Acetylome reprograming participates in the establishment of fruit metabolism during polyploidization in citrus

Abstract

Polyploidization leads to novel phenotypes and is a major force in evolution. However, the relationship between the evolution of new traits and variations in the post-translational modifications (PTM) of proteins during polyploidization has not been studied. Acetylation of lysine residues is a common protein PTM that plays a critical regulatory role in central metabolism. To test whether changes in metabolism in citrus fruit is associated with the reprogramming of lysine acetylation (Kac) in non-histone proteins during allotetraploidization, we performed a global acetylome analysis of fruits from a synthetic allotetraploid citrus and its diploid parents. A total of 4,175 Kac sites were identified on 1,640 proteins involved in a wide range of fruit traits. In the allotetraploid, parental dominance (i.e. resemblance to one of the two parents) in specific fruit traits, such as fruit acidity and flavonol metabolism, was highly associated with parental Kac level dominance in pertinent enzymes. This association is due to Kac-mediated regulation of enzyme activity. Moreover, protein Kac probably contributes to the discordance between the transcriptomic and proteomic variations during allotetraploidization. The acetylome reprogramming can be partially explained by the expression pattern of several lysine deacetylases (KDACs). Overexpression of silent information regulator 2 (CgSRT2) and histone deacetylase 8 (CgHDA8) diverted metabolic flux from primary metabolism to secondary metabolism and partially restored a metabolic status to the allotetraploid, which expressed attenuated levels of CgSRT2 and CgHDA8. Additionally, KDAC inhibitor treatment greatly altered metabolism in citrus fruit. Collectively, these findings reveal the important role of acetylome reprogramming in trait evolution during polyploidization.

Acetylome reprogramming plays an important role in trait evolution during polyploidization through regulation of enzyme activity and protein abundance.

Introduction

Polyploidization is not only a major force in plant evolution (Otto and Whitton, 2000; Comai, 2005) but also has great potential for crop improvement (Allario et al., 2013; Yu et al., 2021). The majority of flowering plants have suffered from at least one round of genome duplication during their evolution. Polyploids are widespread in major crops including wheat (Triticum aestivum), rapeseed (Brassica napus), cotton (Gossypium hirsutum L.), potato (Solanum tuberosum), banana (Musa acuminata), kiwifruit (Actinidia deliciosa), persimmon (Diospyros kaki), and strawberry (Fragaria vesca).

Efforts have been devoted to understanding mechanisms that underpin the new traits that appear during polyploidization, especially the genetic and epigenetic variations in massive polyploids (Xiong et al., 2011; Jiang et al., 2021). Autopolyploidization causes few genomic changes. In contrast, a much greater chance of detectable genetic variations, such as chromosomal rearrangements and DNA sequence elimination, occur during allopolyploidization. Varying degrees of epigenetic changes, such as DNA methylation and histone modifications, can be detected during both autopolyploidization and allopolyploidization (Ding and Chen, 2018; Liu et al., 2021). To date, few studies provide evidence that these genetic and epigenetic variations are directly responsible for the new traits that appear during polyploidization. Another central focus of these efforts has aimed at understanding the regulation of gene expression and protein abundance in polyploids. However, the variations in gene expression and protein abundance have never provided a good explanation for the phenotypic variations that occur during polyploidization. For instance, few genes and proteins are differentially expressed in polyploids relative to their parents (Allario et al., 2011; Tan et al., 2015; Vu et al., 2017). Meanwhile, several studies compared the proteomes of polyploids and their parents and revealed that the proteomes of polyploids might not always mirror their transcriptomes (Marmagne et al., 2010; Yan et al., 2017; Li et al., 2020). These data are consistent with an important role for the post-translational regulation of proteins in the establishment of new traits during polyploidization. Whether polyploidization influences post-translational modifications (PTM) of proteins has remained an open question.

Glycosylation, phosphorylation, acetylation, and methylation regulate the structure and function of proteins (Macek et al., 2019). In plants, PMTs are important for growth, development, and adaptation to the environment (Millar et al., 2019; Møller et al., 2020). The acetylation of lysine residues is a reversible and widespread PTM that was discovered in histones (Gershey et al., 1968) and was demonstrated to regulate both chromatin structure and gene expression (Kurdistani and Grunstein, 2003). Recently, many proteomic studies have shown that the majority of acetylation events occur on non-histone proteins and contribute to key cellular processes (Narita et al., 2019). Acetylation can affect protein stability, enzymatic activity, protein–protein interactions, protein–DNA interactions, subcellular localization, and crosstalk with other PTMs (Millar et al., 2019; Narita et al., 2019).

Citrus is one of the most important fruit crops that is grown worldwide because of its great economic value and its positive influence on human health. Citrus fruits provide an important source of vitamin C (VC), carotenoids, phenolic acids, and flavonoids to the human diet. Although citrus cultivars and its relatives are mainly diploids, the production of polyploids is useful for citrus breeding because polyploidization in citrus leads to desirable and novel traits, such as increased fruit size, altered fruit color and quality, and enhanced abiotic and biotic tolerance (Bassene et al., 2010; Ruiz et al., 2018). However, the mechanisms responsible for these novel fruit traits in polyploid citrus are largely unknown.

In this study, we determined whether there is a relationship between the establishment of fruit traits and variations in the acetylation of lysine residues during allotetraploidization in citrus. We tested this idea using a synthetic citrus allotetraploid named “NH” that was derived from a diploid tangelo named “Nova” (Citrus clementina hort. × [Citrus paradisi Macfad. × Citrus reticulata Blanco]), and a diploid pummelo named “HBP” (Citrus grandis Osbeck) (Grosser et al., 1998). We used mass spectrometry to investigate the acetylomes of the allotetraploid NH and its diploid parents. In the citrus allotetraploid, parental dominance (i.e. the resemblance to only one of the two parents) in specific fruit traits, such as fruit acidity and flavonol metabolism, was highly associated with the parental Kac levels of particular enzymes and the Kac-mediated regulation of enzyme activity. Moreover, metabolic profiling of the transgenic citrus callus that overexpressed silent information regulator 2 (CgSRT2) and histone deacetylase 8 (CgHDA8) and lysine deacetylase (KDAC) inhibitor-treated citrus fruit pulp revealed an important role for KDACs in the regulation of fruit metabolism. The acetylated proteins that we identified contribute to a wide range of citrus fruit traits and include numerous well-studied proteins. Thus, the data reported here will facilitate research on the post-translational regulation and function of proteins associated with fruit traits.

Results

Morphological characterization and fruit quality evaluation in the “NH” allotetraploid and its diploid parents

As diploid cultivars, the “Nova” tangelo and “HBP” pummelo have distinct fruit traits (Figure 1 and Supplemental Figure S1). To improve the traits of citrus fruit, the two cultivars were selected as parents and used to produce an allotetraploid using the protoplast fusion approach (Grosser et al., 1998; Figure 1, A and B). The allotetraploid derived from “Nova” and “HBP” was named “NH” and although “NH” was found to harbor nuclear genomes from both “Nova” and “HBP,” “NH” inherited its chloroplast and mitochondrial genomes from “HBP” (Supplemental Figure S1A). The fruit morphology of the allotetraploid is completely different from its parents. For example, the fruit size was intermediate, juice vesicle was round, and peel and flesh colors were both light yellow (Supplemental Figure S1B; Figures 1C and 2A). The key ripening indicators of total sugar content (TSS, %) and total acid content (TA, %) indicated that the fruit from the NH allotetraploid and its parents ripened during a similar period of time (Supplemental Figure S1C).

Figure 1.

