanin biosynthetic signaling and structural genes

Purple Broccoli (Brassica oleracea L. var italica) attracts growing attention as a functional food. Its purple coloration is due to high anthocyanin amounts. Light represents a critical parameter affecting anthocyanins biosynthesis. In this study, ‘Purple Broccoli’, a light-responding pigmentation cultivar, was assessed for exploring the mechanism underlying light-induced anthocyanin biosynthesis by RNA-Seq. Cyanidin, delphinidin and malvidin derivatives were detected in broccoli head samples. Shading assays and RNA-seq analysis identified the flower head as more critical organ compared with leaves. Anthocyanin levels were assessed at 0, 7 and 11 days, respectively, with further valuation by RNA-seq under head-shading and light conditions. RNA sequences were de novo assembled into 50,329 unigenes, of which 38,701 were annotated against four public protein databases. Cluster analysis demonstrated that anthocyanin/phenylpropanoid biosynthesis, photosynthesis, and flavonoid biosynthesis in cluster 8 were the main metabolic pathways regulated by light and had showed associations with flower head growth. A total of 2,400 unigenes showed differential expression between the light and head-shading groups in cluster 8, including 650 co-expressed, 373 specifically expressed under shading conditions and 1,377 specifically expressed under normal light. Digital gene expression (DGE) analysis demonstrated that light perception and the signal transducers CRY3 and HY5 may control anthocyanin accumulation. Following shading, 15 structural genes involved in anthocyanin biosynthesis were downregulated, including PAL, C4H, 4CL, CHS, CHI, F3H and DFR. Moreover, six BoMYB genes (BoMYB6-1, BoMYB6-2, BoMYB6-3, BoMYB6-4, BoMYBL2-1 and BoMYBL2-2) and three BobHLH genes (BoTT8_5-1, BoTT8_5-2 and BoEGL5-3) were critical transcription factors controlling anthocyanin accumulation under light conditions. Based on these data, a light-associated anthocyanin biosynthesis pathway in Broccoli was proposed. This information could help improve broccoli properties, providing novel insights into the molecular mechanisms underpinning light-associated anthocyanin production in purple vegetables.

Multiple parameters including genetic, developmental and environmental factors control anthocyanin biosynthesis. Light indexes, including intensity and quality, represent critical factors affecting anthocyanin accumulation (Albert et al., 2009). In lettuce and turnip, UV-A and UV-B increase anthocyanin contents via upregulation of DFR and CHS in the anthocyanin biosynthetic pathway (Zhou et al., 2007;Park et al., 2007). In the presence of light, photoreceptors are activated, including PHYs (PHYA to E) that absorb red/far-red light; CRYs (CRY1 to 3) and PHOTs (PHOT1 and 2) that sense blue/UV-A light, and UVR8 that absorbs UV-B (Zoratti et al., 2014), interacting with the ubiquitin E3 ligase COP1 (CONSTITUTIVE PHOTOMORPHOGENIC1) that controls the degradation of target transcription factors, including ELONGATED HYPOCOTYL5 (HY5). HY5 is associated with induced CHS, CHI and flavonoid production under light and UV-B radiation conditions in Arabidopsis (Shin, Park & Choi, 2007). It also binds the MYB75/12/111 promoter to increase their expression and modulate anthocyanin biosynthesis (Stracke et al., 2010;Shin et al., 2013;Nguyen et al., 2015).
The present work aimed to assess the global transcription of regulatory, structural and hormone signal transduction genes which might positively or negatively regulate broccoli's anthocyanin biosynthetic pathway. The light sensitivity of pigment biosynthesis makes the broccoli 'Long Jing' an optimal plant for evaluating anthocyanin synthesis and explore the underpinning mechanisms under light conditions. Indeed, anthocyanin accumulation accompanied with broccoli head growth and development under natural light conditions. However, broccoli heads show reduced coloration under shading conditions. Here, the 'Long Jing' cultivar was employed for analyzing light-associated anthocyanin biosynthesis and the underlying mechanisms by RNA-seq. Based on the association of shading time with anthocyanin amounts, anthocyanin accumulation was assessed at 0 d, 7 d (highest production rate) and 11 d (peak amounts). The present findings provide insights into the molecular mechanisms underpinning light-associated anthocyanin production in broccoli, facilitating genetic engineering for increasing anthocyanin amounts in vegetables.

