Two cultivars of the sugar beet (Beta vulgaris) were used for this study and were described by Zhang et al. (2017). In brief, two representative cultivars (SD13829 and BS02) were obtained by the screening of numerous cultivars. The SD13829, a mono-germy diploid cultivar with high yield, was purchased from Strube GmbH & Co. KG (Solingen, Germany). The taproot of SD13829 exhibited a high growth rate and high fresh weight, but low sucrose content. The BS02 cultivar, a pluri-germy diploid cultivar with low yield, was bred by the Sugar Beet Physiological Research Institute, Inner Mongolia Agricultural University, China. The taproot of BS02 showed a high sucrose content, but low growth rate and fresh weight. Both of the cultivars were grown on the farm of Inner Mongolia Agricultural University with the plant spacing of 25 cm and row spacing of 50 cm, and we had obtained permission from the farmer to collect plant samples. The sampling strategy was based on our previous results; the time point of beginning rapid growth (59 DAE) and the highest growth rate of taproot (82 DAE) were collected in both cultivars. Nine taproots (three biological replicates) at 59 and 82 DAE were collected and frozen in liquid nitrogen for further analysis.

Arabidopsis thaliana seeds of the Columbia-0 (Col-0) genotype were surface-sterilized with a solution of 2% sodium hypochlorite and 0.5% Tween 20 and then sown on MS medium. After vernalization at 4°C for 3 d, the seeds were cultured at 22°C under a photoperiod of 16-h light/8-h dark and a photon flux density of 45 μmol m⁻² s⁻¹.

The plant tissues were broken in lysis buffer (with enzyme inhibitors) using the tissue lyser machine and centrifuged at 25,000 × g for 20 min. Five volumes of cold acetone was added to the recovered supernatant and stored at−20°C for 2 h. After centrifuging again, the pellets were dissolved with lysis buffer, and the 10 mM DTT was added into the solution at 56°C for 1 h to disrupt disulfide bonds. The solution was mixed with 55 mM IAM with the dark treatment for 45 min. After five volumes of cold acetone mixture and centrifugation, the pellet was dissolved in lysis buffer and the supernatant was stored at −80°C for further analysis.

Total proteins were digested by Trypsin Gold (Promega, Madison, WI, USA) with a ratio of 20:1 at 37°C for 12 h. The digested proteins were labeled with the iTRAQ tags using the eight-plex iTRAQ reagent (Applied Biosystems, Foster City, CA, USA) according to the manufacturer's protocol, and the lyophilized labeled peptides were resuspended with the solution A (95% H₂O, 5% acetonitrile [ACN]; pH 9.8) and loaded onto Gemini C18 column (4.6 × 250 mm). The peptide fractions were eluted with buffer B (5% H₂O, 95% ACN; pH 9.8) at a flow rate of 1 mL/min. The elutions have monitored the absorbance at 214 nm and collected 20 fractions and lyophilized. The fraction samples were performed by the LC-ESI-MS/MS analysis as described previously (Li et al., 2017). Briefly, the data acquisition was performed using the Triple TOF 5600 System (AB SCIEX, Concord, ON, Canada) fitted with a Nanospray III source (AB SCIEX) at the conditions of 15 psi nebulizer gas, 30 psi curtain gas, 2.5 kV ion spray voltage, and an interface heater temperature of 150°C. The total cycle time was fixed to 3.3 s, the Q2 transmission window for 100 Da was 100%, and the dynamic exclusion was set for 1/2 of peak width (15 s).

