Gene Expression Patterns Regulating Peanut Reproductive Phenology

ABSTRACT

Peanut reproduction is foundational for crop yield, breeding, and evolution. However, gene regulation underlying peanut flowering pattern and timing has received limited attention. Cultivated peanut ( Arachis hypogaea L.) shows two distinct flowering patterns between two subspecies, with ssp. hypogaea lacking flowers on the main stem and ssp. fastigiata having them. Understanding the gene regulatory networks that control peanut flowering will inform the genetic pathways impacting peanut reproduction, phenology, and yield. To this end, we measured whole‐transcriptome gene expression of leaves and shoot tips (meristem) at six plant growth stages from Tifrunner, a peanut cultivar belonging to ssp. hypogaea, and GT‐C20, a peanut germplasm belonging to ssp. fastigiata. Overall gene expression was distinct between the two genotypes in both tissue types. Flowering regulators including AhFT, AhSOC1, AhAGL42, and AhSPL3 were differentially expressed in both the main and lateral stem at the time of flowering initiation (T3—first bloom). This indicates that positive regulation of these flowering regulators drives the distinct pattern of flowering on the main stem in GT‐C20. Meanwhile, the differential expression of two RING‐finger E3 ubiquitin ligases was identified between the two genotypes, indicating that the PAF1‐complex (PAF1C) may contribute to the lack of flowering on the main stem of Tifrunner. Gene co‐expression network analysis indicates that gibberellic acid (GA) and jasmonic acid (JA) pathways are involved in reproductive regulation. These results provide insight into how flowering physiology is differentially controlled between the two peanut subspecies and provide a launching point for additional research in peanut floral development.

1. Introduction

Peanut is an important protein and oilseed crop, with annual production of 51.2 million metric tons worldwide in 2025 (PS&D 2025). Cultivated peanut ( Arachis hypogaea L.), also known as groundnut, originated from South America, with archaeological evidence found in the Zana Valley in Northern Peru dating back approximately 8500 years (Bertioli et al. 2016, 2019). The Arachis genus is distributed across Argentina, Brazil, Bolivia, and Paraguay, although commercial production of peanut is widespread around the world (Hammons et al. 2016).

Cultivated peanut is a self‐pollinated species with a tetraploid genome (2n = 4x = 40, AABB‐type, genome size of ~2.7 Gb), most likely derived from a recent single hybridization/chromosome duplication event between the wild diploid species A. duranensis and A. ipaensis (Bertioli et al. 2016, 2019). A. hypogaea is divided into two subspecies, A. hypogaea ssp. hypogaea and ssp. fastigiata. These two subspecies have distinct flowering patterns in which ssp. hypogaea never produces flowers on the main stem but ssp. fastigiata does (Pattee and Stalker 1995). The variety Tifrunner (Holbrook and Culbreath 2007) belongs to ssp. hypogaea and was the genotype used to develop the first tetraploid peanut genome assembly (Bertioli et al. 2019). It has a runner‐type growth habit and produces flowers on lateral stems until the end of the growing season. GT‐C20, a Chinese genotype, is a Spanish‐type peanut in ssp. fastigiata with a bunch‐type growth habit. It blooms profusely in mid‐season on both the main stem and lateral stems, and flowering ceases a few weeks later (Figure 1A). The flowering pattern between cultivars from the two subspecies does not differ from the selected cultivars for this study; therefore, the contrasting reproductive and morphological characteristics between these two cultivars, GT‐C20 (Figure 1B) and Tifrunner (Figure 1C), allow for a comparative genetic analysis of mechanisms regulating divergent peanut flowering patterns.

FIGURE 1.

Peanut Arachis hypogaea ssp. fastigiata and ssp. hypogaea flowering phenotype. (A) Number of blooms along days after planting. Tifrunner is represented by a gray line, and GT‐C20 by an orange line. The bloom time is indicated by the day after planting for both genotypes in the X‐axis and the number of blooms in the Y‐axis; The bars for each time point indicate standard error. Picture of first bloom of GT‐C20 (ssp. fastigiata) (B) and Tifrunner (ssp. hypogaea) (C).

Plant architecture is formed by the development of a shoot apical meristem (SAM), lateral meristems (LM), axillary meristems (AM), and floral meristems (FM), which give rise to the main stem, lateral stems, secondary branches, and floral organs, respectively (Wang et al. 2018). Plant meristems produce new cells that differentiate into organs as the plant grows. When a meristem is genetically predetermined to produce a specific structure with a defined size and shape, such as a flower, it is considered “determinate.” In contrast, if a meristem retains the potential for continuous vegetative growth, it is considered “indeterminate.” The determinacy of plant meristems plays an important role in the formation of plant architecture (Sablowski 2007). In some species, such as Arabidopsis thaliana , flowering occurs only on the main stem due to the transition of the SAM from vegetative to reproductive status, resulting in the formation of an inflorescence meristem (Amasino 2010). It is therefore considered to have a determinate growth habit. In other species, such as peanut, the SAM remains vegetative, and floral organs form on AM (Périlleux et al. 2019). Such species are considered to have indeterminate growth habits. Molecular mechanisms regulating vegetative and floral meristem determinacy are well studied in some species, such as rice and Arabidopsis (Sugiyama et al. 2019; Chang et al. 2020), but less so in legumes including peanut.

In peanut, both flowering time and the number of flowers have an impact on number of mature pods, which in turn affects peanut yield (Kaba et al. 2014). This effect also was reported in soybean (Board and Hall 1984; W. Liu et al. 2011). Despite the economic importance of peanut, considerable research on disease resistance and seed size traits was reported (Thiessen and Woodward 2012; Zhang et al. 2019; Chu et al. 2020; Dang et al. 2021; Wu et al. 2022; Gonzales et al. 2023; Moretzsohn et al. 2023). However, there are fewer studies focusing on phenotyping of peanut flowering traits (Emery et al. 1981; Lee et al. 1984; Bagnall and King 1991; Cheong et al. 2002; Kaba et al. 2014). As for molecular studies focusing on gene regulation of peanut flowering (Li et al. 2016; Jin et al. 2019; Wang et al. 2020; Zhong et al. 2022), phosphatidyl ethanolamine‐binding protein (PEBP) family genes were characterized in the wild peanut species, Arachis duranensis and Arachis ipaensis (Jin et al. 2019) where 33 members of this family, including FLOWERING LOCUS T–Like (FT‐Like) and its antagonistic TERMINAL FLOWER 1–Like (TFL1), were identified. Another study explored the role of PEBP genes in flowering regulation in cultivated peanut (Zhong et al. 2022). The authors provided evidence that the PEBP genes arahy.M2THPA, arahy.EM6VH3, arahy.4GAQ4U, arahy.IZ8FG5, arahy.G6F3P2, arahy.LUT2QN, arahy.DYRS20, and arahy.BBG51B are crucial for flowering induction of different genotypes in response to stimuli such as photoperiod and temperature. In Arabidopsis, PEBP genes play a pivotal role in both flowering induction (FT) and repression (TFL1), as well as coordinate meristem identity and control plant architecture (Zhu et al. 2021). As in Arabidopsis, the FT and TFL balance can be determinants of flowering on the peanut main stem (Kunta et al. 2022). If TFL is predominant in the main stem, we hypothesize that it will repress FT flowering induction. The SQUAMOSA‐PROMOTER BINDING PROTEIN–LIKE (SPL) gene family in A. hypogaea has also been characterized (Li et al. 2016). As shown in other species, SPL is a target of miRNA156 and miRNA157 to control flower initiation (Wang et al. 2009; A. Yamaguchi et al. 2009; Jung et al. 2016; Zheng et al. 2019; Jeyakumar et al. 2020). In addition, three QTLs related to flowering time have been identified, though not at the gene level (qIFDB04.1, qIFDB07.1, and qIFDB08.1) (Wang et al. 2020).

Given the limited knowledge about the molecular mechanisms of flowering regulation in peanut, the present study used a transcriptomics approach to compare samples collected from four tissue types at six different developmental stages of two contrasting peanut genotypes Tifrunner (ssp. hypogaea) and GT‐C20 (ssp. fastigiata). The results of this study enhance our knowledge of molecular regulation of peanut flowering, especially concerning the absence of flowers on the main stem of ssp. hypogaea and their presence in ssp. fastigiata.

2. Results

2.1. Tifrunner and GT‐C20 Transcriptomic Profile

RNAseq libraries (132 total) were constructed from two genotypes (Tifrunner and GT‐C20), four tissue types (main stem shoot tip, lateral stem shoot tip, main stem leaf, lateral stem leaf), and six time points (T1 to T6). Sequencing resulted in 3,159,792,008 total raw reads with an average length of 152 bp per read. Lateral stems were accessible at T2, and flowering commenced at T3 (Table 1). Upon read trimming and rRNA sequence removal, 1,709,511,558 reads (54%) were retained for downstream analysis (Table 1). Among these cleaned reads, 1,430,762,396 reads (84%) came from leaf samples, and 278,749,162 reads (16%) came from shoot tip samples.

TABLE 1.

Sample description, abbreviated names, and RNA‐seq reads process information. Vegetative and reproductive stages are as described by Boote (1982).