Plant materials and RNA preparation
The purple broccoli cultivar 'Long Jing' was assessed as the experimental material. After flower head formation, a total of seven stages were defined as 0, 3, 5, 7, 9, 11, and 14 days. Flower heads were collected for phenotype observation and anthocyanin level measurement at each developmental stage of the head under light and dark (shading using a sunshade net over the whole heads or leaves) conditions. In addition, light and darkness treated heads at 0, 7 and 11 days, and darkness treated leaves at 7 days were collected for RNA-Seq from the flower heads of the purple cultivar, respectively. Three biological replicate specimens were obtained. All specimens underwent snap freezing in liquid nitrogen and storage at −80 • C. To assess cells displaying purple coloration, head flowers underwent transverse sectioning by hand and analysis under a Zeiss Axioscope photo microscope.

RNA purification and library generation for transcriptomics
Total RNA was extracted from head flowers in purple broccoli at different times under both light conditions with mirVana miRNA Isolation Kit (Ambion) as directed by the manufacturer. RNA integrity was assessed on an Agilent 2100 Bioanalyzer (Agilent Technologies). Specimens with RNA Integrity Number (RIN) ≥7 were further evaluated. The libraries were generated with TruSeq Stranded mRNA LT Sample Prep Kit (Illumina, USA) as instructed by the manufacturer.

Transcriptome sequencing, de novo assembly and functional annotation
The obtained libraries underwent sequencing by the PE strategy on a HiSeqTM 2500 or Illumina HiSeq X Ten; cDNA fragments approximated 125 or 150 bp. The raw reads obtained underwent pre-processing with Trimmomatic; those with ploy-N or showing low quality were excluded, leaving clean reads. These clean reads underwent assembly into contigs and de novo assembly into transcripts using Trinity (version: 2.4) (Grabherr et al., 2011) by the paired end method. The longest transcripts were selected as unigenes for further assessment. Original sequencing data were deposited in SRA (Short Read Archive; accession number PRJNA560282).

Unigene quantification, assessment of differentially expressed genes (DEGs) and gene annotation
Fragments per kilobase of transcript per Million (FPKM) and read counts for each unigene were assessed with Bowtie 2 and eXpress. DEG identification was carried out with the DESeq functions estimate Size Factors and negative binomial Test. P < 0.05 and fold Change >2 or <0.5 were thresholds for determining significant differential expression. Hierarchical clustering of DEGs was carried out to assess the transcripts' expression patterns. The assembled unigenes were assessed with R according to hypergeometric distribution.

