Genome-Wide Identification of Tannase Genes and Their Function of Wound Response and Astringent Substances Accumulation in Juglandaceae

Tannins are important polyphenol compounds with different component proportions in different plant species. The plants in the Juglandaceae are rich in tannins, including condensed tannins and hydrolyzable tannins. In this study, we identified seven tannase genes (TAs) responsible for the tannin metabolism from walnut, pecan, and Chinese hickory, and three nut tree species in the Juglandaceae, which were divided into two groups. The phylogenetic and sequence analysis showed that TA genes and neighboring clade genes (TA-like genes) had similar sequences compared with other carboxylesterase genes, which may be the origin of TA genes produced by tandem repeat. TA genes also indicated higher expressions in leaf than other tissues and were quickly up-regulated at 3 h after leaf injury. During the development of the seed coat, the expression of the synthesis-related gene GGTs and the hydrolase gene TAs was continuously decreased, resulting in the decrease of tannin content in the dry sample of the seed coat of Chinese hickory. However, due to the reduction in water content during the ripening process, the tannin content in fresh sample increased, so the astringent taste was obvious at the mature stage. In addition, the CcGGTs’ expression was higher than CiGGTs in the initiation of development, but CcTAs continued to be down-regulated while CiTA2a and CiTA2b were up-regulated, which may bring about the significant differences in tannin content and astringent taste between Chinese hickory and pecan. These results suggested the crucial role of TAs in wound stress of leaves and astringent ingredient accumulation in seed coats of two nut tree species in the Juglandaceae.


INTRODUCTION
Tannins are the fourth largest plant secondary metabolites after cellulose, hemicellulose, and lignin (Lekha and Lonsane, 1997). They are plant polyphenols with a large molecular weight widely distributed in various plant organs (Salminen, 2018). Classically, tannins are mainly divided into condensed and hydrolyzable tannins. Condensed tannins are polymers of flavan-3-ol and/or flavan-3,4-diol, also known as proanthocyanidins because their acidic hydrolysates are anthocyanidins (de Jesus et al., 2012;Combs, 2016); hydrolyzable tannins are gallate esters of polyols (usually D-glucose). Hydrolyzable tannins can be further divided into gallotannins and ellagitannins, and the latter will produce ellagic acid after being hydrolyzed while the former will not (Lamy et al., 2016). It is the most important characteristic of tannins that there are a large number of hydroxyl groups, which can bind to proteins in the form of hydrogen bonds, causing changes in protein conformation. Tannins also play an important role in biotic and abiotic stresses of plants, such as protecting plants from predators and pathogens (Treutter, 2006;Nakabayashi et al., 2014). In addition, a variety of phenolic substances in tannins also have great impacts on human health.
Tannase (TA), also known as tannin acyl hydrolase (EC 3.1.1.20), could hydrolyze galloyl ester bonds in hydrolyzable tannins and other gallate esters but does not act on condensed tannins . Tannases were mainly found in microorganisms such as fungi and bacteria, and the enzymatic properties and protein structure of microbial tannases have been extensively studied (Ren et al., 2013;Jana et al., 2014;Abdel-Naby et al., 2016). At present, there are few studies on plant tannases: some researchers found the phenomenon of substrate hydrolyzed via the enzymatic tests in vitro on UDP-glucose-dependent glycosyltransferases related to gallate tannin biosynthesis, which may be due to some unknown esterase involvement (Weisemann et al., 1988;Cammann et al., 1989;Denzel and Gross, 1991). Subsequently, Niehaus and Gross (1997) isolated and purified this esterase from leaves of pedunculate oak (Quercus robur), which indeed could hydrolyze galloylglucose and was similar to fungal tannase, and classified it as plant tannase. However, the gene encoding plant tannase has not been characterized for many years due to the limitation of analytical techniques. Recently, plant TA genes from tea trees were first discovered and characterized, and TA genes were also distributed in some plants rich in tannins (Dai et al., 2020).
The tree species of Juglandaceae such as walnut (Juglans regia), pecan (Carya illinoinensis), and Chinese hickory (Carya cathayensis), as well as some precious timber species such as black walnut (Juglans nigra) have high economic values . They have been cultivated, domesticated, and utilized by human beings for a long time (Guo et al., 2020). The plants in the Juglandaceae are rich in tannins, both hydrolyzable and condensed tannins, especially in seeds, barks, and roots (Gong and Pegg, 2017;Jia et al., 2018;Jahanban-Esfahlan et al., 2019). However, there are some differences in the composition of tannins among different species in seeds: walnut and Chinese hickory nuts may have more hydrolyzable tannins, especially ellagitannins, while pecan has more condensed tannins (Regueiro et al., 2014;Gong and Pegg, 2017).
It is known that there are TA genes in walnut, which may regulate tannin composition (Dai et al., 2020), and whether other species in the Juglandaceae have TA genes has not been reported. The study of TA genes in the Juglandaceae can help us further understand the important role of tannins in the difference in astringent taste between different species in the Juglandaceae. In this study, we systematically identified TA genes and other homologous genes in nine plants based on the published genomic and transcriptional data. Seven TA genes were identified from walnut, pecan, and Chinese hickory, three important nut tree species in the Juglandaceae. Their motif composition, gene structure, chromosome localization, and miRNA prediction were comprehensively analyzed. At the same time, we measured the expression levels of TA genes in roots, stems, leaves, flowers, peels, testae (seed coats), and embryos. The expression changes of GGTs and TAs in response to leaf injury stress were further analyzed. Finally, the combination of RT-qPCR and HPLC results led us to discover the relationship between TA genes and tannin content changes during fruit development, preliminarily explaining the formation of the difference in astringency between Chinese hickory and pecan. These results revealed that TA genes may play a crucial role in the metabolism of tannins in the Juglandaceae, which will be good for future tannin research of other plants.

