To identify SAD genes in C. cathayensis, the coding sequence and amino acid sequence of SADs from different plants (Arabidopsis thaliana, Oryza sativa L., and Populus L.) were used as queries to search against the Chinese Hickory Genome Database. The whole genomic sequences of hickory (C. cathyensis) were obtained from the Portal of Juglandaceae (Guo et al., 2020; Yang et al., 2016). The SAD protein sequence data of A. thaliana, Populus and O. sativa were downloaded from the Arabidopsis Information Resource (TAIR) database (http://www.arabidopsis.org), the China Rice Data Center (http://www.ricedata.cn/gene/index.htm) and the previous studies, respectively. Using Arabidopsis, rice, and poplar SAD gene family members as queries to search homologous protein sequences in the hickory protein database by using the Blast Zone program (TBtools-blast software, E-value <1e-10) (Chen et al., 2020), respectively.

All candidate SACPD (stearoyl-acyl carrier protein desaturase) proteins were analyzed using HMMER software to identify the FA_desaturase_2 domain with the Pfam accession PF03405 (Finn et al., 2015; Mistry et al., 2020). They were further validated using the NCBI Conserved Domain Database (http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi), SMART (http://smart.embl-heidelberg.de/), and InterProScan programs (http://www.ebi.ac.uk/Tools/InterProScan/) (Jones et al., 2014).

Multiple sequence alignment and visualization analyses were performed based on the amino acid sequences of CcSADs and AtSADs. Then, the phylogenetic tree of all SAD proteins was constructed by MEGA7 using the neighbor-joining (NJ) method with bootstrap values of 1,000 replicates (Kumar et al., 2016). All isolated CcSAD (C. cathayensis SAD) proteins were classified into different subfamilies based on the SADs from different species. The nomenclature of CcSADs are based on the similarity of amino acid sequences to homologous genes in Arabidopsis including SAD and SSI2 (SUPPRESSOR OF SA-INSENSITIVITY2)

The ExPASy: SIB Bioinformatics Resource Portal-Home (https://www.expasy.org/) was used to predict the theoretical isoelectric point (pI) and molecular weight (MW) of CcSAD proteins. The instability index (II), aliphatic index, and grand average of hydropathicity (GRAVY) were analyzed using the ProtParam tool of the ExPASy website (https://web.expasy.org/protparam/) (Gasteiger, 2005). The subcellular locations of CcSAD proteins were predicted using WoLF PSORT and ProtCompV.9.0 Server (http://www.soft-berry.com) (Horton et al., 2007). The presence of signal peptide and transmembrane domains of CcSAD proteins was predicted using SignalP 4.0 and TMHMM v.2.0 online software.

The conserved motifs of CcSAD proteins were analyzed using MEME-Suite 5.4.1 online program (http://meme-suite.org/) (Bailey et al., 2015). The width of the conserved motifs was set to have 6 to 50 amino acids, and the maximum number of conserved sequences was set to be 10. Finally, the conserved motif, gene structure, and conserved domain of the CcSAD gene family were visualized using TB tools (Chen et al., 2020).

To identify the cis-elements of CcSADs, the upstream 2,000 bp sequences from the translation start site were collected from the Chinese Hickory Genome Database and submitted to the PlantCARE tool (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) for cis-element sequence search, The identified cis-regulatory elements were analyzed according to the method of Lescot et al. (2002).

Previous transcriptomic analyses documented expression patterns of CcSAD genes during different stages of embryo development (Huang et al., 2022a) and also in stigma (Xing et al., 2022) and endocarp (Li et al., 2022b) when C. cathayensis plants were grown under normal conditions. Thus, raw RNA-Seq data for C. cathayensis embryo, stigma, and endocarp were downloaded from the NCBI created by our lab (Read Archive accession number: PRJNA687050, PRJNA810757). Poly N, adaptor sequences, and low-quality reads were removed and assembled by Trinity software to obtain clean data. The clean reads from each sample were mapped to reference genome of Chinese Hickory (Chinese Hickory Genome Database) using Hisat2. Expression levels of the genes were measured as fragments per kilobase of transcript per million mapped reads (FPKM). The expression data of SAD genes were log2-transformed to normalize the scale by Tbtools (Scale Method: Zero To One). Subsequently, correlation coefficients between variation in unsaturated fatty acids and SAD genes expression were computed using the R language. The software Mascot 2.3.02 (Matrix Science, UK) was employed for protein analysis, and data were downloaded from the NCBI (Read Archive accession number: PRJNA687050)

The ligand spatial data (FAs: C16:/C17:0/C18:0) were obtained from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/). The protein structure and amino acid sequences of CcSADs were predicted using an online tool AlphaFold Colab (https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/AlphaFold2.ipynb). The protein structure was introduced into Autodock tools to remove the water molecules from the pdb add hydrogens. Using CB-DOCK2 (http://cadd.labshare.cn/cb-dock2/), protein file (pdb format) and ligand file (pdb format) were loaded into the specified area, and molecular docking was predicted (Salmaso and Moro, 2018; Liu et al., 2022).

