In this study, three-year-old southern-highbush blueberries of the cultivar ‘Misty’ were utilized, and the plants were grown in a plant factory system with artificial lighting (PFAL) at Zhuang-hang Comprehensive Experimental Station of Shanghai Academy of Agricultural Sciences, Shanghai, China. The growth conditions in the PFAL were as follows: the temperature was set at 25 ± 2°C during the daytime and 22 ± 2°C at night, the photoperiod was 16-h light and 8-h dark, and the CO₂ concentration and relative humidity were set at approximately 400 μmol·mol^-1 and 65% ± 5%, respectively. The blueberry plants were treated respectively with four light qualities: white (W), full blue (B), full red (R), and 60% red + 40% blue (60R/40B), and the white light treatment was considered the control. The light intensity of all treatments was equal at 200 ± 5 µmol·m^-2·s^-1, each treatment with four experimented blueberry plants. Approximately 30 days after the treatments, the fifth to tenth mature leaves, which were similar developmental stage, had no disease, no mechanical damage, and complete form, were collected from the top branch of each blueberry plant. Approximately 3 g of leaves were collected from each plant and immediately frozen in liquid nitrogen and stored at −80°C for transcriptome profiling and metabolite quantitative analysis. The determination of physiological indicators was performed with four replicates and each plant as a replicate. The transcriptome and metabolome were analyzed with three biological replicates.

Anthocyanin content in the fresh leaf samples was determined according to the method improved by Li et al. (2016). Leaves were harvested from the blueberry plants, quickly weighed, and homogenized with a mortar and pestle; they were then treated with 0.1 mol·L⁻¹ acidized ethanol (10mL) in a microcentrifuge tube and then with water at 60°C for 30 min. The extraction procedure was performed twice and then the samples were centrifuged at 1,500g for 5 min at 4°C, and the supernatants were then collected. The absorbance of the samples was measured at 530, 620, and 650 nm, respectively, using an ultraviolet–visible spectrophotometer (UV-2000) (Spark, TECAN, Switzerland). The values are reported as (1) ΔOD = (OD530 − OD620) − 0.1 (OD650 − OD620) and (2) anthocyanin content = (ΔOD × V × 1000000)/(ξ× m), where V is the dilution volume, ξ is the molar absorption coefficient, and m is the fresh weight of the sample. All measurements were performed with four replicates.

The photosynthesis of the blueberry leaves was measured using a leaf gas exchange system (CIRAS-3; PP Systems). During the measurements, the gas exchange system controlled the environmental conditions within the curette. The temperature of the leaf was 25.0 ± 0.2°C and the CO₂ concentration was 390 ± 5 μmol·mol^-1. To discover the photosynthetic efficiency of the different spectra, light response curves were constructed for the blueberry leaves under corresponding light qualities. Photosynthesis indicators, including the maximum gross assimilation rate (A _{g, max}) and maximum quantum yield for CO₂ assimilation (QY _{m, inc}), were calculated according to the light response curves.

Total RNA from the blueberry leaves was isolated using the Polysaccharides and Polyphenolics-rich RNA Prep Pure Plant Kit (TIANGEN, China), following the manufacturer’s instructions. The integrity of the RNA was verified using an Agilent 2100 Bio analyzer (Agilent Technologies, USA). The RNA-Seq and libraries were constructed using the Biomarker Technologies Corporation (Beijing, China) and the Vcorymbosum_v1.0 genome was used as the reference genome in this study. Before conducting data analysis, it is vital to ensure that the red reads are of sufficient quality. Remove reads containing joints and low-quality reads (N was more than 10%; base number with mass value Q≤ 10 accounted for more than 50%), clean reads with high quality were obtained. StringTie was used to assemble the clean reads above and construct a traffic network using the alignment information to construct a multivariable shear map through using the maximum flow algorithm to assemble reads and evaluate its expression. Using the FPKM (Fragments Per Kilobase of transcript per Million fragments mapped) as an indicator to measure the expression level of transcripts or genes through the maximum flow algorithm. The edgeR was used to analysis significant difference of genes transcript level and gene with an adjusted FDR < 0.01& Fold Change ≥2 was identified as differentially expressed. The functions of the unigenes were determined by aligning them to protein databases using BLASTx, including NCBI non-redundant (Nr) and nucleotide (Nt), Swiss-Prot, Kyoto Encyclopedia of Genes and Genomes (KEGG), Clusters of Orthologous Groups of proteins (COGs), and Gene Ontology (GO) databases. GO classification was performed by mapping the relationship between the Swiss-Prot and GO terms, and the genes were mapped to the KEGG database to annotate their potential metabolic pathways (Kanehisa et al., 2008).