Formation and fruit quality evaluation of the synthetic “NH” allotetraploid. A, Diagrammatic illustration for the formation of the NH allotetraploid. Two diploid cultivars, “Nova” tangelo and “HBP” pummelo, were selected as parents to produce the allotetraploid (Nova+HBP, named “NH”) using the protoplast fusion approach. B, Chromosome numbers of the NH allotetraploid and its diploid parents. Chromosomes were stained with DAPI (blue). Scale bar = 5 μm. C, Fruits from the three genotypes. Scale bar = 5 cm. D, Total sugar, acid, VC (ascorbic acid), total carotenoid, and flavonoid levels in fruits (210 DAF) from the NH allotetraploid and its diploid parents. Total soluble solids (TSS, %) represents total sugar. Titratable acidity (TA, %) represents total acidity. E and F, Major sugar (E) and organic acid (F) levels in fruits from the NH allotetraploid and its diploid parents. DW, dry weight. FW, fresh weight. Bar graphs represent means±sd (n = 3). Different letters indicate significant differences (P < 0.05, one-way ANOVA with Tukey’s post hoc test).

Figure 2.

Characterization of the lysine acetylome in citrus fruit. A, Citrus fruit pulp (juice sacs) (180 DAF) used for acetylome analysis. Scale bar = 1 cm. B, Western blot analysis of lysine acetylated proteins in fruit pulp using an anti-acetyl-Lys (anti-Kac) antibody (left). A Coomassie blue-stained blot is shown as loading control (right). C, Number of lysine acetylation (Kac) sites and acetylated proteins from citrus and two published datasets from Arabidopsis (leaves from 5-week-old plants; data from Hartl et al., 2017) and rice (leaves of 14-day-old seedlings; data from Zhou et al., 2018). D, Comparison of acetylation sites per acetylated protein among citrus, Arabidopsis, and rice (Kruskal–Wallis test, ***P < 0.001). E, Comparison of acetylation sites per acetylated protein in different metabolic pathways between citrus and Arabidopsis (Kolmogorov–Smirnov test, ns, not significant; *P < 0.05; **P < 0.01). Proteins were classified based on MapMan categories. The median is shown with a line inside each box, the lower and upper hinges correspond to the first and third quartiles and the whiskers indicate the 10th and 90th percentiles. Data points are jittered to avoid excessive overlay of data points. F, Distribution of predicted subcellular localizations for proteins with acetylated lysine residues in citrus fruit. G, Venn diagrams showing overlaps among homologous acetylated proteins from citrus, Arabidopsis, and rice. H, KEGG enrichment analysis of citrus-specific acetylated proteins (Fisher’s exact test, P < 0.05). Number of proteins is indicated in parentheses.

The fruit quality indicators of the NH allotetraploid at the fully mature stage (i.e. 210 days after flowering, DAF), such as fruit acidity, VC, total carotenoid, and flavonoid contents, were mostly transgressive (Figure 1D and Supplemental Figure S1C). The total sugar content of the HBP parent was dominant in the NH allotetraploid (Figure 1D). However, we observed a different trend for the levels of the major sugars (i.e. sucrose, fructose, and glucose) in the allotetraploid relative to the HBP parent. Only the fructose content of the HBP parent was dominant in the allotetraploid (Figure 1E). Citric acid, malic acid, and quininic acid are major organic acids in citrus fruit, and citric acid accounts for >90% of the organic acids that accumulate in citrus fruit. Consistent with the trend in total acidity, the levels of these major organic acids were transgressive in NH allotetraploid (Figure 1F). The levels of major amino acids showed different trends in the NH allotetraploid, with transgressively increased levels of serine and aspartic acid (Supplemental Figure S1D).

Lysine acetylome in citrus fruit

To explore the relationship between the establishment of fruit quality and the reprogramming of protein lysine acetylation in the synthetic allotetraploid NH, we performed a mass spectrometry-based proteomics analysis to quantify the abundance of proteins and levels of lysine acetylation (Kac) at specific acetylation sites in fruit pulp from the allotetraploid and its diploid parents. Juice vesicles were collected at 180 DAF from the three genotypes and protein was extracted from the isolated juice vesicles (Figure 2A). Substantial acetylation was detected in the proteins extracted from the three genotypes using an anti-Kac antibody (Figure 2B). The protein extracts were subjected to a large-scale MS-based proteomic analysis combined with tandem mass tag (TMT) protein labeling. The acetylomes obtained from the two biological replicates were highly correlated, indicating good reproducibility (Supplemental Figure S2A). In total, 4,115 peptides containing acetylated lysine residues were identified, which corresponded to 4,175 specific Kac sites in 1,640 protein groups (Figure 2C and Supplemental Table S1) in 25.5% of the proteins that were detected. We found more Kac sites and acetylated proteins in citrus and a higher number of Kac sites per protein than were reported in the acetylomes from the leaves of two model plants, Arabidopsis (Arabidopsis thaliana) and rice (Oryza sativa) (Hartl et al., 2017; Zhou et al., 2018; Figure 2, D and E). After we classified these acetylated proteins into different pathways, we found that citrus fruit had significantly more Kac sites per protein in specific pathways including glycolysis, amino acid metabolism, the glutathione–ascorbate cycle, and secondary metabolism (Figure 2E). These acetylated proteins mostly localized to the cytoplasm (32%), plastid (31%), and nucleus (18%) (Figure 2F).

Proteins that contribute to primary metabolism, such as carbohydrate, energy, and amino acid metabolisms, were overrepresented among these acetylated proteins in citrus fruit (Supplemental Figure S2B). Importantly, many of these acetylated proteins were associated with citrus fruit quality traits including the accumulation of sugar and acidity (e.g. citric acid), VC biosynthesis, flavonoid and carotenoid metabolism, terpenoid biosynthesis, fruit ripening, disease resistance, and storage quality (Table 1).

Table 1.

List of acetylated proteins involved in important citrus fruit traits

Pathway	Gene name
Sugar accumulation	Sucrose-phosphate synthase 1 (SPSA1), sucrose synthase 3 (SUS1, SUS3), tonoplast monosaccharide transporter (TMT2), invertase (INV3, INV4), and fructose-1,6-bisphosphatase (FBP)
Fruit acidity	Biosynthesis related: PEPC (PEPC1, PEPC2, and PEPC3) and CS (CS1).
	Utilization related: ACO (ACO1, ACO2, and ACO3), malate dehydrogenase (MDH), phosphoenolpyruvate carboxykinase (PEPCK1), 2-oxoglutarate dehydrogenase (2-OGDH), fumarate hydratase (FUM1), GAD (GAD1, GAD2), isocitrate dehydrogenase (NADP-IDH1, NADP-IDH3, NAD-IDH1, and NAD-IDH3), succinate dehydrogenase (SDH1-1, SDH2-2, and SDH5), succinate-semialdehyde dehydrogenase (SSADH), ATP-CS (ACLb), fructose 1,6-bisphosphatase (FBPase2), gamma aminobutyrate transaminase (GABAT), GS (GS1 and GS2), malate dehydrogenase (MDHC1, mMDH2, MDHNP, and chlMDH), succinyl-CoA ligase (SUCA1 and SUCB).
	Transport related: V-type proton ATPase (VHA-a2, VHA-d, VHA-A, VHA-B, VHA-C, VHA-D, VHA-E, VHA-F2, VHA-G, and VHA-H2), P-type proton pump (PH8), and pyrophosphate-energized vacuolar membrane proton pump (VHP1 and VHP2).
VC biosynthesis	Aldo-keto reductase (GalUR), pectinesterase (PME), inositol-3-phosphate synthase (IPS), l-galactose-1-phosphate phosphatase (GaIPP), GDP-mannose (GME), glucose-6-phosphate isomerase (PGI), nucleobase-ascorbate transporter (AAT), glucose-6-phosphate isomerase (PGI), GDP-l-fucose synthase (GER), monodehydroascorbate reductase (MDAR), aldonolactonase (Alase), phosphomannomutase (PMM), l-ascorbate peroxidase (APX), mannose-6-phosphate isomerase (PMI), dehydroascorbate reductase (DHAR), glutathione reductase (GR), and l-galactono-1 (GalLDH).
Flavonoid metabolism	Naringenin, 2-oxoglutarate 3-dioxygenase (F3H), flavonoid 3′-monooxygenase (F3′H), flavonol synthase (FLS), chalcone synthase (CHS), glucosyltransferases (PGT), and flavonoid O-methyltransferase (FOMT).
Carotenoid metabolism	1-deoxy-d-xylulose 5-phosphate reductoisomerase (DXR), 9-cis-epoxycarotenoid dioxygenase (NCED), and carotenoid 9,10(9′,10′)-cleavage dioxygenase 1 (CCD1).
Terpenoid biosynthesis	4-hydroxy-3-methylbut-2-enyl diphosphate synthase (HDS), acetoacetyl-CoA thiolase (AACT), hydroxymethylglutaryl-CoA synthase (HMGS), mevalonate diphosphate decarboxylase (MVD), geranyl diphosphate synthase (GPPS), and alpha-copaene synthase (TPS).
Storage quality	Long chain acyl-CoA synthetase (LACS1, LACS4, LACS7, and LACS8), very-long-chain enoyl-CoA reductase (ECR), very-long-chain 3-oxoacyl-CoA reductase (KCR), and aquaporin (PIP1:3α, PIP1:1, PIP2:6, and PIP2:5α).
Fruit ripening	Xanthoxin dehydrogenase (ABA2), serine/threonine-protein kinase (SnRK2.4, SnRK2.6), indole-3-acetic acid-amido synthetase (GH3.6), and auxin signaling f-box 3 (AFB3).
Disease resistance	Brassinosteroid signaling kinase (BSK7), abnormal inflorescence meristem1 (AIM1), acyl-coenzyme A oxidase (ACX1, ACX4), allene oxide synthase (AOS), and 3-keto-acyl-CoA-thiolase (KAT).

