The walnut material used in this study is “XinXin2” from the Ministry of Agriculture and Rural Affairs’ Fruit Tree Science Observation Station in Xinjiang (Yecheng County, Kashgar). From July to September 2023, fruits of ten-year-old walnuts were collected at different stages: rapid growth period (S1, S2), hard core period (S3), rapid oil conversion period (S4), and fruit maturity period (S5). During collection, 30 fruits were selected each time from the middle of the canopy periphery of five test plants, choosing fruits of equal size, without injury or pest damage. After separating and removing the clean pulp, the kernels were quickly frozen in liquid nitrogen and stored in cryopreservation tubes for subsequent gene expression quantification. At the same time, gas chromatography was used to determine the fatty acid content in fruits at different stages. The sample processing and measurement methods refer to previous reports (Liu et al., 2022a).

Arabidopsis and soybean Fats pprotein were used as query sequences to blastp to the whole genome sequence of walnut to screen for candidate members. The E-value threshold for alignment was set to 1e-5. Simultaneously, the HMM model of the Acyl-ACP thioesterase N-terminal domain (PF01643) was employed to conduct a genome-wide search in walnut, with the threshold set to an E-value < 1e-5. The intersection of the two search results was considered as the candidate members. For the screened members, the sequences were manually submitted to the CDD (http://blast.ncbi.nlm.nih.gov) and SMART (http://smart.embl-heidelberg.de/) databases to remove genes with incomplete structure and without conservative domains. The remaining genes were identified as members of the walnut Fat gene family and named. The sequences were submitted to the ExPASy (http://web.expasy.org/compute_pi/) database to calculate the physicochemical properties of the protein, and the WoLF PSORT website (https://wolfpsort.hgc.jp/) was used to predict the subcellular localization of JrFats.

The annotation information of JrFat was manually extracted from the walnut genome annotation file. The MEME website (http://meme-suite.org/) was used to search for conservative base sequences of JrFat, with the minimum motif length set to 6, the maximum motif length set to 50, and the number of motifs set to 10. A phylogenetic tree of JrFat was constructed using MEGA7, using the neighbor-joining method, and bootstrap was set to 1,000. All results were visualized through Tbtools.

A phylogenetic tree of Fats genes from Arabidopsis, soybean, and walnut was constructed using the IQtree software, using the maximum likelihood method, and bootstrap was set to 1,000. MscanX was used to separately screen for the paralogous homology relationship of walnut JrFats genes and their orthologous homology relationship with Arabidopsis and soybean, to judge the replication type of JrFat homologous gene pairs. The KaKs_caculate software (v3) was used to calculate the nonsynonymous substitutions per nonsynonymous site (Ka) and the number of synonymous substitutions per synonymous site (Ks) values between homologous gene pairs (Zhang, 2020), and the divergence time of gene pairs was calculated using the chromosome rearrangement rate (1.30E-08).

The upstream 2,000 bp sequence of each JrFats gene was extracted as its promoter sequence and submitted to the PlantCARE database (https://bioinformatics.psb.ugent.be/webtools/plantcare/html/) to predict the cis-acting elements (CREs) in the promoter region. At the same time, the protein sequence of JrFats was submitted to the STRING database (https://string-db.org/) to search for and predict protein interaction relationships, and the protein interaction network was constructed using Cytoscape software.

The kernels of walnuts at different developmental stages were thoroughly crushed and ground after being frozen in liquid nitrogen, and then the EASYspin Plant RNA Rapid Extraction Kit (Beijing Tiangen Biotech Co., Ltd., Beijing, China) was used to extract RNA from the kernels according to the standard procedure provided. Agarose gel electrophoresis was used to determine the integrity and concentration of RNA, and then the qualified RNA was reverse transcribed into cDNA. qRT-PCR primers were designed for JrFats genes, and the primer list can be seen in the Supplementary Table S1. The primers were synthesized by Sangon Biotech Co., Ltd. (Shanghai, China). Using walnut 18S as an internal reference gene, the reaction was carried out using the Bio-rad CFX96 Real Time System (Hercules, California, United States) according to the standard procedure of the TAKARA SYBR Premix Ex Taq kit. Each experiment included three technical replicates and three biological replicates, and the expression level was calculated using the 2^−ΔΔCt technique. Visualizations and significance calculations were performed using Prism software (p < 0.01). The correlation between expression levels and fatty acid content was calculated using the Pearson correlation coefficient in R.

