The Phylogeny and Functional Characterization of Peanut Acyl-ACP Thioesterases

Fatty acyl-acyl thioesterases (FATs), which hydrolyze the thioester bond linking acyl chains to an acyl carrier protein, thereby terminating their elongation, contribute significantly to the fatty acid (FA) content and composition of seed storage lipids. The peanut (Arachis hypogaea L.) genome was found to harbor 21 FAT (AhFAT) genes, distributed over 12 of the 20 chromosomes. The length of their predicted translation products varied from 74 to 415 residues, and all but one included the 1–2 Acyl-ACP_TE conserved domains. All of the coding sequences were interrupted by at least one intron, with the exon number ranging from two to 12, and five of the genes were liable to alternative splicing. When the RNA-Seq platform was used to assess the transcriptional behavior of the 21 AhFAT genes, transcription of only 13 was detectable in samples of root, leaves, and developing seed; among these, six were transcribed throughout the plant, three were root-specific and one was leaf-specific. A detailed analysis of a pair of homologous AhFATs showed that the coding region of each was split into six exons and that both were transcribed in all of the plant organs surveyed (although the intensity of their transcription was not the same in immature seed). The product of both genes was deposited in the chloroplast outer membrane. The constitutive expression of these genes in either yeast or Arabidopsis thaliana increased the FA content, especially that of saturated FAs. In peanut genome, 21 AhFAT genes were found and two of them were transformed into yeast and Arabidopsis for function identification. Results showed that overexpression of these two genes could increase the FA content, especially the saturated FAs content.


Introduction
Fatty acyl-acyl carrier protein thioesterases (FATs) are pivotal enzymes in the synthesis of fatty acids (FAs). Following the termination of acyl chain elongation, FAs are hydrolyzed by FATs and transported to the cytosol (Beisson et al. 2003;Li-Beisson et al. 2010;Sandelius and Aronsson 2008). As a result, FATs determine the acyl chain length of FAs. Two groups of FAT have been recognized: while FATAs prefer unsaturated fatty acids as a substrate, FATBs prefer unsaturated fatty acids (Jones 1995). Within the FATBs, subgroup 1 enzymes act on long-chain fatty acids (Dörmann et al. 1995;Filichkin et al. 2006;Jha et al. 2010;Moreno-Pérez et al. 2011;Sánchezgarcía et al. 2010), while subgroup 2 enzymes target medium-length chain fatty acids (Filichkin et al. 2006;Jing et al. 2011;Voelker 1996;Voelker et al. 1997). The genetic manipulation of various plant FAT genes has been shown to influence FA composition: for example, the overexpression of the Arabidopsis thaliana gene AtFATB1 results in the accumulation of C16:0 (palmitic acid) (Dörmann et al. 2000). The expression in Chlamydomonas reinhardtii of a FATA gene isolated from the algal species Dunaliella tertiolecta has the effect of producing 63% and 94% more neutral lipids than the wild type, without compromising growth (Tan and Yuan 2017). When a FATB gene isolated from the cigar plant Cuphea lanceolata was expressed in C. reinhardtii, the production of C14:0 (Myristic acid)-containing triacylglycerols increased by up to 1.6-fold (Inaba et al. 2017). The heterologous expression of a Brassica napus FATB gene in yeast raised the content of the saturated C16:0 and C18:0 (stearic acid) by, respectively, 46% and 22%, at the same time reducing that of both the unsaturated C16:1 (palmitoleic acid) and C18:1 (oleic acid) by, respectively, 15% and 31% (Tan et al. 2015).
The peanut (Arachis hypogaea L.) is a leading oil seed crop. Several genes influencing its seed oil content have been identified, among which are a number of FATs. The expression of an AhFATA in both Escherichia coli and certain algal species has been shown to be effective in both altering the FA profile and increasing the content of a number of FAs (Chen et al. 2017). Manipulation of AhFATB1 has been effective in increasing the content of both saturated and unsaturated FAs (Chen et al. 2012(Chen et al. , 2017Wen et al. 2012). With the acquisition of the genome sequence of peanut (Bertioli et al. 2016;Bertioli et al. 2019;Zhuang et al. 2019), it is now possible to identify the species' full complement of AhFAT genes and to characterize the functionality of each of them. The aim of the current research was document the family of AhFAT genes, to clarify their phylogeny and gene structure, and to profile their transcription in different organs of the plant. Moreover, two not previously isolated FATs have been transformed and functionally analyzed in Saccharomyces cerevisiae and A. thaliana hosts.

