Transcriptomic and metabolic analysis uncovers the role of light quality in carotenoid accumulation of grapefruit during ripening

23 Light, a crucial environmental signal, is involved in the regulation of secondary 24 metabolites. To understand the mechanism by which light influences carotenoid 25 metabolism, grapefruits were bagged with four types of light-transmitting bags that 26 altered the transmission of solar light. We showed that light-transmitting bagging 27 induced changes in carotenoid metabolism during fruit ripening. Compared with natural 28 light, red light (RL)-transmittance treatments significantly increased the total 29 carotenoid content by 142%. Based on weighted gene co-expression network analysis 30 (WGCNA), ‘red’, ‘darkred’, ‘yellow’, ‘brown’ and ‘midnightblue’ modules were 31 remarkably associated with carotenoid metabolism under different light treatment. 32 Transcriptome analysis identified the transcription factors (TFs) bHLH74/91/122, 33 NAC56/78/90/100, MYB/MYB308, WRKY7/55, MADS29/AGL61, ERF043/118 as 34 being involved in the regulation of carotenoid metabolism in response to RL. Under RL 35 treatment, these TFs regulated the accumulation of carotenoids by directly modulating 36 the expression of carotenogenic genes, including PSY , Z-ISO2 , ZDS6 , LCYB , LCYE , 37 CHYB , CCD1-1/1-3 , CCD4-2 and NCED2/3 . Based on these results, a network of the 38 regulation of carotenoid metabolism by light in citrus fruits was preliminarily proposed. 39 These results showed that RL treatments have great potential to improve coloration and 40 nutritional quality of citrus fruits. 41 42 up- down-regulation level for genes in

represent an ideal material for investigating carotenoid metabolism. 72 The pathway of carotenoid biosynthesis has been clearly established in plants 8 . 73 The five-carbon prenyl diphosphate isopentenyl diphosphate (IPP) and its double-bond respectively, and this step is the pivotal branch point in carotenoid metabolism. Next, 85 α-carotene is converted into lutein by β-ring hydroxylase (CYP97A) and ε-ring 86 hydroxylase (CYP97C) of the cytochrome P450 family. The production of zeaxanthin 87 from β-carotene is catalyzed by β-carotene hydroxylase (CHYB), and violaxanthin is 88 generated via antheraxanthin by zeaxanthin epoxidase (ZEP). The cleavage of 89 carotenoids is catalyzed by the proteins of carotenoid-cleavage genes (CCD or NCED), 90 producing apocarotenoids such as β-ionone, β-citraurin, and ABA 1,3 .  As an effective method of protecting fruit from insect infestations, bird attack, and 111 sunburn as well as reducing disease incidence rate and chemical residues, fruit bagging