MGF file was obtained from raw data through Proteome Discoverer 1.2 (Thermo Fisher Scientific, Bremen, Germany) software analysis, and the protein identification used the Mascot software (Matrix Science, London, UK; version 2.3.02) against the database of sugar beet genome. The allowable mass tolerance of the intact peptide mass and fragmented ions were 0.1 and 0.05 Da (ppm) in protein identification, respectively, and a missed cleavage was allowed in the trypsin digest. The variable modifications were the Gln- > pyro-Glu (N-term Q), deamidated (NQ), and oxidation (M), and the fixed modifications were the iTRAQ8plex (K), carbamidomethyl (C), and iTRAQ8plex (N-term). The +2 and +3 were the charge states of peptides. The automatic decoy database search was carried out using the Mascot software. The identified peptides have 95% significance confidence interval to reduce the possibility of false peptide identification. All proteins in the proteome and all genes in the transcriptome were compared using BLAST reciprocal best hit analysis (Altschul et al., 1990). DEGs were screened according to the |log2 ratio| ≥ 1 (P < 0.001) and false discovery rate ≤ 0.001. Differentially expressed proteins (DEPs) were screened by the fold change of proteins ≥ 1.5 (P < 0.05). The shared DEGs and DEPs were identified from proteome and transcriptome data. Functional annotations of the DEPs or DEGs were conducted using the Blast2GO program and the non-redundant protein database (NR; NCBI). P ≤ 0.05 was used to confirm the significance of the GO, KEGG pathway, and MapMan analysis results.

The BvXTH8 open reading frame was digested from the pBoLn-T-BvXTH vector with SacI and SalI and subcloned into the SacI/SalI site of the pCAMBIA1300 plasmid to construct pCAMBIA1300-BvXTH8, which was used for transformation of A. thaliana through Agrobacterium-mediated inflorescence infiltration method (Kim et al., 1999). Transgenic homozygous lines were prepared by hygromycin resistance screening and self-pollination. The homozygous transgenic plants were detected by genomic PCR and real-time RT-PCR. The genomic PCR was performed using the following primers: XTH-F1 (CCCTATATGGCTTCCTCCTC) and XTH-R1 (CAGCAAAACTGCAGTCAGAGT). Real-time RT-PCR was performed using the following primers: XTH-F2 (GTCAGCGGTCAACCATACAC) and XTH-R2 (ACCTTGTGTTGCCCAATCAT). β-actin was used as an internal control.

The XTH enzyme was obtained as previously reported (Sulova et al., 1995). Briefly, 2 g samples were ground in liquid nitrogen, 1 mL of the 10 mM sodium phosphate buffer (pH 7.0) was added, and the mixture was centrifuged at 10,000 × g for 25 min at 6°C. The precipitate was washed twice with 10 mM sodium phosphate solution (pH 7.0) and then centrifuged at 6°C 10,000 × g for 18 min. The pellet was resuspended in 10 mM sodium phosphate solution at 4°C for 24 h and then centrifuged at 4°C 10,000 × g for 15 min. The supernatant was stored at 4°C and used to detect enzyme activity.

XTH activity was assayed in a reaction solution containing 50 μL of 2 mg/mL xyloglucan, 50 μL of 2 mg/mL xyloglucan oligosaccharides, 50 μL of 400 mM sodium citrate, and 50 μL of the XTH enzyme solution. XTH activity was expressed as the difference measured in the absence and presence of the XTH enzyme solution. After incubating at 37°C for 30 min, 1 M sodium hydroxide was added to terminate the reaction. A freshly prepared solution of 200 μL KI/I2 and 800 μL 20% Na₂SO₄ was then added to the reaction, left in a dark room for 30 min, and colorimetrically examined at 620 nm (UV752N; INESA, Shanghai, China). For the blank, 50 μL of sodium phosphate solution was used instead of 50 μL XTH enzyme solution.

Transgenic Arabidopsis seeds were grown on MS medium, and the root length and seedling fresh weight were compared at 2 weeks after seed germination. Ten plants were measured for each transgenic line. The root tips of 10-day-old transgenic seedlings were used for microscopic observation. A 2-mm-long sample from the root tip was used for temporary loading. The cells were observed microscopically adjacent to the elongation zone, and the area of all clearly contoured cells in the range of 2 mm was determined. Five biological replicates were measured for each sample.

RNA-seq was performed for SD and BS at 59 and 82 DAE. For each sample, 82.23–83.17% of the clean reads were mapped to the genome of Beta vulgaris (Dohm et al., 2014), and the uniquely matched clean-read percentages ranged from 70.81% to 71.55% (Supplementary Table 1). DEGs were screened according to the |log2 ratio| ≥ 1 (P < 0.001) and false discovery rate ≤ 0.001. A total of 3183 (BS59-VS-BS82), 966 (BS59-VS-SD59), 1164 (BS82-VS-SD82), and 3250 (SD59-VS-SD82) DEGs were found, including 1513, 523, 496, and 1371 upregulated genes and 1670, 443, 668, and 1879 downregulated genes (Figure 1A). Previously, we validated transcriptomic data by qRT-PCR, and the qRT-PCR results were highly consistent with the RNA-seq data, suggesting that the transcriptome data were reliable (Zhang et al., 2017).