Sample description
Time points (T)	T1: Production of the first leaf on the main stem
	T2: Production of the first leaf on first lateral
	T3: First bloom
	T4: First peg formation (R2; 5 mm extension)
	T5: First pod formation (R3)
	T6: Maximum vegetative growth (R4)
Tissue type	MS: Main stem shoot tip
	LS: Lateral stem shoot tip
	ML: Main stem leaf
	LL: Lateral stem leaf
Genotypes	Tifrunner ( Arachis hypogaea ssp. hypogaea)
	GT‐C20 ( A. hypogaea ssp. fastigiata)

RNA‐seq reads
Total libraries	132
Reads length	152 bp
Raw reads	3,159,792,008
Paired end	1,579,896,004
Cleaned reads	1,709,511,558 (54% of raw reads)

Principal component analysis (PCA) was performed on DESeq2‐normalized counts using all samples from shoot tip and leaf tissue at T1 (production of the first leaf on main stem), T2 (production of the first leaf on first lateral), T3 (first bloom), T4 (first peg formation), T5 (first pod formation), and T6 (maximum vegetative growth) (Figure 2A– E). PCA results showed that the first principal component (PC1) was associated with tissue type (shoot tip vs. leaf) and accounted for 85% of the variation. On the other hand, PC2 was associated with plant genotype and accounted for 4% of the variation. These principal components clearly separated the two genotypes and the two tissue types within genotype (Figure 2A). After performing PCA for each separate tissue type, we observed that samples distinctly clustered by genotype and were loosely grouped based on time points (Figure 2B–E). A smaller variation can be observed for the shoot tip samples compared to leaf samples for both the main and lateral stems. Shoot tip samples at the main and lateral stems showed similar variation, 14% for PC1 with time point comparison, and for PC2, variation between genotypes that accounted for 11% in main stem and 10% in lateral stem (Figure 2B,C). As for leaf samples, higher variation across time points was observed for main stem samples with PC1 accounting for 31% of the variation compared to lateral stems with PC1 of 26%. The variation in leaf sample for genotype was higher than shoot tip sample; however, there was not much difference between main stem and lateral stem, showing PC2 of 12% and 13%, respectively (Figure 2D,E).

FIGURE 2.

Tifrunner and GT‐C20 transcriptome pattern for tissue type and time point. (A–E) Principal component analysis (PCA) of DESeq2 variance stabilized RNA‐seq data. The percentages along each axis indicate the proportion of variation captured by the respective principal components. (A) PCA of all samples, where each color represents a tissue type and each shape represents a genotype. (B) PCA of shoot tip samples at the main stem (C) and lateral stem. (D) PCA of leaf sample at the main stem (E) and lateral stem. Each color represents a time point (T1, production of the first leaf on main stem; T2, production of the first leaf on first lateral; T3, first bloom; T4, first peg formation; T5, first pod formation; T6, maximum vegetative growth), and each shape represents a genotype. (F) Differentially expressed (DE) genes between Tifrunner and GT‐C20 comparing each of the six time points for shoot tip and leaf samples. The number and percent of DE genes (from 51,430 total genes) are shown at the top or bottom of each bar.

Given the distinct clustering of genotypes and tissue groups, differential gene expression analysis comparing genotypes at each time point was performed in each tissue type separately (Table S1). Differential expression (DE) analysis between Tifrunner and GT‐C20 revealed that the number of differentially expressed genes (DEGs) fluctuated across the time points in both shoot tip and leaf tissues (Figure 2F). The general trend was that the number of DEGs increased with time, which suggests that genotypic differences became more pronounced as the plants matured. Across all time points, shoot tip samples had fewer DEGs than leaf samples. For leaf samples, the number of total DEGs was below 4000 from T1 to T3, increased to 6000–6500 at T4 and T5, and reached 12,678 at T6. In the shoot tip, the number of total DEGs ranged from approximately 700 to 2600 from T1 to T5 and peaked at 6898 at T6. The observed reduction in the number of DEGs in the shoot tip relative to the leaf may be partially attributed to the significantly fewer number of cleaned reads recovered from this tissue type compared to leaf.

2.2. Gene Co‐Expression

To identify the gene networks associated with peanut reproduction regulation, we conducted gene co‐expression analyses on leaf and shoot tip samples separately across the six time points. Our results revealed different network patterns between the two tissue types (Figure 3A,B). Specifically, leaf samples produced 16 modules (L01 to L16) from 13,381 genes (Figure 3C and Table S2), whereas shoot tip samples formed 10 modules (S01 to S10) from 7955 genes (Figure 3D and Table S3). Among these modules, we identified three (L14, S02, and S03) with hub genes related to flowering regulation, AhFRIGIDA, and two AhJAZs, respectively (peanut base genome IDs: arahy.WFFD1N, arahy.NLIW19, and arahy.76C240) (Choi et al. 2011; Zhai et al. 2015). Gene ontology (GO) enrichment analysis confirmed the involvement of these three modules in plant reproduction, revealing multiple enriched terms linked to the regulation of flowering and hormone biosynthesis processes (Figure 3E). Interestingly, GO enrichment analysis showed that enriched GO terms identified in the modules S03 and L14 were mostly shared between the two modules. On the other hand, module S02 showed a unique set of enriched GO terms, with only a minimal overlap of enriched GO terms with the other two modules (Figure 3E). The log2‐fold change of the hub genes comparing Tifrunner and GT‐C20 main stem shoot tip for AhJAZs (Figure S1) showed higher expression in Tifrunner than GT‐C20 at all time points except for time point 2 (Figure S1B). As for AhFRIGIDA log2‐fold change for main stem leaf samples (Figure S2), Tifrunner showed higher expression only at time point 3 (Figure S2C).

FIGURE 3.

Gene dendrogram and merged module from leaf and shoot tip tissue. (A,B) Gene tree dendrogram from leaf samples (A) and shoot tip samples (B). On the top are the 16 modules genes tree, and on the bottom, gene clustering on topological overlap matrix (TOM)‐based dissimilarity colors. (C,D) Hierarchical clustering and heatmap of the leaf (C) and shoot tip (D). The similarity was determined through the eigengene correlation. Modules are organized by clustering according to gene expression similarity. Modules with flowering‐associated hub genes are indicated with red rectangles. (E) Dotplot of functional enrichment results of modules (S02, S03, and L14) with flowering‐associated hub genes AhJAZ–genome id arahy.NLIW19, AhJAZ–genome id arahy.76C240, and AhFRIGIDA–genome id arahy.WFFD1N (genome IDs from peanut base). The X‐axis represents module IDs, and the y‐axis shows significantly enriched GO terms. Dots are colored based on their adjusted p‐values, and the size of each dot shows the gene ratio, meaning the number of genes associated with that GO term divided by the total number of genes used in the enrichment analysis. The top 15 GO terms for each module are shown, based on adjusted p‐values > 0.05. The gene IDs for each module (S02, S03, and L14) are in Table S4.

We also examined each module for subnetworks and individual nodes potentially linked to flowering regulation. From this, we identified the positive flowering regulator AGAMOUS‐like24 (AhAGL24) (He et al. 2020) within module S02 (Figure 4A), the gibberellin pathway‐associated gene SCARECROW/DELLA in module S04 (Figure 4D), and the abscisic acid (ABA) biosynthesis positive gene AhNCED5 in module S03 (Figure 4G). Functional enrichment analysis revealed that GO terms related to flowering, including “maintenance of flowering meristem identity” and “regulation of reproductive phase transition,” were enriched in the AhAGL24 subnetwork (Figure 4B,C). A similar pattern was detected in the SCARECROW/DELLA subnetwork, where enriched GO terms notably included the term “floral organ morphogenesis” (Figure 4E,F). On the other hand, the AhNCED5 subnetwork displayed a significant enrichment of GO terms associated with stress and defense responses, as well as hormone biosynthesis processes, such as ABA, jasmonic acid (JA), and ethylene (Figure 4H,I). Comparing the Log2‐FC between Tifrunner and GT‐C20 for these genes in shoot tip main stem (Figure S1A–F), it was noted that AhAGL24 was more highly expressed in time points T2 and T4 only (Figure S1B,D). The GA‐related gene SCARECROW/DELLA showed higher expression in Tifrunner at T2 and T6 only (Figure S1B,F); and the ABA precursor gene AhNCED5 showed negative Log2‐FC between Tifrunner and GT‐C20 only at T1 (Figure S1A).

FIGURE 4.

Analysis of nearest‐neighbor gene networks of flowering‐associated genes in shoot tip modules. (A) Nearest‐neighbor gene network of AhAGL24 in module S02. (B,C) Functional GO enrichment analysis of all nearest‐neighbor genes of AhAGL24. (D) Nearest‐neighbor gene network of SCARECROW/DELLA in module S04; (E, F) functional GO enrichment analysis of all nearest‐neighbor genes of SCARECROW/DELLA. (G) Nearest‐neighbor gene network of AhNCED5 in module S03. (H,I) Functional GO enrichment analysis of all nearest‐neighbor genes of AhNCED5. The line between dots means the edge distance of the gene connection network; each gene is identified by the gene ID from the peanut genome database.