Plant and Yeast Materials
The peanut cultivar (cv.) 'Fenghua 1' was used for gene isolation. 'Fenghua 1' was planted in greenhouse with 16-h light/8-h dark photoperiod and 30 °C day/22 °C night temperature cycle. When it got to 12 days, the young mainstem, mainstem leaves, and root tips of ten plants were taken down and frozen in liquid nitrogen. In the full podding bearing stage, seeds of different developmental stages were taken down and frozen in liquid nitrogen. These materials were kept at − 80 °C until use. Genetic transformation experiments involved A. thaliana ecotype Col-0 and yeast (S. cerevisiae) strain W303. Arabidopsis plants were grown in flower pots in a temperature controlled incubator, with 16-h light/8-h dark photoperiod and 21 °C day/16 °C night temperature cycle. W303 yeast strains were kept on YPD medium (Qingdao Hope Bio-Technology Co. Ltd, China) and grown at 30 °C.

Phylogenetic Analysis
The AhFATA and AhFATB1 sequences were used to scan the Phytozome database (https ://phyto zome.jgi.doe.gov/ pz/porta l.html) for plant FATs (Supplemental Table S1). The resulting sequences were used to construct a FAT phylogeny based on ClustalW software (https ://www. genom e.jp/tools -bin/clust alw), with parameters set to default values. Mega6 software (https ://www.megas oftwa re.net/) was used to construct a plant FAT phylogeny, based on the neighbor-joining method; statistical confidence in the clade branching points was obtained by running 1,000 bootstrap replicates (Tamura et al. 2013).

Transcription Profiling Analysis
RNA was extracted from the all roots and leaves of 12-day old cv. 'Fenghua 1' seedlings and from developing seeds harvested either 30 or 50 days after flowering (respectively "seed1" and "seed2"). RNA was extracted using the TRIzol reagent (Invitrogen, Carlsbad, CA, USA) then treated with RNase-free DNase I (New England Biolabs, USA) for 30 min at 37 °C to degrade any contaminating DNA present. The concentration and purity of the resulting RNA preparations were assessed using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Wilmington, DE, USA) and its integrity was checked using an RNA Nano 6000 Assay Kit (Agilent Technologies, CA, USA). A 1.5 μg aliquot of RNA was processed with a Ribo-Zero rRNA Removal Kit (Epicentre, Madison, WI, USA) to remove the rRNA component, and the subsequently prepared sequencing libraries based on the residual RNA, following treatment with an NEBNext® UltraTM Directional RNA Library Prep Kit for Illumina® (New England Biolabs, USA). Index codes were added to enable each sequence to be attributed its organ provenance. Paired-end sequences were generated by an Illumina Hiseq2500 platform. The resulting sequence data, stripped of adapter sequence and low-quality reads, have been deposited with NCBI (SRA ID: SRP093901). The sequences were aligned with the genome sequence of the peanut cv. 'Tifrunner' (https ://www.peanu tbase .org/peanu t_genom e) using TopHat2 software (Kim et al. 2013). Aligned reads were assembled into a full transcriptome using the Cufflinks v2.2.1 program (Trapnell et al. 2010). FPKM values for each transcript were used to estimate transcript abundance. The expression levels of FATBs from Physcomitrella patens, Populus trichocarpa, and Glycine max are obtained from Phytozome v12.1 (https ://phyto zome.jgi.doe.gov/pz/porta l.html).

Sub-cellular Localization
The open reading frames (lacking the stop codon) of AhFATB2-1 and AhFATB2-2 were PCR-amplified using the primer pairs FATB2-1-GFP-F/-R and FATB2-2-GFP-F/-R (Supplemental Table S2), after which the amplicons were treated with Hind III and BamH I, then cloned into the multiple cloning site of the pBSK+-35S-EGFP plant transient expression vector (kindly provided by Weicai Yang, Institute of Genetics and Developmental Biology, Chinese Academy of Science). The constructs were transformed into A. thaliana mesophyll protoplasts using the PEG-calcium transfection method (Yoo et al. 2007). Fluorescence generated by the expression of GFP was observed by confocal laser scanning microscopy.

