Anthocyanin Accumulation Provides Protection against High Light Stress While Reducing Photosynthesis in Apple Leaves

The photoprotective role of anthocyanin remains controversial. In this study, we explored the effects of anthocyanin on photosynthesis and photoprotection using transgenic ‘Galaxy Gala’ apple plants overexpressing MdMYB10 under high light stress. The overexpression of MdMYB10 dramatically enhanced leaf anthocyanin accumulation, allowing more visible light to be absorbed, particularly in the green region. However, through post-transcriptional regulation, anthocyanin accumulation lowered leaf photosynthesis in both photochemical reaction and CO2 fixation capacities. Anthocyanin accumulation also led to a decreased de-epoxidation state of the xanthophyll cycle and antioxidant capacities, but this is most likely a response to the light-shielding effect of anthocyanin, as indicated by a higher chlorophyll concentration and lower chlorophyll a/b ratio. Under laboratory conditions when detached leaves lost carbon fixation capacity due to the limitation of CO2 supply, the photoinhibition of detached transgenic red leaves was less severe under strong white, green, or blue light, but it became more severe in response to strong red light compared with that of the wild type. In field conditions when photosynthesis was performed normally in both green and transgenic red leaves, the degree of photoinhibition was comparable between transgenic red leaves and wild type leaves, but it was less severe in transgenic young shoot bark compared with the wild type. Taken together, these data show that anthocyanin protects plants from high light stress by absorbing excessive visible light despite reducing photosynthesis.

Anthocyanin has an absorption spectrum in visible light, especially in the green light region. Like red and blue light, green light may also efficiently drive photosynthesis, especially at a high photosynthetic photon flux density [5,6]. Therefore, it is generally accepted that epidermal anthocyanin may protect plants from photoinhibition/photodamage by lowering the light

Effects of Anthocyanin Accumulation on Photosynthesis
Dark respiratory rate was comparable between WT and transgenic red leaves (Figu 2A). When the photon flux density (PFD) was higher than 150 μmol m −2 s −1 , the net ph tosynthetic rate was significantly lower in transgenic red leaves compared with WT leav (Figure 2A). Under saturated PFD, the photosynthetic rate was approximately 70% transgenic compared with WT plants. Apparent quantum yield was also lower  Table S1, with three biological replicates; Panel C, five biological replicates were performed for each point. For the statistical analysis, "**" above the bars indicates a significant difference at p < 0.01, t-test. CHS, chalcone synthase; CHI, chalcone isomerase; F3H, flavanone 3-hydroxylase; F3 H, flavonoid 3 -hydroxylase; DFR, dihydroflavonol 4-reductase; ANS, anthocyanin synthase; UFGT, UDPglycose: flavonoid glycosyltransferase; GST, glutathione S-transferase.

Effects of Anthocyanin Accumulation on Photosynthesis
Dark respiratory rate was comparable between WT and transgenic red leaves (Figure 2A). When the photon flux density (PFD) was higher than 150 µmol m −2 s −1 , the net photosynthetic rate was significantly lower in transgenic red leaves compared with WT leaves (Figure 2A). Under saturated PFD, the photosynthetic rate was approximately 70% in transgenic compared with WT plants. Apparent quantum yield was also lower in transgenic red leaves. With the increase in intercellular CO 2 concentration, photosynthetic rates increased in transgenic and WT leaves ( Figure 2B). When the intercellular CO 2 concentration was approximately 500 µmol mol −1 , the photosynthetic rates reached the maximum in transgenic and WT leaves. Apparent carboxylation efficiency was significantly lower in transgenic compared with WT plants.