To identify Fat genes in walnuts, the amino acid sequences of Fat genes from Arabidopsis and soybean were utilized as query sequences for a blastp analysis against the walnut genome. Genes with high recognition rates were further scrutinized for domain determination, culminating in the identification of eight genes as members of the walnut Fat gene family (Table 1). The Fat gene was initially classified into the FatA/B subfamilies based on blast results, subsequently, they were named as JrFatA1-JrFatA2 and JrFatB1-JrFatB6 according to their respective physical loci in the genome. The 8 JrFat genes are dispersed across seven chromosomes. Apart from JrFatB3, which has a shorter amino acid length (223 aa) and lower molecular weight (25,683.22), the remaining JrFat amino acid lengths range from 379 to 459, and their molecular weights range from 42,686.53 to 51,806.35. The pI values of JrFat span widely from 6.5 to 9.41, indicating that all JrFat proteins are hydrophobic proteins. Predictive results suggest that JrFatA proteins may localize to the chloroplast, while JrFatB proteins may localize to the nucleus, cytoplasm, and chloroplast.

To further elucidate the evolutionary status and relationships of JrFat, we incorporated the Fat genes from Arabidopsis thaliana and soybean as outgroups to construct a phylogenetic tree (Figure 1A). Almost all branches obtained high confidence that the tree is stable. All 23 fat genes from the three species were classified into two subfamily, FATA and FATB. The number of FatA is identical across the three species, all of which reside within the same subgroup. In contrast, the FatB subfamily exhibits divergence among the three species. JrFatB1 and JrFatB2 clustered together with AtFatB from Arabidopsis thaliana and GmFatB1/B2 from soybean, forming a clade that likely represents the most conserved members within the FatB family. Only one JrFatB4 grouped with four GmFatB in the same clade, whereas AtFatB is absent, suggesting functional divergence or loss within this subgroup in walnuts, leading to a contraction of family members within this subgroup. JrFatB5/6 and GmFatB5A/5B are positioned on the same branch, with Arabidopsis similarly absent within this grouping. JrFatB3 displayed distinctiveness, standing apart from all other FatB genes, suggesting its unique phylogenetic position and sequence characteristics may be associated with specific functions.

A fundamental gene feature analysis was conducted on the JrFat gene family (Figure 1B). In the phylogenetic tree of Fat within the walnut, JrFat was divided into three clades (C1-C3). JrFatA1 and JrFatA2 both belong to C1, although JrFatB4 is also assigned to this clade despite a relatively low bootstrap value. JrFatB5 and JrFatB6 are situated in C2, while the remaining 3 JrFatB members are in C3. Although members of the JrFat gene family vary in length, their gene structures are relatively simple. All JrFat genes have 5 to 7 exons, with only JrFatA1/A2/B4/B1 possessing complete UTRs. JrFatB5 has the longest gene due to a single long intron, while JrFatB3 has the shortest gene. There appear to be no structural differences between JrFatA and JrFatB genes. Regarding conserved motifs, almost all JrFat genes exhibit similar conserved sequences, with motifs 1–7 distributed in the middle and C-terminal regions of the genes. However, JrFatB3 is an exception once again, reflecting its shorter amino acid length and possession of only four motifs, consistent with its gene and protein structures. Minor differences in motif distribution exist between JrFat genes in different clades. Specifically, motifs 8 and 10 are uniquely present in C3 (except JrFatB3) and C2, respectively, while C1 lacks unique motifs. Additionally, the distribution of motif nine differs between clades, potentially reflecting differences in the localization and substrate recognition of different JrFat proteins.