Total RNA was isolated from the blueberry leaves collected under the four light-quality treatments. For the qRT-PCR analysis, the SYBR RNA RT-PCR Kit (TaKaRa, Japan) was used to synthesis the first-strand of cDNA and the Light Cycler 480 (Roche, USA) was used to perform the qRT-PCR procedure. qRT-PCR was performed with the SYBR-Green PCR kit (TaKaRa, Japan) and the reaction mixture system was prepared according to the manufacturer’s instructions. The qPCR program was previously described by Zhang et al. (2021b). The expression level of each gene was calculated in triplicate based on the 2^-ΔΔCt algorithm, and VcGAPDH was used as the internal reference gene. Snap Gene software (www.snapgene.cn) was used to design the primers for the target genes ( Table S1 ).

Anthocyanins are secondary metabolites in plants. The qualitative and quantitative analysis of the anthocyanins could be measured using the LC-MS/MS system. Freeze-dried blueberry leaves (50 mg) were crushed and dissolved in 500 μL of extract. They were then centrifuged at 1200 g/min for 10min; the extracts were filtrated and absorbed for LC-MS/MS analysis. The instrument system used for data acquisition included an Ultra Performance Liquid Chromatography (UPLC) and Tandem Mass Spectrometry (MS/MS). Qualitative analysis of the anthocyanins detected using mass spectrometry was performed based on the construction of standard databases and the quantification of anthocyanins was identified by employing scheduled multiple reaction monitoring (MRM). The Analyst 1.6.3 software (Sciex) was used to acquire and dispose data and the quantities of all metabolites were analyzed using Multiquanta 3.0.3 software (Sciex). To draw standard curves for the different substances the concentration was used as the abscissa and the peak area as the ordinate. The integrated peak areas for all detected samples were substituted into the standard curve linear equation for calculation and the absolute content values for each sample were obtained.

Targeted anthocyanins and their intermediates were selected based on the following criteria: (1) the obtained compounds should be related to the phenylpropanoid and anthocyanin biosynthesis pathways and their molecular formula and mass information was established using the KEGG and plant metabolic network databases; (2) certain filtrated daughter ions in the anthocyanins could be identified as similar compounds as described in the literary references; and (3) the MS² spectra of the selected metabolites and standard compounds was as found in the METLIN and LIPDMAPS databases. The obtained anthocyanin-related compounds are shown in Supplementary Table S2 .

Pearson’s correlation coefficient (PCC), a statistical method used to measure the linear relationship between two variables with values between −1 and 1, was used to calculate the correlation coefficient between the transcriptome and metabolome data and further integration to analyze the relationships between the gene transcript and metabolite content. Associations between PCC > 0.8and P < 0.05 were selected and the network of genes and metabolites was constructed using Cytoscape software (Cytoscape Consortium, San Diego, CA). Additionally, genes and metabolites related to anthocyanin biosynthesis were mapped using the KEGG pathway database.

For this study, all experiments were performed in triplicate and the results are expressed as the mean ± standard deviation (SD). Analysis of variance (ANOVA) was performed to estimate the significant differences between the treatment means at P < 0.05. A heat map was constructed to exhibit the gene expression pattern and metabolite accumulation and to separate the groups of genes and metabolites with similar expressions and accumulation values. Principal component analysis (PCA) was performed to realize the metabolite variability among the white, blue, red, and red/blue light treatment groups.

