Background

Weed control is essential for agricultural floor management in vineyards, conventional permanent bare ground by continuous soil tillage can result in soil erosion, increasing soil disturbance and lowering organic matter. The use of herbicides increases the risk of damaging the ecological balance of the vineyard [1]. Alternatively, inter-row mulching is an eco-friendly practice to inhibit weed growth via filtering out photosynthetically active radiation (PAR), and besides weed suppression, inter-row mulch can influence grapevine growth and the accumulation of metabolites in grape berries [2,3,4]. In terms of light conditions, reflective mulches such as white or silver reflective films, and glass particles can increase the reflected PAR around the cluster zone, while the black or dark mulches can absorb the incident light and decrease the reflected solar radiation around the cluster zone [2, 3]. In parallel, the cluster-zone temperature could also be altered by inter-row mulching due to its influence on reflected solar radiation.

Previous studies reported the effects of inter-row mulching on grape composition, especially in cool climate regions. In general, reflective mulch shows a tendency to promote grape ripening, and the accumulation of sugars and phenols due to the increased reflective solar radiation. Some studies reported that although reflective inter-row mulching increased reflective solar radiation, no significant influence on grape composition including total soluble solids, titratable acidity, pH, total phenols, anthocyanins, and flavonols was found [2, 3, 5]. The inconsistent effects of mulching treatments on grape composition could be attributed to the complex interactions between mulching materials, grape cultivars, soil properties, trellis system, vineyard management, and vintage climates. Therefore, the mulching strategy needs to be adjusted based on investigations on the plasticity of specific grapes to local terroir and vintage climate characteristics. So far, little information is available regarding the grape performance in dry and hot regions in response to inter-row mulching. Based on the gap above, our previous studies investigated the comprehensive effects of black geotextile inter-row mulching on Cabernet Sauvignon vine growth and grape composition in a semi-arid wine region. In addition, we undermined the potential causes of the influence of black inter-row mulching on grape composition through correlation analysis between metabolite concentrations and climatic parameters [4, 6].

The grapevine microclimate variations on metabolite accumulation in grapes could be attributed to multiple factors such as light, temperature, water status, soil nutrients status, etc. The regulation mechanism of light on metabolite accumulation has been extensively reported. Many studies revealed that sunlight exposure upregulated the expressions of key genes involved in the biosynthesis of anthocyanins, flavonols, terpenoids, and norisoprenoids, such as CHS, DRF, F3H, LDOX, UFGT, AOMT, FLS, CCD, TPS, etc [7,8,9,10]. . . As for temperature effects, moderate temperature is favorable for anthocyanin accumulation, while high temperature could decrease anthocyanins [11, 12]. Regarding volatile compounds, it was reported that relatively low temperatures and high temperatures both upregulated the expressions of VviCCDs, which led to increases in norisoprenoids in grapes [13]. As for water status, many studies investigated the effects of deficit irrigation level and timing on grape composition. Generally, moderate deficit irrigation at the green stage was considered to be favorable for berry ripening and the accumulation of flavor compounds which was not only ascribed to the decreases in berry volume but also the upregulation of relevant key structural genes [14, 15].

Extensive variation in vineyard environments leads to the phenotypic plasticity of grapevines within and across vintages. Under field conditions, the influence of inter-row mulching on grape composition was ascribed to the complex interaction of multiple factors including light, temperature, soil temperature, and soil water content, and the vintage interventions. Particularly, the two key factors, light, and temperature, are often difficult to separate under field conditions. However, inter-row mulch diminishes light in the cluster zone and increases temperature, and becomes an optimal treatment to separate these two factors [4]. Besides, the interaction effect of multiple environmental factors on the global metabolites of grapes and the underlying molecular mechanism is still limited. In this study, we conducted metabolomics and transcriptomics analyses in two vintages, aimed at providing insights into metabolic and molecular responses of Cabernet Sauvignon grapes to microclimate changes caused by black inter-row mulching in a semi-arid climate, and assisting winegrowers to adjust agricultural strategies to cope with the interactions of multi-stress field environment.

Results

The transcriptomic changes of grape berries under inter-row mulching

The overview of transcriptomic profiles of grape berries was presented in Text S1. The linear regression analysis between the expressions assessed by qRT-PCR and RNA sequencing validated that expression profiles obtained by RNA sequencing were reliable (Fig. S1).