The quantitative proteomic analyses were performed for SD and BS at 59 and 82 DAE by iTRAQ platform. A total of 338,982 spectra were generated, 64,267 spectra and 22,197 peptides were matched, and 59,328 unique spectra, 22,197 peptides, and 5,827 proteins were identified with the Q-value ≤ 0.01 (Supplementary Figure 1A). The protein mass distribution, length of peptides, and peptide number distribution were also investigated (Supplementary Figures 1B–D). Differentially expressed proteins (DEPs) were screened by the fold change of proteins ≥1.5 (P < 0.05). A total of 673 (BS59-VS-BS82), 373 (BS59-VS-SD59), 581 (BS82-VS-SD82), and 686 (SD59-VS-SD82) DEPs were found, including 287, 152, 265, and 304 upregulated genes and 386, 221, 316, and 382 downregulated genes, respectively (Figure 1B). The more DEPs were found in comparative groups of two growth stages (59 vs. 82), indicating that these DEPs may play important roles in the process of taproot growth and development.

The correlation analysis of iTRAQ and RNA-seq data was performed, and the correlation coefficient was calculated in four comparison groups (BS59-VS-BS82, BS59-VS-SD59, BS82-VS-SD82, and SD59-VS-SD82; Supplementary Figure 2). The expression levels of proteins and their corresponding transcripts in four comparison groups exhibited lower correlation (r = 0.2316, 0.2304, 0.2312, and 0.2181; Figure 2A), but the DEPs and DEGs showed a higher correlation (r = 0.6189, 0.7714, 0.6803, and 0.7056; Figure 2B). The same or opposite trend correlation coefficients of the DEPs and DEGs also exhibited higher positive or negative correlation (Figures 2C,D). In addition, the lower correlations were observed between DEPs and their corresponding non-additive DEGs, non-additive DEPs and their corresponding DEGs, or non-additive DEPs and their corresponding non-additive DEGs (Supplementary Figure 3).

The correlation analysis showed that 5724 (BS59-VS-BS82), 5726 (BS59-VS-SD59), 5720 (BS82-VS-SD82), and 5711 (SD59-VS-SD82) identification genes were correlated between the proteomic and transcriptomic, respectively (Figure 3A). Proteomic analysis showed that 673 (BS59-VS-BS82), 373 (BS59-VS-SD59), 581 (BS82-VS-SD82), and 686 (SD59-VS-SD82) DEPs were identified. A total of 621 genes (hereafter called cor-DEGs-DEPs), including 190, 71, 140, and 220 in the BS59-VS-BS82, BS59-VS-SD59, BS82-VS-SD82, and SD59-VS-SD82 groups, were regulated at both the mRNA and protein levels, respectively (Figure 3B). Among the 621 cor-DEGs-DEPs, 549 genes exhibited the same trend, including 163, 65, 124, and 198 in the BS59-VS-BS82, BS59-VS-SD59, BS82-VS-SD82, and SD59-VS-SD82 groups, respectively. A total of 72 cor-DEGs-DEPs exhibited the opposite trend (Supplementary Table 2), including 27, 6, 17, and 22 in the BS59-VS-BS82, BS59-VS-SD59, BS82-VS-SD82, and SD59-VS-SD82, respectively.

GO annotation results showed that 373 of the 621 cor-DEGs-DEPs were successfully annotated, including 103, 37, 90, and 143 cor-DEGs-DEPs found in the BS59-VS-BS82, BS59-VS-SD59, BS82-VS-SD82, and SD59-VS-SD82 groups (Figure 4). Among the four groups, 40 GO terms were identified and covered a wide range of cellular components, molecular functions, and biological processes. The two largest subcategories were found in the “biological processes” category, including “cellular process” and “metabolic process.” In the “cellular component” category, “cell” and “cell part” were the most abundant GO terms. In the “molecular function” category, two largest subcategories were “binding” and “catalytic activity.”

