Genome-wide identification of NAC gene family in pecan and its expression patterns during graft healing Authors: Muge Niu, Yunying Wu, Yuanchen Xu, Qihan Zheng, Xinwei Xu, Mingwei Wang, KaiKai Zhu, Pengpeng Tan, Fangren Peng Published: 2026 ABSTRACT The NAC gene family represents a class of plant-specific transcription factors characterized by a conserved NAC domain. These proteins exert pivotal regulatory influences in diverse biological processes, including plant growth and development, stress responses, and hormone signaling. Recent studies have further highlighted their core functions in graft union formation, scion-rootstock interactions, and the establishment of systemic functionality. In this study, we systematically identified the NAC gene family in pecan (Carya illinoinensis), revealing 116 CiNACs members distributed across.s 16 chromosomes. Phylogenetic analysis classified them into 16 subfamilies and 12 distinct groups, indicating evolutionary conservation of their functions. Interspecific collinearity analysis suggested that the family has undergone expansion and differentiation, leading to broad functional divergence. Furthermore, we screened six key CiNACs genes potentially associated with grafting and analyzed their expression patterns across different tissues, under drought stress, and during grafting via RNA-Seq and qRT-PCR. The results showed that, except for CiNAC24, the other five genes were significantly upregulated in stem tissues. Among them, the expression profiles of CiNAC24 and CiNAC72 implied a stronger responsiveness to abiotic stress rather than direct involvement in grafting. In contrast, CiNAC115 exhibited significant upregulation during grafting, suggesting its potential key role in regulating graft union formation. Collectively, this study provides important clues for elucidating the molecular mechanisms of CiNAC115 in grafting and lays a theoretical foundation for further uncovering the functions of the NAC gene family in the grafting process.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12870-026-09080-x. Background Pecan (Carya illinoinensis) is a high‑quality woody oil crop [1] with considerable economic value and ecological benefits [2]. Its thin shell is easy to remove, and the kernel is rich in aroma and fat, with an oil content reaching approximately 70%. These characteristics make it suitable both as a premium edible‑oil source and for direct consumption [3]. Moreover, the fatty acids and multiple bioactive components (e.g., phenolics, phytosterols, vitamin E) present in pecan are considered to have positive effects on the prevention of cardiovascular diseases [4]. Additionally, pecan is one of the world’s important high‑quality hardwood timber species, featuring fine wood texture, good bending properties [5], attractive grain, and elegant color, which enable its wide range of applications. In the early 20th century, pecan was introduced into China and has been cultivated mainly in subtropical climatic regions such as Yunnan, Jiangsu, and Jiangxi. However, its long juvenile growth period hinders early commercial‑scale planting [6]. Consequently, vegetative propagation methods such as grafting and cutting are commonly used for its introduction and promotion. For example, the widely adopted “patch budding” technique in recent years can effectively improve grafting success rates, thereby enhancing planting profitability [6]. Nevertheless, graft compatibility still varies markedly among cultivars; for instance, the ‘Pawnee’ cultivar shows high survival, whereas the ‘Jinhua’ cultivar is difficult to establish [7]. Therefore, identifying genes that influence graft success in pecan is of great importance for germplasm preservation and industry development.`. To improve grafting efficiency across different cultivars, Mo et al. reported a significantly up‑regulated NAC gene during grafting in a de novo transcriptome study of the ‘Pawnee’ cultivar, suggesting its potential role in the grafting process [6]. However, due to the lack of a pecan genome reference at that time and the absence of clear reports on NAC gene family involvement in grafting, the precise function of this gene remained unclear, with only speculative inference about its regulatory participation. Subsequently, the successful assembly of the pecan genome provided an essential foundation for related research. Notaguchi et al. demonstrated in Science that the NAC family member ANAC071 is involved in grafting [8]. Furthermore, Wang et al. noted in a Horticulture Research review on graft‑union formation mechanisms that ANAC071/096 function in rootstock and that ANAC071 also exerts an important influence in the cambium [9]. These advances have renewed our interest in the expression and function of the CiNAC gene family during grafting. The acronym “NAC” derives from the names of three proteins with conserved domains: NAM from petunia (No Apical Meristem), and ATAF1/2 and CUC1/2 from Arabidopsis [10]. Among these, the NAM gene is essential for boundary formation in embryos and floral organs; ATAF1/2 not only functions in stress defense but also participates in the regulation of phytohormone homeostasis under normal conditions; whereas CUC1/2 is considered a core regulator establishing “boundaries” during plant morphogenesis [10]. Subsequent studies have demonstrated that the NAC gene family exerts pivotal influence in plant growth, development, and stress‑response mechanisms [11]. As one of the largest families of transcription factors, NAC genes have been identified and characterized in numerous plant species [12], including staple crops such as rice [13], maize [14], and oat [15], as well as major horticultural crops such as grape [16], tomato [17], and apple [18]. Structurally, NAC proteins exhibit a highly modular organization, consisting of a conserved N‑terminal NAC domain and a divergent C‑terminal transcriptional regulation region [19]. The N‑terminal domain can be further subdivided into five subdomains (A–E). Subdomains A, C, and D are highly conserved across species, whereas subdomains B and E display certain species‑specific variations. Subdomain A, the most conserved region, is involved in protein dimerization and contributes to structural stability. Subdomain B is relatively less conserved, which may correlate with its functional diversity. Subdomains C and D contain predicted nuclear localization signals (NLS), while subdomains D and E are responsible for direct DNA binding [20]. As one of the largest plant‑specific transcription factor families, NAC genes exhibit remarkable functional diversity [21]. They are widely involved in key biological processes such as plant growth and development, organ formation [22], and responses to both biotic and abiotic stresses [23]. Functional studies across different species have revealed substantial versatility within this family: in kiwifruit, NAC genes regulate leaf senescence and organ development [24]; in tomato, they participate in hormone signaling and induce specific defense responses [25]; in Taxus, family members act as critical regulators in the transcriptional control of paclitaxel biosynthesis [26]; in Zanthoxylum, NAC genes respond to drought and other environmental stresses [27]; and in barley, they activate the flavonoid biosynthesis pathway to promote secondary metabolite accumulation [28]. Despite these well‑documented roles in diverse physiological processes, research on NAC functions in grafting—an important agricultural practice—remains relatively limited. NAC genes may exert crucial regulatory influences in scion‑rootstock interactions, vascular tissue regeneration, and signal transduction during grafting, yet the underlying mechanisms are still unclear. Therefore, in this study, we systematically identified NAC family members in the pecan genome, analyzed their phylogenetic relationships, conserved motifs, and expression patterns. The findings provide a foundation for future investigations into the potential functions of this gene family in the regulation of grafting. Materials and methods Plant materials The plant materials used for qRT-PCR were grown at the Pecan Germplasm Base of Nanjing Forestry University (31°52′45″ N, 119°09′06″ E). In early September 2025, uniformly cultivated and vigorous one-year-old seedling container plants were used as rootstocks. Healthy one-year-old branches of the pecan cultivar ‘Pawnee’ were selected as scions, and grafting was performed using the patch budding method. The containers measured 17 cm in upper diameter, 13 cm in lower diameter, and 20 cm in height, each filled with 10 kg of growing substrate consisting of a 1:1:1 volume ratio of peat, vermiculite, and perlite. Following the method described by Mo et al. [6], the graft junctions were photographed, collected, and immediately frozen in liquid nitrogen at 2 h, 8 d, 15 d, and 30 d after grafting. Each time point included three biological replicates, with each replicate consisting of 6–10 plants. The rootstock was labeled “Rootstock” and the scion “Scion.” Phloem, cambium (including exudates), and the outermost 1‑mm layer of xylem were collected from the graft junction to investigate the response strategies of the graft union and surrounding tissues during the grafting process. Identification of NAC genes in pecan The reference genome of C. illinoinensis ‘Pawnee’ was downloaded from the Phytozome database (https://phytozome-next.jgi.doe.gov/info/CillinoinensisPawnee_v1_1) [29]. The hidden Markov model (HMM) profile of the NAM domain (PF02365) was retrieved from the Pfam database (http://pfam-legacy.xfam.org/) [30]. A genome-wide search was performed using the HMMER search program in the graphical interface software SPDE V3.0 to identify all putative NAC protein sequences from the C. illinoinensis genome [31]. All identified NAC genes were subsequently named based on their chromosomal locations. Following the analytical approach described by Zhu et al. [32], the physicochemical properties of the putative pecan NAC proteins were predicted using the ExPASy online tool (https://www.expasy.org/). Subcellular localization predictions were conducted using WoLF PSORT (https://wolfpsort.hgc.jp/). The conserved motifs within pecan NAC proteins were analyzed using the MEME suite (https://meme-suite.org/meme/) with the motif count set to 10. Cis-acting regulatory elements in the promoter regions of CiNAC genes were identified using the PlantCARE database (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) [33]. Based on their functional annotations, these elements were classified into five categories: growth-related, hormone-responsive, light-responsive, stress-responsive, and transcription factor binding sites. The distribution of these elements was visualized using a bubble plot generated with OmicStudio tools (https://www.omicstudio.cn/tool) [34]. Annotation information for the 116 CiNACs genes was extracted from the reference genome annotation file. TBtools-II V2.373 was used to integrate and visualize the results from the above analyses, including gene structures [35]. Phylogenetic tree construction, chromosomal localization, and synteny analysis The amino acid sequences of 110 Arabidopsis thaliana NAC proteins were obtained from the PlantTFDB database (https://planttfdb.gao-lab.org/). A phylogenetic tree was constructed using MEGA V11.0.11, incorporating the 110 Arabidopsis NAC sequences and the 116 identified CiNAC protein sequences [12, 35]. The tree was built employing the maximum likelihood method with 1000 bootstrap replicates for nodal support assessment. The resulting tree was subsequently visualized and refined using the iTOL online platform (https://itol.embl.de/). To unify nomenclature and enhance topological confidence, we merged the initially defined 16 subfamilies into 12 monophyletic groups (Groups I–XII) based on strongly supported branches from the phylogenetic tree. All subsequent analyses were conducted using this classification (see Supplementary Table S1 for the correspondence between the two groupings). Based on the reference genome annotation, the chromosomal positions of the CiNAC genes were mapped and visualized using TBtools-II V2.373. For intra-species synteny analysis, the MCscanX toolkit integrated within TBtools-II was utilized. The syntenic relationships were graphically represented using the software’s Circle Gene View and Advanced Circos visualization modules. To explore the syntenic relationships of NAC genes between pecan and other species, namely Arabidopsis thaliana, Vitis vinifera (grape), and Juglans regia (walnut), their respective reference genome files were downloaded from NCBI (https://www.ncbi.nlm.nih.gov/). Inter-species synteny information and collinear NAC gene pairs were identified using the method described above and visualized accordingly. Protein sequences of NAC genes from C. illinoinensis, Juglans regia, Vitis vinifera and Arabidopsis thaliana were subjected to orthogroup inference using OrthoFinder (v2.5.5) with default parameters (-og flag). The resulting orthogroups were examined to identify those containing CiNAC genes but lacking any member from the other three species. To assess the selection pressure acting on these gene pairs, the non-synonymous (Ka) to synonymous (Ks) substitution ratios were calculated for all collinear CiNAC gene pairs based on their CDS and corresponding protein sequences. This calculation was performed