For more detailed information, see Supplemental Table S2.

The modified lysine residues of the acetylated proteins from citrus are significantly more conserved among different plant species than the lysine residues of the non-acetylated proteins (Supplemental Figure S2C). Additionally, approximately 24.0% (245/1022) and 35.0% (358/1024) of the acetylated proteins in Arabidopsis and rice could be found in the citrus acetylomes, respectively (Figure 2G). Furthermore, the 120 proteins that were acetylated in these three species mostly contribute to core metabolism (Supplemental Figure S2D). Importantly, 1,157 acetylated proteins were citrus-specific and are associated with important fruit quality traits and contribute to important primary metabolic pathways, such as the citrate cycle (TCA cycle), fructose and mannose metabolism, and flavonoid biosynthesis (Table 1 and Figure 2H; Supplemental Table S2). We conclude that protein acetylation is common in citrus and that many of these acetylation sites appear specific to citrus.

Characterization of protein lysine acetylation in the NH allotetraploid

We first compared the acetylomes between the two parents. About 36.3% (1,320/3,633) of total quantifiable Kac sites were different in HBP relative to Nova (Supplemental Figure S3A). We observed increases in lysine acetylation at 1,082 Kac sites (82.0%) on 618 proteins in HBP relative to Nova and decreases in lysine acetylation at only 238 Kac sites (18.0%) on 174 proteins in HBP relative to Nova (Supplemental Figure S3, A and B and Supplemental Table S1).

The number of Kac sites that were differentially regulated in the NH allotetraploid relative to the HBP parent (NH-HBP_dif, 907 Kac sites, 603 proteins) was very similar to the number of Kac sites that were differentially acetylated in the NH allotetraploid relative to the Nova parent (NH-Nova_dif, 870 sites, 559 proteins) (Supplemental Figure S3, A and B and Supplemental Table S1). Thus, the NH allotetraploid was equally divergent from HBP and Nova. We observed that lysine acetylation in a total of 838 proteins with 1,064 Kac sites was differentially regulated between the parents and/or between the NH allotetraploid and the two parents (Nova-NH-HBP_dif).

To better understand the reprogramming of the acetylome in the NH allotetraploid, we classified the 1,064 Kac sites with different levels of acetylation in the different genotypes (Nova-NH-HBP_dif) into 12 bins. This strategy is similar to a strategy that was used to study differentially regulated gene expression in allopolyploids (Yoo et al., 2013; Li et al., 2014). Acetylation level dominance (ALD) was defined as Kac sites or acetylated proteins with acetylation levels equal to only one of the parents. For example, ALD-HBP sites or proteins had similar levels of acetylation in the NH allotetraploid and HBP that were much different in Nova. Overall, we observed ALD in the NH allotetraploid at more than 67% of the Kac sites (713 Kac sites, 545 protein) (Supplemental Table S3, categories II, IV, XI, and IX). We refer to Kac sites or acetylated proteins with acetylation levels that were either much higher or lower relative to both parents as transgressive acetylation. Approximately 23.3% of the Nova-NH-HBP_dif acetylation sites (248 Kac sites, 185 proteins) were transgressively regulated (Supplemental Table S3, categories III, VII, X, V, VI, and VIII).

Parental lysine acetylation dominance is associated with fruit acidity in the NH allotetraploid

We defined a total of 346 Kac sites on 270 acetylated proteins as ALD-HBP sites and ALD-HBP proteins because they were acetylated at similar levels in the NH allotetraploid and the HBP parent (Figure 3A). Notably, the KEGG pathways that were enriched in the ALD-HBP proteins were mostly associated with the citric acid cycle (TCA cycle), which is located in the mitochondria (Figure 3B). Additionally, several cellular component terms associated with mitochondria were enriched in the ALD-HBP proteins (Figure 3C).

Figure 3.

Parental lysine acetylation dominance associated with fruit acidity (citric acid content) in the NH allotetraploid. A, Number of Kac sites and acetylated proteins with ALD of the HBP parent (ALD-HBP) in the NH allotetraploid. ALD-HBP proteins refer to proteins with the same acetylation levels at their Kac sites in the NH allotetraploid and the HBP parent but with significantly different acetylation levels in the NH allotetraploid relative to the Nova parent. B and C, KEGG pathway (B) and GO enrichment (C) analysis of ALD-HBP proteins (Fisher’s exact test, P < 0.05). Number of proteins is indicated in the enriched term. D, Heatmap showing the acetylation levels of proteins that contribute to citric acid metabolism. The color scale bar represents z-score normalized acetylation level. Number of Kac sites is denoted in the parentheses. E, Global acetylation levels of proteins associated with citric acid metabolism in the NH allotetraploid and its diploid parents (Kruskal–Wallis test, *P < 0.05, **P < 0.01, ***P < 0.001, “ns” indicates no significant difference). Number of Kac sites in each panel is denoted in the parentheses. The protein classification is the same as in (D). The boxplot shows the median of the data between the first and the third quartiles, whiskers indicate the 10th and 90th percentiles, and the points represent outliers. F, Effect of site-directed mutagenesis on GS2 and NADP-IDH1 activity. Lysine residue (K) is substituted with arginine (R) as a mimic of non-acetylated lysine residue, or with glutamine (Q) as a mimic of acetylated lysine residue. Immunoblotting with anti-His antibodies were used as a loading control. Bar graphs represent the means ± sd (n = 3). Different letters indicate significant differences (P < 0.05, one-way ANOVA with Tukey’s post hoc test).

Citric acid metabolism is largely responsible for the acidity of citrus fruit (Yamaki, 1989) and involves proteins that contribute to the biosynthesis, utilization, and transport of citric acid (Supplemental Figure S4A and Table 1). The biosynthesis of citric acid takes place in the mitochondria and involves two enzymes, citrate synthase (CS) and phosphoenolpyruvate carboxylase (PEPC) (Guo et al., 2016; Lin et al., 2016). The quantitative acetylation data showed that the global level of acetylation for proteins involved in the biosynthesis of citric acid was similar in the NH allotetraploid and HBP, but significantly higher than that in Nova (Figure 3, D and E). However, several investigations provide evidence that the utilization is critical for the accumulation of citric acid in citrus fruit rather than the regulation of enzymes (PEPC and CS) and genes that contribute to biosynthesis (Sinclair, 1984; Sadka et al., 2001; Guo et al., 2016). Therefore, the acetylation level of these two enzymes might have limited effect on citric acid accumulation.