To better understand the function of cor-DEGs-DEPs, the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis was performed using a P-value of less than 0.05 as the cutoff, and the result showed that 138 of the 190 (BS59-VS-BS82), 51 of the 71 (BS59-VS-SD59), 112 of the 140 (BS82-VS-SD82), and 170 of the 220 (SD59-VS-SD82) were mapped to 72, 38, 70, and 79 KEGG pathways, respectively (Supplementary Table 3). Two KEGG pathways were highly annotated at both the transcriptional and translational levels in the four groups, including “starch and sucrose metabolism” (ko00500) and “amino sugar and nucleotide sugar metabolism” (ko00520). Moreover, the KEGG pathways of “phenylpropanoid biosynthesis” (ko00940), “biosynthesis of amino acids” (ko01230), and “glutathione metabolism” (ko00480) were significantly enriched in the BS59-VS-SD59 and BS82-VS-SD82 groups, indicating that these processes can be highly differentiated between the SD and BS. In the BS59-VS-BS82 and SD59-VS-SD82 groups, “RNA transport” (ko03013), “protein processing in endoplasmic reticulum” (ko04141), “glyoxylate and dicarboxylate metabolism” (ko00630), “glycine, serine, and threonine metabolism” (ko00260), and “plant hormone signal transduction” (ko04075) were apparently enriched among the cor-DEGs-DEPs. Two cor-DEGs-DEPs were found in the ko04075 pathway. Bv_25960_dhcc.t1 encodes leucine-rich repeat receptor kinase (BRI1), which acts as a brassinolide receptor involving in brassinosteroid signal transduction. Bv2_044480_uecx.t1 encodes gibberellin receptor (GID1) that interacts with DELLA proteins and participates in ubiquitin-mediated proteolysis. These results suggest that the brassinosteroid and gibberellin signal transduction may play an important regulatory role in the taproot growth and development of sugar beet.

GO analysis of the DEGs and DEPs was performed in four comparative groups at the transcriptome and proteome levels, respectively. The results showed that several subcategories that are associated in the “cellular component” category were “external encapsulating structure,” “membrane,” “cell periphery,” “cytoplasmic part,” and “cell wall.” In the “molecular function” category, the correlation subcategories were “catalytic activity,” “oxidoreductase activity,” “hydrolase activity,” “kinase activity,” and “cation binding.” For the “biological process” category, “response to stress,” “single-organism metabolic process,” and “metabolic process” were correlated among the four groups (Figure 5).

To better understand the regulatory mechanism of taproot growth and development, GO significant enrichment analysis for the DEGs and DEPs was performed using a P-value of less than 0.05 as the cutoff, and the correlation analysis of significantly enriched GO terms was conducted between transcriptome and proteome (Supplementary Figure 4). In the BS82-VS-BS59 groups, four of the significant GO enrichment terms in the proteome and transcriptome were mainly in the “cellular component” category. Ten terms were observed in the SD82-VS-SD59 group, including six in “cellular component,” two in “molecular function,” and two in “biological process.” Five terms were observed in the SD82-VS-BS82 group, including three in “cellular component” and two in “molecular function.” No terms were found in the SD59-VS-BS59 group (Supplementary Figure 5). We systematically integrated these cor-DEGs-DEPs, which were significantly enriched in both the transcriptome and the proteome. The results showed that 10, 32, and 68 cor-DEGs-DEPs were found in the BS82-VS-BS59, SD82-VS-SD59, and SD82-VS-BS82 groups, respectively (Figure 6). Among the three groups, only one cor-DEG-DEP was shared, encoding ribonuclease 1-like protein. Six cor-DEGs-DEPs were shared in the BS82-VS-BS59 and SD82-VS-SD59 groups, including the DEAD-box ATP-dependent RNA helicase (Bv9_207310_uxns.t1), TolB protein-like protein (Bv3_060640_zokm.t1), heat shock protein 83-like (Bv3_053900_qhgs.t1), aspartyl protease family protein 2 (Bv1_003130_spiz.t1), 20 kDa chaperonin, chloroplastic (Bv4u_091190_ygqi.t1), and xyloglucan endotransglucosylase/hydrolase protein 8 (Bv8u_204710_otoo.t1). Eight cor-DEGs-DEPs were shared in the SD82-VS-BS82 and SD82-VS-SD59 groups, encoding the aspartyl protease AED3 (Bv2_043900_thoh.t1), endoglucanase 2 (Bv_38810_ipip.t1), alpha-glucosidase (Bv3_053660_hmht.t1), probable polygalacturonase (Bv5_094840_upzj.t1), basic endochitinase (Bv1_008140_uzgx.t1), acidic mammalian chitinase (Bv8_202140_kacq.t1), ribosome-inactivating protein lychnin (Bv9_211960_arex.t1), and antiviral protein alpha isoform X2 (Bv9_206800_xxdr.t1) (Supplementary Table 4). The results suggested that these cor-DEGs-DEPs might be involved in the expansion of sugar beet taproot.