The relative spectra under different wavelengths were significantly different ( Figure 1I ) and the different light qualities influenced the color gradation and anthocyanin synthesis in the blueberry leaves. The color gradation in the blueberry leaves was deepest with the blue light, followed by the red/blue, red, and white lights, and the corresponding anthocyanin content was highest under the blue light, followed by red/blue, red, and white lights. The anthocyanin content of the leaves under blue, red/blue, and red lights was approximately 4.6-, 3.7-, and 1.6-fold higher, respectively, than that under the white light (the control group) ( Figure 1II ). A phenomenon was identified in which the blue and red lights could promote the accumulation of anthocyanin, but differed significantly in their impact levels. The promotion of anthocyanin biosynthesis was more prominent under blue light than red light, while the accumulation was in between these two levels with the red/blue light combination. Leaf photosynthesis exhibited adverse effects on anthocyanin accumulation, which were strongest under white light but lowest under blue light. The photosynthesis indices, including A _{g, max} and QY _{m, inc}, presented a consistent trend, with the greatest index under the white light, followed by red, red/blue, and blue ( Figure 1III ). Blue and red lights can promote anthocyanin accumulation but impair photosynthesis in blueberry leaves. In contrast, white light prejudiced anthocyanin biosynthesis but facilitated photosynthesis.

To elucidate the molecular genetics mechanisms underlying anthocyanin biosynthesis in blueberry leaves when under different light conditions, a comparative transcriptome analysis was performed. The Illumina sequencing platform was used to construct and sequence twelve RNA-Seq libraries (each treatment with three biological replicates). The rigorous quality estimation and data cleaning resulted in 21.48–26.23 M clean reads with Q30 bases that were identified as high-quality reads for further analysis. Their GC contents and Q30 ranges were 46.93%–47.36% and 94.69%–94.93%, respectively ( Table S3 ). Moreover, the N percentage was 0.00% in all samples.

To substantiate the effects of light quality on the transcript levels of anthocyanin biosynthesis genes in blueberry leaves, DEGs between samples from the light quality treatments were identified using pairwise comparisons of the four cDNA libraries (blue vs. white, red vs. white, red/blue vs. white, and blue vs. red), and the fragments per kilo-base per million mapped reads (FPKMs) of all DEGs were calculated. A total of 1812 (640 up-regulated and 1172 down-regulated), 596 (250 up-regulated and 346 down-regulated), 3283 (1123 up-regulated and 2160 down-regulated), and 3428 (1974 up-regulated and 1457 down-regulated) differentially expressed genes (P < 0.05, fold change > 2) were found during the comparisons of the blue and white, red and white, red/blue and white, and blue and red libraries, respectively, whereas 13 DEGs were shared among the four comparisons ( Figure 2I ; Table S4 ). The number of DEGs was substantially higher in the blue vs. white comparison when compared with red vs. white, which indicated a strong gene expression disturbance experienced by the blueberry leaves under blue light. The blueberry leaves were more sensitive and actively responded to blue light to promote anthocyanin accumulation when compared to red light. GO term annotation was performed to improve our understanding of the functions of these DEGs in blueberry leaves under different light conditions. The DEGs were considered enriched in GO-terms if at least one of the terms was categorized as a biological process, cellular component, or molecular function ( Figure 2II ). The results of the GO-biological process (GO-BP) analysis revealed that significantly enriched DEGs were associated with metabolic processes, mainly phenylalanine and secondary metabolisms, among the four comparisons. To further explore the enriched metabolic pathways related to anthocyanin accumulation in blueberry leaves, we mapped all DEGs to reference canonical pathways in the KEGG database. A total of 4097 DEGs (blue vs. white:844; red vs. white:307; red/blue vs. white:1440; and blue vs. red:1506) with annotated KEGG results were classified into five main categories, including organismal systems, metabolism, genetic information processing, environmental information processing, and cellular processes, based on the type of KEGG pathways ( Figure 2III ). During metabolism classification, phenylpropanoid biosynthesis and starch and sucrose metabolism (photosynthesis) were the primary enriched pathways throughout the four comparisons. Additionally, DEGs were annotated by searching against the COG dataset. COG-annotated putative proteins were functionally classified into at least 26 categories, including cellular structure, biochemical metabolism, molecular processing, and signal transduction ( Figure S1 ). In addition, the DEGs were collected and classified into 17 groups based on their annotated functions ( Table 1 ). Among all the categories, phenylalanine metabolism, secondary metabolism, starch and sucrose metabolism, and carbohydrate metabolism (photosynthesis) were highly significant and enriched pathways, as well as being enriched they also had a greater number of DEGs. The results indicated that these DEGs were associated with metabolic pathways that could contribute to the biosynthesis and accumulation of anthocyanin in blueberry leaves under different light qualities.