Real-time quantitative reverse transcription-PCR
To confirm the results of Illumina analysis, qRT-PCR was performed for multiple genes. Total RNA isolation from specimens collected in various developmental stages under light or dark conditions was carried out. Reverse-transcription used PrimeScript RT Master Mix Perfect Real Time Kit (Takara) as directed by the manufacturer. Finally, qRT-PCR was carried out on a QuantStudio 5 Real-Time PCR System (Fisher Scientific, USA) with SYBR Premix Ex Taq (TaKaRa, Japan) in triplicate. Data were normalized to Actin 2 amounts, and the comparative CT method was employed for analysis.
To study light-response factors in anthocyanin production in the broccoli cultivar 'Long Jing', we shaded the whole head and leaves using a sunshade net during the developmental stages of the head from 0 d to 14 d, with light conditions employed as a control treatment. In comparison with controls, flower buds grown under head-and leaves-shading all faded, with head shading exerting more pronounced effects (Fig. 1C). The relative amounts of total anthocyanin were measured, and the head-shading treatment group showed lower levels compared with the leaf-shading treatment group, and both of these groups had lower values than the control group ( Fig. 1B; Table S1). In addition, decrease in relative anthocyanin levels was more pronounced under head-shading (2.04 mg g −1 FW) compared with leaf-shading (1.47 mg g −1 FW), indicating that shading during the head development in broccoli significantly affected the relative contents of total anthocyanin.
Quantitative analysis was further conducted to identify the key developmental stage under light during anthocyanin production. Spectra were obtained at 200-600 nm, and chromatograms at 520 nm (Fig. 1B). In the control group, relative anthocyanin amounts rose during head development (0 d to 11 d), showed a peak growth rate at 7 d, and then remained steady after 11 d, while in shaded plants they increased slowly during head's developmental stages from 0 d to 14 d. Therefore, 0 d, 7 d and 11 d were considered critical times for response to light during anthocyanin production. Microscopic examination of sections of flower buds from head shading, leaf shading and control plants revealed that the prominent purple color extended to more bud tissue cells, with anthocyanins accumulating closer to the outer layer of buds (Fig. 1D), in accordance with previously published data on purple Graffiti cauliflower (Chiu & Li, 2012). Compared with the control treatment, head and leaves shading conditions resulted in lighter purple pigments in upper epidermal layers. The above results suggested light had a pivotal function in anthocyanin production, and shading treatment significantly repressed anthocyanin accumulation during head development in broccoli.

Organ responses to light during anthocyanin biosynthesis
In order to assess differences in organ response to light during anthocyanin production in broccoli, RNA-Seq was performed under head-shading, leaves-shading and normal night conditions at the fastest point (7 d), respectively. The flowers were collected to construct nine libraries for transcriptomics in three biological replicates.
Comparing the expression amounts of differently expressed genes (DEGs) under head-shading and leaves-shading treatment, a total of 2,223 and 2,558 DEGs were up-and down-regulated, respectively. To identify the functions of these down-regulated DEGs, Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses were carried out (Du et al., 2010;Conesa et al., 2005;Kanehisa et al., 2008). Thirty GO terms were found as enriched biological processes based on the DEGs (Table S2; Fig. S1): ''nucleus'', ''response to abscisic acid'', ''DNA-binding transcription factor activity'', ''response to water deprivation'' and ''sequence-specific DNA binding'', which were the major GO terms. The significantly enriched pathways were identified using KEGG analysis ( Fig. 2A). Among the significantly enriched pathways, ''Starch and sucrose metabolism'', ''Plant hormone signal transduction'' and ''Protein processing in endoplasmic reticulum'' were the major public pathway-related database.
Comparing the expression amounts of differently expressed genes (DEGs) under shading and normal light treatment, a total of 378 and 660 DEGs were up-and down-regulated under head-shading treatment, respectively. For DEGs under leaf-shading treatment, we found that a total of 1532 and 1628 DEGs were up-and down-regulated, respectively (Tables  S3 and S4; Fig. S2 and S3). In GO analysis, DEGs encoding proteins related to response to ''light stimulus'', ''chloroplast and photosystem I'' were down-regulated under both head-and leaf-shading treatments. In KEGG analysis, DEGs were grouped in 20 functional classes. Under leaf-shading treatment, DEGs were significantly enriched in ''amino acid biosynthesis'', ''carbon metabolism'', ''sulfur metabolism'', ''cysteine and methionine metabolism'', and ''photosynthesis-antenna proteins'' (Fig. 2B), while downregulated DEGs with previously described functions were associated with ''photosynthesis'', ''peroxisome'' and ''flavonoid biosynthesis'', indicating such pathways/processes might be affected by head shading (Fig. 2C). The KEGG analysis showed the critical pathways in response to light. The above results indicated that photosynthesis and anthocyanin biosynthetic process were markedly inhibited by head shading while photosynthesis and carbon-nitrogen-sulfur metabolism were significantly repressed by leaf shading by transcriptional regulation in response to light.
According to flower head phenotypes, pigment synthesis and associated DEGs, the head might constitute the main organ showing a response to light during anthocyanin biosynthesis, in accordance with previous data on chrysanthemum in which the capitulum was the key organ responding more to light compared with the leaf during anthocyanin production (Hong et al., 2015).