using the Ka/Ks Calculator tool in TBtools-II V2.373, and the results are provided in Supplementary File S2. Spatiotemporal expression pattern analysis of CiNACs Following the methodology described by Zhu et al. [36], transcriptomic data from various pecan tissues were obtained from the NCBI BioProject database (Accession: PRJNA799663). Additionally, transcriptome data of ‘Pawnee’ grafted seedlings under drought stress at 0, 3, 6, 9, 12, and 15 days were downloaded from the Gene Expression Omnibus (GEO) database (Accession: GSE179336). The expression abundance of CiNAC genes was quantified using the Fragments Per Kilobase of exon model per Million mapped reads (FPKM) values (results provided in Supplementary Files S3 and S4). The FPKM data were log2-transformed. Heatmaps were subsequently generated using the ggplot2 and pheatmap packages in R software (v3.0.3). Schematic diagrams illustrating the expression patterns of selected CiNAC genes across different tissues were created using Adobe Illustrator 2020. Transcriptome data from the graft union of ‘Pawnee’ at 0, 8, 15, and 30 days after grafting were retrieved from the Sequence Read Archive (SRA) database (Accessions: SRR6065910–SRR6065921), as previously described [6]. As the original data were derived from a non-reference transcriptome study, we used the published reference genome [29] for sequence alignment and subsequent analysis. Clean reads were aligned to the reference genome using HISAT2 software for fast and precise mapping [37]. The aligned reads were then assembled and a transcriptome was reconstructed using StringTie [38] for downstream expression quantification. Consistent with the aforementioned method, CiNAC gene expression levels were calculated as FPKM values (results provided in Supplementary File S5), and heatmaps were generated accordingly. Finally, hierarchical clustering (H_Cluster) analysis was performed using the pheatmap package in R (v3.0.3) with Euclidean distance and complete linkage, and the number of clusters was set to nine. To evaluate cluster robustness, silhouette coefficients were calculated based on a distance matrix derived from 1 − Pearson correlation of row-normalized expression values. Clusters with mean silhouette > 0.5 were considered to have acceptable internal consistency [39]. qRT-PCR analysis of CiNAC genes at different grafting time points Total RNA was isolated and purified from samples collected at different grafting stages using the TRIzol reagent. RNA quantity and purity were assessed using a NanoDrop ND-1000 spectrophotometer (NanoDrop, Wilmington, DE, USA). RNA integrity was verified with a Bioanalyzer 2100 system (Agilent, Santa Clara, CA, USA). Subsequently, first-strand cDNA was synthesized from the qualified total RNA using the HiScript III RT SuperMix for qPCR (+ gDNA wiper) kit (R333, Vazyme, Nanjing, China). Quantitative real-time PCR (qRT-PCR) was performed on a FQD-96 C real-time PCR detection system (BIOER, Hangzhou, China) using the ChamQ Universal SYBR qPCR Master Mix kit (Q711, Vazyme). Each experiment consisted of three independent biological replicates, with each sample analyzed in triplicate (technical replicates). CiActin (CiPaw.03G124400) was used as the internal reference gene [36]. The relative expression levels of target genes were calculated using the 2−ΔΔCt method and visualized with GraphPad Prism software (version 10.1.2). Primer sequences used in this study are listed in Supplementary File S6. Statistical significance was determined by one‑way ANOVA followed by Tukey’s post‑hoc test (P < 0.05). Results Identification and phylogenetic analysis of the NAC gene family in C. illinoinensis Through identification of the NAM domain and phylogenetic analysis, a total of 116 CiNAC genes were obtained and designated as CiNAC1 to CiNAC116 based on their chromosomal locations (Fig. 1B). All CiNAC proteins contain the NAM domain (PF02365). We analyzed the physicochemical properties of the 116 CiNAC proteins, including amino acid number, molecular weight, isoelectric point, grand average of hydropathicity (GRAVY), and subcellular localization (Supplementary File S1). Their amino acid lengths ranged from 125 (CiNAC17) to 1047 (CiNAC26), and their molecular weights varied from 14,554.78 Da (CiNAC17) to 119,998.46 Da (CiNAC26). Among the 116 CiNAC family members, 38 proteins were basic (pI > 7), while the others were acidic (pI < 7), with isoelectric points ranging from 3.75 (CiNAC57) to 9.52 (CiNAC34). All 116 CiNAC proteins were hydrophilic, with GRAVY values between − 0.362 and − 1.04. Subcellular localization prediction indicated that 108 members were localized to the nucleus, while 8 were not predicted to be nuclear; 19 members were predicted to localize to two or more organelles. Fig. 1Phylogenetic and chromosomal distribution of the NAC gene family in C. illinoinensis. A Phylogenetic tree of NAC proteins from C. illinoinensis and A. thaliana. Red circles and blue stars represent proteins from C. illinoinensis and A. thaliana, respectively. B Chromosomal distribution of the identified CiNAC genes. The density of genes on each chromosome is indicated by the gradient of the chromosomal bars Phylogenetic and chromosomal distribution of the NAC gene family in C. illinoinensis. A Phylogenetic tree of NAC proteins from C. illinoinensis and A. thaliana. Red circles and blue stars represent proteins from C. illinoinensis and A. thaliana, respectively. B Chromosomal distribution of the identified CiNAC genes. The density of genes on each chromosome is indicated by the gradient of the chromosomal bars To investigate the evolutionary relationships of CiNAC genes, a phylogenetic tree was constructed using Arabidopsis and pecan sequences (Fig. 1A). Following the classification methods of Liu et al. [12] and Ooka et al. [40], the 116 CiNAC members were initially divided into 16 subfamilies. However, according to previous naming conventions, the ANAC063 and ONAC022 subfamilies did not cluster together. Variations in clustering criteria across iterative algorithms may lead to changes in grouping outcomes [41]. Moreover, the continuous expansion of reported Arabidopsis NAC members in recent years may also contribute to alterations in cluster structure [42]. To better represent the relationships among CiNAC members, we divided them into 12 groups (designated I–XII) based on the clustering results. The grafting-related ANAC071 and ANAC096 clustered into Group VI (ANAC011 subfamily). Another adjacent cluster containing CiNAC78 and CiNAC87 may also play roles in the grafting process of C. illinoinensis. Domain, motif, and gene structure analysis of CiNAC proteins The NAC protein contains a conserved NAM domain, which is