After synthesis, citric acid can be partially utilized in mitochondria through citric acid cycle or in cytosol by a series of enzymes (Supplemental Figure S4A and Table 1). It is noteworthy that the majority of these proteins had abundant Kac sites (Supplemental Figure S4A and Supplemental Table S2). Many of them have been confirmed to influence citric acid accumulation of fruit, including aconitate hydratase (ACO), isocitrate dehydrogenase (NADP-IDH) (Sadka et al., 2000; Wang et al., 2018), glutamate decarboxylase (GAD) (Liu et al., 2014), ATP-citrate lyase (ACL), and glutamine synthetase (GS) (Bao et al., 2015). It is clear that the parent Nova had very low Kac level particularly for these critic acid utilization associated proteins compared with that in NH and HBP (the difference was more pronounced than that in biosynthesis and transportation) (Figure 3, D and E). Coincidentally, the NH allotetraploid inherited mitochondrial genome form HBP parent, which might explain the parental lysine acetylation dominance of HBP on those citric acid utilization associated proteins from mitochondria and why Nova had lower acid content than the NH allotetraploid and HBP (Figure 1D), since Kac has been reported to weaken mitochondrial enzymes activity (Baeza et al., 2016).

As another pathway, citric acid can be translocated to the vacuole for storage by tonoplast-associated proton pumps and inward-rectifying channels (Supplemental Figure S4 and Table 1). In contrast, the Kac levels of these proteins involved in citric acid transport were not significantly different among the three genotypes (Figure 3, D and E).

To validate that the higher citric acid contents in NH and HBP were associated with the higher Kac levels of those enzymes involved in citric acid utilization, we selected CgGS2 and CgNADP-IDH1 to study the effect of Kac on their activities. These two enzymes have been verified to promote the utilization of citric acid (Sadka et al., 2000; Bao et al., 2015). Both CgGS2 and CgNADP-IDH1 have multiple Kac sites and high lysine residue conservation in different species (Supplemental Figures S5 and S6). There were six and nine quantifiable Kac sites in CgGS2 and CgNADP-IDH1, respectively. In CgGS2, three Kac sites (K106, K305, and K326) were acetylated at higher levels in the NH allotetraploid and HBP relative to Nova. In contrast, the other sites were not acetylated at different levels among three genotypes. In CgNADP-IDH1, three Kac sites (K117, K205, and K376) were acetylated at much higher levels in the NH allotetraploid and HBP relative to Nova. In contrast, the other sites were not acetylated at different levels among the three genotypes. K106 and K205 are located in the active sites of CgGS2 and CgNADP-IDH1, respectively (Supplemental Figure S4, B and C). We used site-directed mutagenesis to change these lysine residues to arginine (R) to inhibit acetylation or glutamine (Q) to mimic acetylation. Using quantitative enzyme assays, we demonstrated that the activities of the hyper-acetylation-mimics, CgGS2^K106Q and CgNADP-IDH1K^205Q, were markedly reduced relative to the wild-type (WT) enzymes. In contrast, the activities of the hypo-acetylation-mimics, CgGS2^K106R and CgNADP-IDH1^K205R, were similar to those of WT (Figure 3F). These findings indicate that the Kac of CgGS2 and CgNADP-IDH1 can inhibit their activities.

In sum, these data indicate that higher levels of Kac in proteins involved in the utilization of citric acid might inhibit the consumption of citric acid and thus might lead to higher levels of citric acid accumulating in the NH allotetraploid and HBP relative to Nova (Figure 1F).

Parental lysine acetylation dominance associates with flavonol accumulation in the NH allotetraploid

We defined a total of 367 Kac sites on 288 acetylated proteins as ALD-Nova sites and ALD-Nova proteins because they had similar levels of acetylation in the NH allotetraploid and the Nova parent (Figure 4A). The KEGG pathways that were enriched in the ALD-Nova proteins were associated with phenylpropanoid biosynthesis and flavonoid biosynthesis (Figure 4B). Therefore, we tested whether there is a relationship between ALD and flavonoid metabolism in the NH allotetraploid.

Figure 4.

Parental lysine acetylation dominance associates with flavonol content in the NH allotetraploid. A, Number of Kac sites and acetylated proteins with ALD from the Nova parent (ALD-Nova) in the NH allotetraploid. ALD-Nova proteins refer to proteins with similar levels of acetylation at Kac sites in the NH allotetraploid and the Nova parent but significantly different levels of acetylation relative to the HBP parent. B, KEGG pathway enrichment analysis for ALD-Nova proteins (Fisher’s exact test, P < 0.05). The number of proteins is indicated in parentheses. C, Levels of total flavonols and two major flavonols in the NH allotetraploid and its diploid parents. Data are presented as the mean ± sd (n = 3). D, Relative acetylation levels at two Kac sites (K121 and K127) and levels of mRNA and protein for CgFLS (flavonol synthase). The horizontal gray dashed line indicates the mid-parent value (MPV) for CgFLS. E, Sequence conservation (upper) and protein structure model (lower) for CgFLS. S. lycopersicum, Solanum lycopersicum. CgFLS structure was modeled using the crystal structure of anthocyanidin synthase (PDB ID: 1GP6) from A. thaliana. The active site is shown with ball-and-stick representations. K121 is located in the active site. F, Effect of various site-directed mutations on FLS activity. Lysine residue (K) is substituted with arginine (R) as a mimic of non-acetylated lysine residue, or with glutamine (Q) as a mimic of acetylated lysine residue. 2KR or 2KQ, double substitution mutants at sites K121 and K127. Bar graphs represent the means ± sd (n = 3). For (C), (D), and (F), different letters indicate significant differences (P < 0.05, one-way ANOVA with Tukey’s post hoc test).

The flavonoid content in fruit of the NH allotetraploid was close to the Nova parent but much higher than the HBP parent (Figure 1D). Flavonol is one of the most important classes of flavonoids. Consistently, the total flavonol content of the NH allotetraploid was similar to that of the Nova parent and significantly higher than that of the HBP parent (Figure 4C). Consistent with the total flavonol data, quercetin 3-O-rutinoside (Rutin) and quercetin 7-O-rutinoside—the two most abundant flavonols in citrus fruit—accumulated to similar levels in the NH allotetraploid and Nova but accumulated to five-fold and ten-fold higher levels in the NH allotetraploid relative to the HBP parent (Figure 4C).

The final step in flavonol production, the conversion of dihydroflavonols to flavonols, is catalyzed by flavonol synthase (FLS), which is the key enzyme for flavonol biosynthesis (Verhoeyen et al., 2002). FLS was one of the ALD-Nova proteins. Two Kac sites (K121 and K127) were found with similar levels of acetylation in the NH allotetraploid and the Nova parent but much lower levels than the HBP parent (Figure 4D and Supplemental Table S1). More interestingly, based on the transcriptomic and proteomic data, both the CgFLS mRNA and protein accumulated to much lower levels in the NH allotetraploid and the Nova parent, relative to the HBP parent (Figure 4D). Additionally, the CgFLS gene was expressed at levels that were more similar in the NH allotetraploid and the Nova parent than in the HBP parent during all four fruit ripening stages (Supplemental Figure S7A). These data indicate that the higher levels of flavonol accumulation in the NH allotetraploid and Nova parent are probably not caused by differences in the expression of the CgFLS mRNA or protein. Thus, it is reasonable to hypothesize that the lower levels of Kac on one or two Kac sites might elevate CgFLS enzyme activity and thus lead to more flavonols accumulating in the NH allotetraploid and Nova. Consistent with this hypothesis, two lysine residues were found to be highly conserved in plants and one of them (K121) was predicted to localize in the active site of CgFLS (Figure 4E and Supplemental Figure S7B), which provides evidence for a regulatory mechanism.