Based on the leaf reflection spectra, MdMYB10 overexpression leaves showed more absorption in visible light, especially in green light (550-600 nm, Figure 3A). Stomatal phenotypes were similar between transgenic and WT plants ( Figure S4). The synthesis of anthocyanin increased the contents of chlorophylls, especially chlorophyll b. As a result, the chlorophyll a/b ratio decreased in transgenic red leaves ( Figure 3B). transgenic red leaves. With the increase in intercellular CO2 concentration, photosynthetic rates increased in transgenic and WT leaves ( Figure 2B). When the intercellular CO2 concentration was approximately 500 μmol mol −1 , the photosynthetic rates reached the maximum in transgenic and WT leaves. Apparent carboxylation efficiency was significantly lower in transgenic compared with WT plants. Based on the leaf reflection spectra, MdMYB10 overexpression leaves showed more absorption in visible light, especially in green light (550-600 nm, Figure 3A). Stomatal phenotypes were similar between transgenic and WT plants ( Figure S4). The synthesis of anthocyanin increased the contents of chlorophylls, especially chlorophyll b. As a result, the chlorophyll a/b ratio decreased in transgenic red leaves ( Figure 3B). The chlorophyll a fluorescence transient and the 820 nm modulated reflection curves were also altered by the accumulation of anthocyanin ( Figure 4). Compared with WT, transgenic leaves showed higher fluorescence signals from the O to I step but a slower decrease in the 820 nm reflection from MRo to MRmin, indicating that the photosynthetic electron transport capacity between the two photosystems was lower in transgenic red leaves [29].
Regarding the differentially expressed genes between transgenic and WT leaves, KEGG enrichment analysis revealed that many pathways were affected by anthocyanin synthesis, including primary and secondary metabolism, transcription, translation, membrane transport, and signal transduction, among others (Table S2). However, none of the gene involved in photosynthesis showed different expression levels between transgenic and WT leaves (Table S1). Interestingly, the activities of Rubisco, GAPDH, PRK, A6PR and SPS, key enzymes involved in photosynthesis, were significantly lower in MdMYB10 overexpression leaves ( Figure 5; Table S5). To further assay the effects of anthocyanin accumulation on photosynthesis, a targeted metabolomics assay was performed via LC-MS/MS to quantify the changes in metabolites directly involved in the Calvin cycle. The produced photosynthates, including sorbitol, sucrose, and intermediate products such as 3-phosphoglycerate, sorbitol-6-phosphate, sucrose-6-phosphate, adenosine diphosphate glucose, ribulose-1,5-bisphosphate, and others, were all decreased in transgenic red leaves compared with WT leaves ( Figure 5; Table S4). Thus, the decreased contents of these metabolites further confirmed that the overall photosynthetic capacity was impaired in the transgenic lines. The chlorophyll a fluorescence transient and the 820 nm modulated reflection curves were also altered by the accumulation of anthocyanin ( Figure 4). Compared with WT, transgenic leaves showed higher fluorescence signals from the O to I step but a slower decrease in the 820 nm reflection from MRo to MRmin, indicating that the photosynthetic electron transport capacity between the two photosystems was lower in transgenic red leaves [29].  Regarding the differentially expressed genes between transgenic and WT leaves, KEGG enrichment analysis revealed that many pathways were affected by anthocyanin synthesis, including primary and secondary metabolism, transcription, translation, membrane transport, and signal transduction, among others (Table S2). However, none of the gene involved in photosynthesis showed different expression levels between transgenic and WT leaves (Table S1). Interestingly, the activities of Rubisco, GAPDH, PRK, A6PR  phenylalanine, serine, and alanine, whereas the contents of glutamine and asparagine were significantly higher in transgenic red leaves. Other amino acids, such as methionine, valine, arginine, tyrosine, threonine, glutamate, ornithine, and aspartate, showed comparable contents in WT and transgenic leaves.  (Table S1), only genes involved in respiration and other carbon and nitrogen metabolism were shown in Figure 5. For the expression of each gene, please see Supplemental Tables S1 and S3.