responsible for DNA binding and is crucial for its biological functions. In this study, comparative analysis of the domain sequences of the CiNAC family members in C. illinoinensis was conducted, further dividing the NAM domain into five subdomains (A–E) (Fig. 2). The results indicated that most CiNAC proteins possess all five conserved subdomains, suggesting high sequence conservation during evolution. However, subdomain deletions were observed in some genes, including CiNAC01, CiNAC22, CiNAC42, CiNAC48, CiNAC53, and CiNAC61. Fig. 2Multiple sequence alignment of the conserved NAC domain in 116 CiNAC proteins. Sequences with incomplete or absent N-terminal domains were excluded from the alignment. Unshaded columns indicate sites where sequence similarity among family members is below 35% Multiple sequence alignment of the conserved NAC domain in 116 CiNAC proteins. Sequences with incomplete or absent N-terminal domains were excluded from the alignment. Unshaded columns indicate sites where sequence similarity among family members is below 35% To further characterize the sequence features of CiNAC proteins, ten conserved motifs were systematically identified, with lengths ranging from 15 to 50 amino acids (Fig. 3). Among them, Motif 2, Motif 4, Motif 1, Motif 6, and Motif 3 correspond to subdomains A through E, respectively. The composition of conserved motifs was generally consistent within the same subfamily, with most subfamilies containing one to six motifs. Notably, certain motifs were specific to particular subfamilies; for example, motif 9 was exclusively present in NAC2 subfamily and CiNAC12 gene, while motifs 7, 8, and 10 were unique to the ONAC003 subfamily. Fig. 3Gene structure and conserved motifs of the CiNAC gene family. The exon-intron structure of each CiNAC gene is shown on the left. Conserved protein motifs identified in the corresponding CiNAC proteins are displayed on the right Gene structure and conserved motifs of the CiNAC gene family. The exon-intron structure of each CiNAC gene is shown on the left. Conserved protein motifs identified in the corresponding CiNAC proteins are displayed on the right To further investigate structural variations among CiNAC genes, the distribution and number of exons and introns were analyzed (Fig. 3). The number of introns ranged from 1 to 4, and exons from 2 to 8. CiNAC66 contained the highest number of introns (4), whereas CiNAC45 possessed the most exons (8). Most genes harbored two introns and three exons. Genes within the same subfamily exhibited similar exon/intron counts, reflecting high structural conservation within the NAC gene family during evolution. Gene duplication and collinearity analysis of CiNACs To investigate the contribution of gene duplication to the expansion of the CiNAC gene family, intra-species synteny within C. illinoinensis was annotated and analyzed (Fig. 4). We identified 83 syntenic gene pairs involving nearly all CiNAC family members, indicating that most CiNAC genes were retained following genome duplication events. Chromosomes 3 and 4 harbored the highest number of syntenic CiNAC genes, suggesting their potential key roles in the expansion of the NAC family. Furthermore, two tandemly duplicated genes (CiNAC3 and CiNAC4) were detected in the C. illinoinensis genome. Both segmental and tandem duplications contribute to gene family expansion and evolution [33], thereby increasing the diversity of the CiNAC family. Fig. 4Intra-species synteny analysis of the NAC gene family in C. illinoinensis. Red lines connect syntenic CiNAC gene pairs. Chromosome numbers are labeled within the gray boxes Intra-species synteny analysis of the NAC gene family in C. illinoinensis. Red lines connect syntenic CiNAC gene pairs. Chromosome numbers are labeled within the gray boxes To further explore the evolutionary relationships of NAC genes across species, inter-species synteny analyses were performed among Arabidopsis thaliana, Vitis vinifera, and Juglans regia (Fig. 5). Strong syntenic relationships were observed between corresponding chromosomes of the different species. C. illinoinensis and J. regia each had 16 chromosomes and shared 228 syntenic gene pairs (compared to 116 pairs between C. illinoinensis and V. vinifera, and 114 pairs between C. illinoinensis and A. thaliana), indicating a closer genetic and evolutionary relationship between pecan and walnut. Fig. 5Inter-species synteny analysis of NAC genes among C. illinoinensis, Arabidopsis thaliana, Vitis vinifera, and Juglans regia. Gray lines in the background represent collinear blocks between C. illinoinensis and the other species. Colored lines highlight syntenic NAC gene pairs Inter-species synteny analysis of NAC genes among C. illinoinensis, Arabidopsis thaliana, Vitis vinifera, and Juglans regia. Gray lines in the background represent collinear blocks between C. illinoinensis and the other species. Colored lines highlight syntenic NAC gene pairs Additionally, the Ka/Ks ratios for all syntenic NAC gene pairs were calculated (Supplementary File S2). In comparisons with more distantly related species, a substantial number of Ka/Ks values could not be calculated due to high sequence divergence (pS ≥ 0.75), a phenomenon indicating that the sequence evolutionary rate exceeds the comparative capacity of standard models [43]. Specifically, 50 and 9 Ka/Ks values were missing in the syntenic pairs between C. illinoinensis and A. thaliana, and between C. illinoinensis and V. vinifera, respectively. In contrast, only one missing Ka/Ks value was found in the comparison with the most closely related species, J. regia. For gene pairs with calculable ratios, most Ka/Ks values were less than 1, suggesting that these genes have predominantly undergone purifying selection. Notably, four gene pairs with Ka/Ks values greater than 1 were identified exclusively in the C. illinoinensis–J. regia comparison, namely CiNAC16–rna‑XM_018971994.2, CiNAC30–rna‑XM_035694353.1, CiNAC63–rna‑XM_018977264.2, and CiNAC67–rna‑XM_019003236.2 (see Supplementary File S2 for details), hinting at potential positive selection acting on these genes. To identify NAC genes unique to C. illinoinensis, we performed orthogroup clustering using OrthoFinder across four species (C. illinoinensis, J. regia, V. vinifera, and A. thaliana). A total of 79 orthogroups were obtained (Supplementary Files S8). All CiNAC genes were assigned to orthogroups that also contained at least one ortholog from each of the other three species. No orthogroup was found to contain CiNAC genes exclusively (i.e., no CiNAC member was absent in J. regia, V. vinifera, and A. thaliana simultaneously). These results indicate that the NAC gene family is highly conserved among