Negative Regulation of GmNAC35 by miR164 Enhances Drought Tolerance in Soybean

Abstract

Soybean (Glycine max (L.) Merr.) is a critical crop in China, serving as a primary source of food, oil, and animal feed. Drought stress significantly impacts soybean growth and yield. MicroRNAs (miRNAs) play crucial roles in plant drought responses. The miR164 family is highly conserved across plant species and has been shown to participate in drought responses in a range of plants, yet the function of miR164 in soybean remains unclear. In this study, we identified GmNAC35 as a direct target of miR164 through published degradome sequencing data and 5′ RLM-RACE assays. Under drought stress, miR164 members (e.g., MIR164a, MIR164f, and MIR164k) rapidly down-regulated, reaching their lowest expression at 2 h and returning to basal levels within 6 h. Conversely, GmNAC35 showed an inverse pattern, indicating negative regulation by miR164. Overexpression of GmNAC35 enhanced drought tolerance in transgenic soybean plants, as evidenced by higher survival rates and reduced water loss. Transcriptomic analysis revealed that GmNAC35 modulates stress-responsive pathways, including ABA signaling and phenylpropanoid biosynthesis. Our findings indicate that miR164 negatively regulates GmNAC35, a positive regulator of drought tolerance. This enhances our understanding of the molecular mechanisms of drought tolerance in soybean and may inform strategies for breeding drought-resistant varieties.

1. Introduction

Soybean (Glycine max (L.) Merr.) is an annual herbaceous plant belonging to the class Dicotyledoneae. Taxonomically, it is placed within the family Fabaceae, subfamily Papilionaceae (Faboideae), genus Glycine Willd., and subgenus Soja [1]. As a vital source of protein and oil for both human consumption and livestock feed, soybean also serves as a key raw material for the food and feed industries [2]. It holds substantial agricultural and socioeconomic significance in China and worldwide [3].

China, the birthplace of soybean cultivation, was the world’s leading producer and exporter of soybean until the 1990s [4]. However, since 1996, with the rapid expansion of soybean production across the Americas, China’s soybean industry has fallen significantly behind that of the United States, Brazil, and other nations [5,6]. The long-term security of China’s soybean supply now faces serious challenges, highlighting the urgent need to revitalize domestic production and reduce reliance on imports [7]. Soybean, a typically thermophilic and heliophilic crop, requires substantial water for growth and development [8]. However, under the stresses imposed by global climate change, drought has emerged as a major constraint limiting yield improvements in China [9]. Breeding drought-resistant varieties and enhancing water-use efficiency are, therefore, critical strategies for ensuring the sustainable development of the soybean industry [10].

Transcription factors (TFs) are regulatory proteins that modulate gene expression by specifically recognizing and binding to the promoter regions of target genes [11]. Their activity is tightly controlled by post-translational modifications (PTMs), among which phosphorylation is a critical mechanism. Phosphorylation can alter TF conformation, stability, and DNA-binding affinity, thereby fine-tuning their regulatory functions [12]. Typically, TFs act as central nodes at the terminus of plant cell signaling pathways. Several TFs associated with drought stress responses have been identified, including members of the DREB, CBF, bZIP, MYC/MYB, WRKY, and NAC families [13]. Under drought conditions, signal transduction cascades and protein kinase-mediated phosphorylation activate constitutively expressed TFs within plant cells. This modification initiates downstream regulatory networks, ultimately reshaping the transcriptional landscape of stress-responsive pathways [14]. For instance, overexpression of the maize bZIP transcription factor gene ZmbZIP72 in Arabidopsis thaliana induces ABA-responsive genes (Rab18, Rd29B, and His1-3), thereby enhancing drought tolerance [15]. Similarly, the MYB transcription factor gene PIMYB94 from Populus trichocarpa exhibits a pronounced drought-induced expression. Transcriptional profiling revealed that PIMYB94 expression is strongly up-regulated under drought stress. Compared to wild-type plants, PIMYB94-overexpressing lines display two distinguishing traits: elevated endogenous ABA levels and significantly improved drought resilience [16]. NAC transcription factors also play a key role in plant responses to drought stress. Studies have shown that soybean NAC transcription factors, such as GmNAC2, GmNAC3, and GmNAC4, are crucial in mediating drought stress responses. Specifically, GmNAC3 and GmNAC4 are activated through the ABA signaling pathway, while GmNAC2 may respond to osmotic stress via a non-ABA pathway. These transcription factors enhance drought tolerance by regulating downstream stress genes, such as LEA. GmNAC3 and GmNAC4 are homologous to Arabidopsis ANAC019, ANAC055, and ANAC072 and are capable of binding to drought response elements, such as the promoter of the ERD1 gene [17,18].