To test the hypothesis, we used site-directed mutagenesis to change one or two lysine (K) residues in CgFLS (i.e. K121 and K127) to arginine (R) to inhibit acetylation or glutamine (Q) to mimic acetylation. We performed quantitative enzyme assays and found that the FLS activity of the hyper-acetylation-mimic CgFLS^K121Q was markedly reduced relative to CgFLS^WT. In contrast, the FLS activity of the hypo-acetylation-mimic CgFLS^K121R was upregulated relative to CgFLS^WT (Figure 4F). However, we found that when K127 was changed to either R or Q, there was no effect on FLS activity. We obtained similar data when we changed only K121 and when we changed both K121 and K127. These data make the case that the acetylation of K121 but not K127 inhibits the activity of FLS. Collectively, these results indicate that the higher levels of flavonols in the NH allotetraploid and Nova relative to HBP were partially explained by lower levels of Kac in FLS that regulate CgFLS activity rather than by affecting CgFLS mRNA or protein levels.

Transgressive acetylation is negatively associated with protein abundance in the NH allotetraploid

The NH allotetraploid had more sites and proteins with transgressive hyper-acetylation (154/104) than transgressive hypo-acetylation (94/81) (Figure 5, A and B). The proteins with transgressive hyper-acetylation were predominantly associated with the ribosome. In contrast, the proteins with transgressive hypo-acetylation were associated with diverse functions (Supplemental Figure S3, C and D and Supplemental Table S3).

Figure 5.

Transgressive acetylation associates with protein abundance in the NH allotetraploid. A and B, Number of Kac sites and acetylated proteins with transgressive hyper-acetylation (A) and hypo-acetylation (B) in the NH allotetraploid. Transgressive hyper-acetylation and hypo-acetylation refers to significantly higher and lower levels of acetylation relative to both parents. C, Global abundance of proteins with transgressive acetylation. Other proteins represent non-transgressively acetylated proteins. Number of proteins in each panel is denoted in the parentheses. Different letters indicate significant differences (P < 0.05, Kruskal–Wallis test). The boxplot shows the median of the data between the first and the third quartiles, whiskers indicate the 10th and 90th percentiles, and the points represent outliers. D, Scatter plot of log₂ FC in mRNA levels (allotetraploid/MPV) versus log₂ FC in protein levels (allotetraploid/MPV) for all acetylated proteins (left) and transgressive hyper-acetylated (middle) and hypo-acetylated proteins (right). Data for all proteins in the proteome are plotted in the background (gray points). Black dashed lines indicate the linear fit for corresponding acetylated proteins. Pearson’s correlation (r) coefficient is indicated. E, Protein structure model of CgRPL5. The structure was modeled using the crystal structure of the translating 80S ribosome from T. aestivum (PDB ID: c3iz5Q). The binding site is shown with a ball-and-stick representation. Except for K151 on CgRPL5, all other Kac sites are quantifiable. F, Degradation assay for the WT CgRPL5 (CgRPL5^WT) and mutants (CgRPL5^K43R/Q and CgRPL5^K240R/Q). Lysine residue (K) is substituted with arginine (R) as a mimic of non-acetylated lysine residue, or with glutamine (Q) as a mimic of acetylated lysine residue. Flag-tagged proteins were transiently expressed in N. benthamiana. Immunoprecipitated proteins were incubated in citrus peel extracts and assayed at the indicated time points. Anti-actin was used as a loading control. Protein levels above the band were quantified using ImageJ.

Considering that Kac had been reported to have either a negative or positive effect on protein abundance (Du et al., 2010; Jiang et al., 2011; Narita et al., 2019), we studied the relationship between transgressive acetylation and protein abundance in the NH allotetraploid. We analyzed the mRNA and protein levels of all detectable proteins with Kac in the transcriptome and proteome of fruits at 180 DAF from the NH allotetraploid and its diploid parents (Supplemental Figure S8, A and B and Supplemental Tables S4 and S5). Notably, transgressively hyper-acetylated proteins accumulated to lower levels in the NH allotetraploid relative to its two parents and mid parent. In contrast, the transgressively hypo-acetylated proteins accumulated to higher levels in the NH allotetraploid relative to its two parents and mid parent (Figure 5C).

We next compared the transcript and protein abundance of the transgressively acetylated proteins in the NH allotetraploid and mid parent. Strikingly, we found a weaker correlation between protein and mRNA abundance for transgressively hyper-acetylated proteins (Corr = 0.36), followed by all acetylated proteins (Corr = 0.47). We found the best correlation between the abundance of transgressively hypo-acetylated proteins and their mRNAs (Cor = 0.60) (Figure 5D). For comparison, when all acetylated proteins were subdivided into three categories according to the number of acetylation sites (1–2 sites, low acetylated proteins; 3–4 sites, middle; >4, high), no obvious difference was found in the correlations between the abundance of the proteins and their mRNAs (Supplemental Figure S9, A and B). Thus, the changes in the levels of mRNA that occurred during polyploidization were poorly correlated with the levels of protein for transgressively hyper-acetylated proteins but were much better correlated for transgressively hypo-acetylated proteins. Moreover, all acetylated proteins were subdivided into nine categories based on changes in mRNA and protein levels (Supplemental Figure S9C and Supplemental Table S6). Proteins with lower abundance were significantly hyper-acetylated, whereas proteins with higher abundance were hypo-acetylated (Supplemental Figure S9D). Accordingly, we hypothesized that transgressive hyper-acetylation leads to lower protein abundance probably by decreasing protein stability.

To test whether transgressive hyper-acetylation decreases protein stability, a ribosomal protein L5 (CgRPL5)—one of the transgressively acetylated proteins—was selected for further study. Five quantifiable Kac sites (K27, K43, K177, K240, and K241) were detected in CgRPL5 and all of them were transgressively acetylated in the NH allotetraploid (Supplemental Table S3). Two quantifiable Kac sites (K27 and K43) were predicted to locate in the binding site of CgRPL5 (Figure 5E). First, we used site-directed mutagenesis to change all five of the K residues in CgRPL5 to arginine (R) or glutamine (Q) residues to mimic differences in acetylation status. Protein stability assays demonstrated that the hypo-acetylation mimic (CgRPL5^5KR) was more stable (Supplemental Figure S9E) and that hyper-acetylation mimic (CgRPL5^5KQ) was less stable. Given the central location of K43 in the binding site of CgRPL5 and our finding that the greatest difference in Kac levels in CgRPL5 between the NH allotetraploid and its parents was observed at K43 (Figure 5E andSupplemental Table S3), we evaluated the effect of the acetylation status of K43 on protein stability. Interestingly, CgRPL5^K43R was more stable than WT (CgRPL5^WT) and CgRPL5^K43Q was less stable than CgRPL5^WT. However, the degradation of CgRPL5^K240Q was similar to that of CgRPL5^WT and CgRPL5^K240R (Figure 5F). Based on these data, we conclude that the acetylation status of K43 contributes to the stability of CgRPL5 and that the negative association between transgressive acetylation and protein abundance in the NH allotetraploid is probably explained by transgressive acetylation regulating protein stability.