Yeast Transformation
AhFATB2 fragments were PCR-amplified using the primer pairs FATB2-1-BamHI-F/-KpnI-R or FATB2-2-BamHI-F/-KpnI-R (Supplemental Table S2), after which the amplicons were treated with Kpn I and BamH I, then inserted into the multiple cloning site of the galactose-inducible yeast expression vector pESC-URA (Agilent Technologies). The resulting constructs plasmids were transformed into S. cerevisiae strain W303 competent cells which were grown on a selective SD-URA medium. Pooled colonies from each strain were grown in a selective liquid medium, then diluted into a liquid medium containing galactose and raffinose to induce protein expression. Cells harboring an empty plasmid served as the negative control. The cells were dried, then analyzed for their fatty acid content, following a gas chromatography protocol described by Zheng et al. (2017).

A. thaliana Transformation
The AhFATB2 open reading frames were PCR-amplified using the primer pairs FATB2-1-KpnI-F/-SacI-R or FATB2-2-KpnI-F/-SacI-R (Supplemental Table S2) and the resulting amplicons inserted first into the binary vector PRI101 (kindly provided by Minqin Wang, Shandong University, China) and thence into Agrobacterium tumefaciens strain LBA4404 (Poirier et al. 2000). A. thaliana plants were transformed using inflorescence infection method (Clough and Bent 1998). Progeny of putative transgenics were plated on half-strength MS medium (Murashige and Skoog 1962) containing 100 mg/L kanamycin in order to select for transgenepositive T 1 individuals, and this procedure was repeated to obtain the T 2 and T 3 generations. The progeny of T 3 plants were analyzed for their fatty acid content, following a gas chromatography protocol described by Zheng et al. (2017).

The AhFAT Gene Content of the Cultivated Peanut Genome
In total, 21 potential AhFATs were uncovered from A. hypogaea genome (Supplemental Table S3); four loci mapped to each of chromosomes Arahy.12 and Arahy.13, three to Arahy.01, two to Arahy.02, and the other eight each mapped to different chromosomes. Two of the 21 genes were FATA s, namely Arahy.YI95N7.1 (on chromosome Arahy.04) and Arahy.QR9F0J.1 (Arahy.14); the 19 AhFATB genes were distributed across ten chromosomes. On the basis of their sequences, Arahy.YI95N7.1 was deemed to be a synonym of the previously described gene AhFATA, as was Arahy. KT8BMX.1 of AhFATB1 (Chen et al. 2012;Wen et al. 2012). The length of the predicted translation products ranged from 74 (Arahy.L512HM) to 415 (Arahy.SJ2WML) residues. By sequence alignment in NCBI, we found Arahy.L512HM appeared to be an incomplete protein. By comparing the gene sequences with their genomic sequences, we found five of the genes produced more than one isoform: while both Arahy.QR9F0J and Arahy.L8EKIA produced three splicing variants, Arahy.H8W5PF, Arahy.LCII34, and Arahy.3BV5AA each produced two. Alternative splicing (AS) led to the formation of one transmembrane domain of Arahy.QR9F0J.3. Arahy.QR9F0J.3 was the only one which had transmembrane domain, whereas the other AhFATs had none. It had no effect on the products' sub-cellular localization, with all 21 products predicted to be deposited in the chloroplast. With the exception of the very short Arahy.L512HM.1, all of the proteins possessed a 1-2 Acyl-ACP_TE domain, with six also possessing an Acyl-thio_N domain.
There was extensive variation with respect to the gene structures of the AhFATs, with the number of exons ranging from two to twelve (Fig. 1). The shorter genes Arahy. L512HM.1, Arahy.C2SFPD.1, Arahy.0VM2VP.1, Arahy. QR9F0J.3, and Arahy.5L01PW.1 each featured only two or three exons, while the others' coding sequences were split into between six and 12 fragments. The Arahy.L8EKIA.1 coding sequence was interrupted by 11 introns and featured three alternative splicing variants (one complete, one comprising eight exons, and one of seven exons); these isoforms derived from exon skipping, alternative transcription start sites and alternative transcription termination sites. The three isoforms generated by Arahy.QR9F0J were of length eight, six, and four exons. Intron retention was observed for both Arahy.QR9F0J.1 and Arahy.QR9F0J.2.