Plant Materials and Growth Conditions
Chinese hickory (landrace "ZAFU-1") and pecan (cultivar "Mahan") plants, planted at the farm of Zhejiang Agriculture and Forestry University (Hangzhou, China), were used as experimental materials. Mature female flowers were collected from late April to early May, and roots, stems, buds, leaves, peels, embryos, and testae (seed coats) were collected in June to October. In order to mimic the effect of chemical defenses on herbivory, the upper third compound leaves of each leaflet were cut off with scissors before and after treatment for 3, 6, 12, 24, and 48 h, respectively. For the tannin determination of seed coat, we collected five periods of Chinese hickory (CcS1-CcS5) as well as ripe period pecan and walnut (CiS5 and JrS5). The walnut (cultivar "Xiangling") was sampled from Tiantongyuan Company (Tianshui, Gansu, China). All plant samples were frozen in liquid nitrogen prior to storage at −80 • C until use.

Identification of TA Genes in Juglandaceae and Other Plants
The genomes of Chinese hickory (C. cathayensis, Cc) and pecan (C. illinoinensis, Ci) were downloaded from the GIGADB database 1 , and the genome of walnut (J. regia, Jr) was downloaded from Xuehui Huang Lab 2 . The CsTA protein sequence of tea (Camellia sinensis, Cs) was downloaded from the National Center for Biotechnology Information (NCBI). The genome and protein sequence of other species was downloaded from the Phytozome database, including clementine (Citrus clementina, Ccl), persimmon (Diospyros kaki, Dk), strawberry (Fragaria × ananassa, Fa), woodland strawberry (F. ananassa, Fv), pomegranate (Punica granatum, Pg), aspen (Populus tremula, Pt), and grape (Vitis vinifera, Vv). The protein of CsTA was used as a query to search against the protein database of other plants, and the putative proteins were obtained by BLASTP search with a bit score of more than 200 and an E-value less than 1 × 10 −20 . For Chinese hickory and pecan, in order to obtain all potential TA genes, the protein sequence of CsTA was also used to blast the novel transcripts from previous transcriptome data. All potential TA protein sequences were examined by searching for abhydrolase_3 (PF07859) and COesterase (PF00135) domains using the Pfam database 3 and SMART database 4 (Letunic et al., 2012;Finn et al., 2016). All candidate sequences were searched on the whole genome to find the possible genome location of TA genes through TBLASTN.

Sequences Alignment and Phylogenetic Analysis
The protein sequences of previously reported tannase (from plant, bacterial, and fungal) and homologous genes were obtained from the NCBI protein database (Dai et al., 2020). The potential TA protein sequences in the other published genomes were identified by the method described earlier. The multiple sequence alignment of all proteins was performed using MAFFT version 7 8 (Nakamura et al., 2018). A maximum likelihood (ML) phylogenetic tree of full-length protein of all sequences was constructed with 1,000 bootstrap replicates using MEGA10.0 (Tamura et al., 2011). The best model "WAG + G" was predicted by MEGA10.0 due to its lowest Bayesian Information Criterion (BIC) score.

Analysis of Gene Structure, Conserved Motif, and Cis-Acting Elements
The conserved motifs of TAs were searched using MEME 5.1.1 9 (Bailey et al., 2009). It was performed with the following parameters: 20 different motifs, a motif width of 6-50 amino acids, and any number of repetitions. The potential function of motifs was searched against Pfam database (see Text Footnote 3). Plant CARE software 10 was used to predict the cis-acting elements within 2000 bp upstream of all TA genes (Lescot, 2002), and PLACE 11 was used as a supplement to cis-acting elements of brassinosteroid and cytokinin (Higo et al., 1999). The illustrations of gene structures, motifs, and cis-acting elements were then generated using TBtools .

MiRNA Predicted in the Juglandaceae TA Genes
The psRNATarget Server 12 was used to search and predict potential miRNAs of the coding sequences of the Juglandaceae TA genes with default parameters (Dai and Zhao, 2011). Cytoscape software was used to visualize the predicted interaction between miRNA and TA genes in Juglandaceae (Shannon, 2003).

Expression Analysis of TA Genes
Total RNA was extracted from the samples using the Quick RNA isolation Kit (Huayueyang, China), and cDNA was synthesized using the PrimeScript1st TM Strand cDNA Synthesis Kit (Takara, Japan) according to the manufacturer's instructions. The expression levels of TA genes in the roots, stems, leaves, female flowers, peel, young embryos, mature embryos, and seed coats at five developmental stages were measured by RT-qPCR using the CFX96 real-time PCR Detection system (Bio-Rad, United States) with TB Green R Premix Ex Taq (TaKaRa, Japan). The reaction condition was 95 • C for 3 min, followed by 40 cycles at 95 • C for 10 s, and 55 • C for 30 s. Meanwhile, the expression levels of TA genes treated with mimicking herbivory were also measured with the same method. The RT-qPCR primers of TA genes listed in Supplementary Table 5 were obtained by online software primer 3 13 . The relative expression was calculated based on the 2 − CT method (Livak and Schmittgen, 2001), and the expression of actin gene, which had the same sequences in Chinese hickory and pecan, was monitored as an internal control. Three biological replicates were performed in the RT-qPCR experiment, and three technical replicates were performed in each biological replicate.