Gene expression and subcellular localization of lysine acetyltransferases and KDACs in the allotetraploid

Nonhistone acetylation is generally catalyzed by lysine acetyltransferases (KATs) and removed by KDACs (Narita et al., 2019). To gain mechanistic insight into the lysine acetylation patterns that occur during allotetraploidization including parental acetylation dominance, transgressive hyper-acetylation, and transgressive hypo-acetylation, we evaluated the expression of genes that encode KATs and KDACs in the NH allotetraploid and its diploid parents. Considering that the majority (>80%) of acetylated proteins do not accumulate in the nucleus in agreement with previous findings (Hartl et al., 2017), we focused only on KATs and KDACs that do not localize to the nucleus.

A total of eight putative KATs were identified in citrus (Supplemental Figure S10A). Among them, two proteins (CgHAG3 and CgHAF2) were predicted to accumulate in the cytoplasm and to dual-localize to the chloroplast and the nucleus (Supplemental Figure S10B). However, the genes encoding CgHAG3 and CgHAF2 were expressed at similar levels in the NH allotetraploid and its parents (Figure 6A and Supplemental Figure S10E). Regarding the KDACs, 14 putative KDACs were identified in citrus (Supplemental Figure S10C). Unlike the KAT proteins, the majority of the KDACs (10 of the 14 proteins) were predicted to accumulate in the cytoplasm (Supplemental Figure S10D). Among them, CgSRT2, CgHDA8, and CgHDA14 were expressed at similar levels in the NH allotetraploid and the HBP parent but at higher levels in the Nova parent (Figure 6A and Supplemental Figure S10E). There were 5 and 35 single nucleotide polymorphisms (SNPs) that distinguished the HBP and Nova parents, in genomic region of CgSRT2 and CgHDA8, respectively. The NH allotetraploid possessed the SNPs from both parents. Therefore, the lower expression levels of CgSRT2 and CgHDA8 in the NH allotetraploid were not likely due to the loss of the genomic sequences (Supplemental Files S1 and S2). A subcellular localization analysis revealed that CgSRT2-GFP was dual-localized to the mitochondria and the plastid (Figure 6, B and C). In contrast, CgHDA8 was dual-localized to the nucleus and cytoplasm (Figure 6D).

Figure 6.

Gene expression and subcellular localization of KATs and KDACs in the NH allotetraploid. A, Semi-quantitative RT-PCR analysis of KATs and KDACs expression in the NH allotetraploid and the two diploid parents. These KATs and KDACs were not predicted to reside in the nucleus. The asterisks represent the obvious differences of gene expression among the allotetraploid NH and the parents. B–D, Subcellular localization of CgSRT2 and CgHDA8 in N. benthamiana leaf mesophyll cells. PTRK, MTRB, and NLRK were used as plastid, mitochondrial, and nuclear markers, respectively. CgSRT2 was dual-localized to the plastid (B) and mitochondria (C). CgHDA8 was dual-localized to the nucleus and cytosol (D). Scale bars, 15 μm (B) and 10 μm (C and D).

Overexpression of CgSRT2 and CgHDA8 partially restores a metabolic status to the allotetraploid NH

As described above, the NH allotetraploid and HBP parent produced fruit with similar qualities that included similar levels of fruit acidity, sugar, and carotenoids relative to the Nova parent (Figure 1D). To test whether the expression profiles of the KDACs were associated with the similarities in fruit quality, we overexpressed CgSRT2 and CgHDA8 separately in citrus callus and obtained two independent transgenic lines for each gene (Figure 7, A and B). We found that overexpressing both CgSRT2 and CgHDA8 led to decreases in the levels of the majority of primary metabolites, including the major organic acids (citric acid and malic acid), amino acids (serine, proline, aspartic acid, alanine, and gamma aminobutyrate [GABA]), and sugars (glucose, fructose, and sucrose) (Figure 7, C–E). Moreover, the overexpression of CgSRT2 and CgHDA8 significantly increased the levels of carotenoids and flavonoids (Figure 7F). These data may explain the darkening of the overexpression lines (Figure 7, A and B). The metabolic changes provide evidence that CgSRT2 and CgHDA8 have similar physiological roles in the regulation of metabolism in fruit.

Figure 7.

Overexpression of CgSRT2 and CgHDA8 leads to metabolic flux remodeling in citrus callus. A and B, Phenotypes of CgHDA8-overexpressing (A) and CgSRT2-overexpressing (B) citrus callus. EV, transgenic citrus callus containing the empty vector. Two independent lines were shown. Semi-quantitative RT-PCR analysis was used to confirm the overexpressing lines (lower). Actin was used as an internal control. C–E, Levels of major primary metabolites, amino acids (C), sugars (D), and organic acids (E) in overexpressing lines and EV. F, Levels of major secondary metabolites, carotenoids, flavonoids, and flavonols in overexpressing lines and EV. All of the above results are mean values ± sd from three biological replicates. Asterisks indicate statistically significant differences relative to the EV (Student’s t test, ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001).

Importantly, the overexpression of CgSRT2 and CgHDA8 partially restored the metabolic status of the NH allotetraploid and HBP parent because they expressed lower levels of CgSRT2 and CgHDA8 relative to the Nova parent and accumulated higher levels of organic acids and amino acids, but lower levels of sugars and carotenoids relative to the Nova parent (Figure 1, D–F). Collectively, these results provide evidence that the altered patterns in the expression of the genes encoding KDACs during allopolyploidization probably contributed to establish the metabolic traits of the NH allotetraploid.

Treatments with KDAC inhibitors alter citrus pulp color and metabolic flux

To further study the biological functions of KDACs in the regulation of fruit metabolism, two KDAC inhibitors, sirtinol (an inhibitor of the Sirtuin family) and sodium butyrate (NaB, a HDACs inhibitor) were used to treat the fruit pulp from Nova and citrus callus. Interestingly, after a treatment with two KDAC inhibitors, we noticed an abnormal accumulation of red spots in the citrus pulp and that the yellow juice vesicles turned red after they were lyophilized (Figure 8A). A similar but weaker phenotype was observed after a treatment with only sirtinol and only NaB (Supplemental Figure S11). Interestingly, pigments accumulated abnormally in citrus callus after they were treated with the same inhibitors (Supplemental Figure S12).

Figure 8.

Influence of KDAC inhibitors on citrus pulp color and metabolic flux. A, Phenotypes of fruit pulps from the Nova parent co-treated with sirtinol (500 μM) and NaB (10 mM) or mock treated with DMSO in culture medium for 25 days. White triangles indicate the appearance of red spots in fruit pulp after the inhibitor treatment. Left panel shows the lyophilized fruit pulps after the treatment. B–D, Levels of major primary metabolites, amino acids (B), sugars (C), and organic acids (D) in fruit pulp after inhibitor treatments. E, Levels of major secondary metabolites, anthocyanins, carotenoids, flavonoids, and flavonols in fruit pulp after inhibitor treatments. Bar graphs represent means ± sd (n = 3). Asterisks indicate statistically significant differences relative to the mock-treated samples (Student’s t test, ns, not significant; **P < 0.01; ***P < 0.001; ****P < 0.0001).

A metabolic profiling experiment demonstrated that the inhibitor treatments significantly decreased the levels of the majority of primary metabolites (i.e. amino acids, sugars, and organic acids), total carotenoids, and flavonoids in the citrus fruit pulp and callus (Figure 8, B–E and Supplemental Figures S11 and S12). Surprisingly, the inhibitor treatment greatly enhanced the anthocyanin content (Figure 8E and Supplemental Figures S11C and S12C), which is consistent with the red color that appeared in the fruit pulp and callus during the treatment. In fact, anthocyanin could be detected only in a few citrus varieties, not including HBP and Nova. Generally, the inhibitor treatment didn’t exert an opposite effect on metabolism relative to the overexpression of CgSRT2 and CgHDA8 (Figure 7), which was probably due to the complexity that comes from the inhibition of additional KDACs during the inhibitor treatments. These results together with data from the overexpression of CgSRT2 and CgHDA8 in callus make a strong case that KDACs play important roles in metabolic regulation in fruit, including the regulation of primary metabolism and the production of carotenoids, flavonoids, and anthocyanins.