In examining the leaf modules, we found AhJOINTLESS within module L01, AhbHLH63 in module L16, and AhSOC1 in module L12 (Figure S3A–F). Additionally, analysis of leaf modules revealed seven genes (AhMYC4, AhPYR, AhAOS3, AhMYB109, AhbHLH35, AhWRKY15, and AhWRKY40) within module L14 associated with flowering regulation (Yu, Huaxia, et al. 2012; Wang et al. 2017; Li et al. 2018; So et al. 2020; Villar et al. 2020; Wang et al. 2023) (Figure 5A). All these genes are involved in the JA biosynthesis pathway and response (Itoh et al. 2002; Chen et al. 2010; Fernández‐Calvo et al. 2011; Lackman et al. 2011; Vanderauwera et al. 2012; Jiang et al. 2019; So et al. 2020; Delgado et al. 2021).

FIGURE 5.

Analysis of nearest‐neighbor gene networks of flowering‐associated genes in leaf modules. (A) Nearest‐neighbor gene network of allene oxide synthase 3 (AhAOS3) and jasmonate‐zim‐domain protein 1 (AhTIFY10b), basic helix–loop–helix 35 (AhbHLH35, AhbHLH, and AhMYC4), transcription factor (TF) family (AhMYB109), abscisic acid receptor (AhPYR1), AhWRKY family TF (AhWRKY40 and AhWRKY15). The line between dots means the edge distance of the gene connection network; each gene is identified by the gene ID from the peanut genome database. The gene names of the hub genes of each network are labeled with gray rectangles. (B) Heatmap of gene expression profiles of flowering‐related genes from each network across all time points. The samples on the columns are grouped by location, time point, and genotype. Stars on each gene ID indicate the p‐value variation across samples based on DESeq2 result (***p < 0.001, **p < 0.01, *p < 0.05). No star indicates no significant expression variation across samples. (C) Comparison of functional enrichment results of each nearest‐neighbor gene network. The X‐axis represents the hub genes of each network, and the y‐axis shows significantly enriched GO terms. Dots are colored based on their adjusted p‐values, and the size of each dot represents the gene ratio. The gene ratio represents the number of genes associated with that GO term divided by the total number of genes used in the enrichment analysis. The top 10 GO terms based on adjusted p‐values for each network are shown.

The analysis of the expression profiles of the nearest‐neighbor genes of all these main node genes indicated distinct patterns of expression between Tifrunner and GT‐C20 at time points T5 and T6, with the exception of arahy.DPKT20, arahy.Y6BJG8, and arahy.JWU5I5 (Figure 5B). GO enrichment analysis of these subnetworks identified multiple GO terms consistently enriched across all of them, highlighting potentially shared biological functions or processes (Figure 5C). Among these common terms, “flower development” and “JA biosynthetic process” were particularly notable (Figure 5C). Other common terms identified as related to stress response include “response to fungus” and “hyperosmotic salinity response” (Figure 5C). The GO term “ethylene biosynthetic process” was also noteworthy, showing enrichment across all subnetworks except AhbHLH35 (Figure 5C). Although the plants were not under induced stress, they were grown in the field where a certain level of uncontrolled stress may have been encountered and could have contributed to stress response gene expression. The log2‐fold change results from the comparison between Tifrunner and GT‐C20 leaf main stem (Figure S2) showed varying expression levels at each time point for both genotypes. At time point T1, most genes were more highly expressed in GT‐C20, except for AhMUB109 and AhMYC4 (Figure S2A). At time point T2, AhPYR1 and AhAOS3 were more highly expressed in Tifrunner (Figure S2B). As for time point T3 (Figure S2C), which occurred at the first bloom, AhbHLH35 was more highly expressed in GT‐C20, whereas the other genes showed minimal variation in expression between the genotypes. At the remaining time points, T4, T5, and T6 (Figure S2D–F), most genes were more highly expressed in Tifrunner, except for AhPYR1, which was more highly expressed in GT‐C20.

2.3. DEGs Associated With Flowering by Pairwise Comparison of T1, T2, and T3

DE analysis showed that most DEGs were specific to each genotype in pairwise time point comparisons (Figure 6 and Tables S4 and S5). In the main stem shoot tip tissues, Tifrunner exhibited at least 10‐fold greater number of DEGs than GT‐C20 in the transitions from T1 to T2 or T1 to T3 (Figure 6A–D), whereas GT‐C20 showed more DEGs than Tifrunner when T3 and T2 were compared (Figure 6E,F). In the main stem leaf tissues, GT‐C20 showed a higher number of DEGs than Tifrunner in the comparisons of T3 versus T1 and T3 versus T2 (Figure 6C–F). In lateral stem tissues, GT‐C20 showed a much higher number of downregulated DEGs in the shoot tip compared to Tifrunner (Figure 6G,H).

FIGURE 6.

Venn diagrams showing significantly differentially expressed genes (DEGs) across time point comparisons for each genotype–tissue type pair. The number of upregulated and downregulated DEGs identified in the T2 versus T1 comparison in main stem tissues (A,B), in the T3 versus T1 comparison in main stem tissues (C,D), in the T3 versus T2 comparison in main stem tissues (E,F), and in the T3 versus T2 comparison in lateral stem tissues (G,H). GTC20.LL, GT‐C20 lateral leaf; GTC20.LS, GT‐C20 lateral shoot tip; GTC20.ML, GT‐C20 main stem leaf; GTC20.MS, GT‐C20 main stem shoot tip; Tifrunner.LL, Tifrunner lateral leaf; Tifrunner.LS, Tifrunner lateral shoot tip; Tifrunner.ML, Tifrunner main stem leaf; Tifrunner.MS, Tifrunner main stem shoot tip. DEGs with an adjusted p‐value ≤ 0.05 are considered significant. Each genotype–tissue type pair is represented by a different color.

Through DE analysis of time point comparisons within genotypes, we identified a set of DEGs that are likely involved in the regulation of flowering, some of which overlapped with the results of co‐expression analysis. This analysis demonstrated that the ABA receptor gene AhPYR1 (arahy.Y6BJG8 and arahy.JWU5I5) (Figure S4A), along with various genes implicated in flowering processes such as AhAGL24 (Figure S4A,B,F), AhSOC1, AhJOINTLESS (Figure S4C), AhSPL12 (Figure S4D), AhGIGANTEA (Figure S4E), and AhbHLH130 (Figure S4F) exhibited upregulation in the Tifrunner cultivar. The analysis also identified that the transcription factors AhMYB109‐like1 and 2 (Figure S5A,C) and AhbHLH63 (Figure S5C), alongside AhSOC1 (Figure S5C,F), AhFT (Figure S5C), and CAULIFLOWER (AhCAL) (Figure S5C), were among the genes exhibiting an upregulation in the GT‐C20 cultivar. This set of upregulated genes in GT‐C20 also included TIME FOR COFFEE‐like (AhTIC‐like) (Figure S5B), which acts as a flowering repressor involved in the JA signaling pathway, and SCARECROW/DELLA–AhGAIP (Figure S5D), which is involved in the gibberellin (GA) signaling pathway. Additionally, AhSWEET3 (Figure S5B), functioning as a sugar transporter, AhNCED5 (Figure S5D), associated with the ABA signaling pathway, and SCARECROW‐like 3 (AhSCL3‐like) (Figure S5F) were also identified as upregulated in GT‐C20.

In Tifrunner, across time point comparisons, we identified several DEGs that were downregulated and are potentially linked to the negative regulation of flowering. These genes include AhAOC4, AhAOS3 (Figure S4B), and AhMYC4 (arahy.0RH9QK), which are involved in the JA biosynthetic pathway (Figure S4C). Additionally, genes such as AhEIN3‐Like1 and 2 (Figure S4B), along with AhERF12, associated with the ethylene (ET) signaling pathway (Figure S4C), were also found to be downregulated. Furthermore, genes AhGA20ox1‐like 1, 2, and 3, involved in the negative regulation of gibberellin (GA) biosynthesis, were identified as downregulated in Tifrunner (Figure S4B). Additionally, in Tifrunner, DE analysis identified a subset of flowering activator genes that were downregulated. These include transcription factors AhMYB4, AhMYB30, and AhbHLH35 associated with stress response alongside AhPIF3 (Figure S4C). In the genotype GT‐C20, DE analysis revealed the downregulation of several genes acting as potential repressors of flowering, including AhJAZ/TIFY10b and AhAOS3, related to the JA signaling pathway (Figure S5A,C), AhERF106 from the ET signaling pathway (Figure S5A), and SCARECROW/DELLA genes from the GA signaling pathway (Figure S5D). Moreover, time point comparisons highlighted the downregulation of certain flowering activator genes in GT‐C20, such as AhWRKY15, AhWRKY40, AhbHLH35, and AhbHLH13, which are associated with stress response (Figure S5A,C), and AhGA20ox1, involved in the GA signaling pathway was differentially expressed (Figure S5D,E).