Total RNA was extracted from the endosperm of immature seeds using EasyPure® Plant RNA Kit (Transgen, Beijing, ER301-01) and was reverse-transcribed to cDNA using the EasyScript® First-Strand cDNA Synthesis SuperMix (Transgen, Beijing, AE301-02). Using TransStart® FastPfu Fly DNA Polymerase (Transgen, Beijing, AP231-21), PCR was performed using the cDNA as a template to generate CcSAD amplicons for subsequent sequencing analysis. An EasyPure® Quick Gel Extraction Kit (Transgen, Beijing) was used to purify the PCR products. All primer sequences used in this paper were in the Supplementary File Table S1 .

The CcSADs without the stop codon were cloned into the BamHI site of the pCambia1300-35S-EGFP vector to generate a 35S-CcSAD-GFP construct so that these genes could fuse with the GFP protein driven by the 35S promoter when expressed in tobacco (N. benthamiana) leaves (Craig et al., 2008). Primers with homologous arms and related enzyme digestion sites were used to amplify the full-length CDS of CcSAD, which was then inserted into the pCambia1300-35S-EGFP vector with a pEASY®-Basic Seamless Cloning and Assembly Kit (Transgen, Beijing, CU201-02). In addition, the coding sequences of CcSADs were amplified by PCR to generate pDR195-CcSAD yeast expression vectors using the pEASY®-Basic Seamless Cloning and Assembly Kit (Transgen, Beijing) on a BamHI-treated pDR195 vector.

The correctly sequenced 35S-CcSADs-GFP construct was transferred into Agrobacterium tumefaciens strain GV3101, which was transformed into tobacco leaves through infiltration. Two days after infiltration, the transient expression of GFP in tobacco leaves was observed using a Confocal Laser Scanning Microscopy (ZEISS LCM-800, Carl Zeiss, Oberkochen, Germany) with 470 nm excitation filter and 525 nm emission filter.

Mesophyll protoplasts were isolated from A. thaliana leaves. Briefly, the cut leaf strips were mixed with an enzymatic digestion solution containing mannitol, CaCl2·2H2O, MES, macerozyme R10, and cellulase R10. The mixture was incubated in the dark at room temperature for at least 3 hours. After digestion, the protoplasts in the solution were examined under a microscope. After the incubation, protoplasts were harvested by filtration with 0.65-μm nylon filters and centrifugation at 100 g for 5 min. The supernatant was removed, and protoplasts were gently resuspended in pre-chilled W5 solution (2 mM MES, 154 mM NaCl, 125 mM CaCl2, and 5 mM KCl) on ice. The resulting pellet was resuspended in MMG solution (Mannitol, MgCl2·6H2O, and MES). The vitality of the protoplasts was observed using a microscope, and the concentration of protoplasts was adjusted to 2 ×10⁵ protoplasts/mL. Protoplasts were transiently transformed using 5 mg of 35S-CcSAD-GFP plasmid with chloroplast marker RUB1sp-mCherry by PEG solution (40% PEG6000, 100 mM glucose, 10 mM CaCl2 and 0.7 mM KH2PO4 at pH 5.8).

Using the PEG/lithium acetate method (Gietz and Schiestl, 2007), the recombinant plasmids of pDR195 containing PMA1::CcSADs and pDR195 control (an empty vector) were transformed into S. cerevisiae strain BY4389, respectively to express proteins in transgenic yeast. S. cerevisiae transformants were selected on SD/-Ura minimal medium, and the positive clones with pDR195 (the control vector) or pDR195/PMA1::CcSADs were grown in SD/-Ura liquid medium at 30°C with sharking. When the OD600 value reached 0.8, OA or LA was added synchronously as a substrate for yeast cells. After being cultured at 30°C for 72 h, the cells were centrifuged at 4000 ×g for 10 min and washed with sterile water three times. The thalli were freeze-dried at -80°C overnight, and 10 mg of freeze-dried yeast was weighed into a 1.5 mL centrifuge tube, 200 μL of a 1 mg/mL C17:0 hexadecanoic acid hexane solution and 5 mg of glass beads were added. The 1.5 mL centrifuge tube was shaken for five min and then sonicated for 10 min for total lipid extraction. Samples were then centrifuged at 12000r/min for 10 min, and the supernatant was collected and stored at -80°C for FA content analysis.