these species, and the functional divergence of CiNAC genes likely arises from differential expression regulation rather than lineage-specific gene gain/loss. Analysis of Cis‑acting elements in CiNAC promoters To predict potential upstream regulators of CiNAC genes, a 2000 bp upstream sequence of each of the 116 CiNAC genes was analyzed using the PlantCARE database (Fig. 6, Supplementary File S8). Identified cis-acting elements were classified into five functional categories: growth-related, hormone-responsive, light-responsive, stress-responsive, and transcription factor binding sites. The most abundant category was light-responsive elements (1,553), followed by hormone-responsive (843) and stress-responsive elements (556). Growth-related elements (302) and transcription factor binding sites (43) were the least represented. Among the light-responsive elements, Box 4 (501), Gap-box (351), and G-box (394) were predominant, suggesting that CiNAC genes may be responsive to light and stimuli. Within hormone-responsive elements, 372 ABRE and 336 CGTCA-motif elements were detected, indicating potential regulation by multiple plant hormones. The stress-responsive category contained 225 ARE elements, implying a role in plant responses to adverse environments. Among growth-related elements, 79 WUN-motifs (wound-inducible elements) were most frequent, further supporting a potential function in processes such as grafting. These results suggest that CiNAC genes (particularly those in subfamilies like NAM and ATAF) may be involved in diverse regulatory pathways related to hormone signaling, environmental stress, and growth/development, exerting significant influences in plant physiology and developmental processes, especially grafting. Fig. 6Enrichment and distribution of cis-acting regulatory elements in the promoters of CiNAC genes. The size of each dot corresponds to the count of a specific element type, with larger dots indicating a greater number Enrichment and distribution of cis-acting regulatory elements in the promoters of CiNAC genes. The size of each dot corresponds to the count of a specific element type, with larger dots indicating a greater number Expression profiles of CiNAC genes in different tissues To gain insight into the potential functions of CiNAC genes, a heatmap was generated to visualize the FPKM values of all 116 members across various tissues, including female flowers, male flowers, fruit, seeds, roots, stems, and leaves (Fig. 7A; Supplementary File S2). Expression patterns of six representative genes were illustrated separately (Fig. 7B). Five genes (CiNAC11, CiNAC72, CiNAC78, CiNAC87, and CiNAC115) were upregulated in stems. Four genes (CiNAC11, CiNAC78, CiNAC87, and CiNAC115) were highly expressed in roots. Only CiNAC72 showed upregulated expression in leaves. Notably, several genes exhibited upregulation in vegetative tissues. For instance, CiNAC11, CiNAC24, and CiNAC78 were upregulated in both female and male flowers. CiNAC24 and CiNAC72 displayed high expression levels in male flowers. CiNAC24, CiNAC72, and CiNAC115 were upregulated in fruit. In contrast, all six genes showed relatively low expression in seeds. Fig. 7Expression profiles of CiNAC genes across different tissues. A Heatmap depicting the FPKM values (log2-transformed) of 116 CiNAC genes in various tissues. Values represent the mean of three biological replicates. The phylogenetic tree on the left provides subfamily classification for each member (consistent in subsequent heatmaps). B Schematic representation of expression patterns for selected CiNAC genes of interest. The top panel illustrates the different tissues sampled; colors are used for visual distinction only and do not reflect actual tissue pigmentation Expression profiles of CiNAC genes across different tissues. A Heatmap depicting the FPKM values (log2-transformed) of 116 CiNAC genes in various tissues. Values represent the mean of three biological replicates. The phylogenetic tree on the left provides subfamily classification for each member (consistent in subsequent heatmaps). B Schematic representation of expression patterns for selected CiNAC genes of interest. The top panel illustrates the different tissues sampled; colors are used for visual distinction only and do not reflect actual tissue pigmentation Expression analysis of CiNAC genes under drought stress As a large and important family of transcription factors, NAC genes are widely involved in responses to abiotic stress [44]. Given the high sensitivity of C. illinoinensis to soil water conditions, the transcriptional response of 116 CiNAC genes in leaf samples under natural drought stress were analyzed (Fig. 8; Supplementary File S3), with six genes highlighted. CiNAC11, CiNAC78, CiNAC87, and CiNAC115 did not show a clear expression trend in response to drought. However, the expression levels of CiNAC24 and CiNAC72 gradually increased with the duration of drought stress. This suggests that CiNAC24 and CiNAC72 may exert significant influences in the drought stress response. Fig. 8Expression analysis of CiNAC genes under drought stress. T3D, T6D, T9D, T12D, and T15D denote leaf samples collected at 3, 6, 9, 12, and 15 days of drought treatment, respectively. Numbers following underscores indicate biological replicate IDs. Selected genes of interest are marked with gray circles Expression analysis of CiNAC genes under drought stress. T3D, T6D, T9D, T12D, and T15D denote leaf samples collected at 3, 6, 9, 12, and 15 days of drought treatment, respectively. Numbers following underscores indicate biological replicate IDs. Selected genes of interest are marked with gray circles Expression analysis of CiNAC genes during grafting in C. illinoinensis Recent studies have shown that NAC transcription factors exert crucial influences in processes essential for successful grafting, such as callus formation and vascular regeneration [45]. To investigate the expression patterns of the NAC gene family during the grafting of C. illinoinensis, an expression heatmap was generated for the cultivar ‘Pawnee’ at 0, 8, 15, and 30 days post-grafting (F0, F1, F2, and F3, respectively; Fig. 9B, Supplementary File S4). The 116 CiNAC members did not exhibit fixed expression patterns according to their phylogenetic subfamily or group affiliation during grafting. This likely reflects the broad functional diversification that occurred during the expansion and evolution of the NAC family [18]. Although family members share sequence homology, they may have undergone subfunctionalization or neofunctionalization, leading to specific or complementary roles in response to grafting stress, which accounts for their complex expression trends. Fig. 9Expression and