Our previous study showed that the taproot of both SD and BS grew slowly before 59 DAE, and the growth rate of taproot was highest at 82 DAE. In addition, the growth rate of taproot in SD was significantly higher than that of BS at 82 DAE (Zhang et al., 2017). Some regulatory processes at the RNA level were significantly enriched in both the proteome and transcriptome data by comparison of SD82-VS-BS82, SD82-VS-SD59, and BS82-VS-BS59. For example, one shared enriched protein encoding a ribonuclease 1-like protein (Bv6_153260_pdmn.t1) was found in three comparative groups (Figure 6). This protein catalyzed the hydrolysis of ester linkages within ribonucleic acid by creating internal breaks to promote RNA catabolism, which also plays a role in remobilizing phosphate, particularly when cells senesce or when phosphate is limited. In the SD82-VS-SD59 and BS82-VS-BS59 groups, one shared enriched protein encoded a DEAD-box ATP-dependent RNA helicase (Bv9_207310_uxns.t1), which is necessary for mRNA export from the nucleus and can positively regulate the CBF/DREB transcription factors to enhance plant chilling and freezing tolerance (Gong et al., 2002, 2005). Two shared enriched proteins were found in the comparative groups SD82-VS-SD59 and SD82-VS-BS82 (Figure 6). These proteins were Bv9_211960_arex.t1 and Bv9_206800_xxdr.t1, encoding ribosome-inactivating protein lychnin and antiviral protein alpha isoform X2, which act as RNA glycosylases to catalyze the hydrolysis of N-glycosidic bonds in an RNA molecule, and play an important role in negative regulation of translation, restricting the formation of proteins in many processes (Figure 7). These results suggested that the transcriptional regulatory processes involving these proteins might play a crucial role in the growth and development of sugar beet taproot.

Based on a combined analysis of the transcriptome and proteome, we found that some protein metabolism regulatory processes were significantly enriched by comparison of the SD82-VS-BS82, SD82-VS-SD59, and BS82-VS-BS59 groups. Four shared enriched proteins were found in the SD82-VS-SD59 and BS82-VS-BS59 groups: Bv3_060640_zokm.t1, Bv3_053900_qhgs.t1, Bv1_003130_spiz.t1, and Bv4u_091190_ygqi.t1, encoding TolB protein-like protein, heat shock protein 83-like, aspartyl protease family protein 2, and 20 kDa chaperonin, respectively (Figure 6 and Supplementary Table 4). TolB protein-like proteins can cleave peptide bonds to hydrolyze proteins into smaller polypeptides and/or amino acids. The heat shock protein 83-like maintains the structure and integrity of a protein, prevents it from degrading or aggregating, and promotes folding of single-chain polypeptides or multisubunit complexes into the correct tertiary structure. The aspartyl protease family protein 2 may act as an aspartic-type endopeptidase involved in protein catabolic processes of BAG6 and plant basal immunity (Li et al., 2016). The 20 kDa chaperonin is involved in chaperone cofactor-dependent protein refolding to assist in the correct posttranslational non-covalent assembly of proteins, which is also involved in the positive regulation of superoxide dismutase activity (Bonshtien et al., 2007; Kuo et al., 2013) and negative regulation of the abscisic acid-activated signaling pathway (Zhang et al., 2013) (Figure 7). In the SD82-VS-SD59 and SD82-VS-BS82 groups, Bv2_043900_thoh.t1 was a shared enriched protein, which encodes aspartyl protease AED3 and acts as aspartic-type endopeptidase, catalyzing the hydrolysis of internal alpha-peptide bonds in a polypeptide chain resulting in protein degradation (Figures 6, 7). This protein is also involved in the regulation of programmed cell death. These results suggested that these protein metabolism processes might play a crucial role in the expansion of sugar beet taproot.