In total, 717/485, 1736/1869 and 11/13 genes were up/down-regulated by inter-row mulching treatment at E-L 33, E-L 35.5, and E-L 38 stages respectively in 2016, and 190/144, 44/25 and 7/2 genes were up/down-regulated by inter-row mulch at E-L 33, E-L 35.5, and E-L 38 stages respectively in 2017 (Fig. S2). The numbers of DEGs between M and C at E-L 33 and E-L 35.5 stages were higher than those at E-L 38. Besides, the number of differentially expressed genes (DEGs) at E-L 33 and E-L 35.5 stages in 2016 was far higher than those in 2017.

Consistent DEGs revealed the variables of microclimate and physiology under inter-row mulching

At E-L 33 stage, the Venn plot showed that inter-row mulching significantly upregulated the expression of heat shock transcription factor HSFA6B at E-L 33 stage in two years (Fig. S3). Specifically, M upregulated VviHSF18 (VIT_204s0008g01110), VviHSF02 (VIT_200s0179g00150), and VviHSF01 (also known as VviHSFA6B, VIT_210s0597g00050) in two vintages (Fig. 1). Consistently, 37 heat shock proteins in total presented higher level in M than C grapes in both vintages. Further correlation analysis on VviHSP and VviHSF showed that VviHSF02 and VviHSF18 had the strongest correlation with VviHSP. In terms of 12 downregulated genes, transcription factor TSRF1 involved in the ethylene-mediated signaling pathway, as well as the photosystem II protein D2 involved in photosynthesis were negatively affected by M in both vintages (Fig. S3). At veraison, M significantly upregulated 23 genes in two years (Fig. S4), including genes related to photosynthesis, and genes encoding NADPH: protochlorophyllide oxidoreductase in chlorophyll biosynthesis.

Fig. 1
figure 1

Effects of inter-row mulching on expression profiles of genes related to heat shock protein, and correlation analysis between heat shock transcription factors and heat shock protein. Data of gene expression heatmap represents the log2-transformed fold changes of gene expressions between inter-row mulching and control group. The black border of the heatmap cell represents significant differences in gene expressions between the inter-row mulching and control group (student’s t-test, p < 0.05). The color of each cell (blue to red) in the correlation heatmap represents the correlation coefficient (-1 to 1). The same as below

Full size image

The further k-means and KEGG analysis were conducted based on DEGs identified in each vintage respectively, which were described in Text S2. Overall, based on the DEGs in two years and the microclimate we previously reported [4], we concluded that genes related to photosynthesis and heat shock proteins in grapes at E-L 33 and 35.5 stages were upregulated by M, while C grapes experienced higher light exposure and severe drought stress. Besides, the influences of gene expression in grapes were highly dependent on vintages, and thus limited consistent DEGs were identified in two vintages.

The profiles of sugars and organic acids revealed the ripening difference under inter-row mulching

To investigate the effects of inter-row mulching on grape composition, untargeted metabolome analysis was conducted. In this study, M slightly inhibited the accumulation of glucose, fructose, and sucrose at E-L 33, but accelerated sugar accumulation at veraison (Table 1). At harvest, most sugars presented at a lower level than C grapes, especially in 2017. M also inhibited most of the organic acids at E-L 33, especially the citric acid and malic acid in 2017, however at harvest, M grapes presented higher levels of organic acids in both vintages.

Table 1 Log2 fold change of relative contents of sugars and organic acids
Full size table

Grapes of inter-row mulching accumulated more amino acids

Regarding the amino acids involved in glutamate metabolism (Fig. S5), M showed a tendency to increase the concentrations of glutamate, glutamine, ornithine, arginine, and proline at veraison and harvest. In particular, the concentrations of ornithine and arginine in M grapes were significantly higher than in C grapes at veraison in 2016, and so were the glutamate at veraison in 2017, and proline at veraison in both vintages. In amino acids of the shikimate pathway, M tended to increase the concentrations of tryptophan and tyrosine at E-L 33 stage. At veraison, M slightly decreased the tryptophan concentrations, whereas significantly increased phenylalanine concentrations. At harvest, M showed a tendency to increase the concentrations of tryptophan and phenylalanine. As for aspartate metabolism, M had no significant influence on the concentrations of asparagine, aspartate, and lysine at veraison and harvest in two vintages.