Discussion

In this study, we investigated the relationship between the establishment of fruit traits and the reprogramming of the acetylome during allopolyploidization in citrus. We found that (a) a considerable proportion of acetylated proteins were differentially acetylated in the allotetraploid relative to the diploid parents and the mid parent; (b) parental ALD is closely related to the formation of fruit traits in the allotetraploid by regulating enzyme activity; (c) transgressive hyper-acetylation decreases protein stability and leads to lower protein abundance in the allotetraploid; (d) the gene expression of several KDACs, such as CgSRT2 and CgHDA8, is likely responsible for the acetylome reprograming during allopolyploidization; and (e) overexpression of CgSRT2 and CgHDA8 partially restores the metabolic traits that develop during allopolyploidization.

Protein Kac modification may open up new strategies for fruit quality improvement

We report a high-resolution acetylome map of 4,175 Kac sites on 1,640 proteins from citrus fruit. Many proteins that make important contributions to fruit quality are acetylated, including proteins responsible for major commercial traits, such as fruit acidity and nutritional value (Table 1). Identification of these acetylation sites will lead to a new approach for studying the regulation of proteins that contribute to fruit quality and other agronomically important traits.

One of the most important traits is fruit flavor, which is determined by complex interactions between sugars (mainly glucose, fructose, and sucrose) and organic acids (mainly citric and malic acid) (Gaston et al., 2020). We found that the majority of the proteins involved in the accumulation of sugar and citric acid metabolism had abundant Kac modifications. These Kac modifications can probably regulate enzyme activities and protein abundance (Figures 3–5). For instance, using site-directed mutagenesis to change Kac residues to amino acid residues that mimic a particular acetylation status can attenuate the activities of CgGS2 and CgNADP-IDH1 (Figure 3). In the last several years, the cytosine and adenine base editors (ABEs) have emerged as efficient tools for precise genome modification (C to T or A to G) in plants (Cai et al., 2020; Hua et al., 2021; Huang et al., 2022). These Kac sites are attractive targets for precise genomic editing experiments that aim to improve the quality of fruit. Moreover, the functions of these Kac sites can also be verified using a targeting-induced local lesions in genomes (TILLING) platform based on seeds that are mutagenized with ethylmethanesulfonate (EMS) treatments, which primarily induces G/C to A/T transitions (McCallum et al., 2000; Wang et al., 2008; Okabe et al., 2011).

Moreover, numerous KATs and KDACs exist in plants. The subcellular locations and biological functions of many of them have not been determined (Shen et al., 2015). Some KAT and KDAC proteins are known to influence metabolism in yeast, animal, and plants (Vidal-Laliena et al., 2013; König et al., 2014; Mortenson et al., 2015). For example, an Arabidopsis sirtuin (AtSRT2) interacts with a few protein complexes mainly involved in energy metabolism and metabolite transport. Loss of AtSRT2 function leads to an increase in the acetylation of several of these proteins and affects the sugar and amino acid content (König et al., 2014). In our study, the overexpression of CgSRT2 and CgHDA8 led to increased levels of carotenoids and flavonoids and decreased levels of organic acids and sugars. Therefore, the functional analysis of these KATs and KDACs will lead to new strategies for fruit quality improvement.

Trait establishment associates with the reprogramming of protein acetylation during allopolyploidization

Most of proteins that are associated with the utilization of citric acid are mitochondrial enzymes. Consistent with our findings, higher levels of acetylation generally negatively regulate the activity of mitochondrial enzymes, such as MDH and IDH1 in vivo (Wang et al., 2020; Balparda et al., 2022). Thus, the elevated levels of acetylation that we observed in the proteins that contribute to the utilization of citric acid should attenuate both their activities and the consumption of citric acid in the NH allotetraploid and the HBP parent. This interpretation is consistent with the significantly higher levels of fruit acidity in the NH allotetraploid and the HBP parent (Figure 1D). Mitochondrial enzymes can be deacetylated by sirtuin-type deacetylases, such as SIRT2 in Arabidopsis (Finkemeier et al., 2011; Wang et al., 2020). The lower expression of CgSRT2 might explain the higher levels of acetylation in the proteins associated with the utilization of citric acid in the NH allotetraploid and HBP parent. Consistent with this notion, overexpressing CgSRT2 leads to decreases in the levels of most organic acids (Figure 7E).

The novel phenotypes produced by polyploidization cannot be fully explained by variations in genetics, gene expression, and protein abundance (Yu et al., 2010; Allario et al., 2011; Vu et al., 2017). In our study, the expression level and protein abundance of CgFLS are not consistent with the higher flavonol content in the NH allotetraploid. In contrast, the lysine acetylation levels of CgFLS appear to provide a better explanation (Figure 4). Based on our findings, we propose that the establishment of plant traits during allopolyploidization is associated with the reprogramming of protein acetylation that occurs during allopolyploidization (Figure 9).

Figure 9.

Working model for the influence of acetylated lysine residues on the remodeling of plant traits during polyploidization. The allopolyploid developed novel acetylation patterns for lysine residues relative to the two parents. The non-additive expression of KDACs (e.g. downregulation of HDA8 and SRT2) can lead to novel acetylation patterns, which is associated with the development of novel traits (e.g. high fruit acidity and flavonol content, mRNA/protein discordance) in allopolyploids because of the influence of acetylation on enzyme activity (e.g. GS2, IDH, and FLS) and protein abundance.

PTM might contribute to the discordance between transcriptomes and proteomes in allopolyploids

Many studies performed on polyploids have found that transcriptomes poorly explain proteomes (Marmagne et al., 2010; Li et al., 2020). In this study, we found Kac level likely affects the correlation between mRNA and protein abundance. The proteins with transgressive hyper-acetylation in the allotetraploid exhibit a poor correlation between mRNA and protein abundance, whereas the proteins with transgressive hypo-acetylation exhibit a greater correlation (Figure 5).

Ribosomal proteins make up the majority of the proteins with transgressive hyper-acetylation (Supplemental Table S3). The stability assay of CgRPL5 supports the notion that hyper-acetylation promotes the protein degradation (Figure 5F). A recent study performed in rice demonstrates that HDA714 (a KDAC) loss-of-function mutations increased Kac levels but reduced abundance of ribosomal proteins (Xu et al., 2021). The increased Kac level weakened the stability of rice ribosomal proteins, like RPS3, RPS6, and RPL7a, through ubiquitination-mediated degradation (Xu et al., 2021). Kac also has been found to trigger the ubiquitination for some other proteins (Du et al., 2010; Jiang et al., 2011). For example, DNA methyltransferase 1 (DNMT1) is destabilized by acetylation by the acetyltransferase, Tat-Interactive protein (Tip60), which triggers ubiquitination by the ubiquitin-like with plant homeodomain (PHD) and ring finger domains 1 (UHRF1), thereby targeting DNMT1 for proteasomal degradation (Du et al., 2010).

Moreover, other PTMs such as sumoylation, phosphorylation, and lysine methylation also have been proven to affect protein stability without altering mRNA levels (Truman et al., 2012; Nukarinen et al., 2017; Lehning and Morrison, 2022). Further efforts are needed to elucidate the effects of polyploidization on the abundance of these PTMs and in turn how this influences the protein abundance in polyploids.

Materials and methods

Plant materials and sampling

The study was conducted on fruit pulp from “Nova” tangelo (C. clementina hort. × [C. paradisi Macfad. × C. reticulata Blanco]) (abbr. Nova), Hirado Buntan pummelo (C. grandis [L.] Osbeck) (abbr. HBP), and their allotetraploid somatic hybrid abbreviated as NH (C. reticulata Blanco + C. grandis [L.] Osbeck), which was produced by symmetric protoplast fusion (Grosser et al., 1998). Fruits were harvested at 15-day intervals from December 2016 to February 2017 at stage III of fruit development. Three individual trees of each genotype were used as biological replicates. Five fruits were used as one sample and after harvesting were immediately frozen in liquid nitrogen and stored at −80°C.