differential analysis of CiNAC genes during grafting. A Morphological features of the graft union in the C. illinoinensis cultivar ‘Pawnee’ at 0, 8, 15, and 30 days post-grafting (DAG). The red-marked region indicates the sampling site. B Heatmap showing the expression levels of 116 CiNAC genes at 0 (F0), 8 (F1), 15 (F2), and 30 (F3) dpg. Numbers following underscores indicate biological replicate IDs. C Hierarchical clustering analysis of expression patterns for the 116 CiNAC genes across the grafting time course. D qRT-PCR validation of expression patterns for selected CiNAC genes during grafting. Black and gray bars represent expression levels in the rootstock and scion, respectively. Data are presented as mean ± SD. Different letters indicate significant differences (P < 0.05) Expression and differential analysis of CiNAC genes during grafting. A Morphological features of the graft union in the C. illinoinensis cultivar ‘Pawnee’ at 0, 8, 15, and 30 days post-grafting (DAG). The red-marked region indicates the sampling site. B Heatmap showing the expression levels of 116 CiNAC genes at 0 (F0), 8 (F1), 15 (F2), and 30 (F3) dpg. Numbers following underscores indicate biological replicate IDs. C Hierarchical clustering analysis of expression patterns for the 116 CiNAC genes across the grafting time course. D qRT-PCR validation of expression patterns for selected CiNAC genes during grafting. Black and gray bars represent expression levels in the rootstock and scion, respectively. Data are presented as mean ± SD. Different letters indicate significant differences (P < 0.05) To identify key CiNAC members potentially involved in grafting, the FPKM values of all 116 genes across the grafting time course were subjected to cluster analysis (Fig. 9C). Based on their expression patterns, the genes were classified into nine clusters. Silhouette analysis confirmed that Cluster 6 had a mean coefficient of 0.585 (Supplementary Files S10), exceeding the 0.5 threshold for acceptable cluster separation, supporting its internal consistency. Cluster 6 conformed to the expected pattern: relatively low and reproducible expression at F0 (pre-grafting), higher and reproducible expression during the critical healing phases F1 and F2, and subsequently lower but reproducible expression at F3 when the graft union was essentially formed. The screening criteria for defining a core cluster were therefore set as: (i) mean silhouette coefficient > 0.5, (ii) a low‑high‑low expression pattern across grafting stages, and (iii) containing at least three genes. Cluster 6 uniquely satisfied all three criteria. Hence, members of Subcluster_6 were selected for focused analysis. These members were CiNAC11 (subfamily ONAC003, group I), CiNAC24 (subfamily NAM, group XII), CiNAC72 (subfamily ONAC022, group VII), and CiNAC115 (subfamily NAC7, group IX), representing diverse subfamilies and groups. To identify CiNAC genes potentially involved in pecan grafting, members of Subcluster_6, along with CiNAC78 and CiNAC87 (both from subfamily ANAC011, group VI, noted earlier as possibly grafting-related), were selected for experimental validation. Graft union tissues of ‘Pawnee’ were sampled (Fig. 9A), separated into scion and rootstock parts, and subjected to qRT-PCR to analyze the expression of these six genes during grafting (Fig. 9D). The results showed that five genes, except CiNAC87, exhibited an expression pattern that increased initially and then decreased, whereas CiNAC87 showed a declining trend. The expression levels of CiNAC11, CiNAC78, and CiNAC115 were higher in the scion than in the rootstock. CiNAC72 and CiNAC87 showed similar expression levels in both scion and rootstock, while CiNAC24 expression was higher in the rootstock. Notably, the expression level of CiNAC115 peaked at 15 days post-grafting, reaching 785.7 in the scion and 90.15 in the rootstock. Therefore, we speculate that CiNAC115 exerts a significant influence in the graft healing process, primarily functioning in the scion. Discussion As a high-value nut tree species, the healthy development of the C. illinoinensis industry depends on the grafting propagation of superior cultivars. Recent studies have shown that the NAC transcription factor family plays a key role in vascular tissue regeneration and rootstock-scion interaction during graft healing [45, 46]. Therefore, characterizing the NAC gene family in C. illinoinensis provides an important entry point for elucidating its grafting biology. In this study, a total of 116 NAC family members were identified genome-wide, distributed across 16 chromosomes. Based on phylogenetic tree analysis, the CiNAC gene family can be classified into 16 subfamilies and 12 groups. Wang et al. [9] noted in their review that ANAC071/096 genes exert important influences in grafting. ANAC071/096 belongs to the ANAC011 subfamily and, as a member of the secondary wall-associated NAC (SWN), regulates cortical cell wall thickening that mediates mechanoresponsiveness of the root cap in gramineous plants [47–49]. Within this subfamily, seven C. illinoinensis NAC members are present (Fig. 1), namely CiNAC12, CiNAC27, CiNAC78, CiNAC85, CiNAC87, CiNAC109, and CiNAC116.Among these, CiNAC78 and CiNAC87 cluster closely with ANAC071/096 and may thus serve important functions during grafting in C. illinoinensis. Subsequently, CiNAC78 and CiNAC87, together with CiNAC11, CiNAC24, CiNAC72, and CiNAC115—whose expression trends during grafting exhibit a “low‑high‑low” pattern, aligning with expectations (Fig. 9C)—were selected as focal genes for further investigation. Interestingly, based on the genome-wide phylogenetic analysis (Fig. 1A), the six genes are distributed across five distinct subgroups (ONAC003, NAM, ONAC022, ANAC011, and NAC7). No single subgroup dominates among the grafting-responsive NACs. This distribution pattern is consistent with the highly pleiotropic roles of NAC transcription factors, which regulate diverse biological processes including stress responses, vascular development, hormone signaling, and cell wall remodeling [21–23]. Grafting is a complex process that simultaneously activates wound healing, vascular reconnection, regulation of water/nutrient transport, and defense responses [9]. Consequently, NACs from different subgroups may contribute to distinct aspects of grafting. Liu et al. showed that NAC7 and ANAC011 subgroups participate in abiotic stress responses and regulate secondary cell wall thickening [47–49], whereas the ONAC022 and ONAC003 subgroups are involved in regulating stress responses [50, 51]. Their differential expression under drought conditions and in a tissue-specific manner (Figs. 7 and 8) suggests distinct regulatory