Overall transcriptomic analysis under head shading and control treatments
To assess how light induces anthocyanin production in broccoli, RNA-Seq was performed for three triplicate groups at 0 d, 7 d and 11 d under head-shading and normal light treatments, respectively (Fig. 3A). Averagely 20 million clean reads were produced per sample, including 81∼85% which were mapped to the Brassica oleracea genome (Table S5). Cumulatively, 90231 transcripts were detected across all six treatments with an FPKM ≥1.
The full annotation and the expression levels of all genes (FPKM values) are found in Tables S6 and S7.

Thousands of genes are activated in broccoli in response to light
To identify genes responding to light, differential gene expression analyses were performed at 0 d, 7 d and 11 d under both head-shading and control treatments. Transcriptomics revealed 1461 (7 versus 0 day) and 5132 (11 versus 0 day) DEGs under control treatment, and 1752 (7 versus 0 day) and 1859 (11 vs. 0 day) DEGs under head-shading treatment.
The number of shared DEGs increased over time, probably due to the total number of DEGs increasing between treatments. Relative to the control treatment, 378 upregulated and 660 downregulated DEGs (P < 0.05) were detected at 7 d, while 512 upregulated and 924 downregulated DEGs were found at 11 days, respectively (Fig. 3B).
To gain a deeper understanding regarding the associated biological processes, transcripts were assigned to nine clusters per treatment group (Figs. S4 and S5). Then, KEGG analysis was performed for identifying biological pathways enriched in clusters of comparably regulated genes (Figs. 4A and 4B). Cluster 1, 6 and 8 were under both light and head-shading treatments. Cluster 1 encompassed genes downregulated throughout the study, including those controlling Plant hormone signal transduction, Starch and sucrose metabolism, Glycero-phospholipid metabolism and Fructose and mannose metabolism. Cluster 6 comprised genes positively regulated throughout the entire study, including those associated with Carbon metabolism, Nitrogen metabolism and Fatty acid metabolism. Cluster 8 contained genes significantly upregulated from 0 d to 7 d, many of which were involved in Phenylpropanoid biosynthesis, Photosynthesis, and Flavonoid biosynthesis. These findings indicated that cluster 8 DEGs were expressed in early phases of light induction and played roles in regulating anthocyanin biosynthesis. As many as 2400 DEGs showed differential expression between the light and shading libraries in cluster 8, in which 373 and 1377 individual DEGs showed specific expression under head-shading and normal light conditions, respectively, and 650 genes were co-expressed under both conditions ( Fig. 4C; Table S8, S9 and S10). Thus, we focused on the 650 co-expressed and 373 specifically expressed under head-shading, which most likely represented light-induced genes.

Different expression patterns of anthocyanin biosynthesis structural genes
The expression profiling of structural genes was performed to assess their roles in anthocyanin biosynthetic pathway after shading. In this study, fifteen such genes, e.g., PAL, C4H, 4CL, CHS, CHI, F3H and DFR were regulated by light (Fig. 5). Most of the genes had comparable expression profiles; their expression amounts were low in the A53CK sample at 0 d, and gradually increased in the B53CK sample exposed to normal light for 7 d, peaking in the C53CK sample treated by normal light for 11 d. The expression levels of all transcripts were suppressed in the head-shading treatment group, corroborating the reduced anthocyanin amounts in plants under head-shading. Most genes showed higher expression levels in the C53WJ group treated by head-shading for 11 d compared with the B53WJ group treated by head-shading for 7 d, except assembly18039 (PAL), assembly 39669 (C4H ) and assembly 52832 (CHS), which showed higher expression levels in the B53WJ group. Thus, these genes were considered critical structural genes associated with the effects of light on anthocyanin production.