Concerning the expression of genes involved in respiration and other carbohydrate metabolism, 23 genes were upregulated, and 11 were downregulated ( Figure 5, Table S3). With respect to the levels of genes involved in amino acid metabolism, 15 genes were upregulated, and 8 were downregulated. For metabolites involved in carbon metabolism, the contents of 15 kinds of non-structural carbohydrates and organic acids were decreased, including 2-phosphoglycerate, aconitate, 2-oxoglutarate, malate, myo-inositol, glucose, ribose, erythritol, phosphoenolpyruvate, glycerate, xylitol, isocitrate, shikimate, citrate, and succinate, while those of the other 12 kinds of metabolites, such as xylulose, pyruvate, maltose, galactose, maltitol, fumarate, rhamnose, quinate, maleic acid, fructose, or xylose, remained unchanged in transgenic red leaves compared with WT leaves ( Figure 5, Table S4). For nitrogen metabolism, nine kinds of free amino acids had reduced contents, including histidine, glycine, proline, isoleucine, γ-aminobutyrate, leucine, phenylalanine, serine, and alanine, whereas the contents of glutamine and asparagine were significantly higher in transgenic red leaves. Other amino acids, such as methionine, valine, arginine, tyrosine, threonine, glutamate, ornithine, and aspartate, showed comparable contents in WT and transgenic leaves.

Effects of Anthocyanin Accumulation on Photoprotection
Without high light treatment, the accumulation of anthocyanin did not affect the xanthophyll cycle pool size and its de-epoxidation ( Figure 6). Upon exposure to strong red or green light, the pool size remained unchanged, whereas the de-epoxidation level was significantly lower in transgenic red leaves compared with WT leaves. For the transcriptome assay, three biological replicates were performed; for the enzyme activities and metabolomics assay, five biological replicates were conducted. For the statistical analysis, "*" or "**" indicates a significant difference at p < 0.05 or p < 0.01, respectively, t-test.

Effects of Anthocyanin Accumulation on Photoprotection
Without high light treatment, the accumulation of anthocyanin did not affect the xanthophyll cycle pool size and its de-epoxidation ( Figure 6). Upon exposure to strong red or green light, the pool size remained unchanged, whereas the de-epoxidation level was significantly lower in transgenic red leaves compared with WT leaves. after strong red or green light treatment. Panel A, the inserted heat maps with blocks from left to right are the values of the log 2 fold change of the xanthophyll cycle pool size of WT, L8, and L9 before and after strong red (R) or green (G) light treatment, respectively. Five biological replicates were performed for each point; Panel B, '0 h', 'R3h' and 'G3h' represented before (0 h) and after strong green (G3h) or red (R3h) light treatment for 3 h, respectively. For the statistical analysis, "*" or "**" indicates a significant difference at p < 0.05 or p < 0.01, respectively, t-test.
The accumulation of anthocyanin did not affect the activities of APX, MDHAR, and DHAR ( Figure 7). However, the activities of SOD and GR and the contents of ascorbate and glutathione were decreased in transgenic compared with WT leaves. Under strong red or green light treatment, the activities of antioxidant enzymes and the contents of ascorbate and glutathione changed in similar patterns in WT and transgenic red leaves, except that the activity of MDHAR showed a greater decrease in transgenic leaves following exposure to red light. represented before (0 h) and afte strong green (G3h) or red (R3h) light treatment for 3 h, respectively. For the statistical analysis, "* or "**" indicates a significant difference at p < 0.05 or p < 0.01, respectively, t-test.
The accumulation of anthocyanin did not affect the activities of APX, MDHAR, an DHAR ( Figure 7). However, the activities of SOD and GR and the contents of ascorbat and glutathione were decreased in transgenic compared with WT leaves. Under stron red or green light treatment, the activities of antioxidant enzymes and the contents o ascorbate and glutathione changed in similar patterns in WT and transgenic red leave except that the activity of MDHAR showed a greater decrease in transgenic leaves follow ing exposure to red light. The inserted heat maps with blocks from left to right are the values of the log2 fold chang in enzyme activity or metabolite content of WT, L8, and L9 before and after strong red (R) or gree (G) light treatment, respectively. Five biological replicates were performed for each point. SOD, su peroxide dismutase; APX, ascorbate peroxidase; MDHAR, monodehydroascorbate reductas DHAR, dehydroascorbate reductase; GR, glutathione reductase. Five biological replicates were pe formed for each point. For the statistical analysis, "*" or "**" indicates a significant difference at p 0.05 or p < 0.01, respectively, t-test.