The 35S-CcSAD-GFP construct recombinant plasmids were transferred into an Agrobacterium tumefaciens strain (GV3101) using the freeze-thaw method. The strain was then cultured in 20 mL of Luria-Bertani (LB) medium at 28°C for 16 hours with shaking at 200 rpm. Walnuts were submerged in 10% sodium hypochlorite (NaClO) and placed in a vacuum pump for 15 minutes. Afterward, the walnuts were washed 2-3 times with sterile water. The fruits were then peeled, and the embryos were placed in WPM medium. Embryos cultured in WPM medium were transferred into the bacterial liquid of A. tumefaciens with an OD600 value of 0.8. After 15-20 minutes of co-cultivation, the embryos were transferred to paper towels to dry. Once dry, the embryos were transferred to MS medium containing 100 µM As for incubation.

Chinese hickory stems, roots, leaves, and fruits, somatic embryos of walnut, and tobacco leaves were harvested for GC–FID analysis, all these plants were grown in normal conditions. The collected samples were transmethylated in 1 mL methanol containing 2% H₂SO₄ (v/v) and 50 μg butylated hydroxytoluene, and 300 μg of heptadecanoic acid (C17:0) was added to each sample as an internal standard. Transmethylation was maintained in a heated water bath (85°C) for 1.5 h and rapidly cooled in an ice box. Then 2 mL of 9% NaCl solution (m/v) and 2 mL of n-hexane were added for extracting FA methyl esters. Finally, the supernatant was used for analyzing FA components using DB-FASTFAME (Agilent). Nitrogen was used as the carrier gas in constant pressure mode at 28 psi. 1 µL of the sample was injected into the column and repeated three times. The inlet temperature was set at 250°C, the detector temperature was 260°C, and the GC oven was initially set at 80°C (hold for 0.5 min), ramped up to 175°C at 65°C/min, to 185°C at 10°C/min (hold for 0.5 min), and finally to 230°C at 7°C/min. Fatty acid fractions were estimated from the retention times of the fatty acid standards, and data were collected by peak area normalization using the C17:0 internal standard.

All the data were presented as means ± SE (n = 3) and statistically analyzed using SPSS 22.0 (SPSS Incorporated, USA). Mean differences between groups were analyzed using Student’s t-test.

Five SAD proteins were identified in Chinese hickory and named as CcSADs based on their similarity to Arabidopsis SAD proteins. They were further characterized based on their molecular weight, isoelectric point, amino acid, instability index, aliphatic index, and grand average of hydropathicity, respectively ( Table 1 ). The deduced CcSADs lengths varied from 289 to 396 amino acids. The molecular weight of CcSAD proteins ranged from 32.48 kDa to 45.22 kDa. CcSSI2-2 has a higher molecular weight of 45.22 kDa, followed by CcSSI2-1 with 45.57 kDa. CcSAD6 has the lowest molecular weight of 32.48 kDa among the CcSAD proteins.

To characterize the phylogenetic relationship among SADs from different plants and Chinese hickory, a neighbor-joining (NJ) tree was constructed using the MEGA software based on the alignment of five CcSADs in C. cathayensis, seven AtSADs in Arabidopsis thaliana, five PtSADs in Populus, and SAD proteins from other plants ( Figure 1A ). Topology of the phylogenetic tree indicated that the SAD gene family could be divided into two subgroups indicated by blue and green shading, respectively. We also performed a progressive alignment of SAD gene sequences from C. cathayensis and Arabidopsis, resulting in a multiple alignment ( Figure 1B ). The putative two conserved peptides (DETGASP and DYADILE) and the E/DEXXH motifs are denoted with boxes.