The transcriptome analysis under light conditions showed that the DEGs related to phenylalanine, secondary, starch and sucrose, and carbohydrate (photosynthesis) metabolism processes may help blueberry leaves to accumulate anthocyanins. The physiological indicators suggested that there were significant differences in anthocyanin accumulation and photosynthesis in blueberry leaves due to the distinct light qualities. The DEGs implicated in the crucial steps of these processes were consequently screened. Moreover, the blueberry leaf was treated with distinct spectra, the DEGs associated with photoreceptor responses to the different light qualities were searched for. Core genes related to anthocyanin biosynthesis, photosynthesis, and photoreceptors were studied in detail, and the results demonstrated that most of the uni-transcripts showed significant changes in their expression levels ( Figure 3 ; Table S5 ). An opposite dynamic trend was observed between the expression patterns of the anthocyanin accumulation-related and photosynthesis-related DEGs. The expression levels for most of the anthocyanin biosynthesis-related DEGs were higher in the blue light when compared with the other lights, but the DEGs associated with photosynthesis were down-regulated with the blue light. There may be feedback regulation among the DEGs related to anthocyanin accumulation and photosynthesis under blue light. DEGs encoding phytochrome, as receptor proteins of red and far-red light (600–750 nm), were up-regulated under red or red/blue light. Consistently, DEGs encoding cryptochrome, which mediated several responses to blue light and UV-A (320–500 nm), were up-regulated under blue light ( Figure 3I ). In relation to photosynthesis, there were DEGs among the different spectra. Photosynthesis-related DEGs were down-regulated under blue light when compared with the other treatments ( Figure 3II ), which was consistent with the weakness of photosynthesis under blue light. These results suggest that blue light prejudiced photosynthesis in the blueberry leaves, which is in sharp contrast to the results for anthocyanin accumulation under blue light. In many cases, changes in anthocyanin biosynthesis correspond to changes in the expression of genes that encode pathway enzymes. Among the related anthocyanin accumulation processes, 22 DEGs (including C4H, 4CL, CHS, DFR, F3’H, F3’5H, F3H, LDOX, OMT, and UFGT) were found to participate in vital steps in the anthocyanin pathway. Among these, LDOX, UFGT, and OMT were expressed well under blue light, whereas DFR and OMT were highly expressed under red/blue light. These DEGs contributed to the cyanidin, pelargonidin, malvidin, delphinidin, petunidin, and peonidin accumulation in the anthocyanin biosynthesis pathway ( Figure 3III ). Additionally, the expression of the DEGs related to lignin and flavonoid metabolism, which are also significant secondary metabolite pathways in plants, decreased in the blue light in comparison with that in white, red, and red/blue lights. These results indicate that blue light could promote the accumulation of anthocyanin but hinder the photosynthesis and secondary metabolism process of lignin and flavonoids in blueberry leaves when compared with other light sources via the direct or indirect regulation of the expression levels of related DEGs.

Blueberry leaves under different light qualities showed significantly different TFs, belonging primarily to 20 different families ( Figure 4I ; Table S6 ). The differentially expressed TFs belonged mostly to the AP2/ERF-ERF, bZIP, bHLH, MYB, and NAC families. The number distribution mole for the TFs in the blueberry leaves under different light qualities was consistent with the quantitative distribution of the gene function annotation. There were more differentially expressed genes or TFs under the red/blue light, indicating that the red/blue light combination was beneficial for changes in gene or TF expression levels.