The stomata of apple leaves are mainly distributed on the abaxial side. In this study when treated under different high light stresses, the abaxial surface of apple leaves wa closely attached to wet gauze. In this way, the stomata were sealed with water and th carbon fixation capacity of detached leaves was almost fully inhibited due to the limitatio of CO2 supply (Under this condition, the Fv/Fm values decreased in similar patterns i transgenic red leaves and WT leaves with strong UV-B and UV-A irradiation ( Figure 8A However, the values of Fv/Fm decreased at a slower rate in transgenic red leave (WT) and its MdMYB10 overexpression transgenic lines (L8, L9) before and after strong red or green light treatment. The inserted heat maps with blocks from left to right are the values of the log 2 fold change in enzyme activity or metabolite content of WT, L8, and L9 before and after strong red (R) or green (G) light treatment, respectively. Five biological replicates were performed for each point. SOD, superoxide dismutase; APX, ascorbate peroxidase; MDHAR, monodehydroascorbate reductase; DHAR, dehydroascorbate reductase; GR, glutathione reductase. Five biological replicates were performed for each point. For the statistical analysis, "*" or "**" indicates a significant difference at p < 0.05 or p < 0.01, respectively, t-test.
The stomata of apple leaves are mainly distributed on the abaxial side. In this study, when treated under different high light stresses, the abaxial surface of apple leaves was closely attached to wet gauze. In this way, the stomata were sealed with water and the carbon fixation capacity of detached leaves was almost fully inhibited due to the limitation of CO 2 supply (Under this condition, the Fv/Fm values decreased in similar patterns in transgenic red leaves and WT leaves with strong UV-B and UV-A irradiation ( Figure 8A). However, the values of Fv/Fm decreased at a slower rate in transgenic red leaves compared with WT leaves upon exposure to strong white, green, or blue light ( Figure 8A and Figure S5). Moreover, the values of Fv/Fm decreased at a faster rate in red leaves following exposure to strong red light.
In field conditions, the values of Fv/Fm were slightly but significantly higher in red leaves than in WT leaves in the morning ( Figure 8B). However, the values became comparable between WT and transgenic leaves in the afternoon. For young branch barks, the values of Fv/Fm were higher in transgenic red barks than in WT in the morning ( Figure  8C). In the afternoon, the values of Fv/Fm remained unchanged in transgenic red barks but decreased significantly in WT.  Figure S5. Five biological replicates were performed for each point, with two leaves or branches from one plant assayed for each replicate. For the statistical analysis, different lowercase letters below symbols or above bars indicate a significant difference at p < 0.01, LSD.
In field conditions, the values of Fv/Fm were slightly but significantly higher in red leaves than in WT leaves in the morning ( Figure 8B). However, the values became comparable between WT and transgenic leaves in the afternoon. For young branch barks, the values of Fv/Fm were higher in transgenic red barks than in WT in the morning ( Figure 8C). In the afternoon, the values of Fv/Fm remained unchanged in transgenic red barks but decreased significantly in WT.

