Cell wall remodeling and inositol metabolism coexpression modules associated with nut size in Carya illinoinensis cvs. ‘Mahan’ and ‘Tiny Tim’

Abstract

Pecan is a tree nut crop native to the United States and Mexico, with a global market of over 2 billion USD. Nut size has been the most important target trait for crop improvement during the very limited breeding cycles. However, relatively little is known about the molecular basis of pecan nut ontogeny and the mechanisms underlying pecan nut sizing. Besides nut size, pecan fruit faces myriad physiological disorders throughout the growing season, making knowledge of essential genes at each growth stage a necessary first step in developing new cultivars and management practices to overcome these issues. To develop a deeper understanding of pecan fruit development and identify candidate genes underlying the large fruit phenotype, a time-course transcriptomic study of pecan fruit in two genotypes, ‘Mahan’ and ‘Tiny Tim’, was conducted. Weighted Gene-Coexpression Network Analysis (WGCNA) was employed to group transcripts into functional clusters, and hub transcripts were identified through module correlation analysis to select those that are potential drivers of these functional clusters. Modules related to cell wall biosynthesis, cell wall organization, and inositol metabolism in ‘Mahan’, and proteolysis and abscisic acid response in ‘Tiny Tim’ were found to be potentially associated with nut size.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-38292-2.

Introduction

Pecan (Carya illinoinensis) is a large tree native to the Central United States and the hilly regions of Mexico (T. E. Thompson & Grauke, 1991) known for its large drupe-like fruits, commonly referred to as nuts¹. The nuts are enclosed in husks derived from the floral involucre, commonly referred to as the “shuck” or “husk”, which are inedible and are generally discarded^1,2. Pecan fruit development can be broadly split into two phases: fruit enlargement and kernel filling. Pecan fruit enlargement is generally linear for the first half of fruit development, slowing considerably during the latter half of development³. Kernel filling during the latter half of fruit development is typically subdivided into the water, gel, and dough stages, based on the consistency of the developing endosperm³.

Pecan fruits are susceptible to various physiological disorders and diseases during these phases, making an understanding of pecan fruit development critical for identifying best practices for orchard management. For example, pecan scab is far more devastating to yield and quality when it initiates earlier in the season⁴, while shuck split is most likely to occur during the water stage^5,6, and vivipary is a late growing season issue, especially under conditions of irrigation and fertilization⁷. Susceptibility to these various physiological maladies varies by cultivar, yet the underlying molecular basis remains unknown.

While pecan fruits have been collected for sustenance for thousands of years by Native Americans, the history of their cultivation is less than 300 years old⁸. Pecan trees have experienced few generations of selection under cultivation due to a relatively long immature stage, with a minimum time of about 7 years to fruit bearing⁹. Despite this, nut size is one of the most striking traits that have been selected for in modern pecan cultivars, with the mature fruit of some cultivated pecans exceeding 10 g^10,11, as opposed to the 2–6 g of their uncultivated counterparts¹².

Genes associated with the regulation of fruit size in higher plants are incredibly diverse in function, including hormone metabolism, transcription factors, kinases, central carbon metabolism, and polysaccharide anabolism, among others¹³. In tomato, the best-studied model for fruit development, auxin is a driver of cell division, while gibberellins are drivers of cell expansion, both influencing fruit size¹⁴.

In the related Persian walnut (Juglans regia), cell wall-related metabolic processes appear to be loci of interest in explaining the genetic determination of nut size^15,16. Genome size also appears to be correlated with fruit size in Persian walnut, but the exact mechanism driving this relationship remains unexplored¹⁷. The wide range of possible mechanisms influencing fruit size makes the identification of genes responsible for increased fruit size difficult and makes a transcriptomic approach appealing. Previous transcriptomic studies of pecan fruit have primarily focused on the later stages of development^18,19, with a focus on lipid and protein content, though one study tracked fruit development biweekly through the whole growing season in the cultivar ’Annong3’²⁰. Genotype-specific differences at the transcriptomic level, however, remain entirely unexplored, and genes responsible for fruit size have not been identified.

To characterize the genetic basis of pecan fruit development, we conducted a developmental time course RNA Sequencing (RNASeq) experiment on the fruits of the pecan cultivars ‘Mahan’ and ‘Tiny Tim’, followed by Weighted Gene Co-expression Network Analysis (WGCNA). ‘Mahan’ was selected as a cultivar with particularly large fruit, and it is the progenitor of many commercial pecan cultivars, including ‘Pawnee’ and ‘Wichita’. ‘Tiny Tim’ was selected as a representative of small nut native pecans. Fruit was collected from two individuals of each genotype biweekly from the start of fruit development in May to fruit maturity in October.