2.4. Analysis of DEGs Between the Main Stem and Lateral Stem at T3

Unlike GT‐C20, Tifrunner exclusively produces flowers on its lateral stems, never on the main stem. Because the comparison of gene expression patterns at pre‐ and post‐flowering time points did not clearly reveal a pathway behind the lack of main stem flowering in Tifrunner, we performed a DE analysis to compare gene expression levels in the main and lateral stem tissues of Tifrunner and GT‐C20 at a single time point (T3—first bloom). DE analysis was conducted separately on leaf and shoot tip samples. Initially, our analysis prioritized the identification of downregulated DEGs between the main and lateral stem tissues within each respective tissue type. Through this analysis, we identified several genes related to flowering, such as AhAGL42, AhFT, AhSOC1, AhFKF1, AhAP3‐like, AhSPL3 (Dorca‐Fornell et al. 2011; Huang et al. 2005; Jung et al. 2016; Kim et al. 2012; C. Liu et al. 2008; Moon et al. 2005, 2003; Yan et al. 2020); all these genes were downregulated in the main stem shoot tip of Tifrunner, compared to the lateral stem shoot tip at T3 (Figure 7A and Table S6). As for GT‐C20 shoot tip, AhTIC (arahy.5R3DYZ) was upregulated (Figure S6A and Table S8), and after GO term analysis (Table S9), no other flowering‐related genes were differentially expressed between main stem and lateral stem shoot tips. To empirically validate the results of this DE analysis, we carried out qRT‐PCR assays on four of the downregulated flowering‐associated genes in Tifrunner: AhAGL42, AhFT, AhSOC1, and AhSPL3. The qRT‐PCR findings confirmed that these genes exhibited decreased expression in the main stem shoot tip of Tifrunner in comparison to the lateral stem shoot tip (Figure 7B). Additionally, we calculated the Pearson correlation between the RNA‐seq and qPCR Log2‐FC values for these four genes, and an R = 0.78 indicates a good analysis consistency between methods (Figure 7C). GO term enrichment analysis was performed on Tifrunner downregulated DEGs (Figure 7D,E and Table S7). GO enrichment analysis revealed that the downregulated DEGs were significantly enriched in multiple flowering‐related GO terms (Figure 7D). These terms included positive regulation of flower development (GO:0009911), regulation of vegetative phase change (GO:0010321), inflorescence development (GO:0010229), and phosphatidylethanolamine binding (GO:0008429). Additionally, we identified that another potential flowering‐related GO term, response to gibberellin (GO:0009739), was also enriched by the downregulated DEGs (Figure 7D), notably including two AhMYB‐like transcription factors: arahy.RM3VXC and arahy.WVAI17. This term has previously been linked to the regulation of flowering in other species (N. Yamaguchi et al. 2014; Zhang et al. 2023).

FIGURE 7.

Downregulated genes in Tifrunner shoot tip between main stem and lateral stem at time point 3. (A) The volcano plot of downregulated genes in Tifrunner main stem shoot tip compared to lateral stem shoot tip at time point 3. Flowering‐related genes are labeled. Green points show significantly downregulated genes, and dashed lines represent log2‐fold change and p‐value cutoffs (0.58 and 10⁻³), respectively. Blue triangles, red crosses, and orange rectangles represent genes identified as not insignificant either because of log2‐fold change cutoff, p‐value cutoff, or both. (B) Log2‐FC comparison of arahy.ATH5WE (AhAGL42), arahy.M2THPA (AhFT), arahy.679QY2 (AhSOC1), and arahy.18KTSH (AhSPL3) relative expression in a sample from shoot tip, greenhouse. Comparison with the difference of expression is indicated by an asterisk (Z‐value < 0.05). A positive value means down‐expressed genes in the main stem shoot tip based on higher qRT‐PCR cycle threshold (CT) values. (C) Pearson correlation between RNA‐seq and qRT‐PCR Log2‐FC values for AhAGL42, AhFT AhSOC1, and AhSPL3. (D) GO enrichment dot plot for downregulated genes in main stem shoot tip compared to lateral stem shoot tip at time point 3. The p‐values are adjusted using Benjamini–Hochberg correction, and the adjusted p‐value cut‐off is 0.05. The gene ratio represents the number of genes associated with that GO term divided by the total number of genes used in the enrichment analysis. The top 15 GO terms based on adjusted p‐values were shown. (E) Gene‐concept network (cnet plot) plot of enriched GO terms. Tan‐colored nodes represent the GO terms, whereas gene nodes are colored based on log2fold change values. Size represents the number of differentially expressed genes connected to that enriched GO term. Fold change (the color of the gene node) means the log2‐fold change difference between Tifrunner main stem shoot tip compared to the lateral stem shoot tip at time point 3. The top 15 GO terms based on adjusted p‐values are shown.

DE analysis was also performed on leaf tissues. This analysis revealed a set of downregulated DEGs in the Tifrunner main stem leaf tissues compared to the lateral stem at T3 (Figure 8A and Table S6). We observed that the number of downregulated DEGs identified in leaf tissues (Figure 8A) was significantly fewer than those identified in the shoot tip tissues (Figure 7A) when comparing the main stem to the lateral stem. Despite this, DE analysis identified several downregulated DEGs associated with flowering, such as FT (arahy.EM6VH3), AhAGL42 (arahy.ATH5WE), and (arahy.H0ZZJY) AhAGL17‐like (Han et al. 2008). The downregulation of AhFT (arahy.EM6VH3) was further validated through a qRT‐PCR assay conducted on plants grown under both field and greenhouse conditions (Figure 8B). GO term enrichment analysis identified that these downregulated DEGs were enriched in the GO terms associated with flowering, including positive regulation of flower development (GO:0009911), phosphatidylethanolamine binding (GO:0008429), and plant ovule development (GO:0048481) (Figure 8C,D and Table S7). In contrast to the shoot tip samples of GT‐C20 in which no flowering‐related genes were DE, three genes—AhTPS (arahy.G88L7W), AhCXE (arahy.Q6VMSA), and AhFT (arahy.EM6VH3)—were downregulated in GT‐C20 leaf between the main stem and lateral stem (Figure S6B and Table S8). GO enrichment analysis showed terms for trehalose biosynthesis process (GO:0005992) for AhTPS, and positive regulation of gibberellic acid (GA) mediated signaling pathway (GO:0009939) for AhCXE (Table S9).

FIGURE 8.

Downregulated genes in Tifrunner leaf between main stem and lateral stem at time point 3. (A) The volcano plot of downregulated genes in Tifrunner main stem shoot tip compared to lateral stem shoot tip at time point 3. Flowering‐related genes arahy.ATH5WE (AhAGL42), arahy.H0ZZJY (AGL17‐like), and arahy.EM6VH3 (AhFT) are labeled. Green points show significantly downregulated genes, and dashed lines represent log2‐ fold change and p‐value cutoffs (0.58 and 10⁻³), respectively. Blue triangles, red crosses, and orange rectangles represent genes identified as not insignificant either because of log2‐fold change cutoff, p‐value cutoff, or both. (B) Log2‐FC comparison of arahy.M2THPA/EM6VH3 (AhFT) relative expression in a sample from shoot tip, greenhouse. Comparison with the difference of expression is indicated by an asterisk (Z‐value < 0.05). A positive value means down‐expressed genes in the main stem shoot tip based on higher qRT‐PCR cycle threshold (CT) values. (C) GO enrichment dot plot for downregulated genes in main stem shoot tip compared to lateral stem shoot tip at time point 3. The p‐values are adjusted using Benjamini–Hochberg correction, and the adjusted p‐value cutoff is 0.05. The gene ratio represents the number of genes associated with that GO term divided by the total number of genes used in the enrichment analysis. The top 13 GO terms based on adjusted p‐values were shown. (D) Gene‐concept network (cnet plot) plot of enriched GO terms. Tan‐colored nodes represent the GO terms, whereas gene nodes are colored based on log2‐fold change values. Size represents the number of differentially expressed genes connected to that enriched GO term. Fold change (the color of the gene node) means the log2‐fold change difference between Tifrunner main stem shoot tip compared to the lateral stem shoot tip at time point 3. The top 13 GO terms based on adjusted p‐values are shown.