TFs regulate gene expression patterns by acting as activators or repressors to induce or inhibit gene promoter activity. There is a potential regulatory mechanism by which differentially expressed TFs (DETFs) may influence anthocyanin biosynthesis by adjusting the expression levels of the DEGs implicated in anthocyanin accumulation under different light qualities. To verify this hypothesis, co-expression networks were established between the TFs and genes using gene expression profile data based on the correlation coefficient. The resulting co-expression networks indicated that different TFs connected distinct genes, and there were correlations among them ( Figure 4II ). TFs (MYB, NAC, and bHLH) connected more genes related to anthocyanin accumulation, and these TFs were highly expressed in blueberry leaves under blue light.

To further validate the comparative transcriptome results, the transcript level variances of some putative genes or TFs related to anthocyanin accumulation were evaluated by qRT-PCR analysis with three biological replicates ( Figure 5 ). Structural genes, including F3H, LDOX, OMT, and UFGT, involved in the anthocyanin biosynthesis pathway exhibited similar dynamic trends, with the highest transcript levels under blue light, followed by red/blue, red, and white. The expression level of DFR was the highest under red/blue light, and the effect of inducing DFR transcripts was the least significant under white light. The transcript level for UFGT was the lowest under red/blue light, and its peak occurred under blue light. Moreover, the transcript level for the OMT was second highest under the red/blue light. These expression trends agreed with the alterations in gene expression detected by the transcriptome analysis, corroborating the RNA-sequencing data. These results coincided with the dynamic changes in anthocyanin content under different light qualities (they were highest in blue light, followed by red/blue, red, and white light), indicating that these structural genes may play a role in anthocyanin biosynthesis in blueberry leaves in response to the different spectra. The expression levels of the TFs (AP2/ERF-AP2, bZIP, NAC, and WRKY) were higher in the blue, red, and red/blue lights than in the corresponding control (white light). Moreover, a correlation analysis of the gene expression results was obtained using qRT-PCR and RNA-Seq data under different light qualities and a close connection was identified between them (correlation coefficient: R^{2 =} 0.908). The high consistency between the RNA-Seq and qRT-PCR results suggested that the RNA-Seq data were reliable for evaluating the regulation of gene expression under different light quality treatments in blueberry leaves.

To compare the metabolite compositions associated with anthocyanin accumulations in the blueberry leaves under different light conditions, the major anthocyanins were determined using LC-MS/MS. In total, 108 anthocyanidins were detected and grouped into eight categories ( Figure 6I ). The largest category (pelargonidin) contained 19 metabolites, followed by cyanidin (n=17), peonidin (n=17), delphinidin (n=16), and malvidin (n=13). The PCA for the anthocyanin derivatives was used to classify twelve samples into four distinct clusters, accurately reflecting the four light quality treatments for the blueberry leaves ( Figure 6II ). In this model, the secondary principal component (PC2; 22.66% of the total variables) was clearly separated between blue and red/blue light. The differences between the white and red light resulted from PC1 and PC2 in this model (48.09% and 22.66% of the variables, respectively). An Orthogonal Partial Least Squares-Discriminant Analysis (OPLS-DA) model was performed to identify the differentially accumulated metabolites responsible for metabolic differentiation among the various light quality treatments for the blueberry leaves. A total of 43 (25 up-accumulated; 3 down-accumulated), 41 (7 up-accumulated; 4 down-accumulated), 44 (24 up-accumulated; 3 down-accumulated), and 42 (22 up-accumulated; 1 down-accumulated) metabolites were selected for comparison analysis as follows: blue vs. white, red vs. white, red/blue vs. white, and blue vs. red lights, respectively ( Table S7 ). The hierarchical clustering and correlation analysis exhibited a compact dependence of the anthocyanin metabolites on the different light qualities of the blueberry leaves ( Figure 6III ). The overall accumulation of the anthocyanin metabolites was higher under blue light than under the other light conditions, and the accumulation of cyanidin, pelargonidin, malvidin, peonidin, and petunidin was significantly higher under the blue and red/blue lights when compared with the white and red lights.