We hypothesize that the previously defined stages of pecan development will correlate with transcript modules, and that hub and other genes with high module membership can be identified as potential drivers of development-stage-specific physiology, worthy of further investigation. Differences between genotypes may also be useful for developing hypotheses to understand the basis of the recent domestication of pecan and for identifying directions for future pecan improvement.

Methods

Sample collection

Pecan tissue was collected from two grafted trees (scions serving as biological replicates) per genotype, ‘Mahan’ (PI 518075) and ‘Tiny Tim’ (BCAR 1250), USDA-ARS National Plant Germplasm System - Brownwood pecan and hickory collection near College Station, TX²¹. Tree location identifiers were CSV 6–6 and 18 − 12 for ‘Mahan’ and CSV 9–10 and 20 − 8 for ‘Tiny Tim’. A minimum of three nuts were harvested at approximately two-week intervals from May 9, 2022, to October 31, 2022 (Table S1), immediately frozen in liquid nitrogen and subsequently stored at -80 °C. A total of 16 ‘Mahan’ and 20 ‘Tiny Tim’ sample collection dates occurred.

RNA isolation

Pecan tissue was cryomilled using a SPEX Freezer Mill 6870 (Metuchen, NJ) for 4 min at 15 cpm. A minimum of three fruits and a maximum of eight fruits from a minimum of three clusters were pooled into each sample (representing one tree at one time point) and milled together. Milled tissue was immediately stored at −80 °C until further use. A CTAB and lithium chloride-based RNA isolation method was used to prepare DNA-free samples for RNA sequencing. Approximately 100 mg of milled pooled tissue was added to 1 mL of CTAB Isolation Buffer and incubated at 60 °C for 30 min, with vortexing for 5 s every 5 min. An equal volume of 24:1 chloroform: isoamylalcohol was added to each sample, and vortexed for 10 s, followed by 5 min of rotational mixing. Samples were centrifuged at 8000 g for 6 min at 4 °C, and then the supernatant was moved to a fresh tube. Chloroform extraction was repeated once. Once the aqueous phase was transferred to a fresh tube, 0.5 volumes of 7 M DEPC-treated LiCl was added to each sample. Samples were gently mixed and then incubated at −20 °C for 3 days. The samples were then centrifuged at 15,000 g for 20 min at 4 °C.

All subsequent isolation steps were performed in a laminar flow hood to ensure sterility. Supernatant was decanted, taking care not to disturb the pellet. Samples were washed twice with 70% DEPC-treated ethanol, with intervening centrifugation at 15,000 g for 10 min at 4 °C. The samples were then resuspended in 80 µL of DEPC-treated ultrapure water. Samples were tested for RNA integrity and RNA quantity using the RNA IQ and RNA BR assays with a QuBit4 fluorometer. RNA purity was assessed using a Nanodrop 8000 UV spectrophotometer. Any samples with RIN values < 7260 nm/280nm absorbance ratios < 1.8, or RNA concentrations < 20ng/µL were re-isolated until a sample with sufficient quality and quantity was generated.

RNA sequencing

RNA samples were processed at the Texas A&M AgriLife Genomics and Bioinformatics Service for RNASeq library preparation and sequencing. Quality control was performed on 11 randomly selected samples before library preparation using a Fragment Analyzer system (Agilent, USA). Library preparation was carried out on all samples using the PerkinElmer Next Flex RNA kit. The prepared library was run on a single lane of a NovaSeq S4 XP 2 × 150 flow cell in a NovaSeq S4 6000 instrument.

Data analysis

The collected sequence data were assessed using FastQC to determine the optimal trimming parameters. CLC Genomics Workbench version 24 was used for read trimming and mapping of reads to the Pawnee v1.1 reference transcriptome²², accessed at Phytozome on December 15, 2023. Reads were trimmed to remove the first 15 nucleotides, remove the adapters, and remove ambiguous or low-quality sequences (quality limit = 0.05). TPM counts of transcript abundance were log2 normalized in R. Normalized counts underwent principal component analysis, and two outlier samples were excluded from further analysis.

The WGCNA package in R was used for network construction and visualization²³. A soft thresholding power was selected for signed network construction. A soft thresholding power of 16 was selected, as this value maximized the scale-free topology model fit at a value of 0.8520. WGCNA network construction was completed with the following parameters: maxBlockSize = 15,000, minModuleSize = 30, power = 16, networkType=signed. Module membership was calculated for all genes, and gene correlations were calculated based on month within the growing season (May to October) and genotype (‘Mahan’ and ‘Tiny Tim’). Predicted protein sequences from ‘Pawnee’ v1.1 reference were processed by DeepGO-SE with default parameters to predict gene ontology (GO) terms for functional analysis of modules²⁴.