Phylogeny of the AhFATs
After removing gene fragments and incomplete sequences, the scan of plant FATs delivered a set of the 179 sequences from 26 species (Supplemental Table S1) used to construct a FAT phylogeny (Fig. 2). The representative from C. reinhardtii was located closest to the root of the tree, while the remainder formed two large groups (the FATAs and the FATBs). The two subgroups FATB1 and FATB2 were recognized: the former consisted entirely of genes harbored by angiosperm species, and comprised four branches (I through IV). The dicotyledoneae FATBs (including some of the peanut FATBs) clustered within branch I, whereas branch II consisted entirely of monocotyledoneae FATBs, branch III only of peanut FATBs, and branch IV of FATBs from Prunus persica and Populus trichocarpa. With respect to the FATB2 subgroup, genes harbored by primitive land plant species located close to the root, and the remainder clustered into two groups (branches V and VI); and branch V contained all of the other primitive land plant genes, and branch VI the angiosperm genes. Peanut FATBs were distributed over the four branches I, III, V, and VI, each branch consisted of four, seven, two, and two AhFATBs, respectively. The FATAs produced by Marchantia polymorpha (a liverwort), Selaginella moellendorffii (a lycophyte), and Physcomitrella patens (a bryophyte) were located close to the root, and the remainder formed branches VII and VIII of the phylogenetic tree. Branch VII members were from monocotyledoneae and some dicotyledoneae species, while branch VIII clustered genes exclusively from dicotyledoneae (including the AhFATAs). The most closely related sequences to the AhFATs were from other Leguminaceae, such as soybean and Medicago truncatula.

Transcription Analysis
When the abundance of the various AhFAT transcripts was assessed using RNA-Seq data, only 13 of the 21 genes were represented (Fig. 3) (Wen et al. 2012). The two genes with red lines are the genes identified in this paper seed1, root, and leaf (with FPKM values of 0.5, 0.24, and 0.95, respectively) but no expression in seed2. Combined with RNA-Seq data (Fig. 3) and phylogenetic tree analysis (Fig. 2), the results demonstrated that the AhFATB genes in branch I were ubiquitously transcribed in all tissues, whereas the FATB genes in branches III-VI were expressed with tissue-specific patterns.
To verify the universality of this phenomenon, we analyzed the expression patterns of FATBs from P. patens, P. trichocarpa, and G. max (Fig. S1A-C). The three species have many FAB genes in their genomes, similar to peanut. Moreover, they exhibit different levels of evolution. The P. patens genome harbored eight FATBs, two of which (i.e., Pp3c15_18050V3.1 and Pp3c17_23790V3.1) were located closest to the root of the FATB group, while the others were located in branch V (Fig. 2). Pp3c19_20600V3.1 and Pp3c18_15010V3.1 were ubiquitously transcribed in all tissues at high levels, and the others were expressed in different tissues (Fig. S1A). The P. trichocarpa genome harbored 11 FATBs, which were distributed over branches I, IV, V, and VI, with each branch consisting of four, two, two, and three FATBs, respectively (Figs. 2, S1B). Four members in branch Fig. 2 Phylogenetic tree analysis of plant FATs. The tree was conducted with Mega6 software based on the neighbor-joining method. Statistical confidence in the clade branching points was obtained by running 1,000 bootstrap replicates I were ubiquitously transcribed in all tissues at high levels, while the others were differentially expressed (Fig. S1B). The G. max genome harbored 10 FATBs, which were distributed over branches I, V, and VI, with each branch consisting of four, four, and two FATBs, respectively (Figs. 2, S1C). Four members in branch I were ubiquitously transcribed at high levels in all tissues, while the others were differentially expressed (Fig. S1C). The sub-functionalization of FABs started early, such as in the P. patens period. Comparatively, the evolution of the FATA genes was conservative. Most plant genomes harbored one to four FATA genes (Fig. 2), but only one or two expressed ubiquitously in all tissues. The rest were expressed with different expression patterns (Fig. S1).