The homologous relationships of JrFat are investigated. Within the JrFat gene family, there are three pairs of paralogs: JrFatA1-JrFatA2, JrFatB1-JrFatB2, and JrFatB5-JrFatB6, all of which cluster together in the phylogenetic tree (Figure 2A). JrFatB3 and JrFatB4 lack paralogs (Figure 2A). Additionally, utilizing whole-genome alignment, orthologs between walnut, model plants Arabidopsis thaliana, and soybean, the latter being a high-oil content species with a well-characterized Fat, have been screened. Except for JrFatB3, the remaining 7 JrFat genes exhibit direct orthologous relationships with GmFat genes (Figure 2B). JrFatB1/B2 each have four orthologs, namely GmFatB1/B2 genes, maintaining both evolutionary conservation and orthologous relationships. JrFatB4 has two orthologs, GmFatB3A and GmFatB3B, indicating potential gene loss or functional divergence of its paralogs, similar to the situation observed for JrFatB3, which lacks both paralogs and orthologs. JrFatA1/A2 are only orthologous to GmFatA1B, with no direct orthologous relationship with GmFatA1A, despite their high sequence similarity and placement within the same branch of the phylogenetic tree, suggesting that the JrFatB subgroup may be more evolutionarily conserved than the JrFatA subgroup.

We calculated the ratio of Ka to Ks for the three pairs of paralogs within the JrFat gene family (Table 2). The Ka/Ks ratios for JrFatA1-JrFatA2, JrFatB1-JrFatB2, and JrFatB5-JrFatB6 were 0.18, 0.15, and 0.21, respectively, all significantly below 1. This indicates strong negative selection acting on these gene pairs, resulting in protein sequences that tend to be stable with slow evolutionary rates. All gene pairs are located on different chromosomes, suggesting segmental duplication. Furthermore, we estimated the divergence times between different gene pairs. JrFatA1-JrFatA2 diverged approximately 15.71 million years ago (Mya), JrFatB1-JrFatB2 diverged approximately 14.04 Mya, and JrFatB5-JrFatB6 diverged approximately 13.18 Mya. The divergence times of JrFat genes are relatively close to each other and significantly more recent than the divergence time between J. regia and M. rubra (31 Mya) and the last whole genome duplication event (24.48 Mya). However, they are considerably earlier than the divergence time between J. regia and its related species (at the latest 3.45 Mya) (Zhang et al., 2020).

The promoter sequences, defined as the upstream 2,000 bp sequences of each JrFat gene, were analyzed to identify cis-regulatory elements (CREs). Apart from core and basic CREs, other elements were classified into three functional categories: phytohormone response, light response, and stress response (Figure 3). These categories of elements are broadly distributed across all JrFat gene promoters. Although in JrFatB1 and JrFatB2, CREs tend to concentrate more towards the distal end of the promoter (between 400 and 2,000 bp), this pattern appears somewhat indistinct. Overall, there is no apparent specific or clustered distribution of CREs in the promoter regions of JrFat genes. In terms of element quantity, JrFatB1/B3/B6 exhibit a relatively higher abundance of ARE elements, which are considered essential for anaerobic induction. Apart from these three genes, the remaining 5 JrFatA and JrFatB gene promoters contain a higher number of ABRE, G-Box, and Box4 elements. ABRE is known to be involved in abscisic acid responsiveness, while G-Box and Box4 elements are associated with plant light responsiveness. There is no significant difference in the quantity and distribution of CREs among different clades. All JrFat genes, whether FatA or FatB, appear to be extensively involved in walnut’s response to various environmental stimuli, suggesting minimal differences in regulatory expression among JrFat genes.