Results and discussion

Sequencing results

A total of 534.9 GB of sequencing data was generated for 70 samples. Median sequencing depth per sample was 7.5 GB, and no sample had less than 6.3Gb of sequencing depth (Table S2). Principal components analysis of TPM abundances across all samples shows a general progression through the growing season shared by both genotypes, accounting for most of the observed variation (Figure S1).

Fruit development

WGCNA analysis

In this study, WGCNA was employed to identify functional gene clusters that could be organized based on their relationships to genotype and the time of year (Figure S2). A total of 105 modules were detected, with a median size of 80 and a mean size of 412 transcripts (Table S3). 50% of the present transcripts were found in the top 6 modules, and 90% of the present transcripts were found in the top 38 modules. The module grey, representing transcripts that could not be organized into modules, contained 1192 transcripts (2.7% of the 44,626 transcripts in the dataset). A minority of transcripts in the dataset exhibited a significant genotype correlation ( p < 0.05). A total of 5468 transcripts (12.5%) showed significant associations with ‘Mahan’, while 8386 transcripts (19.2%) showed significant associations with ‘Tiny Tim’. Correlations between module eigengenes (MEs) and traits are enumerated in Fig. 1 for modules with 100 or more members. Smaller modules tended to be composed exclusively of transcripts with very low expression levels. Most modules showed a correlation with at least one genotype or month of the growing season.

Fig. 1.

Module-trait correlations for all modules containing more than 100 transcripts. In the “Genotype” column, negative values correspond to ‘Mahan’, while positive values correspond to ‘Tiny Tim’. Number of transcripts per module indicated to the right of the module name. Significance markers: * - p < 0.05, ** - p < 0.01, *** - p < 0.001.

Early season: diverse function for cell development

For both ‘Mahan’ and ‘Tiny Tim’, the early season is a time of rapid fruit development, developing from an ovary to a fruit of mostly final dimensions within about 3 months. Module turquoise, the largest module (10668 transcripts), is rich in genes of diverse biosynthetic functions. Module blue, the second-largest module, is highly enriched for cell cycle-related genes (Fig. 2a and b). The gene functions represented in these modules reflect the rapid cellular division and catabolic processes occurring within pecan fruit early in the growing season. Module lightyellow is highly enriched in protein synthesis and ribosomal genes, further displaying the first months of pecan development, which is dominated by central metabolic and biosynthetic processes (Fig. 2c). As with many plant growth-related processes, auxin is a key regulatory hormone, with module turquoise’s hub gene being a highly expressed auxin response factor.

Fig. 2.

Gene ontology term enrichment analysis for biological processes in early season specific modules (a) turquoise, (b) blue, and (c) lightyellow.

Module greenyellow’s hub gene has a predicted function as an IAA methyltransferase (Table 1), reinforcing the centrality of auxin to early fruit development in pecan, independent of genotype. While this early stage of fruit development has remained unexplored in other pecan transcriptome papers, in walnut, the earliest stages of nut development are associated with diverse biosynthetic functions, cellular division, and auxin response, as well^25,26.

Table 1.

Hub transcripts of each module containing more than 100 members.