AhFATB2 Isolation and Analysis of Its Sequence
When cDNA prepared from seeds harvested 30 days after flowering was amplified using the primer pairs FATB2-1-F/-R and FATB2-2-F/-R (Supplemental Table S2), the two sequences thereby generated proved to be highly similar (99.4%) to those of Arahy.L4EP3N.1 and Arahy.4E7QKU.1. (Their sequences, designated AhFATB2-1 and AhFATB2-2, have been archived in GenBank under accession numbers MH105081 and MH105082). The two sequences, differing from one another at just eight nucleotides, comprised a 1,245 nt open reading frame, encoding the identical 414 residue polypeptide. Both shared the three conserved catalytic residues Asp311, Asp313, and Cys348, as well as the Acyl-ACP_TE domain. Their level of similarity with the sequence of AhFATB (GenBank number EF117305.1) was 91.4%. AhFATB2-1 mapped to chromosome Arahy.10 and AhFATB2-2 to chromosome Arahy.20, a pair of chromosomes recognized as homologous; they shared a different gene structure (six and seven exons, respectively) and variation in their intronic sequence (Fig. 1).

Semi-quantitative RT-PCR Analyses of AhFATB2-1 and AhFATB2-2
The very high similarity between the AhFATB2-1 and AhFATB2-2 sequences made it impossible to distinguish their transcripts from one another using conventional RT-PCR. To achieve the necessary discrimination, a Taqman-PCR method was employed. A comparison of their transcript abundance in the root, stem, leaf, and flower, sampled at 10, 20, 30, 40, 50, and 60 days after flowering, showed that they shared a very similar transcriptional profile. Both were more abundantly transcribed in the flower than elsewhere (Fig. 4a). In seeds, the expression level of AhFATB2-1 and AhFATB2-2 increased with the seeds development from 10 to 30 d, then peaked at 30 days, and decreased along with the seeds maturation. The period from 20 to 40 d was the most important stage of peanut seed development, the high expression level of AhFATBs was in accordance with the rapid oil accumulation in this stage (Fig. 4b).

Sub-cellular Localization of AhFATB2-1 and AhFATB2-2
When the two genes were expressed in A. thaliana protoplasts, the transgene was expressed most strongly in the vicinity of the chloroplasts, specifically in their outer membranes (Fig. 5).