Non-coding RNAs (ncRNAs), including microRNAs (miRNAs) and long non-coding RNAs (lncRNAs), are pivotal regulators of gene expression in response to drought stress [19]. Under drought conditions, miRNA expression typically exhibits dynamic fluctuations rather than consistent upregulation or downregulation [20]. Up-regulated miRNAs often target negative regulators, while down-regulated miRNAs primarily target positive regulators. Through the suppression of negative regulators or the stabilization of positive ones, plants can enhance their resilience to drought [21,22]. For instance, in Arabidopsis thaliana, drought stress suppresses miR167 expression, resulting in increased levels of its target gene, IAR3, which promotes auxin accumulation and lateral root development [23]. Overexpression of soybean miR169c in Arabidopsis increased drought sensitivity by suppressing NF-Y transcription factors AtNFYA1/5, leading to greater leaf water loss and lower survival rates under drought stress [24].

miR164 is a highly conserved miRNA family in plants. According to the annotation data from miRBase version 22.1 (http://www.mirbase.org, accessed on 15 March 2025), members of this gene family are widely distributed across species, including Arabidopsis thaliana, Oryza sativa (rice), Zea mays (maize), Glycine max (soybean), Sorghum bicolor, Medicago sativa (alfalfa), Populus euphratica, and Nicotiana tabacum (tobacco). In wheat (Triticum aestivum L.), the miR164-TaNAC14 regulatory module plays a crucial role in modulating root system architecture and abiotic stress responses. Functional characterization revealed that TaNAC14 acts as a positive regulator, promoting root growth and enhancing drought tolerance in wheat seedlings. Conversely, miR164 negatively regulates these processes by specifically targeting TaNAC14 transcripts for degradation, thereby suppressing root development and compromising drought/salt stress tolerance [25]. Recent study reveals how plants utilize the miR164:NAC module to combat drought stress. In soybean, drought-tolerant varieties exhibit decreased levels of miR164k and concomitantly increased levels of GmNAC1, which in turn activate the stress-response genes. Although the core regulatory mechanism of this module is conserved across plant species, its expression patterns vary significantly depending on species and genotype. The findings suggest targeting the miR164:NAC module could be a promising strategy for developing drought-resistant crops [26]. However, current research on the mechanisms by which miR164 and NAC transcription factors coordinately regulate the biological phenotypes of soybean, particularly under stress conditions, such as drought, remains limited.

In this study, we investigated the role of the miR164-GmNAC35 regulatory module in soybean drought tolerance. Through published degradome sequencing data and 5′ RLM-RACE assays, we identified GmNAC35 as a direct target of miR164, which exhibits dynamic expression patterns under drought stress. We further characterized the function of GmNAC35 by generating overexpression transgenic soybean lines and demonstrated its positive role in drought resistance through physiological assays and transcriptomic analysis. Our findings reveal that miR164 negatively regulates GmNAC35 to modulate drought-responsive pathways, including ABA signaling and phenylpropanoid biosynthesis. This study provides new insights into the molecular mechanisms of drought tolerance in soybean and highlights potential targets for breeding drought-resistant varieties.