Transcript	Module Color	Positive Correlations	Module Correlation	Genotype	May	June	July	Aug	Sept	Oct	Panther Annotation
CiPaw.05G102300.2	turquoise	May, June, July	0.98	−0.04	0.29*	0.41***	0.27*	−0.14	−0.5***	−0.45***	AUXIN RESPONSE FACTOR 4-RELATED
CiPaw.13G105600.1	blue	May, June	0.99	−0.05	0.66***	0.34**	−0.02	−0.32**	−0.36**	−0.38**	OS05G0121800 PROTEIN
CiPaw.09G216900.1	brown	July	0.98	−0.01	−0.12	0.16	0.39**	0.21	−0.36**	−0.41***	D-ALANINE–D-ALANINE LIGASE FAMILY PROTEIN
CiPaw.16G011600.1	yellow	Tiny Tim, May, June	0.97	0.71***	0.22	0.29*	0.11	0	−0.33**	−0.39**	LEUCINE-RICH REPEAT-CONTAINING PROTEIN
CiPaw.15G016400.1	green	September, October	0.98	0.13	−0.21	−0.37**	−0.33**	−0.04	0.55***	0.54***	TRIHELIX TRANSCRIPTION FACTOR GT-2
CiPaw.08G177100.6	red	Mahan	0.99	−0.84***	0.08	0.07	0.17	−0.04	−0.14	−0.2	LEUCINE-RICH REPEAT-CONTAINING PROTEIN
CiPaw.01G154500.1	black	July	0.95	0.08	0.12	0.14	0.29*	−0.06	−0.23	−0.35**	SHIKIMATE DEHYDROGENASE
CiPaw.07G043800.2	pink	Tiny Tim, September, October	0.98	0.28*	0.15	−0.21	−0.24*	−0.21	0.23	0.38**	Polysaccharide Lyase Family 4
CiPaw.04G156100.1	magenta	Tiny Tim, September	0.96	0.63***	0.1	−0.08	−0.27*	−0.16	0.29*	0.2	LEUCINE-RICH REPEAT RECEPTOR-LIKE PROTEIN KINASE-RELATED
CiPaw.04G041200.1	purple	August, September	0.93	0.01	−0.55***	−0.37**	0.12	0.26*	0.36**	0.23	UV EXCISION REPAIR PROTEIN RAD23
CiPaw.12G133800.2	greenyellow	May	0.99	−0.11	0.84***	−0.05	−0.19	−0.25*	−0.24*	−0.17	INDOLE-3-ACETATE O-METHYLTRANSFERASE 1
CiPaw.02G073400.1	tan	Tiny Tim, July	0.96	0.33**	0.13	0.32**	0.26*	−0.05	−0.45***	−0.29*	FLOWERING TIME CONTROL PROTEIN FCA-RELATED
CiPaw.02G152500.2	salmon	Tiny Tim, July, August	0.93	0.18	−0.47***	−0.14	0.16	0.52***	0.03	−0.18	RING FINGER DOMAIN-CONTAINING
CiPaw.04G111200.1	cyan	October	0.99	0.12	0.25*	−0.06	−0.06	−0.07	−0.06	−0.03	40 S RIBOSOMAL PROTEIN
CiPaw.04G137800.2	midnightblue	Tiny Tim	0.92	0.43***	0.01	0.07	0.17	−0.05	−0.15	−0.06	SERINE/THREONINE-PROTEIN KINASE
CiPaw.12G022500.1	lightcyan	Tiny Tim, May	0.94	0.34**	0.56***	0.1	−0.24*	−0.47***	0.04	0.05	ENHANCED DISEASE SUSCEPTIBILITY 1
CiPaw.11G132500.1	grey60	September, October	0.98	−0.13	−0.24*	−0.24	−0.24*	0.14	0.37**	0.28*	CCT MOTIF FAMILY PROTEIN
CiPaw.10G158400.1	lightgreen	Tiny Tim	0.97	0.94***	0.16	0.1	−0.04	−0.07	−0.11	−0.05	DISEASE RESISTANCE FAMILY PROTEIN/LRR FAMILY PROTEIN-RELATED
CiPaw.08G008700.1	lightyellow	May	0.95	−0.19	0.38**	0.16	0	−0.35**	−0.08	−0.12	40 S RIBOSOMAL PROTEIN S11 FAMILY MEMBER
CiPaw.04G066300.1	royalblue	July, August	0.95	−0.14	−0.44***	−0.09	0.33**	0.4***	−0.01	−0.31**	RHOMBOID-RELATED
CiPaw.02G081900.1	darkred	Mahan, June, July	0.97	−0.28*	0.04	0.28*	0.34**	0.05	−0.39**	−0.44***	WD40 REPEAT PROTEINPRL1/PRL2-RELATED