Expression of genes associated with light signal perception and transduction
Photoreceptors (PHYA, PHYB, PHYC, PHYD, PHYE), cryptochromes (CRY1, CRY2, CRY3), phototropins (PHOT1, PHOT2) and UV RESISTENCE LOCUS8 (UVR8) are four classes of photoreceptors contributing to light response (Chaves et al., 2011;Christie, 2007;Jenkins, 2014). Compared with head shading samples, flower head response to light was mediated by three CRY3 photoreceptors, including assembly 37255, assembly 86925 and assembly 18166 (Table 1), which were all downregulated under head-shading treatment. Under light conditions, the expression levels of assembly 37255 and assembly HY5 induces photomorphogenesis under all light conditions and exerts direct regulatory effects on light-responsive genes (Chattopadhyay et al., 1998). Here, we found BoHY5 (assembly 42644) was downregulated under head-shading treatment in comparison with light conditions. Assembly 42644 showed rapid upregulation in the B53CK group, peaking in C53CK, with reduced amounts under head-shading treatment.

DEGs Encoding Transcription Factors and their interaction with Hormone-Related Genes
To assess the complex network of signaling pathways in light-induced anthocyanin biosynthesis, we further compared the expression profiles of transcription factors. A total of 133 genes were assigned to the MapMan ''transcription factors'' bin and more than half were down-regulated (Fig. 6). Members of the GATA, Trihelix, bZIP and C3H families were  The interaction partners of these TFs activated as molecular responses are key components of signal transduction pathways that take place during anthocyanin biosynthesis. To investigate the functions of plant hormones in light-induced anthocyanin biosynthesis, the expression patterns of genes involved in plant hormone response as receptors and response factors were assessed by heat map analysis (Fig. 7). The results showed that multiple genes involved in abscisic acid, auxin, salicylic acid, ethylene and jasmonic acid signaling pathway were mostly upregulated in samples treated for 11 d in comparison with those treated for 0 d and 7 d.
In the auxin transduction pathway, the expressions levels of genes encoding auxin influx carrier/auxin-responsive protein IAA (AUX/IAA) (assembly 37159 and assembly 75889) and auxin responsive GH3 gene family (GH3) (assembly 40896) peaked in the C53WJ sample treated by head-shading for 11 d. Meanwhile, the other two GH3 genes (assembly45596 and assembly 87995) were significantly up-regulated in the C53CK sample treated by normal light for 11 d. In the abscisic acid transduction pathway, genes encoding protein phosphatase 2C (PP2C) (assembly 49106), serine/threonine protein kinase SRK2n (SnRK2) genes (assembly17083) and ABRE binding factors (ABF) (assembly 77239) were downregulated under head-shading conditions in comparison with normal light conditions. In the jasmonic acid transduction pathway, the transcriptional level of jasmonate ZIM (JAZ) domain-containing gene (assembly133) was highest in the B53WJ sample treated by headed-shading for 7 d, while gene encoding jasmonic acid resistant 1 (JAR1) (assembly23539) was highly expressed in the C53WJ sample treated by headshading for 11 d. In the salicylic acid (SA) signaling pathway, SA receptors Non-Expresser of Pathogenesis Related Gene 1 (NPR1) genes (assembly 66422), TGA factor (assembly 89977) and pathogenesis-related 1 (PR-1; assembly 26640) genes were upregulated in

Validation of the expression of genes of the Anthocyanin Biosynthetic Pathway
To confirm gene expression data revealed by transcriptomics, seven regulatory genes (BoMYB6-1, BoMYB6-2, BoMYB6-3, BoMYB6-4, BoTT8_5-1, BoTT8_5-2 and BoEGL5-3) and two structural genes (PAL1, 4CL-1) contributing to anthocyanin production in broccoli underwent amplification from specimens obtained in head's developmental stages under shading and light conditions, respectively, by RT-qPCR (Fig. 10), with Actin2 employed as a reference gene. All 9 genes displayed identical expression trends obtained by RNA-seq. Most of the transcription factors assessed also showed identical expression trends as determined by transcriptomics. Moreover, the light and shading groups were significantly different (P < 0.05).