2. Materials and Methods

2.1. Bioinformatics Analysis

(1)

Sequence Retrieval and Evolutionary Tree Construction of miR164

The sequence information of the miR164 gene family, including precursor and mature sequences, was retrieved and downloaded from the miRBase database (http://www.mirbase.org/, accessed on 15 March 2025). The precursor sequences were aligned using DNAMAN software (version 8.0.8.789). The evolutionary tree was reconstructed in MEGA11.0 using the neighbor-joining method, and 1000 bootstrap replicates were performed to assess the confidence of the evolutionary relationships among the precursor sequences of the miR164 gene family.

(2)

Analysis of Cis-Acting Elements in the Promoter Region

The 2000 bp upstream promoter sequences of the miR164 family members were obtained from the Ensembl database (https://plants.ensembl.org, accessed on 15 March 2025). Subsequently, the PlantCARE online tool (https://bioinformatics.psb.ugent.be/webtools/plantcare/html/, accessed on 15 March 2025) was used to analyze the cis-acting regulatory elements within the promoter regions.

(3)

Target Gene Prediction and Analysis

Potential target genes of soybean miR164 were predicted using the psRNATarget database (https://www.zhaolab.org/psRNATarget/, accessed on 15 March 2025). Genes with an expected value < 2.0 were selected as candidate target genes of miR164. The target genes for miR164 degradation were analyzed and identified based on the sequencing data of the degradome of roots and leaves under drought stress in soybean, which has been reported by previous authors [27].

2.2. Expression Profiles of miR164 Precursors in Soybean During Dehydration Stress

The larger particles and impurities in the nutrient soil were removed using a 6-mesh sieve, followed by sterilization in an oven at 121 °C for 3 h. The sterilized nutrient soil was then mixed with vermiculite at a 1:1 ratio. Uniform seeds of the low-latitude soybean variety Huachun 6 (HC6) were selected, and six seeds were sown per pot at a depth of 0.5 cm. The pots were covered with lids and placed in an artificial climate chamber at 28 °C under a 16 h light/8 h dark photoperiod for cultivation. After germination, the soybean plants were maintained under adequate moisture conditions until they reached the 10-day growth stage. At this stage, uniform and healthy seedlings were selected for dehydration treatment. The whole seedlings were carefully uprooted, their roots gently cleaned and placed on weighing paper for natural dehydration at room temperature (25 ± 1 °C). The fresh weight of the seedlings was measured at 0 h, 0.5 h, 2 h, 4 h, and 6 h of dehydration. At each time point, whole seedlings were flash-frozen in liquid nitrogen and stored at −80 °C for subsequent RNA extraction and expression analysis. Specific primers are detailed in Table S1.

2.3. Hairy Root Formation in Soybean Mediated by miR164 Overexpression

To isolate the precursor sequence of miR164k, we designed specific primers targeting approximately 200 bp upstream and downstream of the miR164k stem-loop structure. These primers were then used to amplify the genomic fragment containing the precursor sequence via PCR. The resulting sequence was then cloned into the modified pFGC5941 vector (pFGC5941m) under the control of a dual CaMV 35S promoter (2 × 35S) to create a miR164 overexpression construction (OE-miR164). Soybean transgenic hairy roots with an overexpressing miR164 vector were generated through Agrobacterium rhizogenes-mediated hairy root transformation. HC6 germinating seedlings were used for hairy root transformation according to the previously described methods [28,29]. Agrobacterium rhizogenes strain K599 was used for the soybean hairy root transformation experiment.

2.4. 5′RLM-RACE

RNA ligase-mediated rapid amplification of cDNA ends (RLM-RACE) was performed according to an earlier study [30]. Primers used in nested PCR are listed in Table S3. Total mRNAs were isolated from 20 mg of total soybean RNA by using TRIzol (VAZYME, Nanjing, China) and ligated to the RNA oligo adapter. The first-strand cDNA was synthesized by (TransGen Biotech, Beijing, China) according to the manufacturer’s instructions. Nested PCR was carried out using the nested adapter primer and other primers specific to the target gene. Amplification products were gel-purified and cloned in the pEASY-T5 Zero Cloning vector (TransGen Biotech) for sequencing.