CiPaw.15G177100.1	darkgreen	Mahan, September, October	0.97	−0.96***	−0.11	−0.14	0.06	−0.12	0.18	0.2	DISEASE RESISTANCE PROTEIN-RELATED
CiPaw.01G211800.1	darkturquoise	August, September	0.94	−0.11	−0.63***	−0.42***	0.05	0.52***	0.39***	0.09	OS06G0193300 PROTEIN
CiPaw.04G162800.1	darkgrey	September	0.93	0.2	−0.03	−0.07	0.04	−0.26*	0.23	0.16	PEROXISOMAL FATTY ACID BETA-OXIDATION MULTIFUNCTIONAL PROTEIN AIM1
CiPaw.07G040300.1	orange	Mahan, July	0.95	−0.13	0.06	0.26*	0.24*	0.19	−0.33**	−0.59***	LRR RECEPTOR-LIKE SERINE/THREONINE-PROTEIN KINASE HSL2
CiPaw.06G119500.1	darkorange	Tiny Tim	0.94	0.76***	−0.31*	−0.2	−0.02	0.24*	0.19	0.11	MECHANOSENSITIVE ION CHANNEL PROTEIN 6-RELATED
CiPaw.05G139700.1	white	August, September, October	0.97	0.07	−0.58***	−0.45***	−0.07	0.4***	0.47***	0.27*	2-oxoadipate dioxygenase/decarboxylase
CiPaw.14G004000.2	skyblue	July	0.94	0.03	−0.26*	−0.08	0.2	0.12	0.01	0.01	Section 31, ISOFORM B
CiPaw.15G053600.1	saddlebrown	July, August	0.97	−0.04	−0.42***	0.07	0.29*	0.48***	−0.19	−0.35**	OS02G0158600 PROTEIN
CiPaw.09G008200.1	steelblue	June, July, August	0.97	0.04	−0.17	0.28*	0.35**	0.28*	−0.32**	−0.58***	HISTONE DEACETYLASE HDT2-RELATED
CiPaw.13G089000.1	paleturquoise	August	0.98	−0.18	−0.21	−0.19	−0.18	0.37**	0.28*	−0.13	FRINGE-LIKE PROTEIN
CiPaw.05G021100.1	violet	August, September	0.97	−0.18	−0.4***	−0.38**	−0.25*	0.44***	0.46***	0.14	AMINO ACID PERMEASE 1-RELATED
CiPaw.09G029300.1	darkolivegreen		0.94	0.14	−0.03	0.08	0.14	0.03	−0.04	−0.24*	PROTEIN TRICHOME BIREFRINGENCE-RELATED
CiPaw.14G037300.1	darkmagenta	September, October	0.96	0.07	0.23	−0.17	−0.44***	−0.34**	0.38**	0.49***	CYCLIC DOF FACTOR 2-RELATED
CiPaw.01G000700.1	sienna3	May	0.99	−0.13	0.25*	−0.05	−0.06	−0.07	−0.05	−0.04	PROTEIN MIZU-KUSSEI 1
CiPaw.03G172000.1	yellowgreen	October	0.97	−0.13	−0.06	−0.05	−0.06	−0.07	−0.05	0.39***
CiPaw.11G154900.1	skyblue3	May, October	0.95	0.36**	0.55***	−0.07	−0.32**	−0.38**	0.07	0.22	KUNITZ TRYPSIN INHIBITOR 2
CiPaw.04G110000.1	plum1	October	0.98	0.2	0.14	−0.11	−0.13	−0.16	0.04	0.31**	1-AMINOCYCLOPROPANE-1-CARBOXYLATE OXIDASE 3-RELATED
CiPaw.02G086200.1	orangered4	May, June	0.95	0.16	0.54***	0.31*	−0.11	−0.52***	−0.19	−0.01	NITRATE, FROMATE, IRON DEHYDROGENASE
CiPaw.02G186600.1	mediumpurple3		0.99	−0.13	−0.06	−0.05	−0.06	0.23	−0.05	−0.04
CiPaw.15G135500.2	lightsteelblue1	Mahan, May	0.97	−0.32**	0.59***	0.04	−0.08	−0.23	−0.21	−0.15	CYCLIC NUCLEOTIDE-GATED ION CHANNEL 1
CiPaw.07G001300.1	lightcyan1	August	0.99	0.12	−0.05	−0.06	−0.06	0.23	−0.06	−0.02	IONOTROPIC GLUTAMATE RECEPTOR
CiPaw.01G001500.1	ivory	October	0.99	−0.13	−0.06	−0.05	−0.06	−0.07	−0.05	0.39***