Discussion
Thioesterases are ubiquitous in both prokaryotic and eukaryotic organisms. Based on their catalytically active site, their oligomerization, and their substrate specificity, they have been allocated EC numbers from 3.1.2.1 to 3.1.2.27, with many remaining unclassified (EC 3.1.2.-) (Cantu et al. 2010). The FATs (EC 3.1.2.14) hydrolyze FAs synthesized by the FA synthase complex. The FATs found in bacteria and plants have been classified into ten subfamilies (Jing et al. 2011); here, a collection of 179 documented plant FATs, which included both FATAs and FATBs, fell into eight clades (Fig. 2). Peanut FATBs were distributed over the four branches I, III, V, and VI. Of particular interest Fig. 6 Analysis of FA content in transgenic and wild type yeast lines. Note: compared with control, *P < 0.01. Three replicates were conducted Fig. 7 Analysis of the FA content and percentage in wild and transgenic Arabidopsis seeds. a FA content; b FA percentage. Note compared with WT, *P < 0.05, **P < 0.01. Three replicates were conducted was the finding that one of these (branch III) was composed entirely of FATBs synthesized by Arachis spp. (seven by A. hypogaea, five by A. ipaensis and four by A. duranensis) (Fig. 2). A likely explanation for this rather surprising outcome is that these genes have evolved from a single ancestral copy by duplication. Of the seven A. hypogaea branch III FATBs, only two (Arahy.3BV5AA.1 and Arahy.BCJ5R8.1) appeared to be active (Fig. 3): transcripts of the former gene were particularly abundant in the root, whereas the latter was transcribed-at a rather low level-exclusively in the leaf.
The large number of FAT genes present in the genome varies widely between species. This variation cannot be accounted for by suggesting that their number simply accumulates during evolution, since although the algal species C. reinhardtii encodes just one FAT, the genomes of the primitive land species M. polymorpha (a liverwort), S. moellendorffii (a lycophyte), and P. patens (a bryophyte) harbor, respectively, four, six, and ten FATs, a greater number than is harbored by most angiosperm species (Supplemental Table S1). Nor can the large number of FAT genes harbored by peanut (21AhFATs) does not be explained by its being a tetraploid, since the hexaploid species bread wheat harbors just ten genes. Finally, there is no discernible relationship between the number of FAT genes and genome size, as can be seen from a comparison between the small genome species A. thaliana and the large genome species bread wheat. A possibility which remains to be tested is that selection for high oil content in the seed tends to favor the accumulation of FAT genes.
Based on RNA-Seq data, only 13 of the 21 peanut FATs were transcribed in root, leaf, and seed tissue (Fig. 3), and only six of these were ubiquitously transcribed. This observation implies a degree of sub-functionalization among the genes, as well as suggesting a high rate of gene silencing. However, the present experiment did not cover many other organs in the plant, such as the flower, the stem, and the pod. Furthermore, some FATs appear to be involved in the stress response (Li et al. 2017). Given the importance of lipid synthesis in the seed, it was unexpected that only one in three of the active genes were transcribed in the seed, while almost half were active in the root and leaf. The implication is that a wider spectrum of FAs is required by vegetative organs than is required in the seed. Lipids account for about half of the peanut seed's dry weight, from which about 80% is in the form of C18:1 or C18:2 compounds. The remarkable gene Arahy.KT8BMX.1 expressed at extremely high level in seed1 stage may be responsible for the major lipids synthesis. Lipid storage is of minor importance in both the root and the leaf, but a wider range of FATs may be required to support the complexity of physiological reactions relying on membrane lipids.
Based on substrate preference, the FATs have been divided into two groups: while the FATAs prefer unsaturated FAs, FATBs prefer saturated FA, as shown by observation that co-expressing in E. coli the Cinnamomum camphora gene CcFatB1 with E. coli fadR and Bacillus megaterium P450 BM3 boosts the content of ω-3-OH-C14:1 (Xiao et al. 2018). Three FATBs produced by coconut share a similar substrate specificity, producing predominantly C14:0 and C16:1 compounds (Jing et al. 2011). At the same time, different FATs produced by a given same species do not necessarily differ in their function, since three FATA s and two FATBs encoded by Sorghum appeared to show a very similar substrate specificity when heterologously expressed in E. coli (Jing et al. 2011). Similarly, when AhFATA and AhFATB1 were expressed in the cyanobacterium Synechocystis sp. PCC6803, both increased the content of C16:1, C18:0, C18:2, and C18:3 compounds (Chen et al. 2017). Here, the two AhFATB2s constitutively expressed in both yeast and A. thaliana appeared to prefer saturated over unsaturated FAs (Figs. 6, 7).
In response to the world energy crisis and global climate change, biodiesel has attracted renewed interest in the exploration for sustainable development (Zinoviev et al. 2010). Free fatty acids (FFAs) have been extensively applied in the manufacture of biofuels, cosmetics, and pharmaceutical drugs (Shin et al. 2016). Given that the overexpression of FATs has considerably increased the production of FFAs in prokaryotes and eukaryotes (Chen et al. 2017;Li et al. 2017;Lin and Lee 2017;Shin et al. 2016;Tan and Yuan 2017), FATs remain the first choice in the production of oleochemicals. The substrate specificity of FATs often determines the composition of oils. Thus, many plant FATs have been isolated and remolded to produce oleochemicals with desired chain lengths (Lozada et al. 2018). Peanut is an important oil crop (with seed oil content up to 50%), and its genome possesses 21 FATs, which are far more than any other species (Supplemental Tables S1, S3). Not all AhFATs are suitable for oleochemical production. At least four AhFATs, namely, Arahy.KT8BMX.1, Arahy.L8EKIA.1, Arahy.L4EP3N.1, and Arahy.4E7QKU.1, with high expression levels in branch I are good choices for biofuel production. Studies have demonstrated that all these AhFATs can increase the oil content of the transformed microalgae (Chen et al. 2012(Chen et al. , 2017Li et al. 2017;Wen et al. 2012), yeast, and plants (Figs. 6, 7). In addition, the FATs in branch I from other plants are good candidates for biofuel production because of their high expression level (Supplemental Fig. S1).

Conclusions
In summary, we analyzed the peanut FAT gene family in peanut genome, and analyzed their phylogeny relationship with other plant FATs. Here, two new AhFATB genes from cultivated peanut were cloned and transformed into S. cerevisiae and A. thaliana for function verification. Results showed that both of them could increase the FA content significantly, especially that of saturated FAs.