Fig. 2
figure 2

Effects of inter-row mulching on expression profiles of genes related to amino acid metabolism

Full size image

At the transcription level, M significantly upregulated the expressions of VviGS, VviASSY, and VviAOT at veraison in 2016 and showed a similar upregulation tendency in 2017 (Fig. 2). In terms of aromatic amino acids, the 3-deoxy-D-arabinheptanoic acid-7-phosphate (DHAP) synthase (DHAPS) gene VviDHAPS, VviSDH, and 2 shikimate kinase genes were upregulated by M at veraison in 2016, most of them also presented at higher level than C grapes in 2017 (Fig. 2). Besides, M significantly downregulated VviCM at veraison in 2016 and showed a similar tendency in 2017.

The inter-row mulching downregulated genes related to the synthesis, modification, and transport of phenylpropanoid metabolism

In metabolites of the phenylpropanoid pathway, five stilbenes were identified in grapes, including piceid, astringin, piceatannol, resveratrol, and ε-viniferin (Fig. 3). Most stilbenes were enhanced by M at E-L 33 stage, whereas they presented at a lower level at harvest. Notably, M showed a tendency to decrease ε-viniferin concentrations at E-L 33 and 38 stages in 2016 and at three stages in 2017. These results suggested that M inhibited the accumulation of stilbenes in grapes post veraison. Regarding flavonoids, M decreased the concentrations of kaempferol-3-O-glucoside, quercitrin, and syringing at veraison and harvest, while having little influence on flavan-3-ols. In terms of anthocyanins, M significantly decreased malvidin-3-O-glucoside concentrations at harvest in 2016, and it also showed a tendency to decrease the concentrations of acetylated anthocyanins. However, M had little influence on anthocyanins in grapes in 2017.

Fig. 3
figure 3

Effects of inter-row mulching on metabolites in phenylpropanoid pathway

Full size image

At the transcriptomic level. M showed downregulation tendency for 2 VviPAL genes (VIT_216s0039g01110, VIT_216s0039g01300) at E-L 33 stage, whereas upregulated 3 other VviPAL genes (VIT_208s0040g01710, VIT_211s0016g01520, VIT_211s0016g01660) at veraison (Fig. 4). M upregulated VviC4H expressions in 2016, and they also showed a similar tendency at E-L 33 stage in 2017. However, Vvi4CL, VviCHS, and VviCHI presented at a lower level in E-L 33, and/or E-L 35.5 in two years. Stilbene synthase (STS) is a key enzyme for the biosynthesis of stilbenes. Notably, STSs were negatively affected by M in most of the phenology stages, however, it presented a significant higher level than C grapes at veraison in 2016.

Fig. 4
figure 4

Effects of inter-row mulching on expression profiles of genes in phenylpropanoid metabolism

Full size image

In the flavonoid pathway, M downregulated 5 VviF3’5’H genes in 2016, with the most significant influence observed at veraison (Fig. 4). In flavonol synthesis, M significantly downregulated 3 VviF3H genes, VviFLS4 (VIT_218s0001g03470), and VviGT6 at E-L 35.5 in 2016. In anthocyanin metabolism, M significantly upregulated the expression of VviLDOX in 2016, two genes encoding UFGT at E-L 35.5 in 2016, but significantly downregulated their expression in 2017. Anthocyanin acyltransferase (3AT) controls the biosynthesis of stable acylated anthocyanins. In this study, M significantly downregulated Vvi3AT at veraison in 2016.

Regarding the transportation of the anthocyanins, M downregulated the genes encoding AnthoMATE2, VviABCC1, and VviGST 5 at veraison or at harvest (Fig. 4). The expressions of VviGST4 and VviGST5 in grapes were relatively low at E-L 33 stage while drastically increased at veraison (FPKM range in 300–600), and slightly declined at harvest. However, M had no significant influence on VviGST4 expressions, but the expressions of VviGST5 in M grapes were significantly lower than in C grapes at harvest in 2016.

For the transcription factors reported, M significantly upregulated the expressions of MYB family, including VviMYB14, VviMYB15, VviMYBF1, VviMYB4A, VviMYB4B, VviMYBC2-L1, VviMYBC2-L3, and VviWRKY03 at E-L 33 and/or veraison in two vintages (Fig. 4). In addition, M tended to downregulate VviWRKY24 and VviHY5 at veraison and E-L 33 stage in 2016, respectively.