SDS-PAGE Analysis of Salivary Protein Precipitated by Seed Coats Extracts
Saliva was collected from six healthy non-smoking volunteers and 2 ml of saliva from each volunteer was used to make a saliva pool (whole saliva). The collection time was standardized from 2 to 3 pm to reduce the concentration changes associated with circadian rhythm secretion. Samples were collected by draining saliva into a cold tube. All samples were collected and centrifuged at 4,000 g for 20 min at 4 • C to remove all insoluble matter. The obtained supernatant was divided equally and immediately frozen at −80 • C, which was called whole saliva (WS) (Ramos-Pineda et al., 2020).
The 20 mg of seed coat lyophilized sample was dissolved in 1 ml of distilled water as the sample solution, and the concentration of which was noted as 20 mg seed coat dry weight/ml and diluted with distilled water to three concentrations (0.625, 1.25, and 2.5 mg/ml). The 200 µl of whole saliva was mixed with 200 µl of seed coat extracts or distilled water, followed by vortexing for 10 s, and incubating for 20 min at 37 • C. The mixtures were centrifuged at 12,000 g for 5 min, and the precipitate was discarded. Then one-quarter volume of 5× sample loading buffer (250 mM Tris-HCl, 10% SDS, 0.5% Bromophenol blue, and 50% Glycerol, pH 6.8) was added to the supernatant.
The detection of salivary proteins after reaction with seed coat extracts was performed by SDS-PAGE using the DYCZ-24B vertical electrophoresis system (LiuYi, China). Forty microliters of each treated sample was electrophoresed on an 8 cm × 7 cm and 1.5-mm-thick, 13.5% w/v denaturing polyacrylamide gel, covered with a 5% w/v polyacrylamide stacking gel. Protein markers (Bio-Rad, United States) in the molecular weight range (10-250 kDa) were also loaded. Electrophoresis is performed at a constant voltage of 75 V for 30 min, followed by switching the voltage to 120 V until the tracking dye (bromophenol blue) reaches the bottom of the gel. The gels were stained with Coomassie Brilliant Blue R250 Staining Solution (0.2% Coomassie blue R250, 10% acetic acid, and 45% methanol) and rinsed overnight using a destaining solution (10% acetic acid and 23.75% ethanol).

Astringency Evaluating Assay
An astringency evaluation method based on the precipitation of tannins by protein was used to evaluate astringency, modified from Llaudy et al. (2004) and Jauregi et al. (2016). We modified this method by replacing ovalbumin with bovine serum albumin (BSA) and adjusting the buffer. Solutions of BSA at 0.4-3.2 mg/ml, solutions of tannic acid at 0.2-1.0 mg/ml, and 2.0 mg/ml seed coat extracts were prepared using 100 mM acetate buffer solution (pH = 5.0), respectively. Two hundred microliters of tannic acid solution and 200 µl of ovalbumin solution were mixed and vortexed for 10 s, and after 10 min, 12,000 g was separated for 10 min. One hundred microliters of supernatant was diluted 50-fold, and the absorbance value at 280 nm was detected using a UV-2600 UV-vis spectrophotometer (Shimadzu, Japan).

Determination of Seed Coat Phenolic Compounds
After manual peeling, the seed coats were freeze-dried for 48 h with a lyophilizer (Christ Alpha 2-4 LD plus, Germany) and ground into powder. Take 20 mg of the sample, add 1.4 ml of 80% aqueous acetone solution, and leave it overnight at 4 • C, followed by ice bath of ultrasonic extraction for 2 h. After centrifugation (12,000 g, 5 min), the supernatant was concentrated in a Rotational Vacuum Concentrator (RVC 2-25 CD plus, Germany) for 1 h to remove acetone, the sample residue was extracted once with a new aqueous acetone solution, and the supernatant was pooled twice and vacuum concentrated to remove acetone. The sample solution was then freeze-dried for 24 h to obtain the lyophilized powder, and 1 ml of methanol was added to redissolve it.

Subcellular Localization Analysis
The protein subcellular localization was performed by Agrobacterium tumefaciens-mediated transient expression in Nicotiana benthamiana leaves. The full-length CDS of TAs was amplified using the gene-specific primers and subcloned into pENTR-D-TOPO (Invitrogen, United States). Sequences of CiTA2a and CiTA2b were highly similar, and one universal primer was designed for analysis. After validation by the sequencing, full-length TAs were cloned into pK7FWG2 vector with EGFP reporter gene by LR reaction. The recombinant plasmids were introduced into A. tumefaciens strains GV3101 competent cell and cultured on the LB medium with 50 µg/ml gentamicin (Geta), 50 µg/ml rifampicin (Rif), and 50 µg/ml kanamycins (Kana) at 28 • C in the constant temperature incubator. Two days later, a single colony was transferred into lipid LB medium and cultured for another 2 days at 28 • C. Then, the cultures of A. tumefaciens (OD600 = 0.5-0.6) were centrifuged at 5,000 rpm at room temperature for 10 min and re-suspended in MMA buffer (10 mM MES, 10 mM MgCl 2 , and 200 µM acetosyringone, pH = 5.6) to an OD 600 of 1.0 and then incubated at room temperature in the darkness for 2-3 h. Subsequently, we injected the agrobacterium cultures into the 4-week-old and well-growing N. benthamiana leaves using a 1-ml syringe. Two days after the culture, GFP fluorescence was observed and examined using laser confocal fluorescence microscopy (excitation: 488 nm; emission: 495-515 nm) (LSM 800, Zeiss, Germany). The experiments were repeated three times.

Statistical Analysis
For the expression of each gene in the figures, multiple comparisons among different samples were performed using Frontiers in Plant Science | www.frontiersin.org Tukey's honestly significant difference (Tukey's HSD) with HSD.test function in R package "agricolae." Different letters above the columns indicate statistically significant differences between groups (P < 0.05).