functions. Therefore, genome-wide subgroup classification and phylogenetic relationships do not simply assign grafting-responsive genes to a single clade; instead, they provide an evolutionary and structural framework to dissect the functional specialization of CiNAC genes in the multifactorial process of grafting. During the evolution of many gene families, diverse gene structures facilitate the creation of novel functions to adapt to the environment [12]. The CiNAC gene family has largely maintained the conservation of the core NAM domain and overall gene structure during evolution (Fig. 2). However, some members exhibit variations such as subdomain loss, unique motifs, and alterations in gene structure, which may indicate functional divergence within the family. Cis‑element analysis revealed that the promoter regions of the 116 CiNAC genes collectively contain 79 copies of the WUN‑motif. Among the six candidate genes, CiNAC24 carries two copies, CiNAC72 and CiNAC115 each carry one copy, whereas CiNAC11, CiNAC78 and CiNAC87 completely lack this motif. Although the WUN‑motif has been demonstrated in Arabidopsis to be involved in wound‑induced expression [52], CiNAC24 carries two copies (among the higher counts in this family, although a few members such as CiNAC19, CiNAC67, and CiNAC68 contain three copies; see Supplementary Table S9)—responds predominantly to drought stress rather than being highly expressed in stems. This indicates that the presence of a single motif cannot independently explain complex expression patterns. Further analysis revealed distinct combinatorial features of cis‑elements among the candidate genes. The promoters of CiNAC24 and CiNAC72 co‑occur with the WUN‑motif alongside stress‑ and jasmonate‑responsive elements such as ARE and CGTCA‑motif, suggesting a potential synergistic role in the early wound‑related signals (e.g., rapid activation of JA signaling) during grafting. CiNAC115 and CiNAC72 are both enriched in hormone‑responsive elements such as ABRE, implying their involvement in later stages of grafting, including callus differentiation and vascular reconnection. In contrast, genes lacking the WUN‑motif, such as CiNAC11 and CiNAC87, are enriched in meristem development‑related elements like the CAT‑box in their promoter regions, suggesting that they may be indirectly activated downstream of the grafting signaling pathway. To gain deeper insights into the evolutionary history of the C. illinoinensis NAC gene family, we conducted a systematic comparative analysis with both closely and distantly related species. This comparison revealed complex evolutionary dynamics (Supplementary File S1). Analysis with the close relative Juglans regia (walnut) showed that the vast majority of syntenic gene pairs have undergone purifying selection (Ka/Ks < 1). This aligns with findings in species such as Madhuca longifolia [53] and Rhododendron delavayi [54], suggesting that the core functions of this family are generally conserved under purifying selection across evolution. In contrast, comparisons with the distant species Arabidopsis thaliana and Vitis vinifera (grapevine) resulted in a significant number of missing Ka/Ks values due to high sequence divergence (pS ≥ 0.75). Notably, the number of such cases increased with phylogenetic distance, indicating that some members of this family may have undergone accelerated sequence evolution and potential functional diversification over long-term evolution. Furthermore, a few gene pairs with Ka/Ks > 1 were identified in the comparison with J. regia, namely CiNAC16–rna‑XM_018971994.2, CiNAC30–rna‑XM_035694353.1, CiNAC63–rna‑XM_018977264.2, and CiNAC67–rna‑XM_019003236.2 (Supplementary File S1). This observation hints at localized adaptive evolution even within closely related lineages. Consistent with findings from a pan-genome study of the NAC family in barley [55], which reported that core NAC genes are typically conserved while lineage‑specific genes are more prone to relaxed selection and functional divergence, we speculate that these four positively selected genes may have acquired lineage‑specific functions in pecan. In addition to these positively selected pairs, another NAC gene, CiNAC87 (CiPaw.12G061800), showed high sequence divergence from its putative ortholog (rna‑NM_001341249.1) (Supplementary File S1). This suggests that this gene may have also undergone functional divergence during evolution. Overall, the combination of purifying selection on most NAC genes and localized positive selection on a few specific members (CiNAC16/30/63/67) reflects a complex evolutionary landscape, where conserved functions are maintained while adaptive innovations emerge in a subset of the family. Analysis of gene expression patterns is a crucial approach for investigating gene function [12, 32]. Among the six focal genes in this study, five showed up-regulated expression in the stem tissues of C. illinoinensis, with the exception of CiNAC24. Under drought stress, the expression levels of CiNAC24 and CiNAC72 increased progressively with the duration of stress, suggesting their potential key roles in drought response. Given the functional diversity of the NAC transcription factor family [56], these expression patterns imply that CiNAC24 and CiNAC72 may be more specialized in abiotic stress response than in graft-related processes. We subsequently employed qRT-PCR to analyze the expression patterns of these genes during the grafting process in the ‘Pawnee’ cultivar (Fig. 9D). The results showed that, among the six genes of interest, CiNAC78 and CiNAC87 exhibited opposite expression patterns, which conflicted with their co-assignment to Subcluster_7 in the RNA-Seq clustering analysis. The classification of CiNAC78 into Subcluster_7 most likely represents a technical artifact caused by multi-mapping errors [57]; we therefore no longer interpret its expression pattern based on this clustering and instead relied on qRT–PCR as the primary evidence. In contrast, the clustering assignment of CiNAC87 was independently validated by qRT–PCR, and the concordance of its expression trends between the two platforms confirms the reliability of this RNA-Seq clustering result. CiNAC87 displayed a gradually decreasing expression pattern; we therefore hypothesize that, in Carya illinoinensis, CiNAC87 may have lost the wounding‑/grafting‑induced expression characteristic of its Arabidopsis ortholog and instead participates in other physiological processes such as abiotic stress responses. Indeed, rapid evolution and functional divergence of the NAC transcription factor family have been widely documented: in hexaploid wheat, numerous NAC homologs have undergone sub‑ or neo‑functionalization to cope with