The inter-row mulching inhibited MEP synthesis of isoprenoid metabolites and limited glycosylation process

In the terpene biosynthesis pathway, M showed a tendency to decrease (E,E)-geranyl linalool concentrations which was significantly lower in M grapes than in C grapes at harvest in 2016 (Fig. 5). Carotenoid metabolism relates to the terpene biosynthesis pathway via geranylgeranyl diphosphate (GGPP), which is a precursor of norisoprenoids. In this study, M significantly increased GGPP concentrations at veraison, while decreasing its concentrations at harvest. In terms of other metabolites in carotenoid metabolism, M decreased β-carotene levels in 2016, while the opposite results were observed in 2017. As for xanthoxin, M tended to decrease its level at E-L 33 stage, especially since its level in M grapes was significantly lower than in C grapes in 2017, while M had little influence on xanthoxin at veraison and harvest.

Fig. 5
figure 5

Effects of inter-row mulching on metabolites in terpenoids biosynthesis, carotenoid metabolism, and lipoxygenase pathway

Full size image
Fig. 6
figure 6

Effects of inter-row mulching on expression profiles of genes in terpenoid biosynthesis pathway. Expression profiles of genes related to carotenoid metabolism are shown in Fig. S6

Full size image

At the transcriptomic level, M upregulated a gene encoding DXS (VIT_205s0020g02130) at 35.5 stages in both vintages (Fig. 6), whereas VviDXR and VviGPPS were significantly downregulated by M at veraison in 2016. In terms of terpene synthases. At E-L 33 stage, M downregulated three VviTPS-b/g genes in 2016, and 4 were also downregulated in both vintages at E-L 35.5. M also downregulated a VviTPS-a gene at E-L 33 and 35.5 stages in 2016, and at E-L 35.5 and 38 stages in 2017. In grapes, a large fraction of terpenoids are present as nonvolatile terpene glycosides, M showed a tendency to downregulate VviGT14 and VviGT15 at E-L 33 and 35.5 stages in 2016, particularly since the expressions of VviGT14 in M grapes at veraison in 2016 were significantly lower than those in C grapes. In 2017, M moderately downregulated VviGT14 at veraison, and VviGT15 at E-L 33 and 38 stages.

In carotenoid metabolism, M showed a tendency to upregulate VviGGPPS at three developmental stages in 2016 and 2017 (Fig. 6), as well as VviLECY (Fig. S6). ABA biosynthesis pathway is a downstream flux of carotenoid metabolism, M downregulated 2 genes (VIT_210s0003g03750, VIT_219s0093g00550) encoding 9-cis-epoxycarotenoiddioxygenase (NCED) at E-L 33 and E-L 35.5 in 2016. However, in 2017, M showed a tendency to upregulate three NCED genes (VIT_210s0003g03750, VIT_219s0093g00550, VIT_205s0051g00670).

The biosynthesis of norisoprenoids is another downstream flux of carotenoid metabolism, in which carotenoid cleavage dioxygenases (CCD) catalyze the cleavage of carotenoids at certain positions to release norisoprenoids. M significantly upregulated VviCCD1 (VIT_213s0064g00840) and moderately upregulated VviCCD4a (VIT_202s0087g00910) at veraison in 2016 (Fig. S6).

ABA as a plant hormone, is a key endogenous messenger in responses to abiotic stress. In this study, M significantly upregulated 2 ABA receptor VviPYR/PYL genes (Fig. S7), while it significantly downregulated 17 type 2 C protein (PP2C) VviPP2C genes at veraison in 2016. Besides, the expressions of 13 out of 17 VviPP2C genes were lower in M grapes than in C grapes in 2017.

The inter-row mulching enhanced the C6 alcohols by regulating alcohol dehydrogenases in the lipoxygenase-hydroperoxide lyase (LOX-HPL) pathway

In terms of the metabolites of the LOX-HPL pathway, M enhanced the accumulation of α-linolenic acid, linoleic acid, and 13(S)-HPOT at E-L 35.5 in both vintages and in E-L 33 of 2016 (Fig. 5). However, at harvest, the difference of these precursors was weakened. Regarding the responsible genes, M had no consistent effect on the expressions of VviLOXs and VviHPLs in two vintages, however, most of them presented at higher transcription levels at harvest. C6 and C9 aldehydes can be reduced to alcohol by alcohol dehydrogenases (ADH), M upregulated most of the VviADHs genes at E-L 33 and 35.5 stages in 2016 and at E-L 35.5 and 38 stages in 2017 (Fig. S8).