Identification and Characterization of TA Genes in the Juglandaceae
Protein blast results revealed that many proteins showed high identification with CsTA in each plant species. Among the similar sequences, we excavated 7 TA genes from the genome of Chinese hickory, pecan, and walnut ( Table 1 and Supplementary  Tables 1, 2). The results indicated that walnut and Chinese hickory had two TA genes, while pecan had three TA genes. These TA proteins in length ranged from 303 to 368 amino acids, with molecular weights from 33.21 to 40.49 kDa and theoretical isoelectric points ranging from 5.52 to 6.17. The average protein length, MW, and hydrophilicity in pomegranate are bigger than in other species. The average pI value of 6.055 in walnut is relatively bigger than others, but the average pI is only 5.51 in strawberry. The GRAVY value of all TA proteins was shown to be less than 0 (varying from −0.307 to −0.111), indicating their hydrophilic feature. Compared with other species, pecan and pomegranate both had a protein, CiTA1 and PgTA2, which had a significantly longer length, bigger molecular weights, and lower PI. Subcellular localization analysis indicated that the vast majority of TA genes are localized in the cytoplasm, except CiTA1 and PgTA2 are located in the plastid ( Table 1). The results of signal peptide analysis indicated that only CiTA1 and PgTA2 contained an N-terminal signal peptide, and TM domain analysis showed that all of TAs do not possess TM domains. To determine the subcellular localization of TA genes, the TA-EGFP fusion proteins in the tobacco leaves were examined under a confocal microscope. As shown in Figure 1, the TA-GFP fluorescent signals of all tannase proteins in three species were observed in the cytosol and nucleus. This result indicated that TAs may be predominantly hydrolyzed substrates in the cytosol and nucleus.

Sequence Alignment and Phylogenetic Analysis
To examine the phylogenetic relationships of plant tannases, a phylogenetic tree of plant carboxylases to which the tannases belonged was constructed (Figure 2). In the present study, plant carboxylesterases were divided into five clades: methyl esterase, caffeoyl shikimate esterase, carboxylesterase I, acetate esterase, plant tannase, and plant tannase-like. The entire phylogenetic tree was divided into three major clades, with caffeoyl shikimate esterase and methylesterase as the first clade, carboxylesterase I as the second clade, and acetate esterase, plant tannase, and plant tannase-like as the third clade. According to protein blast results, except tannases, all genes that shared high identify with CsTA belonged to plant tannase-like clade.
The methyl esterase evolutionary branch contains AtMES1-20 from Arabidopsis thaliana (Yang et al., 2008). The methyl esterase clade was thought to be related to the hydrolyzable MeJA, MeSA, and MeIAA. Caffeoyl shikimate esterase contains AtCSE, CsCSE, and others involved in lignin formation (Vanholme et al., 2013;Dai et al., 2020). Other carboxylesterase genes could be split into four clades. The typical carboxylesterase grouped carboxylesterase I clade contained numerous genes. Phylogenetic analysis showed that plant tannase, plant tannase-like, and acetate esterase clade were closest to the carboxylesterase I clade. These results suggested that these three clade genes may be derived from carboxylesterase I genes and tannase and tannase-like genes may be derived from acetate esterase genes.
In plant tannase clade, most species contained more than one tannase gene and we found 5 TA genes in tea. CsTA was reported in a previous study and CsTA1-4 had not been reported. CsTAs were grouped with TAs in grape and persimmon, which was consistent with their species evolutionary status. In the other group, TAs are further divided into two classes. TAs in strawberry, clementine, and aspen all belong to class 1, while TAs in pomegranate all belong to class 2. Interestingly, for 3 Juglandaceae species (walnut, Chinese hickory, and pecan), all have different TA genes assigned to two classes. This classification may lead to differences in tannase functions in Juglandaceae, such as different catalytic efficiencies for different substrates, or produce different metabolites.

Structure and Conserved Motif Analysis of TA Genes
The exon-intron structure of TA genes was analyzed based on the cDNA and DNA sequences. Results showed that almost all TA genes only contained one exon in different species (Figure 3). The similar single-exon structure was also observed in TA-like genes, and about 91.11% of them are intronless. Only nine genes have more than one exon including one TA gene in pomegranate and eight TA-like genes. In walnut, JrTA-L2 and JrTA-L3 had two and six exons, respectively. In aspen and pomegranate, two genes contained two exons. In strawberry and grape, each had one gene that contained more than one exon.
To understand the diversity of motif compositions among different tannase proteins, the conserved motifs were predicted using MEME. Motifs 1, 2, 3, 4, 6, 7, 8, 9, and 10 were almost distributed in each TA and TA-like protein. Among these nine motifs identified, Motifs 1, 2, and 3 corresponded with abhydrolase_3 domain (PF07859) and Motif 6 corresponded with carboxylesterase domain (PF00135). Motif 5 was also a broadly distributed motif that may play a role in RNA binding and not found in the genes in strawberry, pomegranate, and aspen. According to a previous study (Dai et al., 2020), Motif 7 corresponded with the tannase conserved motif. In alignment result, we also found similar motif 7 in TA-like family, but existed 2 major amino acids change which may lead to their function differentiation (Supplementary Figure 1). Comparing the TA proteins with the TA-like proteins, we found that Motif 14 was more common in TA-like proteins (97.06%) than in TA proteins (33.33%) and this motif was found in all TA proteins of tea and persimmon (Supplementary Table 3 Motif 19 existed in all TA proteins in tea. Further analysis of the three motifs revealed that genes containing Motif 19 are all in tea. These motifs have not found corresponding known domains, which may lead to differences in tannase function in different species.