2.5. Analysis of miRNA Expression

First, small RNAs were enriched, followed by reverse transcription of miRNAs using the VAZYME miRNA Tailing cDNA Synthesis Kit (miRNA 1st Strand cDNA Synthesis Kit, by A-tailing) (VAZYME, Nanjing, China). The template cDNA was a 5-fold dilution of the miRNA reverse transcription product. Subsequent quantitative real-time PCR (qRT-PCR) was performed according to the manufacturer’s instructions of the miRNA Universal SYBR qPCR Master Mix kit (VAZYME, MQ101). U6 snRNA was used as the internal reference gene for normalization, and the relative expression levels of miRNAs were calculated using the 2^−ΔΔCt method [31].

2.6. Northern Blot Analysis

The mature miR164 was detected by Northern blot analysis using the DIG Northern Starter Kit (Roche, Cat. No. 12039672910, Basel, Switzerland), where approximately 15 μg of total RNA was denatured and resolved by electrophoresis on both 15% formaldehyde–agarose and 15% polyacrylamide-7 M urea gels under denaturing conditions, followed by capillary transfer to a nylon membrane (GE Healthcare, Chicago, IL, USA). The membranes were pre-hybridized and hybridized at 42 °C in DIG Easy Hyb Granules buffer. The buffer contained DIG-labeled DNA oligonucleotide probes that were complementary to miR164 or U6 snRNA (used as an internal control), and washed twice at 25 °C with 2 × SSC, 0.1% SDS followed by a higher stringency wash of 0.1 × SSC, 0.1% SDS at 25 °C. For signal detection, the RNA-bound membrane surface was incubated with 1 mL CDP-Star chemiluminescent substrate and imaged using a Luminescent image analyzer (Amersham Imager 600, Ge Healthcare Bio Sciences Ab, Uppsala, Sweden).

2.7. Identification of Phenotypes Under Drought Stress

Total RNA extracted from the wild-type HC6 was reverse transcribed into first-strand cDNA. A 909 bp GmNAC35 (Glyma.15G254000) coding sequence was amplified from the HC6 using the primers GmNAC35-F and GmNAC35-R (manufactured by TsingkeBiotechnology Co., Ltd., Changsha, China) (Supplementary Table S1). The GmNAC35 gene fragment was cloned into the binary pFGC5941-eGFP vector, using Xho I single digestion, placed under the control of 2 × 35S, to form the GmNAC35 overexpression vector. This construction was used to generate the GmNAC35 overexpression vector. The vector was then transferred into Agrobacterium tumefaciens (EHA105). Using the previously described method [32], the GmNAC35 overexpression transgenic soybeans were obtained.

Soybean seedlings were cultivated in a 1:1 (v/v) mixture of soil and vermiculite. Seeds from both GmNAC35 overexpression lines and wild-type controls were sown at a depth of 0.5 cm (6 seeds per pot). On the tenth day, the soil was re-watered with 2 L of water and no further watering was applied. Drought stress was initiated when the soil relative water content reached 17%, and plant phenotypes were monitored throughout the stress period. Following 3 days of rehydration, phenotypic recovery was assessed, and survival rates were calculated. The drought experiment was performed with three biological replications, each including 36 single plants per genotype.

For measurement of water loss in detached leaves, the fully expanded rosette leaves were removed from 3-week-old well-watered plants and exposed to ambient conditions (50% relative humidity, 25 °C). Fresh weight loss was measured at 0 h, 0.5 h, 1 h, 1.5 h, 2 h, 3 h, 4 h, 5 h, and 6 h post-excision to determine the kinetics of water loss. Three independent biological replicates were included for all experiments.