Identification of co-expressed transcriptional regulators and modules associated with black inter-row mulching responsive metabolites

In this study, we applied WGCNA analysis to identify the gene modules closely related to the black inter-row mulching responsive metabolites, the correlation analysis was conducted based on eigenvectors of gene modules and metabolite concentrations (Fig. S9). The absolute coefficient ≥ 0.8 and p value < 0.05 were defined as highly significant correlation, and the absolute coefficient ranging from 0.4 to 0.8, and p value < 0.05 was defined as moderate correlation. The turquoise module was highly correlated with flavonoids and had moderate correlations with sucrose, glutamate, ornithine, arginine, terpenes, and C6/C9 compounds. Brown module was highly correlated with glutamate and showed moderate correlations with ornithine, arginine, phenylalanine, ABA, and terpenes. The yellow module was highly correlated with arginine and had moderate correlations with ornithine, phenylalanine, ABA, terpenes, and C6/C9 compounds. The green module was moderately correlated with sucrose, glutamate, and flavonoids. The red module had moderate correlations with sucrose, flavonoids, and C6/C9 compounds. The black module was moderately correlated with glutamate, ornithine, arginine, and terpenes. The Magenta module had moderate correlations with glutamate, phenylalanine, ε-viniferin, ABA, and norisoprenoids. Besides correlation analysis, we summarized the distribution of differentially expressed genes involved in the interested pathways as shown in Fig. S9.

The hub genes of each module with absolute module membership greater than 0.8 and absolute kME ranking within the top 10% were selected. Those hub genes annotated as transcription factors were regarded as the key regulators to the accumulation of black inter-row mulching responsive metabolites such as flavonoids, terpenes, norisoprenoids, etc. The size and KEGG enrichment analysis for each identified module, the correlation analysis based on eigenvectors of gene modules, and metabolite concentrations were described in Text S3. Based on the above, we summarized the biological significance of each module as follows: the turquoise module was related to photosynthesis, chlorophyll metabolism, glutamate metabolism, flavonoid metabolism, terpene biosynthesis, and LOX-HPL pathway; blue module was mainly related to sucrose metabolism and flavonoid metabolism; brown, yellow, red and magenta modules had close association with glutamate metabolism, LOX-HPL pathway, the biosynthesis of heat shock proteins, and the biosynthesis of stilbenes, respectively.

According to the gene connectivity within each module, we selected hub genes, especially hub transcription factors (hub TFs). In the turquoise module, eleven hub TFs were consistently affected by M in two vintages (Fig. 7), four of them annotated as bZIP11 (VIT_218s0001g13040), SPL2 (VIT_201s0010g03710), and bZIP53 (VIT_207s0005g01450) were downregulated by M at E-L 33 stage, another four hub TFs annotated as Homeobox-leucien zipper protein HB13 (VIT_201s0026g01950), DOF5.3 (VIT_206s0004g03420), MYB141 (VIT_214s0108g01080) were upregulated by M at veraison, and GATA11 was upregulated by M at E-L 33 and 35.5 stages. In the turquoise module, the DEGs co-expressed with VvibZIP11 and VvibZIP53 included VviAOT (VIT_219s0027g01790), Besides, the co-expressed DEGs with VviGATA11 included genes involved in chlorophyll biosynthesis such as VviGLTS, VviCHLD, VviCHLM (Fig. 8).

Fig. 7
figure 7

Effects of inter-row mulching on differentially expressed hub transcription factors of WGCNA modules

Full size image

In the red module, 2 hub TFs, annotated as zinc finger family transcription factor (VIT_204s0008g04480) and heat shock transcription factor HSFA6B (VIT_204s0008g01110) respectively, were identified as hub TFs, and they were downregulated by M at E-L 33 and 35.5 stages (Fig. 7). In addition, a total of 36 heat shock protein genes were included in the red module, and they were highly co-expressed with HSFA6B. It was noted that VviADH3 (VIT_218g0001g15450) involved in the LOX-HPL pathway was included red module, and it was co-expressed with VviHSF6B (Fig. 8).