Analysis of the Promoter Cis-Acting Regulatory Elements of TA Genes
The variances of tannase motif in Juglandaceae may result in the different enzyme activity. The distribution of different cis-acting FIGURE 2 | Molecular phylogenetic analysis of gene families among plant tannase genes and other carboxylesterase genes. The evolutionary history was inferred using the Neighbor-Joining method. The bootstrap consensus tree inferred from 1000 replicates was used to represent the evolutionary history of the tannase genes. Whole tree was divided into five clades, plant tannase clade (with red range), plant tannase-like clade (with blue range), acetate esterase clade (with yellow range), carboxylesterse I clade (with orchid range), caffeoyl shikimate esterase clade (with gray range), and methylesterase clade (with green clade). Previous reported tannase and carboxylesterase genes were marked with a triangle. The Juglandaceae tannase genes in walnut, pecan, and Chinese hickory were marked with asterisks. elements in gene promoters may indicate the differences in their function and regulation when environmental stresses are encountered. To understand the regulatory element of tannases in Juglandaceae, we examined all TA and TA-like genes promoter cis-elements, 2-kb upstream of the ATG start in walnut, pecan, and Chinese hickory (Figure 4 and Supplementary Figure 2). All regulatory elements were grouped into three categories by function, phytohormone responsive, abiotic and biotic stress, and plant growth and development.
The number of motifs between three genes of class 1 and 4 genes of class 2 is almost the same. JrTA2, which was the least motif, only had 35 predicted regulatory elements. The most common motifs found in promoter were E-box (involved in the brassinolide responsiveness) and ARR1AT (involved in the CDSs were shown as green boxes, introns were shown as thin gray lines, and UTRs are shown as yellow boxes. (B) Distribution of conserved motifs among proteins identified using MEME suite program. The motifs, numbered 1-20, were designated with a specific color. The sequence information for each motif was provided in Supplementary Table 3. cytokinin responsiveness). Two motifs, WUN-motif and W-box, are involved in wound-responsive element belonging to abiotic stress and were only found in the gene of class 2. Flavonoid biosynthetic-related motif (MYB) and cell growth promotionrelated motif (MYC) were significantly higher in two genes, CcTA2 and CiTA2a. CiTA2b has more stress-responsive element (STRE) than other genes. This result demonstrates that some identified cis-elements in tannase genes may be involved in phytohormone regulation, wounding, and so on. Two classes of tannase genes may have different regulation ways.

Prediction Target Network of TA Genes and MicroRNA in Chinese Hickory
MicroRNA is a very important mechanism for posttranscriptionally regulation. In order to find the candidate miRNA of TA genes, we predicted the target relationship with psRNAtarget using all plant miRNAs (Supplementary Table 4). The result showed that each TA gene contained multiple sequences that could well-match with miRNA and might be the targets of miRNAs (Figure 5). In total, there were 78 miRNAs that were predicted as candidate regulators of TA genes in  walnut, pecan, and Chinese hickory. The average number of predicted miRNA in each gene was 21 and CiTA1 had the most miRNA target sites. From the result, we found that most miRNAs were found in different TA genes and only a small percentage of miRNAs was unique to each gene. The targeted network showed that two classes of TA genes were basically targeted by different miRNAs. Genes in class 1 had more potential miRNA (50 in total) than class 2 (32 in total), but genes in class 2 had more shared miRNA (18/32) than class 1 (17/50), which implied that genes in class 2 might be more conservative. Notably, there were four miRNAs (miR408, miR909, miR6021, and miR8678) that could target both two classes of genes.

Expression Profiling of TA Genes in Vegetative and Reproductive Tissues
In order to investigate the expression profiles of TA genes, eight main tissues were collected for quantitative real-time PCR, including roots, stems, leaves, female flowers, buds, peels, testae (seed coats), and embryos. Since GGT is a key tannin pathway synthesis gene, we simultaneously quantified its expression pattern (Figure 6 and Supplementary Figure 4). The results showed that the abundance of CcGGT1 in the seed coat was more than 100 times higher than in other tissues and CcGGT2 was both highly expressed in seed coat and leaf. In pecan, CiGGT1 had more than 2000 times higher expression in seed coat than embryo, followed by bud. On the contrary, the abundance of CiGGT2 in leaf, flower, and peel was 50-150 times higher than in seed coat. These results suggest that GGT1 was the main factor to determine the astringent taste in seed coat. GGT2 was involved in the accumulation of tannin in the leaves in addition to the seed coat. This expression pattern suggested that GGT2 played a key role in the resistance of leaves to insect feeding and more tannins may exist in bud and flower in pecan to enhance the response to the environment stress. Compared with the GGT genes with different expression patterns, the pattern of TA genes functioned as tannin acyl-hydrolase was much closer in Chinese hickory and pecan. All five TA genes had high expression in leaves, but low expression in seed coat. Taken together, these results showed that leaves and seed coat were the main tissues of tannin accumulation, and the diverse expression pattern of the synthesis-related gene GGTs and hydrolase gene TAs indicated their important roles in the regulation mechanism of tannins in different tissue.

Role of GGT and TA Genes in Response to Wounding Treatments
According to the expression pattern of GGT and TA genes in different tissues (Figure 6), we found that CcGGT2 and CiGGT2 showed a relatively high expression in leaves. The expression level of TAs was also strongly up-regulated in leaves than in other tissues. In Chinese hickory, the abundance of CcTAs in leaves was only several times higher than other tissues, but in pecan, CiTAs were up to 200 times higher than other tissues. The high expression of two tannin-associated genes in leaves may be related to the resistance mechanism of plants to insect feeding (Barbehenn and Peter Constabel, 2011;Moctezuma et al., 2014). To provide insight into possible physiological roles of TA genes, we simulated insect herbivory under controlled conditions by leaf damage stress to investigate the expression of these genes. All TA and GGT genes in Chinese hickory and pecan were detected using RT-qPCR at 0, 3, 6, 12, 24, and 48 h after wound stress (Figure 7 and Supplementary Figure 5). When leaves were stressed by external damage, CcGGT1 and CiGGT1 quickly reached the maximum expression level in 3 h, and then immediately returned to normal expression. In contrast, CcGGT2 and CiGGT2 were down-regulated initially and returned to normal expression after 24 h. Five TAs shared the same expression pattern: rapidly reached the maximum expression in 3 or 6 h and then decreased to the normal level. Among them, CiTA1 and CiTA2a were up-regulated slightly later and peaked in 6 h. This result suggested that Chinese hickory and pecan might own a rapid tannin-dependent defense mechanism to resistance to insect herbivory. Within 3 h after leaf injury, tannin synthetase and hydrolase genes were highly expressed to synthesize a series of tannin-related substances to resist insects. After 6 h, these genes returned to normal levels, which may mean the end of the defense process. Interestingly, the expression of CcGGT2 and CiGGT2 was completely opposite to GGT1 genes. The expression of GGT2 decreased in 3 h and increased after 24 h later. It is possible that GGT2 does not participate in the defense against insects, or GGT2 was only involved in the reconstruction of defense at the late stage (after 24 h) and this mechanism is worth revealing.