128
Effects of light transmittance on TSS, TA, and CCI during fruit ripening 129 Compared with that in the dark shade treatment (DS), the TSS content of the grapefruits 130 treated with RL, BL, and WL gradually increased during fruit ripening and were 131 significantly higher than that in DS treatment at 215 DAB (p < 0.05) (Fig. 1A). The 132 effect of light treatment on TA content is different (Fig. 1B) were gradually increased in all light treatments (Fig. 1C).
Effects of light transmittance on carotenoid accumulation during fruit ripening 138 Five major carotenoids were identified from 'Huoyan' grapefruit pulp, including β-139 carotene, phytofluene, ζ-carotene, lycopene, and 9-cis-violaxanthin ( Table S1). The 140 carotenoid profiles differed in the four light treatments, and the 'Huoyan' grapefruit 141 pulp was rich in β-carotene and lycopene (Fig. 2). Compared with the control group 142 (DS), the total carotenoid content was the highest in the grapefruits treated with RL   Table S2). The alignment of the clean reads against the reference 153 genome and reference gene sequences generated a total of 25,694 unigenes (Table S3). 154 In four light treatments, the median of gene expression level ranged from 0.76 to 0.88 155 and there were differences between the samples (Fig. S2). The fruit samples in BL  Weighted gene co-expression network analysis (WGCNA) 173 The WGCNA was performed using 12619 unigenes (FPKM > 1, the top 50% of 174 variance), which were classified into twenty-four modules (Fig. 4), of which the 'red', In the carotenoid metabolic pathway (Fig. 5a), a total of eight structural genes, expressed in response to RL during ripening (Table S4)  (2), MADS (2) and GRF (1), were identified as candidate TFs modulating carotenoid 202 biosynthesis in response to RL during fruit ripening (Fig. 5B, Table S5). regulatory genes in co-expression network (Fig. 6). In addition, we found that thirty-214 two TFs, including ERF, WRKY, bHLH, NAC and MYB family members were co-215 expressed with carotenoid cleavage gene CCD4-2 and six TFs had a co-expression 216 relationship with CYP97C1 (Table S6)  induced by red-transmittance bagging treatments (Fig. 2), suggesting that RL played  CCD1-3 (Fig. 2 and 5), which was of PSY and CHYB (Fig. 2, 5 and 7), which accounted for higher phytofluene level, and  (Fig. 2, 5 and 7). By contrast, the significantly reduced expression of AGL61 for 285 'darkred' module during ripening suggested their negative correlation with phytofluene 286 accumulation in grapefruit ( Fig. 5 and 7). Recently, PpERF3 has been shown to be 287 involved in ABA biosynthesis by activating PpNCED2/3 transcription during peach 288 fruit ripening 39 . Here, PpERF3 homolog ERF012 was down-regulated in response to 289 RL, which suggested ERF012 were highly likely to be involved in carotenoid process 290 mediated by RL. On the contrary, RL remarkably promoted transcript of ERF043/118 291 14 and NCED2/3 shared similar expression patterns with them ( Fig.5 and 7). Above 292 analysis indicated ERF TFs differently respond to RL and collaboratively regulated 293 carotenoid accumulation. In the Arabidopsis, suppression of AtRAP2.2 leads to 294 reduction of PSY and PDS transcript 40 . In rice leaves, AP2/ERF genes were negatively 295 associated with carotenoid accumulation under both blue-and red-light treatments 41 . 296 Here, we also found multiple ERF TFs (ERF023-like/025/026) in 'brown' module 297 displayed negative correlation with phytofluence accumulation in response to RL.

298
Another fruit ripening related TF NACs were also reported to be involved in with transcript for these genes, consistent with lycopene accumulation in ripening 313 grapefruit fruit (Fig. 2, 5 and 7; Fig. S4). and further regulated carotenoid biosynthesis 19 . In 'darkred' module, we found that 319 MYB308 induced by RL was also positively correlation with carotenoid accumulation, 320 especially lycopene, during grapefruit ripening ( Fig. 4 and 5). Additionally, seven 321 WRKY TFs were differentially expressed in response to RL during grapefruit ripening 322 (Fig. 5). In Osmanthus fragrans, OfWRKY3 was found to be a positive regulator of the 323 OfCCD4 gene via binding to its W-box palindrome motif 47 . In this study, we also 324 observed that two WRKY TFs, namely WRKY7/55 were gradually down-regulated as 325 grapefruit fruit ripening, accompanied by the reduction of CCD4-2 expression of in RL 326 treatment (Fig. 2, 5 and 7). Besides, RL also notably suppressed expression for C3H

327
(2), bZIP (4) and Dof (2) TFs in 'brown' module, suggested these TFs might involve in 328 carotenoid accumulation (Fig. 5). was established (Fig. 7). These findings not only provide new insight into the regulation 338 of carotenoid metabolism, but also offer an effective approach for enhancing the quality 339 of citrus fruits in agricultural practice. light-transmitting bags (WL), and a dark-shading bag (DS) was as the control (Fig. S1). 352 Fruits of a uniform size were picked at 185 (maturation) and 215 (fully ripe) days after 353 blossom (DAB). Each fifteen fruits were as one replicate and three biological replicates