To clarify the evolutionary relationships, gene structure, conserved motifs, and conserved domains of five CcSAD genes were analyzed. Phylogenetic analysis indicated that CcSADs proteins had two subfamilies, which was similar to the topological structure of a phylogenetic tree in Figure 1A . Except for CcSAD6, all CcSADs had FA desaturase_2, Acyl_ACP_Desat conserved domains and Ferritin like superfamily domain ( Figure 2A ). By comparing the CDS of CcSADs with the genome sequence of the corresponding genes, the UTR, CDS, and intron distribution of CcSADs were analyzed in depth ( Figure 2B ). There was one intron in the two members CcSADs (CcSAD2 and CcSAD6). Members of the same subfamily showed similar exon/intron distribution patterns, such as CcSSI2-1 and CcSSI2-2 having two introns with a long first intron ( Figure 2B ). There were two conserved histidine rich regions which is a typical SAD characteristic, namely EENRHG and DEKRHE where aspartate (D) and histidine (H) provide necessary binding sites for ferric ions in the catalytic active center of CcSAD, ensuring that dehydrogenase has certain catalytic activity. Remarkably, closely related genes in the phylogenetic tree shared similar structure compositions. The online software MEME was used to analyze the motif distribution within the CcSAD genes. A total of 6 motifs were identified, and the results showed that each CcSADs contained 4 or 6 motifs ( Figure 2C ). Some motifs were common to all members, such as motif3, motif4, motif5, and motif6. The motif2 and motif3 were residues conserved that were related closely with the homodimer interface of acyl-ACP desaturase. The motif5 was also residues conserved related closely with fatty acid desaturase_2.

Analyzing promoter cis-elements offers valuable information on tissue-specific gene expression and stress response modes. More than 40 cis-elements were detected which were related to environmental stress, hormone-responsive, light-responsive, development, site binding, and other functions ( Figures 3A , B ), suggesting complex networks in regulation of CcSADs. There were four elements related to responses to environmental stress in CcSADs promoters: ARE (ABA-responsive element), TC-rich repeat, MBS (Myb binding site), and LTR (Low temperature responsive). Hormone-responsive elements were widely distributed in the upstream regulatory sequence of the CcSAD gene, including abscisic acid-responsive elements (e.g., ABRE), auxin-responsive elements (e.g., TGA-element), gibberellin -responsive elements (e.g., P-box, TATC-box), MeJA-responsive elements (e.g., CGTCA-motif, TGACG-motif), and salicylic acid-responsive elements (e.g., TCA-element).

To verify predicted results on the subcellular localization of CcSAD proteins, full-length CDSs of all CcSAD genes were successfully cloned into the pCAMBIA1300-EGFP vector. Results showed that the uninserted GFP protein was expressed in various organelles in N. benthamiana. However, the CcSAD-GFP fusion proteins (CcSAD2, CcSAD4, CcSSI2-1, and CcSSI2-2) were only expressed in the chloroplast, suggesting that these CcSADs were expressed and functional in the chloroplast ( Figure 4A ). In contrast, the green fluorescence signal from CcSAD6-GFP was localized in cytoplasm which was due to the absence of chloroplast signal peptide. In this experiment, the CcSAD-GFP and RUB1sp-mCherry chloroplast marker genes were co-transformed into Arabidopsis protoplasts. The protoplasts were then scanned for green and red fluorescence, respectively. When the SAD-GFP protein was localized in chloroplasts, it overlapped with a chloroplast marker to produce a yellow fluorescence signal ( Figure 4B ). This was due to the combination of the green fluorescence from CcSAD-GFP and the red fluorescence from the chloroplast marker. The results were consistent with the transient expression observed in tobacco.

Molecular docking is a computational technique aiming at predicting the favored orientation of a protein or a ligand to its macromolecular target (receptor) through energy matching, spatial matching, and chemical matching; when the protein and its receptor are bound together to form molecular complexes, the complex structures could be predicted (Salmaso and Moro, 2008). In short, it is the process of placing a ligand molecule into the active site of a protein macromolecule, allowing to observe the conformation of the small molecule bound to the protein and predicting the energy of action. The 3D structure of the CcSAD proteins predicted by Alphafold2 is presented in Figure 5 . Figures 5A, B showed the specific position of stearic acid and palmitic acid binding with CcSAD proteins, respectively. The binding energies of the ligands docked to the target proteins are described in Figure 5C . The lower the binding energy indicates the more substantial the docking of the protein to the ligand. The best binding affinity was predicted to be −4.47 kcal/mol for the conformation represented in CcSAD2- stearic acid-binding. All SAD proteins had a strong binding capacity for stearic acid and palmitic acid, except for CcSAD6, which had a weak affinity for stearic acid. Heptadecanoic acid (17:0) exhibited an insufficient binding capacity to CcSAD4 as it is not present in plants. The amino acid residues presented in the Figures 5A, B were predicted to interact with their substrates, meaning that the residues binding with stearic acid or palmitic acid were considered to be the predicted binding pocket identified through the molecular docking studies.