Discussion
The findings of the present study clearly showed that anthocyanin played two opposing roles in photoprotection under high light stress. On the one hand, anthocyanin accumulation may shade chloroplasts by absorbing visible light, especially green light ( Figure 3A), causing leaves to absorb less light energy and thus having a positive effect on photoprotection. On the other hand, the accumulation of anthocyanin could also decrease photosynthesis and the thermal dissipation dependent on the xanthophyll cycle, especially the former (Figures 2 and 4-6), the major pathways in the consumption of absorbed light energy. This phenomenon had a negative effect on photoprotection. In field conditions when photosynthesis performed normally in both green and transgenic red leaves, the positive and negative effects of anthocyanin on photoprotection might be comparable. The photoinhibition degree was then similar in green and red leaves ( Figure 8B). Under laboratory conditions, as carbon fixation capacity was almost fully inhibited in detached leaves due to the limitation of CO 2 supply, the positive effect of anthocyanin was far greater than the negative effect. Consequently, light attenuation by anthocyanin could alleviate the degree of photoinhibition under white, green, or blue light stress conditions ( Figure 8A). Under red light stress without any light attenuation by anthocyanin, the lower thermal dissipation in red leaves ( Figure 6) might partly account for the more severe photoinhibition in transgenic red leaves in contrast to WT ( Figure 8A). The decreased thermal dissipation was most likely a response to the light-shielding effect of anthocyanin, as indicated by higher chlorophyll concentration and lower chlorophyll a/b ratio (Figure 3). Young branch bark contains chlorophylls and has a normal photochemical reaction capacity. However, the carbon fixation capacity of branch bark is very low similar to other non-leaf green tissues. Under field conditions, the observation that exposed young branch barks of Md-MYB10 overexpression plants lacked any photoinhibition in contrast to WT ( Figure 8C) also supports the two opposing roles of anthocyanin in photoprotection under high light stress.
Compared with that in their green counterpart, the reduced photosynthesis in transgenic red leaves might be related to the lower photochemical reaction and CO 2 fixation capacities (Figures 2, 4 and 5). A stomatal effect could be excluded, as the phenotypes of stomata were similar between the transgenic and WT leaves ( Figure S4). In previous studies, it has been reported that green light might regulate gene expression in plants [12,13]. Moreover, anthocyanin content can also alter the gene expression in plants by attenuating blue light [11]. While purple leaves display an overexpression of genes that promote chlorophyll biosynthesis and light harvesting, genes involved in the stability/repair of photosystems were substantial overexpressed in green leaves [11]. Compared with that in WT, the changes in the expression of more than 1000 genes in transgenic red leaves (Table  S1) could be mainly attributed to the visible light attenuation by anthocyanin. Moreover, the comparable expression of genes involved in photosynthesis between transgenic and WT leaves (Table S1) indicated that anthocyanin accumulation might decrease photosynthesis in apple leaves by regulating photosynthetic genes at the post-transcriptional level. Indeed, the KEGG enrichment analysis revealed that expressions of transcription, translation, membrane transport, and signal transduction genes were changed in the transgenic red leaves (Table S3). The lower production of photosynthates resulted in lower contents of carbon and nitrogen metabolites in transgenic red leaves ( Figure 5), as both carbon and nitrogen metabolism depend on the carbon skeleton supply from photosynthates. The greater accumulation of glutamine and asparagine in transgenic red leaves ( Figure 5) further support this notion, as glutamine and asparagine have a relatively lower carbon/nitrogen ratio compared with glutamate and aspartate. Previous studies have also found that anthocyanin accumulation increases the asparagine content in pear peel [30].
Using Arabidopsis and mutants deficient in anthocyanin synthesis, it has been reported that the light attenuation role of anthocyanin is more important than its antioxidant role in photoprotection under long-term high light treatment [10]. In the present study, although anthocyanin accumulation decreased the antioxidant capacity in apple leaves (Figure 7), it might not be involved in ROS scavenging in vivo under high light stress, as the antioxidant systems changed in similar patterns in transgenic red leaves and WT leaves after green or red high light treatment (Figure 7). This is a reasonable supposition as the major production of ROS occurs at the chloroplast under high light stress [19], while anthocyanin is mainly stored in vacuoles.