In addition to the downregulated DEGs, we also identified DEGs that were significantly upregulated in the main stem tissues compared to the lateral stem in Tifrunner at T3 (Figure S7). This DE analysis was conducted in both the shoot tip (Figure S7A) and the leaf tissues (Figure S7B). GO enrichment analysis of upregulated DEGs in the shoot tip showed only six significantly enriched GO terms (Table S7). Among these terms, only JA biosynthetic process (GO:0009695) stands out as potentially involved in the regulation of flowering (Zhao et al. 2022). This term was enriched by the upregulated DEGs, including arahy.9G9LXM (AhAIM1‐like) and arahy.WI4YKF (Ah4CL‐like) (Figure S7A). The upregulated DEG, arahy.9G9LXM (AhAIM1‐like), was also found to be involved in the GO term flower development (Table S7). However, this term was not significantly enriched. Two other flowering‐related GO terms, negative regulation of flower development (GO:0009910) and stamen development (GO:0048443), were also observed among the GO terms that were not significantly enriched (Table S7). Interestingly, two significantly upregulated DEGs, potentially associated with flowering, were found to be involved in these terms: arahy.VUMD9N (AhPAF1C) and arahy.7585ZU (serine/threonine kinase) (Figure S7A). In addition, we identified two upregulated DEGs, arahy.59P1C9 and arahy.8S0LND, both encoding an AhRING E3 ubiquitin ligase protein (Figure S7A). These genes were significantly enriched in the GO term cellular cation homeostasis (GO:0030003) (Table S7). RING‐finger E3 ubiquitin ligases are known to play various roles in plant growth, stress resistance, and signal transduction (Shu and Yang 2017; Jiménez‐López et al. 2018; Wang, Shi, et al. 2022), as well as the regulation of flowering time (Pavicic et al. 2017). GO term enrichment analysis did not reveal any enriched GO terms among the upregulated DEGs in the leaf. Because the samples for both main stem and lateral stem at each time point were collected at the same developmental stages, the GO enrichment variation result between main stem and lateral stem suggests a more likely association of the variation in gene regulation by tissue type than by developmental stage.

3. Discussion

3.1. Gene Co‐Expression Reveals Potential Pathways for Peanut Flowering Regulation

Significant variation in expression profile was observed between tissue types based on the Tifrunner and GT‐C20 RNA‐seq data; therefore, gene co‐expression analysis was separated by tissue type. The hub genes associated with flowering in our analysis indicate an important direction to investigate the gene cascades underlying flowering regulation. The GO enrichment analysis revealed genes within these key modules that are functionally associated with flowering regulation or biosynthesis of hormones directly or indirectly involved in flowering activation or repression. For example, identifying AhFRIGIDA as one of the hub genes indicates a strong connection with the flowering regulation pathway. FRIGIDA is known to regulate FLC expression in Arabidopsis under cold conditions, which contributes to inhibition of flowering during the winter, followed by flowering in the spring. Although there is no evidence of FLC in peanut, research by Brock et al. (2009) on the molecular evolution of FRIGIDA suggested that species that lack FLC or have a weak FLC allele still undergo a complete cycle of growth, reproduction, and senescence but with altered reproductive timing (Gazzani et al. 2003; Michaels et al. 2003).

3.1.1. Hormone Pathway

Two of the pathways implicated in the reproductive control of peanuts involve JA, which is known for its role in stress response and flowering inhibition (Wang et al. 2017), and GA, which plays roles in both flowering induction and/or delay (Su et al. 2001; Eriksson et al. 2006; Tenreira et al. 2017). The hub genes AhJAZ–arahy.NLIW19 and AhJAZ–arahy.76C240 in S02 and S03 modules indicate a linkage of these two modules by the JA pathway; JAZ proteins can mediate or repress flowering induction (reviewed by Zhao et al. 2022). Previous studies reported that JAZ interacts with AP2, a transcription factor target of TOE1/2 and prevents the repression of FT expression (Zhai et al. 2015). Also, JA mediates the flowering inhibition through the complex JAZ‐MYC2/3/4 to repress the FT transcription and flowering induction in Arabidopsis (Fernández‐Calvo et al. 2011). In addition, some of the subnetwork main genes revealed GO terms associated with the JA biosynthesis process, such as AhNCED5, an ABA biosynthesis precursor in module S03, and Time‐for‐coffee (AhTIC), AhbHLH135, and AhMYB109, in module L14 (Figures 4 and 5). These genes are connected with the JA cascade, such as the modulation of MYB109 expression through MYC2 and JA (Dombrecht et al. 2007). These gene cascades related to the JA pathway may contribute to flowering repression in Tifrunner main stem and promote flowering on the GT‐C20 main stem; however, more detailed studies including protein–protein interaction would help to clarify how this network affects flowering in a positive and negative way in both subspecies. As for the GA pathway, SCARECROW/DELLA‐like found in module S04 (Figure 4D–F) is an important gene associated with the GA‐related flowering induction pathway. DELLA interacts with transcription factors such as PIF4, CO, and NF‐Ys to repress their function in the positive regulation of FT and SOC1; in other words, the transcription factors will not be available to bind the FT and SOC1 promoter region motifs to promote flowering. However, DELLA is degraded by the 26S proteasome through the presence of GA, GID1, and the SCF complex, which makes the flowering regulators PIF4, CO, and NF‐Ys functionally active for the flowering regulation pathway (Kazan and Lyons 2016).

3.1.2. Flowering Regulators

Although the GO terms within the AhAGL24 subnetwork for shoot tips, other than for AhAGL24 itself, did not show a direct function involved in flowering regulation, the connections may indicate a possible correlation between these genes and flowering control, because AGL24 is a key gene in flowering regulation (C. Liu et al. 2008). A similar pattern was observed for the SCARECROW/DELLA subnetwork neighbor genes, which are related more to general biological processes but likely play an important role in the plant reproductive process. However, the nearest‐neighbor genes for the AhNCED5 subnetwork contained hormone pathway‐associated genes with integrated roles in stress response and flowering. ABA, known for its response to drought stress, can either delay or promote flowering in Arabidopsis depending on day length and act as a mechanism of escape from drought (Kazan and Lyons 2016). Ethylene (ET), involved in the transition from vegetative to flowering states in Arabidopsis (Ogawara et al. 2003), may interfere with the promotion of flowering through GA‐mediated accumulation of DELLA proteins (Kazan and Lyons 2016). In addition, two terms were related to trehalose biosynthesis (Figure 4H,I), an important sugar signaling pathway that promotes the transition to flowering in Arabidopsis (Cho et al. 2018). Trehalose is a sugar converted from trehalose‐6‐phosphate (T6P) that is synthesized from UDP‐glucose. The trehalose biosynthetic process is coordinated by trehalose phosphate synthase 1 (TPS1) (Paul et al. 2008). In a previous study conducted with Arabidopsis, it was shown that the absence of TPS1 caused extreme flowering delays by affecting the expression of FT and TWIN SYSTER OF FT (TSF) (Wahl et al. 2013). This result demonstrates the importance of trehalose as a flowering induction signal.

Leaf network modules, which also showed different subnetworks, indicate that there are important correlations to flowering regulation through the genes AhJOINTLESS and AhSOC1 (Figure S3A,B,E,F). However, the occurrence of a separate leaf module that showed seven subnetworks, containing more genes with GO terms associated with JA pathways (Figure 5), indicates a gene cascade towards the JA signaling pathway in leaves. Because JA typically is involved in the inhibition of flowering (Zhao et al. 2022), its repression to a greater level in Tifrunner versus GT‐C20 at time points 5 and 6 suggests its involvement in the protracted flowering of Tifrunner (Figure 1).

3.1.3. Tifrunner and GT‐C20 Differ in Their Flowering Gene Cascade Before and After Flowering Induction

Gene cascades underlying the transition from vegetative to reproductive development can involve many genes associated with multiple pathways (Blümel et al. 2015). Differential gene expression between Tifrunner and GT‐C20 was observed across all six time points, with the greatest difference at time point 6. GT‐C20 is a short duration variety; thus, senescence may already have been initiated at this time point, whereas Tifrunner continued to grow. DEGs associated with flowering regulation likely contributed to their diverse reproductive phenology. However, additional study must be conducted to address whether the gene expression variation between genotypes at the late developmental stage is directly or indirectly related to flowering pattern.

Tifrunner (ssp. hypogaea) does not develop flowers on the main stem, yet several genes involved in flowering (AhAGL24, AhJOINTLESS, AhSOC1, AhGIGANTEA, and AhSPL12) were upregulated in main stem tissues as flowering commenced. Although genes involved in positive flowering regulation at T2 versus T1 and T3 versus T1 were expressed in Tifrunner main stem, repressors of flowering induction, such as the transcription factor AP2/ERF, were also found. The suppression of flowering by AP2 in Arabidopsis can be relieved by overexpression of miR172a, which targets certain AP2 genes (Aukerman and Sakai 2003). It is possible that low expression of miR172 could sustain repression of flowering by AP2 in peanut, although this was not tested. In addition, positional and temporal expression patterns not resolved in this set of experiments could affect flowering induction. For example, although AhAGL24 was upregulated in Tifrunner main stem leaf and shoot tip at T2 versus T1 and lateral shoot tip at T3 versus T2, AhSOC1 was not upregulated at the same time point or in the same tissue, only in main stem leaf T3 versus T1. AGL24 and SOC1 are known to interact for flowering induction signaling (C. Liu et al. 2008).