The plant cell wall is mainly composed of cellulose, hemicellulose, tannin, and pectin and is involved in plant cell wall metabolism throughout the growth and development of plants. In this study, we found that many cell wall metabolism regulatory processes were significantly enriched in the SD82-VS-BS82, SD82-VS-SD59, and BS82-VS-BS59 groups. For example, Bv8u_204710_otoo.t1 was enriched in SD82-VS-SD59 and BS82-VS-BS59 (Figure 6). This protein encodes xyloglucan endotransglucosylase/hydrolase 8 protein (XTH8), can cleave and reconnect xyloglucan polymer through the hydrolysis of xyloglucan or internal transglycosylation, and then participate in the primary cell wall construction of plant growth tissue. In the SD82-VS-SD59 and SD82-VS-BS82 groups, five enriched genes were observed both in the proteome and transcriptome: Bv_38810_ipip.t1, Bv3_053660_hmht.t1, Bv5_094840_upzj.t1, Bv1_008140_uzgx.t1, and Bv8_202140_kacq.t1 (Figure 6). Bv_38810_ipip.t1 encodes an endoglucanase 2, which catalyzes the hydrolysis of beta-D-glucosidic linkages in cellulose, lichenin, and cereal beta-D-glucans, and may be involved in the sloughing (cell–cell separation) of root cap cells from the root tip of cell wall breakdown, which is important in plant development because it assists penetration of the growing root into the soil (Campillo et al., 2004). Bv3_053660_hmht.t1 encodes an alpha-glucosidase that catalyzes the hydrolysis of terminal alpha-D-glucosidic links in alpha-D-glucans, which are essential for stable accumulation of EFR (Burn et al., 2002; Lu et al., 2009). Bv5_094840_upzj.t1 is a probable polygalacturonase involved in cell separation in the final stages of pod shatter, resulting in the breakdown of the cell wall, and the pectin catabolic process, resulting in the breakdown of pectin, a polymer containing a backbone of alpha-1,4-linked D-galacturonic acid residues (González-Carranza et al., 2007; Ogawa et al., 2009). Bv1_008140_uzgx.t1 and Bv8_202140_kacq.t1 encode an endochitinase, which catalyzes the hydrolysis of N-acetyl-beta-D-glucosaminide (1->4)-beta-linkages in chitin and chitodextrins and are involved in cell wall macromolecule, chitin, and polysaccharide catabolism (Figure 7). These results suggested that cell wall metabolism plays important roles in sugar beet root enlargement.

To analyze the functions of BvXTH8 (Bv8u_204710_otoo.t1), we constructed the pCAMBIA1300-BvXTH8 plasmid and introduced it into Arabidopsis plants through Agrobacterium-mediated transformation (Figure 8A). Homozygous transgenic lines were obtained using the self-pollination. Three homozygous lines overexpressing BvXTH8 were validated by genomic PCR (Figure 8B). Real-time RT-PCR results indicate that the BvXTH8 transcriptional levels in transgenic plants were significantly higher than in the wild type (WT) (Figure 8C). Measurement of XTH enzyme levels from WT and transgenic plants revealed that transgenic plants expressing BvXTH8 had higher XTH activity than the WT (Figure 8D). These results suggested that the BvXTH8 is stably expressed at transcript and protein levels in Arabidopsis.

Transgenic Arabidopsis seedlings overexpressing BvXTH8 exhibited higher primary root lengths and fresh weights compared with WT plants at 2 weeks after sowing (Figures 8E–G). The microscopic observation of cells in the elongation zone showed that cells in the root tip elongation zone of transgenic plants were significantly longer than that of WT (Figure 8H), and cells in the elongation zone of transgenic plants were significantly increased compared to the WT (Figure 8I). These results indicate that BvXTH8 can enhance the growth and development of Arabidopsis primary roots by improving cell growth in the root tip elongation zone.