Fig. 8
figure 8

Co-expression networks of transcription factor GATA 11 (A), HSFA6B (B), and WRKY03 (C)

Full size image

In the magenta module, 15 out of 33 genes were annotated as stilbene synthase genes (VviSTSs). Besides, the eigenvector of the magenta module was moderately correlated with ε-viniferin concentrations. Two hub TFs, annotated as zinc finger family transcription factor (VIT_2120028g02530) and WRKY03 (VIT_201s0010g0390) respectively, were identified in the magenta module (Fig. 7), and were upregulated at E-L 33 stages and were downregulated by M at veraison.

Discussion

The transcriptomic profiles and DEGs

Under the same screening condition, there were far more DEGs in 2016 than in 2017, this confirmed that the influence of inter-row mulching on gene transcription levels was highly dependent on vintage. In the present study, vintage 2017 was characterized as stronger light compared to vintage 2016 [4], so the cluster-zone light changes caused by inter-row mulching might had less impact on the gene expression of grapes in 2017 and thus resulted in fewer DEGs in 2017.

Under high temperatures, heat shock transcription factors (HSFs) are recognized as the principal regulator of the heat shock response. In this study, M upregulated the expression of 37 genes encoding HSPs, and 9 genes encoding HSFs were also significantly regulated, supporting that M grapes were subjected to heat stress (Fig. 1). The results of the correlation analysis were consistent with previous studies that VviHSF01, VviHSF05, VviHSF15, and VviHSF18 may be key transcription factors that regulate VviHSP expression under high-temperature stress conditions.

Among the twelve down-regulated genes at E-L 33 stage, transcription factor TSRF1 (VIT_207s005g03230) could respond to ethylene and bind to the GCC box in the promoters of genes related to pathogens and diseases, and further induce these genes’ expression [16]. In addition, TSRF1 could increase the tolerance of rice to osmotic stress and drought stress [17]. Previous study has confirmed that inter-row mulching decreased the evaporation of soil water [2]. Besides, M grapes were more easily subjected to drought stress compared to C grapes, which could result in higher expressions of VviTSRF1 in M grapes than in C grapes. Whether lower ABA levels in M grapes than in C grapes were related to the downregulation of VviTSRF1 in M grapes needs to be further studied.

At veraison, genes related to photosynthesis and genes encoding NADPH were upregulated, besides, the expression of VviFECH (VIT_207s0031g03200 and VIT_204s0008g00800) in 2016 veraison, and VviCLH (VIT_207s0151g00130,) in 2016 E-L 33 was downregulated by M. These results suggested that the more protoporphyrin IX shift to the Mg branch of chlorophyll synthesis, and the degradation of chlorophyll was inhibited. Overall, our results proved that M grapes promoted chlorophyll biosynthesis under more abundant chlorophyll precursors, whereas inhibited the degradation of chlorophyll at E-L 33 and 35.5 stages, which could lead to higher levels of chlorophyll in M grapes than in C grapes.

In summary, the transcriptomic profiles suggested that black inter-row mulching treatment promoted the photosynthesis of grape berries as a compensatory effect after light reduction and suffered heat stress. The biological responses of C grapes revealed the growing conditions of winegrapes in northwest regions and consistently modifying multiple metabolic pathways in grape berries.

Primary metabolites and transcription pathway

The level of sugar and organic acids strongly correlated with the ripening process of grape berries. The efflux efficiency of sucrose reflected photosynthesis rate, and the attenuated solar reflection caused by M led to decreases in photosynthesis rate and further resulted in less accumulation of photo assimilate such as sucrose. This indicated that the up-regulated chlorophyll synthesis did not reverse the decrease in photo assimilate due to the attenuated light, which in turn led to lower concentrations of fructose and glucose at E-L 33, as well as retarded the accumulation of organic acids (Table 1). Higher temperatures observed in M at veraison were beneficial for the sugar accumulation process, however, the lower concentration of TCA intermediates, such as aconitate, fumaric acid, and succinic acid, indicated inhibition of the TCA cycle and resulted in lower organic acids at E-L 35.5. At harvest, the lower sugar and higher acids suggested that M grapes were less ripe than C grapes. Although the elevated temperatures accelerated the accumulation of grape sugars in vintage with weaker light (2016), in years with adequate temperatures, the absence of light resulted in slower ripening. This implied a key role of sunlight rather than temperature in the ripening process of grape berries in semi-arid regions.