We also observed downregulation of AhGA20‐oxidase, which degrades gibberellins and thus prevents flowering. Because flowering does not occur despite these dual signals, it is possible that suppression of FT, which is required to integrate the flowering induction cascade, could be responsible for the absence of flower organ development on the main stem. This hypothesis is supported by the upregulation of FT in the main stem leaf of GT‐C20 at time point 3 (first bloom) versus time point 1, and the downregulation of FT in the Tifrunner main stem when compared to the lateral stem at time point 3 (discussed below). The expression of a strong flowering induction regulator in the main stem of GT‐C20 and its absence in Tifrunner may be primarily responsible for the presence of flowers on the main stem in GT‐C20/ssp. fastigiata and their absence in Tifrunner/ssp. hypogaea. A genetic study supporting this hypothesis characterized AhTFL mutations that caused reduced expression of this flowering antagonist in fastigiata types and thereby increased the FT/TFL ratio to promote flowering (Kunta et al. 2022). Although the RNA‐seq analysis did not show expression variation in any of the comparisons for AhTFL1 (DYRS20), it is clear that it is expressed only in Tifrunner (Figure S8). Moreover, the qRT‐PCR showed significant expression variation between Tifrunner and GT‐C20 leaf at T2 and T3 for main and lateral stems (Figure S5B). This result supports the hypothesis that AhTFL1 can be strongly associated with the flowering inhibition in Tifrunner main stems.

The variation in gene expression between Tifrunner and GT‐C20 along time points preceding flowering (T1 and T2) and at first bloom (T3) may explain the flowering delay in Tifrunner (Figure 1A). The short delay in flower induction contributes to later maturity for ssp. hypogaea. Whereas early maturity of peanut is favored in regions with shorter growing seasons due to seasonal drought or early cold, later maturity is desirable for increasing yield of this indeterminate crop in those regions with longer growing seasons.

3.2. More Floral Regulator Genes Are Suppressed in Tifrunner Main Stem Compared to Lateral Stem Than in GT‐C20

The transition from the vegetative to the reproductive phase during flowering induction is governed by a complex gene regulatory network involving various molecular signaling pathways. This process is characterized by the coordinated regulation of gene expression, involving both the suppression and activation of different genes throughout the plant. Central to this regulatory mechanism are genes that directly influence the timing of flowering, predominantly expressed in the leaves and shoot tips. In the leaves, these genes initiate a regulatory cascade in response to various endogenous and environmental stimuli. A key component of this response is the florigen, FLOWERING LOCUS T (FT), which translocates to the shoot tip from leaves to promote flowering and functions as the floral integrator of several flowering pathways (Amasino 2010; Srikanth and Schmid 2011; Blümel et al. 2015). In the shoot apex, FT functions as a catalyst for the activation of a subsequent floral integrator gene, SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1). This activation precipitates the transcriptional initiation of various floral organ identity genes, including LEAFY (LFY), APETALA1 (AP1), and FRUITFULL (FUL), thereby facilitating the transition to the reproductive phase (Srikanth and Schmid 2011; Blümel et al. 2015). This study demonstrated that the floral integrator genes, AhFT (arahy.EM6VH3 and arahy.M2THPA) and AhSOC1 (arahy.679QY2), were suppressed in the main stem shoot tip tissues compared to lateral stem shoot tip in Tifrunner at the time of first bloom (T3) (Figure 9) and not DE in GT‐C20 at the same comparison. Prior research on peanut flowering identified both AhFT genes arahy.M2THPA and arahy.EM6VH3 as key genes controlling flowering in cultivated peanuts (Zhong et al. 2022). Thus, it is likely that the suppression of these two floral integrator genes is the key factor in the distinctive reproductive behavior observed in Tifrunner, where it never produces flowers on the main stem. This hypothesis is further supported by the results in the leaf tissues, which revealed a downregulation of the florigen FT in the main stem leaf tissues (Figure 9). Based on these findings, we infer that the suppression of FT in the main stem leaf and shoot tip, along with the deactivation of AhSOC1, is responsible for the inhibition of reproductive phase transition in the Tifrunner main stem. Although AhFT was also downregulated in GT‐C20 between main stem and lateral stem leaf tissues (Figure 9), the fact that AhFT was also downregulated in Tifrunner shoot tip and not DE in GT‐C20 shoot tip indicates a much stronger suppression of AhFT in Tifrunner in the main stem.

FIGURE 9.

Putative model of flowering regulation in Tifrunner main stem shoot tip and leaf at T3. The depicted pathways for the regulation of flowering are derived from the Arabidopsis model of flowering time (Srikanth and Schmid 2011). Red gene ID and down arrows represent the downregulated genes, blue gene ID and up arrows represent the upregulated genes. GA, gibberellic acid. The dashed line indicates GA and JA as key hormones.

In addition, we provide evidence that the inhibition of critical floral integrator genes in Tifrunner may be controlled through a gibberellin (GA)‐dependent pathway. GA is one of the major pathways that controls the regulation of flowering time (Izawa 2021; Kim 2020; Wilson et al. 1992; N. Yamaguchi et al. 2014). The GA pathway interacts with other pathways involved in flowering, either through the action of DELLA proteins or by regulating gibberellin homeostasis (Galvao et al. 2012; Bao et al. 2020). This work identified that an AhSOC1‐like gene AhAGL42 was downregulated in the main stem shoot tip compared to the lateral stem shoot tip (Figure 9). Previous research has established that AGL42, together with other SOC1‐like genes, promotes flowering in a gibberellin (GA)‐dependent way and SOC1 directly regulates the expression levels of AGL42 (Dorca‐Fornell et al. 2011). We also identified that AhAGL42 (arahy.ATH5WE) together with an AhAGL17‐like gene arahy.H0ZZJY were downregulated in the main stem leaf compared to lateral stem leaf (Figure 9). In line with this, our work showed that two AhMYB‐like genes, arahy.RM3VXC and arahy.WVAI17, were downregulated in the main stem shoot tip of Tifrunner at T3 compared to lateral stem and were enriched in the GO term response to gibberellin. Previous research indicated that MYB‐like genes may mediate GA signaling and promote flowering by inducing the expression of LFY (Blázquez et al. 1998; Gocal et al. 2001; Bao et al. 2020; Zhu et al. 2020). Although one gene associated with the GA pathway, AhCXE, was found to be downregulated in the leaves of GT‐C20 between the main stem and lateral stem leaves, no other genes related to GA regulation were identified in the DE analysis. Furthermore, despite AhCXE being closely related to the gibberellin receptor GID1 and involved in GA regulation, its primary role is likely associated with the plant's response to stress rather than flowering regulation, as suggested by studies from Yoshida et al. (2018) and Wang, Xie, et al. (2022). The abundance of downregulated genes in Tifrunner compared to GT‐C20 linked to the GA regulation process clearly indicates a more significant negative gene cascade regulation effect in Tifrunner than in GT‐C20. AhTPS, another gene that was downregulated in GT‐C20 main stem versus lateral stem leaves, is not associated with the GA pathway but is involved in the trehalose biosynthesis process through the synthesis of trehalose phosphatase (TP) to T6P (Paul et al. 2008). Although AhTPS was suppressed in GT‐C20 leaf between the main stem and lateral stem, it appears that this suppression was not sufficient to affect flowering regulation. This hypothesis can be supported by the lower number of flowering‐related genes that were differentially expressed in GT‐C20 both in the leaves and shoot tips. However, further study is required to investigate the role of the trehalose pathway in flowering regulation between the two subspecies.

Furthermore, we identified that AhFKF1, which encodes an F‐box protein that promotes flowering in Arabidopsis, was among the suppressed genes in the main stem shoot tip of Tifrunner (Figure 9). FKF1 is regulated by long‐day photoperiod through the circadian clock signaling pathway, where it has an important role in the regulation of CONSTANS (CO) both transcriptionally and post‐transcriptionally (Nelson et al. 2000; Song et al. 2012). CO encodes a zinc‐finger protein that directly affects the FT transcript regulation (Samach et al. 2000; Tiwari et al. 2010). Prior work also indicated that FKF1 promotes flowering by negatively regulating GA signaling DELLA proteins in Arabidopsis under long‐day photoperiod (Yan et al. 2020). In addition, this study demonstrated that SQUAMOSA PROMOTER BINDING‐LIKE (AhSPL) transcription factors, arahy.18KTSH and arahy.6HMZ07, were decreased in the main stem shoot tip compared to the lateral stem (Figure 9). Previous studies showed that SPL is a target of miRNA156 and promotes flowering by activating floral integrator SOC1 and several key floral organ identity genes, including AP1, FUL, and LFY, via the age‐dependent pathway (Kim et al. 2012; Lee et al. 2012; Wang 2014; Jeyakumar et al. 2020). Distinct expression patterns were observed between subspecies, where SPL is downregulated in ssp. hypogaea, possibly due to high miR156 expression in the main stem compared to the lateral stem, although microRNA expression was not analyzed in this study. The miR156 suppression of flowering is highly conserved evolutionarily (Wang 2014). It has also been documented that GA regulates flowering transition through this miR156‐targeted SPL module by direct binding of GA signaling DELLA proteins to SPL (Wang 2014; Yu, Galvão, et al. 2012). These findings collectively indicate that the GA pathway is essential in controlling the expression of floral integrator and floral organ identity genes in Tifrunner, ultimately leading to the inhibition of flower initiation in the main stem tissues. However, the precise roles of the suppression of AhAGL genes in the main stem leaf tissues, as well as the direct relationship between GA and the absence of flowering in Tifrunner main stem tissues, remain unclear. Further investigation on the role of gibberellin in regulating the reproductive phase transition in Tifrunner main stem tissues could yield deeper insights into the mechanisms behind flowering in this crop.