Expression Pattern of TA Genes During Embryo Development in Seed Coats
The seed coat is the main tissue of astringency source in the edible part in Juglandaceae. The astringency FIGURE 6 | Expression analysis of GGT and TA genes in different tissues in Chinese hickory and pecan by RT-qPCR. The analysis was carried out using three biological replicates and three technical replicates for each sample. The error bars represented the standard deviations of nine replicates. Different letters indicated significant differences according to the Tukey-Kramer test (P < 0.05). gradually accumulated with the extension of the seed development period.
The quantitative real-time PCR results of different developmental stages of the seed coat showed that both GGT1 and GGT2 were the highest expressions in the S1 stage in Chinese hickory and pecan (Figure 8). The expression change of GGT1 was much higher than that of GGT2, which indicated that GGT1 may be the most important gene that participated in tannin synthesis in the seed coat. The expression of CiGGT1 was decreased 3,000-fold, while CcGGT1 was decreased only 800-fold. On the contrary, the expressions of CcTAs and CiTAs did not show significant changes. CcTA1 and CcTA2 continued to down-regulate from the S1 to the S4 stage, and slightly increased in S5. Three TA genes in pecan showed two expression patterns. The expression level of CiTA2a and CiTA2b continued to increase, while CiTA1 was lowly expressed in the S1 stage, up-regulated in S2 and S3, and then decreased. Taken together, the above results indicated that the expressions of the synthesis-related gene GGTs in two species had great influence in tannin accumulated especially in early stage of seed coat development, but the hydrolase gene TAs continued to hydrolyzed throughout the developmental period. The expression patterns of GGT genes may lead to the large accumulation of tannins in the early stage of seed coat development, accompanied by the expression of TA genes. However, at the maturity stage, the decrease of GGT expression resulted in tannins that were no longer synthesized in large quantities. At the same time, the stable expression of TA genes resulted in a continuous decrease in the accumulated tannin content. Moreover, compared with the down-regulation of both CcTA genes in Chinese hickory, two of three CiTA genes were up-regulated in the mature stage, which may further enhance the ability to hydrolyze tannins in pecan, resulting in the lighter astringency.
FIGURE 8 | Expression analysis of GGT and TA genes in seed coats in Chinese hickory and pecan by RT-qPCR. The analysis was performed using three biological replicates and three technical replicates for each sample. The error bars represented the standard deviations of nine replicates. Different letters indicated significant differences according to the Tukey-Kramer test (P < 0.05).
FIGURE 9 | Astringency assessment in the seed coats of Chinese hickory and pecan. (A) The difference of precipitate binding by human salivary proteins and the astringent substance in seed coat extracts. WS, salivary protein profile obtained for whole saliva; Cc_1-Cc_3, the residual protein in the supernatant after reaction of saliva and the three concentrations (0.625, 1.25, and 2.5 mg/ml) of mature seed coat extracts in Chinese hickory; Ci_1-Ci_3, the residual protein in the supernatant after reaction of saliva and the three concentrations (0.625, 1.25, and 2.5 mg/ml) of mature seed coat extracts in pecan. (B) SDS-PAGE gel electrophoresis of human salivary proteins in the supernatant of reactions. (C) Influence of serum albumin (BSA) additions on A280 nm from different tannic acid solutions and seed coat extracts. Cc: seed coat extracts in Chinese hickory; Ci: seed coat extracts in pecan. Data were expressed as mean ± SD (n = 3). The asterisk stands for significant difference (p < 0.01) in astringency between Chinese hickory and pecan.

Astringency Assessment in the Seed Coats of Chinese Hickory and Pecan
Furthermore, we detected the astringency in the mature seed coats between Chinese hickory and pecan with two assays. After incubating the seed coat extracts of the two species and human salivary proteins, the results of centrifugation in the bottom of tubes showed that obvious precipitation appeared at different concentrations of the extractions in two species compared with the control (Figure 9A). At the maximum concentration, the precipitation from seed coat extracts in Chinese hickory was obviously more than that in pecan. SDS-PAGE gel electrophoresis also showed that seed coat extracts in Chinese hickory had less salivary protein in the supernatant (Figure 9B), which proved that Chinese hickory had stronger astringency.
The other assay estimated the astringency by the precipitation of tannins, resulting in a decrease in the absorbance value at 280 nm, and the relationship between absorbance value and protein concentration was logarithmic (Llaudy et al., 2004;Jauregi et al., 2016). The slope of the logarithmic equation decreased with increasing tannins, and the calibration curve obtained by plotting the tannin concentration against the slope was linear with a regression coefficient of 0.997. We determined the slope of the logarithmic equation for the seed coat extracts in Chinese hickory and pecan and converted the astringency of the seed coat extracts to the tannic acid standard according to a linear equation. The result showed that the astringency of seed coat in Chinese hickory was 0.333, which was highly significantly greater than 0.281 in pecan (p-value = 0.005) ( Figure 9C). All these two results confirmed our taste feeling that the seed coat of Chinese hickory was more astringent than pecan.

The Phenolic Compounds in the Seed Coats of Chinese Hickory, Pecan, and Walnut
To evaluate the content of astringent phenolic substances in the seed coat of Chinese hickory and pecan, we detected condensed tannins and other low-molecular-weight phenolic compounds (including hydrolyzable tannins, flavonoids, and phenolic acids) in the seed coats of mature seeds in three Juglandaceae species and in the different developmental stages of seed coats in Chinese hickory based on previous research methods ( Gong and Pegg, 2017) (Figure 10). Comparing the other two species, the seed coats of pecan have the highest content of condensed tannins and the lowest content of phenolic compounds with low molecular weight, and the seed coats of walnut had the highest content of phenolic compounds and the lowest content of condensed tannins, while the content of two types of polyphenols in the seed coats of Chinese hickory was at the median level. With the ripening of seeds, the content of phenolic compounds with a low molecular weight in dry samples of seed coats was continuously decreased. The content of condensed tannin was the highest in the S2 period and was decreased from the S3 to S5 period. Notably, the water content of the seed coats varies greatly throughout the fruit ripening stage. Therefore, the trend in fresh seed coats was completely opposite, and the contents of hydrolyzable tannins and condensed tannins showed an upward trend, possibly leading to the increase of astringency in seed coats.