The expression level of each CcSADs gene in different tissues or organs, including endocarp, embryo, stigma as well as stems and leaves ( Figure 6A ) was analyzed. The results revealed that five CcSAD genes had diverse expression patterns among organs during C. cathayensis development. CcSAD6, CcSAD2, and CcSAD4 were mainly expressed in different developmental stages of stigma, and CcSAD2 was highly expressed in leaves. Genes overexpressed in embryo were CcSAD6, CcSSI2-1, and CcSSI2-2. Moreover, CcSAD6 and CcSSI2-1 were highly expressed in the endocarp. These results suggested that these CcSAD genes could play different roles in the development of different organs although their sequences were highly similar. Correlation analysis of unsaturated fatty acid contents during embryo development with the level of SAD gene expression ( Figure 6B ) showed that the accumulation of oleic acid (C18:1) was highly correlated with CcSSI2-1 and CcSAD6 gene expression (r > 0.9) and also correlated with CcSSI2-2 (r > 0.8). The expression patterns of the CcSAD gene differed in different tissues of hickory. CcSSI2-1, CcSSI2-2, and CcSAD6 were highly expressed in the embryo at the S2 stage, which corresponded to the rapid accumulation of hickory oil. Moreover, WRI1, a transcription factor involved in lipid regulation, also exhibited high expression at the S2 and S3 phases of hickory, suggesting that they may potentially regulate SAD genes in hickory (Huang et al., 2022a). The transcripts and proteins of CcSSI2-1, CcSSI2-2, and CcSAD6 were all expressed during embryo development, and they all were expressed at the highest level during the rapid accumulation of oil (S2) and then slowly decreased. The protein expression patterns of the CcSADs were also found to be consistent with RNA-Seq results ( Figure 6C ). Thus, CcSAD6, CcSSI2-1, and CcSSI2-2 were mainly expressed in embryos with the highest expression at the early developmental stage, indicating that these genes played important roles in regulation of oleic acid biosynthesis in embryos.

Fatty acid compositions in roots, stems, leaves, and embryos were analyzed by GC-FID ( Figure 7 ). Unsaturated fatty acids were higher than saturated fatty acids regardless of organs. In roots, saturated fatty acids (C16:0 and C18:0) were approximately 30% of all fatty acids; whereas C18:2 as the major unsaturated fatty acid accounted for more than 30% of total fatty acids. In stem tissue, unsaturated fatty acids were also the major component of the total fatty acids, with both C18:2 and C18:3 accounting for more than 30% of the total fatty acids. The fatty acid composition in leaves was similar to that of the stem tissue. It was noteworthy that the C18:1 content in embryo was almost 70% of the total fatty acids, and the C18:2 content was more than 20%.

To investigate the function of SAD proteins, their expressions in different cells and their effects on fatty acid compositions were examined. The transient expression in tobacco leaves was confirmed with laser confocal microscopy. Results showed that the total fatty acid content in tobacco leaves was low with C18:3 as the primary fatty acid accounting for more than 30% of the total fatty acids in leaves. Transient expression of CcSSI2-1 and CcSSI2-2 resulted in significant increase in C16:1 and C18:1. The expression of CcSAD2, CcSAD4, and CcSAD6 also led to the high unsaturated fatty acid content in tobacco leaves ( Figure 8A ). CcSAD genes were also expressed in yeast BY4389, during which C16:0 or C18:0 was added to the medium as a substrate for detecting changes in fatty acids after co-culture. Figure 8C shows that the expression of the hickory SAD gene in BY4389 reduced the proportion of C18:0 and increased the proportion of C18:1 from 15% to 30%, and CcSAD6 also increased the content of C16:1 ( Figure 8B ). It is known that genetic transformation of woody plants has been a challenging issue. The polyphenol content in the embryo of hickory is extremely high, which makes it impossible to establish an effective tissue culture system. In this study, we transformed SAD genes into walnut embryos. Walnuts are genetically close to hickory, have high oil content in the embryo, and are rich in unsaturated fatty acids. We analyzed transient expression of SAD genes in walnut embryos coupled with fluorescence detection ( Figures 8D , E ). Transient expression of CcSSI2-1 and CcSAD6 in walnut embryos revealed that the major fatty acids in walnut embryos were polyunsaturated fatty acids (C18:2 and C18:3) in which CcSSI2-1 increased the content of C18:2 while CcSAD6 promoted the synthesis of C18:3. Although the level of C18:0 in walnut embryos was low, both CcSSI2-1 and CcSAD6 were able to promote the biosynthesis of C18:1.