different stress conditions [58]. In other words, graft healing may not be the primary functional context for CiNAC87. In contrast, the expression of CiNAC115 peaked at 15 days after grafting, reaching levels of 785.7 in the scion and 90.15 in the rootstock. This expression was significantly higher than that of the other five genes. Notably, its expression in the scion was substantially higher than in the rootstock. This disparity may be attributed to the scion, as the source of new tissue, requiring more active gene expression during the early stages of graft healing to facilitate callus formation and vascular reconnection, with CiNAC115 potentially exerting a key regulatory influence in this process. Its significantly higher expression in the scion further suggests a specific function in scion-driven processes, potentially involving scion-side wound response, hormone signal transduction [59], or cellular reprogramming [60]. This distinct expression pattern provides important clues for elucidating the molecular mechanism of CiNAC115 in graft union formation. Future functional studies are needed to determine whether it directly regulates graft compatibility and tissue regeneration. We acknowledge that, in the absence of functional perturbation assays, the relationship between CiNAC115 expression and graft healing remains correlative. Additionally, the mixed sampling of phloem, cambium, and xylem surface tissues averages transcript signals across cell types and may dilute spatially localized expression patterns relevant to graft union formation. Higher-resolution approaches such as laser-capture microdissection or single-cell transcriptomics will be required in future studies to achieve cell-type-specific precision. Future studies employing laser-capture microdissection or refined tissue separation will be necessary to achieve cell-type-specific resolution. It should be noted that the grafting process inherently initiates with a mechanical wounding event, and the wound response is an integral part of the graft healing cascade rather than a separate confounding factor. Nevertheless, wound stress alone may partially contribute to the observed expression changes of CiNAC genes. To rigorously distinguish wound-responsive genes from graft healing-specific genes, future studies should incorporate mock-grafting controls (incision only without scion grafting) and blank controls. Such controls would help exclude non-specific wound stress interference and further validate the specific role of candidate genes such as CiNAC115 during the grafting process. Functional loss‑ and gain‑of‑function experiments are required to test its necessity and sufficiency for graft union formation. Because a stable transformation system is not yet available for pecan, we propose a two‑step approach: first, transient silencing (e.g., VIGS) in the pecan scion followed by histological assessment of callus bridging and vascular reconnection; second, heterologous overexpression in a tractable woody model (e.g., poplar) with a grafting or wounding assay. Future mechanistic studies should include DAP‑seq to identify genome‑wide targets of CiNAC115 and protein‑interaction assays (e.g., Y2H or BiFC) to map its regulatory network. Until these experiments are performed, the precise molecular role of CiNAC115 in pecan grafting remains to be experimentally tested. Conclusion This study presents a genome-wide characterization of the NAC transcription factor family in pecan (Carya illinoinensis). A total of 116 CiNAC genes were identified, unevenly distributed across 16 chromosomes, and phylogenetically classified into 16 subfamilies and 12 groups. Synteny and selection pressure analyses reveal that the core functions of the pecan NAC family have been largely maintained under purifying selection throughout evolution. Nevertheless, certain members—exemplified by CiNAC87—exhibit substantial sequence divergence and marked shifts in expression profiles relative to distant orthologs, suggesting lineage-specific functional diversification within Carya. By integrating spatiotemporal, drought-stress, and grafting time-course transcriptomes, we identify CiNAC115 as a candidate gene whose expression pattern is strongly associated with the graft healing process. Its exceptionally high and scion-biased expression suggests a potential association with scion-driven wound response, callus proliferation, and vascular reconnection. Future investigations employing loss- and gain-of-function approaches (e.g., transient silencing in pecan scions or heterologous overexpression in poplar) are required to establish causality. Subsequent elucidation of CiNAC115 downstream targets via DAP-seq and its protein interaction network will be essential to transition this correlative observation into a mechanistic framework for pecan grafting biology. Supplementary Information Supplementary Material 1: Supplementary Files S1. ID, Classification, and Physicochemical Properties of CiNAC Genes. Supplementary Files S2. The Ka, Ks, and Ka/Ks values between collinear NAC gene pairs. Supplementary Files S3. Expression levels (FPKM values) of CiNAC in different tissues. Supplementary Files S4. Expression levels of CiNAC under drought stress. Supplementary Files S5. Expression levels of CiNAC at different grafting stages. Supplementary Files S6. Primer sequences used for amplification of CiNAC genes. Supplementary Files S7. The conserved motifs of the genes in the CiNAC gene family aredetermined by MEME. Supplementary Files S8. Orthogroups containing Carya illinoinensis NAC genes and their orthologs in three comparator species. Supplementary Files S9. Statistical Results of Cis-acting Elements in 116 CiNAC Genes. Supplementary Files S10. Silhouette coefficients of the nine CiNAC expression clusters. Supplementary Material 1: Supplementary Files S1. ID, Classification, and Physicochemical Properties of CiNAC Genes. Supplementary Files S2. The Ka, Ks, and Ka/Ks values between collinear NAC gene pairs. Supplementary Files S3. Expression levels (FPKM values) of CiNAC in different tissues. Supplementary Files S4. Expression levels of CiNAC under drought stress. Supplementary Files S5. Expression levels of CiNAC at different grafting stages. Supplementary Files S6. Primer sequences used for amplification of CiNAC genes. Supplementary Files S7. The conserved motifs of the genes in the CiNAC gene family aredetermined by MEME. Supplementary Files S8. Orthogroups containing Carya illinoinensis NAC genes and their orthologs in three comparator species. Supplementary Files S9. Statistical Results of Cis-acting Elements in 116 CiNAC Genes. Supplementary Files S10. Silhouette coefficients of the nine CiNAC expression clusters.