Significance markers: * - p < 0.05, ** - p < 0.01, *** - p < 0.001.

Mid-season: hardening fruit, developing ovule, and resisting summer heat

Growth of the ovule and its filling with free-nucleate endosperm, which will eventually become the nut within the developing pecan fruit, occurs during the middle of the growing season³. Modules violet and royalblue, associated with this time of year, are enriched in floral organ development transcripts, and appear to be responsive to cytokinin and auxin, respectively (Fig. 3a and b). The largest mid-season module, brown, is enriched in transcripts related to polysaccharide biosynthesis and cell wall development (Fig. 3c).

Fig. 3.

Gene ontology term enrichment analysis for biological processes in mid-season season specific modules (a) violet, (b) royalblue, (c) brown, (d) saddlebrown, and (e) steelblue.

Whereas early-season module blue represents the cellular functions necessary for the cell division stage of fruit development, module brown represents many of the functions necessary for cell expansion. Despite this clear division of stages, as in the classic double sigmoid model of fruit development, the growth of the volume of the pecan nut does not appear to follow a double sigmoid model^3,27. Prior work in pecan fruit has shown an elevation of polysaccharide (especially xyloglucan) metabolism-related genes during this time of year, further underlining the importance of this process in mid-season development²⁰.

Module violet, the module correlated most strongly with August, is enriched in transmembrane transport-related functions (Fig. 3a). Module violet’s hub gene itself is an amino acid permease, a class of genes that facilitate the import of amino acids across membranes^28,29. This amino acid transport into the developing embryo increases seed yield and nutrient content in Arabidopsis and rice (Oryza sativa), and it is a key gene in early embryo development^30–32. Developing embryos are regions of high solute concentration and turgor as sink tissues³³. The increasing but incomplete rigidity of the pecan fruit, combined with continued filling mediated by solutes, can result in significant crop losses due to water splitting, particularly if rain is abundant following a dry period^5,34.

In various fruit crops, including apple and cherry, fruit cracking has been linked to genes with functions in cell wall-related processes, polysaccharide metabolism, and ion transport^35,36. Amino acid permeases show an ability to modulate cytokinin levels in rice, with elevated expression being associated with lower cytokinin levels³⁷. In apple, orange, persimmon, and litchi, cytokinin application can aid in the prevention of cracking disorders^38–41. Together with module violet’s cytokinin-related enriched GO terms, this hormone is an interesting candidate molecule for exploring this disorder (Fig. 3a). The interplay between the timing of cell wall hardening, solute transport, cytokinin, and fruit splitting may be worth exploring, with the hub genes of modules violet and brown being key candidates.

A prominent aspect of fruit development in July and August is the presence of modules containing highly expressed heat shock protein genes. The modules saddlebrown and steelblue are all enriched in heat stress response transcripts (Fig. 3d and e). The hub genes of these modules appear to be a homolog of the rice gene Os02g0158600 and a histone deacetylase, respectively (Table 1). The rice gene Os02g158600 has recently been identified as the site of a quantitative trait locus for heat stress tolerance⁴², while histone deacetylases have been shown to mediate heat stress tolerance in rice and Arabidopsis^43,44. Together, this suggests that these heat stress genes and the modules they are associated with may represent widely conserved plant responses to heat stress. Pecans are native primarily to the subtropical region of North America⁴⁵, and they must endure extreme heat regularly during their normal development cycle. In the transcriptome analysis of the Chinese cultivar ‘Annong3’, heat shock proteins were also among the most highly expressed genes during this time of year²⁰. While ’Tiny Tim’ comes from a northern region of the Pecan range, there were no significant genotype differences in these modules.

Late season: rapid nutrient accumulation

The late stage of pecan fruit development is characterized by the solidification of the endosperm and the rapid accumulation of lipids and proteins, resulting in a final nut that is ~ 10% protein and ~ 70% lipid by mass⁴⁶. Previous studies into the transcriptomics of pecan development revealed an abundance of genes related to lipid and storage protein biosynthesis during this time of year^18–20. Module green, the largest late-season module and module most associated with September, is enriched in lipid metabolism transcripts, as well as organic acid metabolism transcripts, which are upstream of lipid synthesis (Fig. 4a). Modules green and grey60 are rich in highly expressed storage proteins, together making up 8 of the 20 most highly expressed transcripts (Table 2). Other late-season associated modules, purple and white, appear to support these nutrient accumulation activities by protein localization and amino acid metabolism, respectively (Fig. 4b and c).

Fig. 4.

Gene ontology term enrichment analysis for biological processes in late season specific modules (a) green, (b) purple, (c) white, (d) plum1, and (e) yellowgreen.

Table 2.

Top 20 transcripts with the highest expression over the course of the growing season.

Name	Module color	Log2Average Expression	Panther Annotation
CiPaw.03G264700.1	grey60	9.145	12 S SEED STORAGE PROTEIN CRA1-RELATED
CiPaw.01G184200.1	grey60	9.023	12 S SEED STORAGE PROTEIN CRA1-RELATED
CiPaw.16G014600.1	turquoise	9.000
CiPaw.01G198100.1	green	8.940	Seed Storage and Functional Proteins
CiPaw.12G126700.1	grey60	8.791	2 S SEED STORAGE PROTEIN 1-RELATED
CiPaw.03G275900.1	green	8.791
CiPaw.03G264300.1	grey60	8.713	Seed Storage and Functional Proteins
CiPaw.08G107900.1	green	8.675	Seed Storage and Functional Proteins
CiPaw.01G186200.1	purple	8.646	Plant Proline-Rich Cell Wall Protein
CiPaw.04G072100.1	orange	8.571	BIFUNCTIONAL INHIBITOR/LIPID-TRANSFER PROTEIN/SEED STORAGE 2 S ALBUMIN-LIKE PROTEIN
CiPaw.09G102900.1	darkturquoise	8.471
CiPaw.05G077600.1	turquoise	8.308	PEPTIDYL-PROLYL CIS-TRANS ISOMERASE
CiPaw.13G044200.1	brown	8.269	DNAJ HOMOLOG SUBFAMILY C MEMBER
CiPaw.10G143900.1	royalblue	8.267	MEDIATOR OF RNA POLYMERASE II TRANSCRIPTION SUBUNIT 37E-RELATED
CiPaw.03G262000.1	brown	8.254	METALLOTHIONEIN-LIKE PROTEIN 2 A
CiPaw.07G058600.1	turquoise	8.206	NARINGENIN,2-OXOGLUTARATE 3-DIOXYGENASE
CiPaw.09G087300.1	darkgrey	8.157	UBIQUITIN
CiPaw.11G151900.1	royalblue	7.986	Plant lipid transfer/defense-related protein
CiPaw.03G264400.1	grey60	7.940	Seed Storage and Functional Proteins
CiPaw.13G096100.1	saddlebrown	7.929	MEDIATOR OF RNA POLYMERASE II TRANSCRIPTION SUBUNIT 37E-RELATED