Moreover, this study suggests that the PAF1‐complex (PAF1C) may act as a potential inhibitor of flowering in the main stem tissues of Tifrunner. We demonstrated that an RNA polymerase II‐associated factor 1 (AhPAF1) gene, arahy.VUMD9N, was significantly upregulated in the main stem shoot tip tissues of Tifrunner compared to the lateral stem tissues at T3 (Figure 9). Previous research demonstrated that PAF1C inhibits the floral transition by activating FLOWERING LOCUS C (FLC) transcription via histone modifications (He et al. 2004; Oh et al. 2004; Cao et al. 2015; Nasim et al. 2022) (Yu and Michaels 2010). FLC is a well‐known inhibitor of flowering in Arabidopsis (Michaels and Amasino 1999) and other species (Ruelens et al. 2013). FLC inhibits flowering by repressing the key floral integrator genes, such as FT and SOC1 (Deng et al. 2011; Mateos et al. 2015). However, peanut appears to lack FLC, as no clear orthologs of this gene have been identified (Mou et al. 2022). In addition, the present work identified that two RING‐finger AhE3 ubiquitin ligases, arahy.59P1C9 and arahy.8S0LND, were significantly upregulated in the main stem shoot tip compared to lateral stem in Tifrunner at T3. Prior research has established the involvement of E3 ubiquitin ligases in flowering regulation by either promoting the degradation of CONSTANS, a key activator of flowering, or through the activation of FLC (Chen and Ni 2006; Cao et al. 2008; Jung et al. 2012; Jang et al. 2015; Lazaro et al. 2015, 2012; Du et al. 2016; Pavicic et al. 2017). Interestingly, earlier research has also indicated that E3 ubiquitin ligases work with the PAF1 complex to activate FLC expression through chromatin remodeling, thereby inhibiting flowering in Arabidopsis (Cao et al. 2008, 2015; Gu et al. 2009; Xu et al. 2009; Linden and Callis 2020). Therefore, it is possible that a similar mechanism might control flowering in legumes, with PAF1C potentially interacting with RING‐finger E3 ubiquitin ligases. Although more investigation would be necessary, we could hypothesize that this interaction could initiate a gene regulatory cascade via histone modifications, ultimately inhibiting the function of floral integrator genes like FT and SOC1. However, the direct target of the proposed PAF1C‐E3 ubiquitin ligase complex and the exact mechanism behind it remains to be elucidated. In a study associated with dehydration stress response in chickpea (Barua et al. 2019), two components of the PAF1 complex, LEO1 and CRT9, were responsive to dehydration stress and caused alteration of flowering time. This work on a legume species helps to support the hypothesis that PAF1C expression in the Tifrunner main stem may negatively regulate flowering. The fact that FLC is absent in peanut but was found in soybean (Lyu et al. 2020) may indicate a closely related regulation process to the FLC in peanuts. However, further analysis in peanut needs to be carried out to identify epigenetic regulators associated with PAF1C that will help interpret its effect on flowering regulation.

The presence of a significantly upregulated AhTIC in the GT‐C20 main stem shoot tip compared to the lateral stem shoot tip at T3 suggests an association with the JA pathway for flowering regulation in GT‐C20. Although the role of TIC in flowering regulation is not fully understood, previous research by Shin et al. (2012) indicates that TIC can negatively impact the JA pathway by suppressing the expression of MYC2, a positive regulator of the JA pathway (Lorenzo et al. 2004). Furthermore, the upregulation of AhTIC in the GT‐C20 main stem shoot tip implies a positive effect on flowering regulation by negatively modulating the JA pathway's role in flowering repression.

4. Conclusion

The comparative gene expression analysis between Tifrunner and GT‐C20 at six plant developmental stages, encompassing the main stem and lateral stem for shoot tip and leaf samples, revealed a potential regulatory mechanism for flowering involving the GA and JA hormone pathways, as well as the flowering regulators AhTFL1, AhFT, and PAF1C, which may contribute to the absence of flowering on the main stem in A. hypogaea subspecies hypogaea. Furthermore, the presence of certain vernalization‐associated components but the absence of others, such as FLC, in peanut suggests a distinct flowering regulation process in this branch of legumes compared to other plant families. Our findings also provide evidence for an unprecedented new gene cascade in peanut flowering regulation, which will facilitate future studies on peanut and legume reproduction.

5. Experimental Procedures

5.1. Plant Material and Growth Condition

For sample collection, 20 plants each of Tifrunner and GT‐C20 were grown in the field spaced 1 m apart in Tifton Georgia (coordinates: 31°27′48″ N 83°30′36″ W). The field was managed with irrigation and cultivated following the recommendations from the University of Georgia Extension. Three intact plants per genotype were used for tissue collection at each time point, and each plant only contributed one sample over the course of the experiment so that prior samplings would not affect later ones. Because flowering signals can be translocated from leaf to shoot tip (Freytes et al. 2021) and main stem flowering is a primary criterion for separating ssp. hypogaea from ssp. fastigiata, four types of tissues were collected at each of six time points. These included main stem shoot tip, lateral stem shoot tip, partially expanded main stem leaf, and partially expanded leaf from the first lateral shoot. Tissue from six time points were collected as follows: T1, production of first leaf on main stem; T2, production of first leaf on first lateral (time just preceding flowering); T3, first bloom; T4, first peg formation (R2; 5 mm extension); T5, first pod formation (R3); T6, maximum vegetative growth (R4). Reproductive stages R2–R4 were as defined by Boote (1982). Three biological replications were collected for each tissue type × time point. The only tissue types that were not collected were lateral shoot tip and lateral leaf at the first time point because the lateral had not yet emerged. Leaf tissue was frozen in liquid nitrogen and stored at −80°C till RNA extraction. The shoot tip (meristem) tissue was dissected with a Number 11 surgical blade (www.feather.com.jp) under a stereo microscope and immediately frozen in liquid nitrogen. Consequently, a total of 132 samples were processed for library construction and sequencing. To validate the outcome of bioinformatic pipelines, samples from Tifrunner and GT‐C20 planted in the greenhouse were used for qRT‐PCR in addition to the RNA samples used for sequencing. Plant tissues from time points T2 and T3 were collected for RNA extraction.

5.2. RNA Extraction and cDNA Synthesis

Plant tissue was ground in liquid nitrogen, and approximately 100 mg of pulverized tissue was aliquoted in a 1.5 mL microtube and stored at −80°C prior to RNA extraction. Total RNA was extracted from three independent biological replicates with the Qiagen Plant RNeasy kit (www.qiagen.com). RNA was treated with Invitrogen DNase I (www.thermofisher.com). RNA integrity was determined by absorbance at 260/280 nm, the 260/230 ratio as measured by a BioTek Synergy HT (www.thermofisher.com) and formaldehyde agarose gel electrophoresis (Mansour and Pestov 2013). RNA concentration was determined using Quant‐iT RiboGreen RNA Reagent (www.thermofisher.com). RNA concentration of all samples was adjusted to 1 μg/μL followed by a cDNA synthesis reaction using the Invitrogen RT SuperScript III kit (www.thermofisher.com).

5.3. RNA‐Seq Library Construction and Sequencing

One microgram of RNA of each sample was used for paired‐end library construction using the KAPA Stranded RNA‐Seq Library Preparation Kit (kapabiosystems.com) resulting in a total of 132 libraries prepared with Illumina TruSeq adapter for 152 bp PE readers. Libraries were sequenced in seven lanes on an Illumina Hi‐seq 2500 sequencer through Hudson Alpha Institute for Biotechnology (Huntsville, AL). All sequenced libraries are deposited in the NCBI under the SRA Bio Project PRJNA1185251.

5.4. RNA‐Seq Data Processing, Alignment, and Transcript Quantification

Sequencing quality and other technical metrics were evaluated with FastQC (Andrews 2010), and the results were visualized with MultiQC (Ewels et al. 2016). To check the analysis integrity such as batch effect, parameters generated through the FastQC analysis were compared among different libraries, such as per sequence GC content, sequence length distribution, per base sequence quality, and adapter content, in addition to a PCA analysis. Raw RNA‐seq reads were trimmed using TrimGalore 0.4.5 with the parameters “‐‐illumina ‐q 20” (https://github.com/FelixKrueger/TrimGalore). Trimmed reads were mapped against 3221 A. hypogaea rRNA sequences (obtained from NCBI) using BBSplit (BBMap 38.76; https://sourceforge.net/projects/bbmap). Unmapped reads were selected and aligned to A. hypogaea L. reference genome version 1 (Bertioli et al. 2019) with the STAR aligner (Dobin et al. 2013) using the two‐pass mode. The first STAR pass was run with the option “‐‐alignIntronMax 10000” to identify the novel splice junctions. The splice junctions from the first pass were used in the second STAR pass using the parameters “‐‐outSAMunmapped Within ‐‐quantMode GeneCounts TranscriptomeSAM ‐‐alignIntronMax 10000.” STAR output Aligned.toTranscriptome.out.bam files (obtained via “‐‐quantMode TranscriptomeSAM” option) were then used with RSEM 1.3.2 (Li and Dewey 2011) with parameters “‐‐no‐qualities ‐‐strandedness reverse ‐‐alignments ‐‐paired‐end” to quantify gene expression levels. All steps described above were automated and performed with Snakemake workflow management system (Mölder et al. 2021).