In previous studies, flavonoids were shown to protect plants from strong UV irradiation stress [31,32]. In the present study, however, the accumulated anthocyanin neither changed the leaf reflection spectrum in the UV range ( Figure 3) nor protected the leaves from strong UV stress under either field or laboratory conditions (Figure 8). We speculate that several reasons might account for these observations. First, the different results were attributed to the different compounds, plants, or UV treatments. Second, in addition to anthocyanin, the changes in other metabolites ( Figure 5) might also affect the absorption of UV light. Third, we used Fv/Fm, an indicator for the status of the photosynthetic apparatus, especially photosystem II [29], to reflect the UV damage to leaves in this study. However, the chloroplast might not be the most sensitive organelle in plants in response to UV stress. Thus, the UV stress damage might not be reflected by the Fv/Fm values over time. Further studies are needed to clarify this phenomenon.
In general, anthocyanin may protect plants from high light stress by absorbing excessive visible light despite reducing photosynthesis. However, anthocyanin cannot protect plants from high UV light stress.

Plant Materials and Treatments
A transcription factor, MdMYB10, which has been demonstrated to regulate the synthesis of anthocyanin in apple plants [2], was overexpressed in 'Galaxy Gala' (WT) to obtain the plants with anthocyanin accumulation. Transformation of 'Galaxy Gala' was performed as previously described [33,34]. Briefly, 'Galaxy Gala' shoots were subcultured on proliferation medium [35] at 4-week intervals, and were then transferred to leaf expansion medium (MS supplemented with 0.1 mg/L naphthyleneacetic acid and 8.3 mg/L 6-(γ,γ-dimethylallylamino) purine 4 weeks before transformation. The youngest unfolded leaf was exercised from in vitro shoots, and the leaf was wounded using a pair of nontraumatic forceps [36] and co-cultivated with EHA105 (pCH32) harboring the binary vector pSAK277-MdMYB10 [2]. After co-cultivation, the leaves were washed with distilled water containing half-strength MS salt mixture and cefotaxim sodium (500 mg/L) to remove excess inoculum and placed on regeneration medium containing 100 ppm kanamycin. The regenerated shoots were propagated in vitro. Wood buds of two transgenic lines (L8 and L9) and wild type (WT) were grafted onto M9-T337 rootstock and grown in the field for 2 years. During the growth period, no blossoms were found on any of the trees. The trees were subjected to standard horticultural management with pest and disease control.
For the high light treatment with artificial light sources, fully expanded leaves were sampled in the morning, punched into leaf discs (diameter 1.5 cm). The abaxial surface of leaf discs was closely attached to four layers of wet gauze, and then treated under high light for 0, 1, 2, and 3 h at room temperature. The UV-B, UV-A, blue, green, and red lights were provided by custom-made LED-light sources, with a wavelength of approximately 280, 370, 450, 520, and 670 nm, respectively ( Figure S1). A custom-made light source with mixed red, green, and blue LED-light beads was considered as white light. During the treatments, the light intensities used to irradiate leaf discs were approximately 90 µW cm −2 for UV-B, 3000 µW cm −2 for UV-A, and 2000 µmol m −2 s −1 for visible light. For each treatment at each sampling time, five replicates were used for the chlorophyll a fluorescence transient and 820 nm modulated reflection measurements with M-PEA (Hansatech, Norfolk, UK) as described by Gao et al. [29]. The wavelength of the red light source of M-PEA is 627 ± 10 nm, which is not absorbed by anthocyanin, thereby excluding any possible interference of anthocyanin in the measurements [37]. The maximum quantum yield of photosystem II (Fv/Fm) was calculated as Fv/Fm = (Fm − Fo)/Fm, where Fm and Fo are the maximum and minimum chlorophyll fluorescence intensity, respectively. Before performing the measurements, leaves were dark-adapted for 1 h. The leaf samples were also collected and stored at −80 • C for biochemical analysis. Fully expanded leaves and new young branches were also collected from the southern canopy of the plants in the field at 8:00 in the morning or 14:00 in the afternoon, transferred to the laboratory, and then immediately used for measurements of chlorophyll a fluorescence transients after dark adaption. During plant growth, the maximum solar UV-B, UV-A, and visible light intensity were 60 µW cm −2 , 1100 µW cm −2 , and 1800 µmol m −2 s −1 , respectively, on sunny days.