Broccoli head is the key light-response receptor in anthocyanin production
The purple broccoli has abundant flavonoids and other bioactive molecules in addition to glucosinolate-derived isothiocyanates, vitamins and minerals, indicating a great nutritional value for this plant (Moreno et al., 2010). In the present study, delphinidin-3-O-galactoside, delphinidin-3-O-glucoside, cyanidin-3-O-galactoside, cyanidin-3-Oglucoside, malvidin-3-O-galactoside and malvidin-3-O-glucoside were identified in broccoli head, in accordance with previously published data on broccoli sprouts, purple Graffiti cauliflower and red cabbage (Moreno et al., 2010;Chiu et al., 2010;Yuan, Chiu & Li, 2009). Light represents a predominant environmental stimulus controlling plant anthocyanin production (Liu et al., 2018). As shown above, shading reduced anthocyanin production in broccoli. Under light conditions, anthocyanins were produced in a rapid manner (Fig. 1). In this study, the relative contents of total anthocyanins under headshading were more reduced than those obtained under leaves shading and normal light conditions. This was similar to anthocyanin accumulation in chrysanthemum under head-shading and leaf shading conditions as reported by Hong et al. (2015).
Usually, more focus is placed on fresh broccoli heads which provide economic benefits directly; however, purple broccoli varieties with purple or green leaves are variable in leaf pigmentation. In order to illustrate this mechanism, RNA-seq was performed under head-shading and leaves-shading treatment. Downregulated DEGs, GO and KEGG analyses were carried out. GO analyses showed that, ''nucleus'', ''response to abscisic acid'', ''DNAbinding transcription factor activity'', ''response to water deprivation'' and ''sequencespecific DNA binding'' were the major GO terms. Moreover, KEGG analyses showed that, ''Starch and sucrose metabolism'', ''Plant hormone signal transduction'' and ''Protein processing in endoplasmic reticulum'' were the major public pathway-related database. ABA played positive role in modulating anthocyanin biosynthesis (Carvalho, Carvalho & Duque, 2010). Starch degradation and Sucrose-specific contribute to anthocyanin biosynthesis, while MdSnRK1.1 interacts with MdJAZ18 to induce sucrose-induced anthocyanin and proanthocyanidin biosynthesis in apple (Liu et al., 2017), however IAA might play an crucial role in anthocyanin accumulation regardless of sugar and starch in ornamental kale (Ren et al., 2019). Endoplasmic reticulum likely function in the biosynthesis and transport of anthocyanin pigments (Wagner, 1987). Therefore, the downregulated genes in hormone signaling, starch and sucrose metabolism, endoplasmic reticulum and transcription factors might affect the anthocyanin accumulation under head-shading treatment in broccoli. Although anthocyanin biosynthesis is well characterized, the associated light-response receptors in broccoli are less clear. Photosynthesis was significantly repressed under both head and leaves-shading conditions, suggesting that shading affects plants during photosynthesis. The DEGs in ''photosynthesis'', ''peroxisome'', ''flavonoid biosynthesis'' were significantly repressed by head shading while ''amino acid biosynthesis'', ''carbon metabolism'', ''sulfur metabolism'', ''cysteine and methionine metabolism'', and ''photosynthesis-antenna proteins'' were significantly suppressed by leaves shading via transcriptional regulation. We can infer that under leaves-shading treatment, carbon fixation and carbohydrate production were affected by less light indirectly leading to less anthocyanin contents (Shao et al., 2014). Under head-shading conditions, some photoreceptors and anthocyanin biosynthesis-associated genes downregulated directly leading to decreased anthocyanin contents in response to light. Therefore, the head might be the more critical role as light-response receptor in anthocyanin production.