Based on the KEGG database annotation, 44 metabolites were found to be involved in the anthocyanin biosynthesis pathway and the identified anthocyanin derivatives and their relevant compounds were rearranged based on their corresponding positions in the anthocyanin biosynthesis pathway, which were established based on the KEGG, PMN, and literature references. Among them, two naringenins (Ng), one dihydrokaempferol (Dk), one kaempferol (Ka), three flavones, nine cyanidins (Cy), four peonidins (Pn), five pelargonidins (Pg), one dihydromyricetin, eight delphinidins (Dp), two petunidins (Pt), four malvidins (Mv), and four proanthocyanidins (Pa) were identified ( Figure 7 ). Each metabolite in the anthocyanin biosynthesis pathway differed significantly under the different light conditions in the blueberry leaves. Interestingly, the content for most metabolites was highest under the blue light but lowest under the white light, suggesting that the blue light was beneficial in promoting anthocyanin accumulation in blueberry leaves. The dynamic changes in the content of metabolites in the anthocyanin biosynthesis pathway were consistent with the variations in the expression levels for the genes related to anthocyanin accumulation under the different light qualities. Among all anthocyanin derivatives, the quantification of Cy-type, Pg-type, and Mv-type derivatives were more advantageous under blue light than under other light qualities. However, the combination of blue and red light could dramatically improve the Dp-type, Pn-type, and Pt-type derivative content.

To understand the regulatory networks for the anthocyanins that were implicated in the differential distribution of anthocyanin derivatives under the different light qualities, we performed an integrated analysis of the core differentially accumulated metabolites (DAMs) and DEGs related to anthocyanin accumulation in the four comparisons (blue vs. white, red vs. white, red/blue vs. white, and blue vs. red). Based on the above results, the DEGs related to secondary metabolism, carbohydrate metabolism processes(photosynthesis), anthocyanin biosynthesis, and anthocyanin derivatives of Cy, Dp, Ng, Pg, Pa, Pn, Pt, and Mv were used to perform an interaction network based on the Pearson correlation analysis. In this interaction network, genes and metabolites were closely related, and positive and negative correlations between gene expression levels and anthocyanin accumulation content were discovered ( Figure 8 ; Table S8 ). In contrast to the comparison between red and white (28), there were much stronger correlations between DAMs and DEGs in both comparisons with blue vs. white (57) and red/blue vs. white (56). This result demonstrated that the blue light or blue/red combination may play a more prominent role in promoting the biosynthesis and accumulation of anthocyanins because there were more DAMs and DEGs related to anthocyanin formation. Because of the changes in DAM content and the DEG transcriptional level, the color and anthocyanin content of the blueberry leaves differed significantly under different light qualities. In the blue vs. white correlation network, there was a close correlation between Cy and LDOX, Mv and OMT, and Pg and UFGT levels. LDOX, OMT, and UFGT are upstream regulatory genes for Cy, Mv, and Pg synthesis in the anthocyanin pathway, and their high expression contributes to Cy, Mv, and Pg accumulation, respectively. Based on the results of the analysis of the DEG transcriptional profile and DAM metabolic profile, the expression levels of LDOX, OMT, and UFGT and the contents of the Cy, Mv, and Pg derivatives were highest under blue light. This validated the idea that blue light induced substantial accumulation of the Cy, Mv, and Pg derivatives via the promotion of LDOX, OMT, and UFGT expression, respectively, further facilitating anthocyanin synthesis. A similar phenomenon was also found in the red/blue vs. white correlation network: there were higher expression levels of DFR and OMT, greater accumulation of Dp, Pn, and Pt derivatives, and an intimate connection between DFR and Dp, OMT and Pn, and OMT and Pt under a combination of red and blue lights. DFR and OMT induced the synthesis of Dp, Pn, and Pt derivatives, leading to an increase in anthocyanin content under the red/blue light in the blueberry leaves. Additionally, in the blue vs. red correlation network, UFGT and OMT were closely connected with Pg and Pn, respectively. Among the DEG transcriptional profiles and DAM metabolic profiles with the comparison of blue vs. red, the transcript levels of UFGT and OMT and the accumulations of Pg and Pn were higher in blue when compared with red. This further indicates that UFGT and OMT play vital roles in promoting Pg and Pn accumulation, respectively.