DISCUSSION
Tannins are an important plant polyphenol and have been classified into two main groups, condensed tannins (CT) and hydrolyzable tannins (HT). The plants in the Juglandaceae are rich in tannins, both CT and HT, and different species have a different component proportion. The plant tannase gene was first discovered in tea plants in 2020 (Dai et al., 2020) and was found to be widely distributed in many species rich in tannins. According to the blast results of CsTA in different plant species, TA genes and the neighboring clade of carboxylesterase genes (named TA-like genes) have very similar sequences compared with other clades (Supplementary Figure 1). In the phylogenetic tree of tannase genes, TA, TA-like, and acetate esterase genes all belonged to one clade. In Arabidopsis, although eight genes were acetate esterase genes, none of them belonged to the TA or TA-like clade. In contrast, persimmon has one TA gene, but no TA-like gene. This phenomenon may be related to the difference in tannin synthesis and degradation in different species. For example, Arabidopsis and rice mainly contain flavonoid-type condensed tannins, while there are no related reports that these two plants contained hydrolyzable tannins (Zhao et al., 2010;Goufo and Trindade, 2014;Shao and Bao, 2015;Demonsais et al., 2020). At the same time, these two species also lack the key genes SDH and GGT for hydrolyzable tannin synthesis as well as the TA genes responsible for the degradation of hydrolyzable tannins. Therefore, we speculate that the TA genes may be distributed in plants rich in hydrolyzable tannins.

Tandem Repeats of TA Genes Indicated Their Evolutionary Origin
According to the genome distribution of TA and TA-like genes from all species we identified, we found that most of these genes were located in a very small region of a chromosome (Supplementary Figure 3 and Supplementary Table 6). These results showed that TA and TA-like genes were tandem repeat genes. In pecan, Chinese hickory, strawberry, and grape, all the TA and TA-like genes were located in a less than 100-kb region, and seven genes were in 15 kb in grape. In pomegranate, TA genes and TA-like genes were distributed on two different chromosomes. In walnut and tea, in addition to one gene, other genes are all located in the same region on the same chromosome. Furthermore, we investigated all the genes in these regions and found that all genes are TA genes or TAlike genes in Chinese hickory and pecan. In other species, most of the genes in the region belonged to the carboxylesterase gene. Tandem repeats of these genes suggested that tannins and tannase were very important for these species. During evolution, carboxylesterase genes produced multiple copies. Some of them kept their original function belonging to CXE genes, and others formed the function of tannase as paralogous genes that belong to TA genes through cumulative mutation. These results demonstrated a strong linkage between CXEs and TAs, which is difficult to separate during plant breeding and needs great attention.
Many metabolites with diversified chemical compounds in plants are produced by the replication, divergence, and selection of metabolic-related enzyme genes. Generally, the more types of metabolites, the more copy of genes are required. In different plants, there are big differences in the number of genes, like triterpenoids (Khakimov et al., 2015;Itkin et al., 2016;Erthmann et al., 2018;Cárdenas et al., 2019;Liu et al., 2019). Tandem repeat is the most important source in the formation of these genes. On the one hand, the copy number of TA genes produced by tandem repeats may affect the ability to hydrolyze tannins in different tissue and even different plants. On the other hand, analyzing the history of tandem repeat formation from the perspective of species evolution may be important for the study of tannin protection mechanisms in plants.

High Expression of Tannase in Leaves May Effectively Resist Herbivores and Microbial Infection
Tannin usually refers to the chemical defense substances against herbivores, which are mainly divided into hydrolyzable tannins and condensation tannins (Boudet, 2007;Miranda et al., 2007;Thipyapong et al., 2007). Tannin mainly forms toxic semiquinone through oxidation and reduces the digestive enzyme activity of herbivores. However, the effects of the two types of tannins are different. Hydrolyzable tannins could resist herbivorous insects, but condensed tannins are ineffective under the condition of high pH (alkaline) in the intestinal tract of insects (Barbehenn et al., 2006;Barbehenn and Peter Constabel, 2011). Although condensed tannins do not affect herbivorous insects, they increase after leaf injury (Osier and Lindroth, 2004;Stevens et al., 2007). The expression of tannase can accumulate more ellagic acid in tissues, further forming ellagic tannins to resist herbivores such as insects. Furthermore, gallic acid produced by hydrolysis of hydrolyzable tannins (HTs) with tannase is an important component, which can effectively inhibit high expression of fungi like Aspergillus flavus, so that tissues have stronger antibacterial ability and reduce fungal infection (Mahoney and Molyneux, 2004).
Leaves are critical to photosynthesis and are the main tissues that plants need to protect. Although the total phenolic content in leaves is low, the main chemical defense substances-condensed tannins and hydrolyzable tannins-have a high proportion. In most plants, leaves are usually the highest tannin content in the whole plant (Barbehenn and Peter Constabel, 2011;Dettlaff et al., 2018). Gallardo et al. (2019) showed that the expression of tannin synthesis-related genes in Quercus ilex leaves increased after mechanical damage treatment, including condensed tannin synthesis-related enzymes like ANR, LAR, ANS, and SDH1, and hydrolyzable tannin synthesis-related enzyme SDH2. After mechanical damage treatment, the content of total phenol, total tannin, and condensed tannin all increased (Gallardo et al., 2019). Another research in Stryphnodendron adstringens also showed that the concentrations of condensed tannins and hydrolyzable tannins all increased, while total phenolics decreased after leaf clipping. Plants showed a trade-off between tannins and total phenols (Tuller et al., 2018). Our quantitative study showed that the expression of tannin-related genes GGTs and TAs in leaves of Chinese hickory and pecan was up-regulated rapidly after 3 h of abiotic stress and began to hydrolyze a large number of substances into small chemicals such as ellagic acid and gallic acid to resist wound stress. After 6 h, the resistance response gradually ended. This result provided a key time point for studying the abiotic stress in Chinese hickory and pecan, and a foundation for further research.