Light-induced anthocyanin biosynthesis is mediated by signal transduction pathways in 'Long Jing'
Plants use many photoreceptors for coordinating responses to environmental light (Ma et al., 2019). When broccoli flowers under head-shading conditions were compared with those under normal light conditions, we found three CRY3 genes were downregulated, in agreement with the tendency of anthocyanin production in broccoli head flower (Fig.  1B). These finding indicate that CRY3 may be an important photoreceptor in broccoli head flower, with critical functions in regulating anthocyanin production (Opseth et al., 2015). Li et al. (2017) andLi et al. (2018) characterized the eggplant photomorphogenic factors CRY3 was upregulated by light. In the current study, HY5 was identified among DEGs during head flower development and its transcription levels quickly rose under light conditions but were reduced under head-shading treatment. In Turnip (Brassica rapa), the upregulation of BrHY5 further induced BrPAP1 expression to produce more anthocyanins under sunlight (Yang et al., 2017). Shin et al. (2013) has reported that HY5 induced anthocyanin accumulation by directly binding the promoter of MYB75/PAP1 transcription factor in Arabidopsis. In apple, MdHY5 also bound on the 5 upstream region of MdMYBA in a yeast system (Peng et al., 2013). This suggests HY5 might represent an inducer of broccoli head flower coloration by binding the promoter of MYB TFs under light conditions (Stracke et al., 2010;Shin et al., 2013;Nguyen et al., 2015).

Plant hormones are involved in light-induced broccoli coloration
Previous studies have shown that phytohormones controlling anthocyanin accumulation can be affected by light (Loreti et al., 2008;Zhang et al., 2016a;Zhang et al., 2016b).
Endogenous application of auxins can inhibit the expression of anthocyanin-related genes (Deikman & Hammer, 1995). In this study, the expressions levels of IAAs and one GH3 genes were higher in the C53WJ sample treated by head-shading for 11 d. However, two GH3s were upregulated in the C53CK sample treated by normal light for 11 d. These distinct expression patterns indicate that auxin signaling has different functions in the regulation of anthocyanin biosynthesis in broccoli through various transduction pathways. In purple ornamental cabbage, Jin et al. (2019) suggested that ABA might increase the intensity of purple pigmentation of the inner leaves. Carvalho, Carvalho & Duque (2010) provided evidence that ABA plays a positive role in modulating anthocyanin biosynthesis in hormone mutants after exogenous application (Carvalho, Carvalho & Duque, 2010). We observed significant upregulation of genes encoding ABA-responsive elements, such as PP2C, SnRK2 and ABF in the normal light treatment (Fig. 7). Thus, we speculate that these ABA signaling factors might promote the expression of anthocyanin-related genes. Wang et al. (2019) indicated that ethylene acts as a negative regulator in red light-regulated anthocyanin biosynthesis in cabbage. Ethylene suppresses the anthocyanin biosynthesis via binding to ETRs in Arabidopsis (Jeong et al., 2010) and peel (Ma et al., 2019). Similarly, we observed that the expression levels of ETR were higher under the head-shading treatments. Previous studies have shown ethylene treatment significantly lowers anthocyanin accumulation, while SA alleviates these effects in canola plants (Brassica napus L.) (Tirani, Nasibi & Kalantari, 2013). ABA, JA and SA pre-treatments could increase anthocyanin accumulation in turnip (Brassica rapa ssp. rapa) and Brassica juncea L. (Thiruvengadam et al., 2016;Sharma et al., 2018). Horváth et al. (2007) reported exogenously applied SA increases the accumulation of anthocyanin in UV-B exposed T. aestivum. Endogenous application of jasmonate can also increase the production of anthocyanin (Memelink). However, we found that all genes encoding JAZ, JAR1 and NPR1 in jasmonic acid signalling pathway as well as TGA and PR-1 in the salicylic acid signaling pathway were upregulated under head-shading treatment, which implies that SA and JA play negative roles in light-induced anthocyanin accumulation in broccoli. In general, some form of hormonal cross-talk may be present in pigment accumulation of broccoli flower head.