Amino acids are the main nitrogen compounds in grape berries. In this study, the underground soil under black inter-row mulching contained higher total nitrogen or efficient nitrogen concentrations than controls at different depths [6], which could favor grapevines of M to assimilate more nitrogen than controls and thus lead to more accumulation of glutamate than controls at veraison and harvest (Fig. S5). Attributed to the much greater supply of precursor (glutamate), the concentrations of glutamine, proline, ornithine, and arginine, derived from the glutamate metabolism pathway, were higher in M berries than in C berries, which were also confirmed by the upregulation of VviGS, VviAOT, and VviASSY in M berries (Fig. 2). In terms of aromatic amino acids, chorismate mutase (CM) controls the flow direction of chorismic acid to the biosynthesis branch of phenylalanine or tyrosine. The upregulation of VviSDH and VviSK caused by M could also result in more accumulation of chorismic acid in M berries, and consequently lead to a higher concentration of phenylalanine in M berries than in C berries, albeit the downregulation of VviCM by M.

Phenolic metabolites in phenylpropanoid metabolism

Based on the metabolome analysis, the stilbenoid ε-viniferin was identified as the sensitive metabolite in response to M treatment. To be specific, M decreased ε-viniferin concentrations at E-L 33 stage in 2016 and 2017, and at E-L 35.5 stage in 2017, whereas increased its concentrations at E-L 35.5 stage in 2016, which was consistent with the influence of M on VviSTSs expressions as mentioned above (Fig. 4). These results confirmed that VviSTS played a key role in stilbenes biosynthesis. Therefore, we attributed the lower concentrations of ε-viniferin in M grapes than in controls to the downregulation of M on VviSTS expressions.

Phenylalanine lyase (PAL) is an entry enzyme of phenylpropanoid metabolism, controlling the conversion of phenylalanine to trans-cinnamic acid. It was noted that the influence of M on the expressions of 2 VviPAL genes (VIT_211s0016g01520, VIT_211s0016g01660) was in parallel with the differences in phenylalanine concentrations between M and C berries (Fig. 4, Fig. S5). In grapes, five flavonol synthase genes (VviFLS1-5) have been identified, previous studies reported that the expressions of VviFLS2, VviFLS4, and VviFLS5 showed downward trends before veraison [18]. The expressions of VviFLS4 and VviFLS5 drastically increased in parallel with the accumulation of flavonols during ripening, suggesting these two genes played key roles in flavonol biosynthesis in berries in the current study. Besides, previous studies revealed that light exposure affected flavonol concentrations in grapes mainly by regulating VviFLS4 expressions [18]. In this study, M significantly downregulated VviFLS4 (VIT_218s0001g03470) expression at veraison in 2016, whereas showed an up-regulation tendency in 2017 despite no statistical significance. GT5 (uridine diphosphate (UDP)-glucuronic acid: flavonol-3-O-glucuronosyltransferase) and GT6 (bifunctional UDP-glucose/UDP-galactose: flavonol-3-O-glucosyltransferase/galactosyltransferase) have been identified as glycoside transferase genes (GT5 and GT6) in grapes which are responsible for the glycosylation of flavonols [19]. In the current study, the downregulation of VviGT6 maybe responsible for the decrease of flavonols in M grapes.