Modules correlated with October plum1 and yellowgreen show enrichment in transcripts related to 1-aminocyclopropane-1-carboxylate metabolism (Fig. 4d and e), a precursor to the ripening and senescence hormone ethylene, which signals the conclusion of fruit development. While ethylene is most closely associated with the ripening of climacteric fruits, it is known that ethylene signals developmental processes in diverse non-climacteric fruits as well^47–49. In walnut, ethylene application preceding harvest stimulates the maturation of the fruit, making the hulling of nuts considerably easier while improving nut quality^50,51. It appears the role of ethylene in final nut maturation is shared in pecan.

Genotype: potential directions for Understanding domestication and fruit size

Of the 43 modules with over 100 transcripts, 9 showed significant correlation with ‘Tiny Tim’ and 5 showed significant correlation with ‘Mahan’. While very few modules showed nearly exclusive expression (genotype correlation > 0.8) in one genotype, 4 modules (red, lightgreen, darkgreen, and darkorange) did. Among these modules, 3 (red, lightgreen, and darkgreen) appear to be highly enriched in biotic response-related transcripts, reflecting individual differences in disease resistance and response. These modules are unlikely to be associated with the large fruit phenotype. Darkorange, however, appears to have a more diverse functional profile (Fig. 5a) and is associated very strongly with ‘Tiny Tim’ (correlation = 0.82). Notably, the most overrepresented biological function in this module is potassium transport, and the hub gene is an ion transporter. Potassium and other ions are known to impact fruit quality and nutrition metrics such as soluble solid content, amino acid content, and vitamin C in different crops, including kiwifruit and pear^52,53. These genes may also impact fruit phenotypes beyond size.

Fig. 5.

Gene ontology term enrichment analysis for biological processes in ‘Tiny Tim’ specific modules (a) darkorange, (b) lightcyan and (c) yellow. Gene ontology term enrichment analysis for cellular components are provided for module d) yellow.

As fruit sizing in pecan is primarily completed in the first half of fruit development³, modules associated with both ‘Mahan’ and earlier months of the fruit development cycle (May, June, July) might be of greater interest in understanding the large fruit phenotype. Three modules fit this categorization: darkred, orange, and lightsteelblue1 (Supplementary Fig. 3). Module lightsteelblue1, strongly associated with May, appears to be enriched primarily in terpenoid biosynthesis (Fig. 6a). Terpenoid biosynthesis is generally associated with plant defense against herbivory and disease, as well as tolerance to abiotic stress, and may play a role in genotype-dependent stress responses. However, evidence for its role in size determination is lacking. Module orange primarily showed enrichment of transcripts related to polysaccharide biosynthesis, cell wall biosynthesis, and cell wall organization (Fig. 6b). This module appears to be a genotype-specific counterpart to module brown, which is also most active in July and highly enriched in transcripts related to cell wall biosynthesis. The chemical and mechanical properties of the plant cell wall are factors that influence the ability of plant cells to grow, and remodeling is an ongoing process in growing plant tissues^54,55. In walnut, alleles in genes with functionality in cell wall modification and loosening during growth were associated with increased nut volume, and in pear, cell wall biosynthesis-related genes were identified as likely candidates for regulating fruit size^15,56.

Fig. 6.

Gene ontology term enrichment analysis for biological processes in ‘Mahan’ specific modules (a) lightsteelblue1, (b) orange, and (c) darkred.