To elucidate the function of JrFat proteins, we screened publicly available data for proteins with high-confidence interactions with JrFat proteins and constructed a protein interaction network, the names or family names of interacting proteins are annotated base STRING (Figure 4). There are seven interaction relationships among the 7 JrFat proteins, including six interactions where JrFatB4 interacts with the other JrFat proteins, and one interaction between JrFatA2 and JrFatB5, indicating JrFatB4 as the central node in the interaction network. Additionally, the 7 JrFat proteins interact with 25 other proteins, either functionally or physically. These interacting proteins mainly belong to families such as Long chain acyl-CoA synthetase-like (ACSL), acyl-activating enzyme (AAE), Beta-ketoacyl-ACP synthases (FabF), short-chain dehydrogenases/reductases (SDR), fatty acid desaturase (FAD), and Caffeoylshikimate esterase (CSE). They are widely involved in fatty acid metabolic or biosynthetic processes and are also implicated in wax biosynthetic processes, cutin biosynthetic processes, and responses to ozone, among others. These results suggest that Fat proteins may synergize with these proteins to mediate fatty acid synthesis and metabolism in different growth and developmental stages of walnuts. Notably, JrFatB3 did not exhibit any interactions with other proteins in the screening.

To clarify the changes in the main fatty acid content at different developmental stages of walnut kernels, we selected walnut fruits at five stages and measured the fatty acid content in their kernels (Figure 5). The content of two saturated fatty acids, Palmitic acid and Stearic acid, first increased and then decreased throughout the kernel development stage, and was higher at the final fruit maturity than in the early development stage. Notably, the content of Palmitic acid only increased rapidly at the S3 stage and then immediately decreased. However, the content of Stearic acid continued to rise from the S2 stage until it began to decrease at the S4 stage. Among the three unsaturated fatty acids, the content of Oleic acid showed almost the same trend as Palmitic acid, also increasing rapidly only at the S3 stage and then decreasing. The content of Linoleic acid also increased rapidly at the S3 stage, but its content remained basically unchanged in the subsequent kernels. The content of Alpha-linolenic acid continued to increase throughout the kernel development stage, especially in the late development stage (S3-S5), with significant growth at each stage. Among the five fatty acids measured, the proportion of unsaturated fatty acids fluctuated within the range of approximately 86%–88%, and fluctuated with the development of the kernel. The proportion of unsaturated fatty acids was highest at the S3 stage, at 88.05%.

To determine the role of the Fat gene family in the synthesis and metabolism of fatty acids in walnut kernels, we performed real-time fluorescence quantification using the same period samples. The expression of all genes was normalized according to the expression of the acting gene to avoid statistical bias caused by the five developmental stage (Figure 6A). The expression trends of JrFatA1, JrFatA2, JrFatB5, and JrFatB6 genes are similar, with their expression significantly upregulated at stages S3 and S5. The difference is that JrFatA1 is downregulated at the S4 stage while the other three genes are not. The expression trends of JrFatB1, JrFatB2, and JrFatB4 are similar, with their expression slightly upregulated at stages S1-S3, and then downregulated at stages S4-S5. The expression of JrFatB3 remains basically unchanged throughout the development stage. To further clarify the association between specific JrFat and fatty acid content, we calculated the correlation between the expression of 8 JrFats and the content of five fatty acids and the ratio of saturated and unsaturated fatty acids (Figure 6B). The results show that the expression of JrFatA1, JrFatA2, and JrFatB6 has a strong positive correlation with the content of Alpha-linolenic acid, with correlations of 52%, 72%, and 58% respectively. They may positively regulate the accumulation of Alpha-linolenic acid content in unsaturated fatty acids. The expression of JrFatB5 has a strong positive correlation with Palmitic acid, with a correlation of 55%. This gene may promote the accumulation of Palmitic acid content in saturated fatty acids. The expression of JrFatB2 is negatively correlated with all fatty acids, with the strongest negative correlation with Palmitic acid and Oleic acid, respectively −73% and −77%. Its expression has a 70% positive correlation with the ratio of saturated and unsaturated fatty acids. This gene may mediate the conversion between different fatty acids during the development of walnut kernels. The correlation between JrFatB1, JrFatB3, and JrFatB4 and the changes in the content of five fatty acids during all kernel development processes is relatively low.