Flavonoids are primarily synthesized in the cytoplasm and then transported to vacuoles. As for the transport and trafficking of flavonoids, the putative mechanisms include vesicle trafficking, tonoplast transport, and the involvement of glutathione S-transferase (GST) in the vacuolar sequestration of anthocyanins [20]. Previous studies confirmed that Multidrug And Toxic Extrusion (MATE) transporters are responsible for the vacuolar transport of acylated anthocyanins in grapes [21]. In this study, M significantly downregulated a gene encoding MATE2 transporters especially at veraison in 2016, besides, the expression of this gene in M grapes was also moderately lower than in C grapes at E-L 35.5 in 2017 (Fig. 4). Tonoplast transport of anthocyanins can also be mediated by ATP-binding cassette (ABC) transporters. Previous studies suggested that an ABC protein, ABCC1, transports malvidin 3-O-glucoside, and ABCC1 is expressed in exocarp throughout berry ripening with a significant increase at veraison [22]. In this study, M significantly downregulated VviABCC1 (VIT_216s0050g02480) at veraison in 2016, and it showed similar effects in 2017. In grapes, five GST genes (VviGST1-5) have been identified. A previous study found that VviGST1 and VviGST4 were positively correlated with anthocyanin accumulation [23]. Another study demonstrated that VviGST3 played a key role in proanthocyanin accumulation, VviGST4 was relevant to the transport of anthocyanins in skins and proanthocyanidin in skins and seeds, VviGST1 was involved in the transport of proanthocyanidin and/or flavonols in green berries [24]. In this study, the expressions of VviGST1 were quite low, especially at veraison and harvest, the expressions of VviGST2 were relatively high and peaked at E-L 33 stage. According to the dynamic changes of flavonoids in grapes, we speculated that VviGST2 might play a key role in the accumulation of proanthocyanidin and flavonols. However, M did not significantly affect the expressions of VviGST2 at E-L 33 stage. The expressions of VviGST4 and VviGST5 in grapes drastically increased at veraison (FPKM range in 300–600), and slightly declined at harvest, which was consistent with the changes of anthocyanins concentration. Also, the lower expression of VviGST5 in M grapes was consistent with anthocyanin concentrations. So, our results showed that GST5 in grapes might be involved in the transport and accumulation of anthocyanin, and its function needed to be further studied. To sum up, M downregulated several key genes involved in flavonoid metabolism in 2016, including VviF3’5’H, VviGT6, VviFLS4, VviUFGT, VviAOMT, VviMATE2, VviABCC1, which could result in less anthocyanins accumulated in grapes. However, in 2017, a vintage with strong light, M only moderately downregulated VviGT6, VviUFGT, VviMATE2, and VviABCC1 in 2017, which indicated that the expression of flavonoid biosynthesis is more likely to be light-driven.

The transcription factors involved in the phenylpropanoid pathway have been widely studied. As for the regulation of stilbenes biosynthesis in grapevines, MYB14 and MYB15 have been demonstrated to specifically activate the promoters of STS genes, VviWRKY24 could function as a singular effector to activate the VviSTS29 promoter, VviWRKY03 could act through a combinatorial effect with VviMYB14 [25,26,Quantitative real-time PCR

Six candidate genes were selected for validation of transcript quantification of RNA-seq data by quantitative real-time PCR (qRT-PCR). Firstly, total RNA isolation was the same as the procedure mentioned above. A total of 1 µg RNA was used for reverse transcription reaction using a HiScript IIQ RT SuperMix for qPCR + gDNA wiper kit (Vazyme, China) following the manufacturer’s instructions. qRT-PCR was performed using a ChamQ Universal SYBR qPCR Master Mix kit (Vazyme, China) in a Bio-Rad CFX96 Real-Time PCR system (Bio-Rad, USA). To be specific, each qRT-PCR reaction (20 µL) contains 2 µL of cDNA template, 0.4 µL of 10 mM forward primer, 0.4 µL of 10 mM reverse primer, 7.2 µL of ddH2O, and 10 µL of 2 × ChamQ Universal SYBR qPCR Mix solution. The cycling conditions were as follows: 95 ℃ for 30 s, followed by 40 cycles of 95 ℃ for 10 s, 60 ℃ for 30 s. The gene-specific primers used for qRT-PCR were listed in Table S1, and VviUbiquitin 1 was applied as the reference gene.

Statistical analysis

A one-way analysis of variance (ANOVA) was conducted using “agricolae” package in R platform (version 3.6.1) to evaluate the statistical differences of metabolites between different treatments employing student’s t-test at p < 0.05. Principal component analysis (PCA) and orthogonal projection to latent structures discriminant analysis (OPLS-DA) were performed using SIMCA 14.1 (Umetrics, Umea, Sweden). Differentially expressed genes (DEGs) were identified using “DEseq2” package in R, the criteria of fold change ≥ 1.5 and false discovery rate (FDR) ≤ 0.01 were set as the thresholds for significant differential expression. Weighted gene co-expression network analysis (WGCNA) was conducted using ‘WGCNA’ package in R, and the resulting network connections between the identified hub genes and the interested DEGS were visualized by Cytoscape 3.5.1 (http://cytoscape.org) using a circular layout.