The enriched functions of the module darkred were more diverse (Fig. 6c), though functions related to RNA modification, inositol metabolism, and phospholipid metabolism are recurrent. Inositol is a known regulator of plant cell growth and elongation⁵⁷. This module shows the highest expression at a time when cell division is slowing down in the developing fruit. Yet fruit expansion is ongoing, making this a promising potential process involved in the development of pecan fruit size. Hub genes and other genes with high module membership are likely to be drivers of module function. Among the transcripts with the highest module membership in modules orange and darkred, many exhibit regulatory function, which may contribute to the large fruit phenotype (Table 1, Supplementary Fig. 4, and Supplementary Tables 4 and 5). Notably, the hub gene of the darkred module is a homolog of pleotropic regulatory locus 1 (PRL1), which has demonstrated relationships with organ size, carbon partitioning, and hormone response in Arabidopsis⁵⁸. These genes and processes are worthy of further investigation, as candidates are potentially associated with the fruit size phenotype.

Examination of early-season ‘Tiny Tim’ associated modules may also give insight into genes which may contribute to small fruit size. Modules yellow, tan, salmon, and lightcyan fit these criteria. Module lightcyan appears to be primarily enriched in transcripts related to biotic stress response (Fig. 5b). Module yellow appears to have many stress tolerance-related transcripts but also has a notable elevation in photosynthesis-related transcripts as well (Fig. 5c and d). The role of photosynthesis in fruit development is quite variable between plant species, but it is generally associated with increased oxygen supply, refixing inorganic carbon in the highly metabolically active fruit tissues, and supplying carbon intermediates to diverse biosynthetic pathways⁵⁹. It is plausible that the small fruit phenotype of ‘Tiny Tim’ exists despite the elevated photosynthetic transcripts, rather than because of it.

Module salmon appears to be associated with proteolysis and catabolism mostly strongly; in Arabidopsis, protein degradation pathways are associated with a decrease in reproductive organ sizes⁶⁰. The hub gene for module tan is FCA (Table 1), which has demonstrated interactions with abscisic acid-related transcription factors, enhancing the effect of ABA on ABA-responsive genes^61,62. ABA response is associated with decreased fruit size, indicating a potential role for this module in producing the small-fruit phenotype of ’Tiny Tim’.

While some of these genotype-correlated genes may be involved in the development of the large fruit phenotype in pecan, selecting only two genotypes limits the broad applicability of these results. It is also possible that biological processes external to the fruit are key drivers of this phenotype, which may not be captured in a fruit-only experiment. In tomato, for example, the final fruit weight may be influenced by the expression of genes during flowering⁶³.

Conclusion

To the best of our knowledge, this study is the first comprehensive evaluation of pecan fruit transcriptomics across multiple genotypes for an entire season. The early season was most strongly associated with broad anabolic processes and cell cycle functions. The mid-season was associated with cell wall development and heat stress response, and the late season was most strongly associated with lipid metabolism, transport, protein localization, and storage protein biosynthesis. While transcript expression varied enormously through the development of the pecan fruit and nut, a much smaller fraction was significantly correlated with genotype, and many of those associated with genotype appear to have roles in biotic stress response. Among these genotype-specific transcripts are potential candidates for further research into the molecular regulation of pecan fruit size, with inositol metabolism, ABA response, proteolysis, and cell wall biosynthesis and organization being particularly noteworthy biological processes.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

AD and WC designed the study. WC and JL collected the samples. JL performed all the experiments and data analysis. JL and AD prepared the first draft of the manuscript. AD supervised the study. All authors reviewed, edited, and approved of the manuscript.

Funding

This research was supported by Texas A&M AgriLife Hatch Project #TEX0-9950-0 and startup funds from Texas A&M AgriLife Research and Texas A&M University to AD. JL acknowledges graduate research assistantship support from the Department of Horticultural Sciences at Texas A&M University. The work was also funded in part by the U.S. Department of Agriculture – Agriculture Research Service National Programs through CRIS project 3091-21000-046-000-D (Crop Germplasm Research Unit, TX) and a USDA ARS NACA #58-3091-3-013 award to WC and AD. The agency was not involved in the study design, collection, analysis, interpretation of data, and/or the writing of this article. However, this manuscript was approved by the agency before submission for publication. This article reports on the results of the research only. Mention of a trademark or proprietary product is solely to provide specific information and does not constitute a guarantee or warranty of the product by the U.S. Department of Agriculture and does not imply its approval to the exclusion of other products that may also be suitable.

Data availability

The RNA sequencing dataset generated and analyzed in this study is available at the National Center for Biotechnology Information Short Read Archive (NCBI SRA) repository under BioProject PRJNA1312749.

Declarations

Competing interests

The authors declare no competing interests.