Metabolite Content and Enzyme Activity Analysis
Anthocyanin, chlorophyll, and xanthophyll cycle pigments were extracted and assayed as described by Zhang et al. [38]. A 20A HPLC (Shimadzu, Kyoto, Japan) was used for the anthocyanin and xanthophyll cycle pigment content assay, and a UV-2450 spectrophotometer (Shimadzu) was used for the chlorophyll content determination. Isolated cyanidin-3-O-galactoside with a purity over 98% at different concentrations was used to generate a standard curve to calculate the concentration of anthocyanin in plants.
Non-structural carbohydrates, organic acids, and free amino acids were assayed as described by Li et al. [39], using an Agilent 7890A GC/5975C MS (Agilent Technology, Palo Alto, CA) and an Agilent 1100 HPLC equipped with a fluorescence detector (Agilent Technology), respectively. Sugar phosphates and other highly polar intermediates were determined on an AB Sciex QTRAP 4000 Mass Spectrometer (Applied Biosystems/MDS Sciex, Waltham, MA) linked to a Shimadzu HPLC according to Ma et al. [40].

Photosynthesis, Leaf Reflection Spectra, and Stomatal Phenotype Assay
On sunny days in June, light-response and CO 2 -reponse curves of photosynthetic rates were measured from 9:00 to 11:00 a.m. using a LI-6800 portable photosynthesis system (Li-Cor Biosciences, Lincoln, NE, USA). For light-response curves measurement, the photosynthetic irradiation light (red:blue = 90:10) was provided by the light source of the LI-6800, changing the intensity from high to low in the following sequence: 1800, 1400, 1000, 750, 500, 250, 150, 100, and 50 µmol m −2 s −1 . Respiration was measured without any irradiation of the leaves. Apparent quantum yield was calculated from the initial slope of the light-response curves. The CO 2 concentration was set as 400 µmol mol −1 . For CO 2 -response curves measurement, the CO 2 concentrations were changed in the following sequence: 400, 300, 200, 100, 400, 550, 700, 900, 1200, and 1400 µmol mol −1 . The light intensity was set at 1600 µmol m −2 s −1 . Fully expanded leaves were kept in each intensity or CO 2 concentration for at least 3 min until the photosynthetic rates became stable. Leaf reflection spectra were assayed using a UV-2600 spectrophotometer equipped with an IS-2600 plus integrated sphere (Shimadzu). The phenotypic assay of stomata was carried out as described by Li et al. [42] using an Olympus BX-63 microscope (Olympus Corp., Tokyo, Japan) with cellSens Entry software (Tokyo, Japan).

RNA-Seq Analysis
Total RNA was isolated from fully expanded leaves using the SDS-phenol method [43] and subjected to Novogene in Beijing (https://en.novogene.com (accessed on 3 December 2019)) for library construction and RNA sequencing with an Illumina NovaSeq 6000. The clean data for each sample were mapped to the GDDH13 genome [44] with Hisat2, and the corresponding read counts were analyzed by Deseq2 to select differentially expressed genes. The resulting p values were adjusted using Benjamini and Hochberg's approach for controlling the false discovery rate, padj < 0.05 and the absolute value of log 2 (fold change) ≥ 1 were set as the threshold for significantly different expression. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways of differentially expressed genes were determined using the clusterProfiler R package at the website https://www.kegg.jp/kegg/pathway (accessed on 15 March 2020).

Statistical Analysis
All data are presented as means ± SE and were analyzed with the t-test or least significant difference (LSD) test using SPSS 16.0 software (SPSS, Chicago, IL, USA). Heatmap representations were produced using the TBtools [45].