Chromosome number determinations were performed using the 4′,6-diamidino-2-phenylindole (DAPI) method as previously described (Wang et al., 2016). Simple sequence repeat (SSR) and cleaved-amplified polymorphic sequence (CAPS) markers were used for genotype analysis for investigating the origin of the allotetraploid hybrid “NH” (Xie et al., 2015) using specific primers (Supplemental Table S7).

Global protein acetylation profiling

Protein acetylation profiling was performed on fruit pulp collected at 180 DAF using tandem mass spectrometry (MS/MS) in Q Exactive Plus (ThermoFisher Scientific) coupled online to a ultra-performance liquid chromatrography (UPLC). Differentially Kac sites were identified separately for each replicate to identify upregulated and downregulated Kac sites. Only Kac sites with >1.5-fold increases or decreases in both replicates were defined as differentially regulated. Details about acetylation profiling and data analysis are described in Supplemental Methods S1.

Enzyme assays

GS2 enzyme activity assays were performed as described by Osanai et al. (2017). NADP⁺-linked IDH enzyme activity was quantified as described previously (Alp et al., 1976). FLS activity was quantified as described previously (Owens et al., 2008).

Subcellular localization

For the colocalization assay, the Agrobacterium tumefaciens (GV3101) strains harboring the GFP fusion constructs were coinfiltrated into Nicotiana benthamiana leaves with Agrobacterium strains harboring a plasmid that contains a transgene that expresses a nucleus marker, 35S:OsGhd7-RFP (Xue et al., 2008), mitochondrial marker 35S:COX4ts-mCherry (Nelson et al., 2007), or a plastid marker, PT-RK (https://nebenfuehrlab.utk.edu/markers/). Nicotiana benthamiana epidermal leaf cells were subjected to a laser-scanning confocal microscopy analysis 48 h after the infiltration (TCS-SP8, Leica, Germany). GFP fluorescence was detected with the following properties: laser, 488 nm; detector, PMT; intensity, 13.0895%; collection bandwidth, 505–550 nm; and gain, 776.2. RFP/mCherry fluorescence was detected with the following properties: laser, 552 nm; detector, PMT; intensity, 8.7987%; collection bandwidth, 590–640 nm; and gain, 827.3.

Sequence alignment and phylogenetic analysis

Putative GS2, NADP-IDH1, and FLS orthologs of different species were identified by BLASTP searches using the corresponding C. grandis protein sequences as query against databases integrated in Phytozome v13.0 (https://phytozome-next.jgi.doe.gov/). Sequence alignment was performed with ClustalW in MEGA7, and the shading of the alignment was generated with GeneDoc software.

KAT and KDAC protein sequences from A. thaliana were obtained from the TAIR database (https://www.arabidopsis.org/index.jsp) and were used to perform a search of the C. grandis genome using BLASTP with an E-value threshold of 1e−5. Meanwhile, the PFAM database (http://pfam.xfam.org/) was used to analyze the sequences of the predicted proteins. The phylogenetic tree was constructed using MEGA7, with default settings and the neighbor-joining method.

Protein degradation assays

For protein stability analysis, N. benthamiana leaves expressing FLAG-tagged RPL5 WT and mutant constructs were harvested 48 h after the infiltration and immunoprecipitation was performed as previously described (Liu et al., 2017). A cell-free protein degradation assay was conducted as previously described (Wang et al., 2009). Immunoprecipitated proteins were incubated in citrus peel extracts and assayed at the indicated time points. Protein abundance was determined using western blotting with an anti-FLAG antibody. The immunodetection of plant actin was used as a loading control. Proteins were detected using the enhanced chemiluminescence Plus reagents (Yeasen, Shanghai, China). The results were quantified using ImageJ.

Data availability

The mass spectrometry proteomics and acetyl-proteomics data were deposited at the ProteomeXchange Consortium using the PRIDE partner repository with the dataset identifiers PXD029778 and PXD029779, respectively. The RNA sequencing data have been deposited in the NCBI BioProject database under the accession number PRJNA776249. The metabolomics data were deposited to EMBL-EBI MetaboLights database with the identifier MTBLS5422.

Other methods

Experimental procedures for metabolite analyses, construction of plasmid vectors, RNA extraction and RT-(q)PCR analysis, homology modeling of protein 3D structure, protein expression and purification, transformation of citrus callus, and deacetylase inhibitor treatments are provided in Supplemental Methods S1. Primers used in this study are listed in Supplemental Table S7.

Accession numbers

Sequence data from this article can be obtained online (http://citrus.hzau.edu.cn/index.php) with the following accession numbers: CgGS2 (Cg6g018300), CgNADP-IDH (Cg3g006900), CgFLS (Cg1g009050), CgRPL5 (Cg2g017280), CgSRT2 (Cg1g024500), and CgHDA8 (Cg8g012880).

Supplemental data

The following materials are available in the online version of this article.

Supplemental Table S1. Proteins with lysine acetylation sites identified in the NH allotetraploid and the two parents.

Supplemental Table S2. Acetylated proteins involved in important citrus fruit traits.

Supplemental Table S3. Twelve bins of differentially Kac sites and acetylated proteins based on acetylation levels among the three genotypes.

Supplemental Table S4. RNA-seq data from the NH allotetraploid and the two parents.

Supplemental Table S5. Details on the proteins identified in the NH allotetraploid and the two parents.

Supplemental Table S6. Genes identified in both the transcriptomes and proteomes from the NH allotetraploid and the two parents.

Supplemental Table S7. Primers used in this study.

Supplemental Figure S1. Characterization of the NH allotetraploid and its two diploid parents with molecular markers and a fruit ripening analysis.

Supplemental Figure S2. Lysine acetylome analysis in citrus.

Supplemental Figure S3. Acetylome comparison analysis between the NH allotetraploid and its two parents.

Supplemental Figure S4. Overview of acetylated proteins involved in fruit acidity metabolism in citrus.

Supplemental Figure S5. Multiple sequence alignment of GS2 from different plant species.

Supplemental Figure S6. Multiple sequence alignment of NADP-IDH1 from different plant species.

Supplemental Figure S7. Gene expression analysis of flavonol synthase (CgFLS) at four fruit ripening stages and multiple sequence alignment of FLS from several other plant species.

Supplemental Figure S8. Summary of transcriptome and proteome data.

Supplemental Figure S9. Acetylation levels of different protein categories classified by transcriptome and proteome data.

Supplemental Figure S10. Phylogenetic analysis of genes encoding KATs and KDACs in citrus.

Supplemental Figure S11. Phenotypic characteristics of citrus pulp after KDAC inhibitors treatment.

Supplemental Figure S12. Phenotypic characteristics of citrus callus after KDAC inhibitors treatment.

Supplemental Methods S1.

Supplemental File S1. DNA sequence alignment for CgHDA8 between the NH allotetraploid and its two parents.

Supplemental File S2. DNA sequence alignment for CgSRT2 between the NH allotetraploid and its two parents.

Funding

This research was financially supported by the Ministry of Science and Technology of China (no. 2018YFD1000200), the National Natural Science Foundation of China (nos. 31820103011 and 32172525), and the Foundation of Hubei Hongshan Laboratory (no. 2021hszd009).

 

W.-W.G. and F.-Q.T. designed the research. M.Z. performed the research with contributions from Y.-J.F., T.-T.W., X.S., K.-D.X. M.Z. analyzed the data. F.-Q.T. and M.Z. wrote the manuscript. W.-W.G., J.W.G., X.-X.D., F.Z., and X.-M.W. improved the manuscript.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys/pages/general-instructions) is: Wen-Wu Guo (guoww@mail.hzau.edu.cn).