TA Genes May Be Regulated by miRNA in Response to Plant Biotic and Abiotic Stresses
According to predicted miRNAs in walnut, pecan, and Chinese hickory, we found that the TAs could be targeted by many miRNAs. This meant that the regulation mechanism of tannase genes was much more complicated than we thought. Based on the targeted network of miRNAs and targeted TAs in three species, it was found that TA genes from class 1 and class 2 were very diverse and they are targeted by different miRNAs. So, it is likely that two classes of TA genes are involved in different biological processes, regulating the tannins by different regulatory pathways. Nevertheless, there are still four miRNAs that can regulate genes in class 1 and class 2 simultaneously. Like miR408, one of the most conserved plant miRNAs was report as a wound-related miRNA in sweet potato and was repressed by wounding and jasmonate (Kuo et al., 2019). It was proved that jasmonate could induce hydrolyzable tannins and participated in wound response in red oak (Allison and Schultz, 2004;Elderd et al., 2013). In Arabidopsis, miR408 participates in seed yield and abiotic stress such as salinity, cold, oxidative stress, drought, and osmotic stress (Ma et al., 2015;Song et al., 2018). In addition, it was proved to be involved in photosynthesis, growth, and other biological processes (Pan et al., 2018). miR6021 is a specific miRNA found in tobacco, which can regulate plant innate immune receptors and was predicted to target a CC-NB-LRR gene, Hcr9 in Solanaceae (Li et al., 2012). In each class, there exists a lot of coregulating miRNAs in different species. That implied that the conservation of regulation of two class TA genes in plants rich in tannins may be regulated by the same miRNAs. In addition to the miRNAs mentioned above, there are other miRNAs in the targeted network that are involved in response to biotic and abiotic stress.

TA Genes Contributed Higher Astringency by Controlling Hydrolyzable Tannin Content in the Seed Coat of Chinese Hickory Than Pecans
Tannin is abundantly accumulated in seed coats in Juglandaceae, which is not only resistant to animal feeding and disease but also the main source of astringent taste when humans eat nuts. However, different plants have different tannin species preferences; for example, the fruits of grapes, persimmons, cocoa, and sorghum are dominated by condensed tannins (Zhu et al., 2019;Wei et al., 2020), while the fruits of pomegranates and walnuts contain more hydrolyzable tannins (Bajec and Pickering, 2008;Akhtar et al., 2015). Xu et al. (2020) carried out a comprehensive analysis of phenolic metabolites in eight tissues of pecan. The results showed that hydrolyzable tannins were the main phenolic metabolites in the seed coat of pecan. Hydrolyzable tannins have complex components and high content. Among them, the highest content in the seed coat is ellagic acid. Through HPLC separation, we found that the hydrolyzable tannin content in Chinese hickory was higher than that in pecan, while the condensed tannin content was lower than that in pecan. By our astringency assessment experiments, the seed coat of Chinese hickory is more astringent than pecan's, indicating that hydrolyzable tannins may be the main source of astringency in the seed coat of Chinese hickory.
In addition, we found that the tannin content in fresh samples showed an upward trend during the seed coat development of Chinese hickory, which was consistent with our astringency assessment assay. However, after excluding the influence of water content change during seed coat development, the tannin content in dry sample decreased continuously. Real-time quantitative PCR results also confirmed that the expression of synthesis gene GGTs and hydrolysis gene TAs continued to decrease during the development of seed coat in Chinese hickory, resulting in corresponding changes in tannin content. At the same time, CcTAs are continuously down-regulated, while CiTA2a and CiTA2b are up-regulated, which may cause a significant difference in tannin content between Chinese hickory and pecan nuts at maturity. Thus, the various expression pattern of TA genes in Chinese hickory and pecan may lead to the difference in the content of hydrolyzable tannins, which are the main source of astringency in the seed coat at the maturity stage.

CONCLUSION
Tannins are a large class of important metabolites of plants with a lot of structures. In 2020, the first plant tannase gene (TA) was discovered in tea (Dai et al., 2020). Tea mainly contains condensed tannins, while the seed coats of Juglandaceae are mainly composed of hydrolyzable tannins. Moreover, although they have similar genetic backgrounds, Chinese hickory and pecan have different levels of astringency in the seed coats. We identified and analyzed two and three tannase genes in Chinese hickory and pecan, respectively. The expression of GGT and TA genes in seed coat during seed development showed that the diversity patterns implied different mechanisms in tannin metabolism. Phenolic compounds separated by HPLC in the seed coats showed that hydrolyzable tannin content in Chinese hickory was higher than that in pecan, while the condensed tannin content was lower than that in pecan. This suggests that the content of hydrolyzed tannin is the main reason for the difference in astringency between the two species. Taken together, the early stage of seed development is a critical period for tannin metabolism in seed coats. Due to the continuous expression of TA genes at the maturity stage, the tannin content in the seed coats decreases, but the percentage of tannins is increased due to the decrease of water content during the maturation process. At the later stages of development, the differential change of expression level in TA genes in Chinese hickory and pecan may be the source of the final difference in astringency between the two species.

DATA AVAILABILITY STATEMENT
The datasets generated for this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
KW, YL, and JHu conceived and designed this study. KW and JW analyzed the data and wrote the manuscript. JW and SL performed the experiments. All authors have read and approved this manuscript.