5.5. DE Analysis

Gene expression levels obtained via RSEM were imported into R 4.2.3 using tximport package 1.26.1 (Soneson et al. 2015). Then, DESeq2 1.38.3 (Love et al. 2014) was used to perform DE analysis. First, PCA was performed using plotPCA function from DESeq2 on variance stabilized (with DESeq2 “vst” function) gene expression counts.

To perform DE analysis, we combined all covariates, including Tissue, Genotype, Time_point, and Tissue_location, into a single factor called “group” that includes all possible combinations (e.g., leafTifrunnerT3n and leafGTC20T2n1). This allowed us to compare any group of interest and easily extract multiple condition effects. We then used the “DESeq” function with the design formula “~group” to estimate size factors and dispersion and fit the negative binomial model. DEGs were identified using a Wald test with “contrast” function from the DESeq2 package for each comparison. Genes with a p‐adjusted value (padj) < 0.05 after the Benjamini–Hochberg correction were considered DEGs. Volcano plots were generated using the Bioconductor package EnhancedVolcano 1.16.0 (Blighe et al. 2019).

5.6. Functional Enrichment Analysis

GO functional enrichment analysis was performed by using the Bioconductor package clusterProfiler 4.6.2 (Yu, Wang, et al. 2012). The “enricher” function for the hypergeometric test was used to find enriched GO terms. The GO annotation file for A. hypogaea cv. Tifrunner was obtained from agriGO (Tian et al. 2017). False discovery rate (FDR) adjusted q‐value < 0.05 and Benjamini and Hochberg corrected p‐value < 0.01 were considered statistically significant. GO enrichment results were visualized using the “dotplot” function from clusterProfiler.

5.7. Gene Co‐Expression Analysis

The gene co‐expression was conducted in R (version 4.2.1‐2022‐06‐23) with the weighted gene co‐expression network analysis (WGCNA) package (Langfelder and Horvath 2008), and the plots through the R package igraph version 1.3.5 (Csardi and Nepusz 2006). The gene count result from RSEM used for the DE analysis was imported to R and converted in count per million (CPM). The samples were separated into leaf and shoot tip datasets to run the network analysis. For both sample groups, only genes with CPM > 10 were considered. The minimum necessary parameters to build the co‐expression network were as follows: SoftPower = 7 for leaf tissue and 6 for shoot tip tissue, based on the scale independence and mean connective graphic result, and the values used to build the topological overlap matrix (TOM). Also, for network modules, size was chosen minModulesize = 30 for both tissue types.

5.8. Validation Analysis by qRT‐PCR

The reverse transcriptase quantitative polymerase chain reaction (qRT‐PCR) was performed with samples collected at time points 2 and 3 collected from both field and greenhouse. qPCR was performed with Power Track SYBR Green Master mix (thermofisher.com) on a LightCycle‐480 II (Roche.com). Candidate DEGs for validation were chosen based on their relevance to genetic pathways and the absence of paralogs as determined through BLASTp search of the peanut genome with E‐value less than or equal to 1e‐5 and similarity > 98%. Primers were designed using PrimerQuest (www.idtdna.com) for the four candidate genes including AhFT1 (sense: 5′‐ATCAAGGGATTACGGAGACC‐3′; antisense 5′‐CTGTGCTCGTAAAAGGGTTC‐3′), AhSPL3 (sense: AGGGCTGAAGGGAAGA; antisense: TTCCTCCTCAGTCTCCTC), AhSOC1 (sense: CACCACAGGCAATTCTAAAG; antisense: CACATCTGAACTTGGACTACT), and AhAGL42 (sense: 5′‐TGATGAACTCGTTGGAATTG‐3′; antisense: 5′‐GTTAATTTGGCATTCTCTAGGA‐3′). Primer efficiency was tested by amplifying target genes diluted at 5, 25, 125, 625, and 3125 times.

For each tissue type collected for qRT‐PCR analysis, three biological and two technical replicates were performed. In addition, a no‐template control (NTC) and a cross‐plate control were included to detect amplification contamination and variation across plates, respectively. A fast‐cycling mode was used to perform the following PCR cycles: 2 min at 95°C, 40 cycles of 5 s at 95°C, and 30 s at 60°C. The dissociation reaction had three steps: 15 s at 95°C ramping at 1.99°C/s; 1 min at 60 C, ramping at 1.77°C/s; and 15 s at 95°C, ramping at 0.075°C/s. Housekeeping genes ubiquitin ribosomal protein S27a (UBI2) and actin depolymerizing factor 1 (ADF) (Chu et al. 2013) were used as reference genes. The relative expression value was transformed by log2 to obtain fold changes. Comparison used the adjusted Ct value of the PCR amplification and a linear mixed model analysis for target and reference genes (Steibel et al. 2009). Statistically significant difference was inferred through the z‐value of the Log2‐FC (fold change) contrast for each sample comparison.

5.9. Phenotype Analysis

To assess the difference in flowering phenology between Tifrunner and GT‐C20, 15 seeds from each genotype were germinated in the greenhouse and then transplanted to the field 6 days after germination. They were planted at the UGA Tifton campus as complete randomized blocks, and the number of blooms was counted every 2–4 days for 117 days after planting.

Author Contributions

C.H.C.: investigation, data curation, formal analysis, visualization, writing – original draft. A.K.C.: investigation, data curation, formal analysis, visualization, writing – original draft. Y.C.: conceptualization, methodology. J.W.: methodology, formal analysis, resources, writing – review and editing. R.H.: conceptualization, funding acquisition, writing – review and editing. P.O.‐A.: conceptualization, funding acquisition, project administration, resources, supervision, writing – review and editing.

Funding

This work was supported by the United States ‐ Israel Binational Agricultural Research and Development Fund (BARD US‐5020‐17) and the Peanut Research Foundation.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Log2‐fold change expression of the modules S02 and S03 hub genes and modules S02, S03, and S04 flowering‐related genes at the six time points between Tifrunner and GT‐C20 for main stem shoot tip.

Figure S2: Log2 fold change expression of the module S14 hub gene and module L14 flowering‐related genes at the six time points between Tifrunner and GT‐C20 for main stem leaf.

Figure S3: Subnetwork modules of leaf samples main modules L01, L16, and L12.

Figure S4: Differential expression analysis results of time point comparisons (T2 vs. T1, T3 vs. T2, and T3 vs. T2) in Tifrunner.

Figure S5: Differential expression analysis results of time point comparisons (T2 vs. T1, T3 vs. T2, and T3 vs. T2) in GT‐C20.

Figure S6: Differential gene expression for GT‐C20 at time point 3 (T3) comparing leaf and shoot tip from main stem and lateral stem.

Figure S7: Differential gene expression (DGE) results of upregulated genes in Tifrunner main stem compared to lateral stem at T3.

Figure S8: AhTFL1 expression comparison between Tifrunner and GT‐C20 shoot tip tissue main stem and lateral stem.

Table S1: Differentially expressed genes (DEGs) between Tifrunner and GT‐C20 at each time point (T1 to T6) for leaf and shoot tip tissues.

Table S2: Gene description for leaf samples co‐expression modules (L01 to L16).

Table S3: Gene description for shoot tip samples co‐expression modules (S01 to S10).

Table S4: Differentially expressed genes (DEGs) for Tifrunner and GT‐C20 samples at three time point comparisons (T2vsT1, T3vsT1, and T3vsT2) for main stem shoot tip and leaf at the two first comparisons and main stem and lateral stem shoot tip and leaf for the last comparison.

Table S5: GO term enrichment results for Tifrunner and GT‐C20 samples at three time point comparisons (T2vsT1, T3vsT1, and T3vsT2) for main stem shoot tip and leaf at the two first comparisons, and main stem and lateral stem shoot tip and leaf for the last comparison.

Table S6: Differential expression analysis results of Tifrunner main stem shoot tip and leaf compared to the lateral stem shoot tip and leaf at time point 3.

Table S7: GO term enrichment results from Tifrunner main stem shoot tip and leaf compared to the lateral stem shoot tip and leaf at time point 3.

Table S8: Differential expression analysis results of GT‐C20 main stem shoot tip and leaf compared to the lateral stem shoot tip and leaf at time point 3.

Table S9: GO term enrichment results from GT‐C20 main stem shoot tip and leaf compared to the lateral stem shoot tip and leaf at time point 3.

Cardon, C. H. , Corut A. K., Chu Y., Wallace J., Hovav R., and Ozias‐Akins P.. 2026. “Gene Expression Patterns Regulating Peanut Reproductive Phenology.” Plant Direct 10, no. 5: e70170. 10.1002/pld3.70170.

Data Availability Statement

The Snakemake pipeline used for RNA‐seq data processing, alignment, and transcript quantification, along with all code used to run differential expression and functional enrichment analysis, can be found on GitHub in the following repository: https://github.com/wallacelab/paper‐corut‐cardon‐peanut‐flowering‐2024.