2.8. RNA-Seq

Two-week-old whole plants of wild-type (HC6) and the GmNAC35-overexpressing transgenic soybean line (GmNAC35 OE-1) were collected for RNA-seq analysis, with three biological replicates per sample. The harvested plant materials were immediately frozen in liquid nitrogen. Library construction and sequencing were performed by Tsingke Biotechnology Co., Ltd. (Beijing, China).

The experimental procedures were as follows: total RNA was extracted from the samples using RNA isolator Total RNA Extraction Reagent (VAZYME, R401), followed by mRNA enrichment via Oligo(dT) magnetic beads. The RNA was then fragmented into short segments, and first-strand cDNA was synthesized using random hexamer primers. Double-stranded cDNA was generated using DNA polymerase, followed by A-tailing and adapter ligation using the Hieff NGS^® Ultima Dual-mode mRNA Library Prep Kit (Yeasen Biotechnology Co., Ltd., Shanghai, China; Cat. No. 12309ES08). The resulting cDNA libraries were amplified by PCR and subjected to sequencing on the BGI sequencing platform (DNBSEQ-T7).

Raw sequencing data were filtered to remove low-quality reads, and the clean reads were aligned to the soybean reference genome (Gmax_Wm82_a2_v1). The mapped data were analyzed using DESeq2 software (http://bioconductor.org/packages/release/bioc/html/DESeq.html, accessed on 10 March 2025) to identify differentially expressed genes (DEGs) between GmNAC35-overexpressing plants and HC6. Finally, functional annotation of the DEGs was performed through Gene Ontology (GO, http://www.geneontology.org, accessed on 15 March 2025) and Kyoto Encyclopedia of Genes and Genomes (KEGG, http://www.genome.jp/kegg/, accessed on 15 March 2025) enrichment analyses.

3. Results

3.1. Analysis of miR164 Family Members in Soybean

The soybean (Glycine max) miR164 gene family comprises 11 members, designated from miR164a to miR164k (Figure 1A). Their mature sequences fall into two variants: miR164b, miR164c, and miR164d share an identical 20-nt mature sequence, while the remaining eight members feature an additional adenosine at the 3′ terminus, resulting in a 21-nt isoform. Chromosomal mapping revealed that these 11 miR164 genes are distributed across 7 chromosomes (Chr2, Chr3, Chr9, Chr10, Chr18, Chr19, and Chr20). Evolutionary analysis performed using MEGA 11.0(Figure 1A) revealed that the 11 members segregate into two distinct clades: Clade I comprising eight members (MIR164a, MIR164b, MIR164d, MIR164e, MIR164f, MIR164g, MIR164j, and MiR164k), and Clade II, containing three members (MIR164c, MIR164h, and MIR164i). Notably, although miR164b (Clade I) and miR164c (Clade II) possess identical mature sequences, they cluster into separate clade, suggesting that divergence at the precursor sequence level may underlie differences in post-transcriptional regulation.

Promoter sequences typically harbor abundant functional biological information, which is primarily reflected in the diversity and abundance of cis-acting elements [33]. To systematically investigate these elements, we analyzed the promoter regions of the 11 miR164 family members, using the PlantCARE database. Furthermore, a systematic analysis of cis-acting elements within the promoter regions of the 11 miR164 family members was performed using the PlantCARE database. Despite their highly conserved mature sequences, the members exhibited marked divergence in cis-regulatory element composition. In addition to the universally conserved core promoter elements, such as the TATA-box and CAAT-box, several hormone- and stress-responsive elements were identified, including motifs responsive to abscisic acid (ABA), salicylic acid (SA), and methyl jasmonate (MeJA) (Figure 1B). These observations suggest that members of the miR164 family may participate in plant stress responses.

3.2. miR164 May Play an Important Role in Soybean Drought Stress Response

To investigate the involvement of the soybean miR164 family members in drought stress responses, we subjected 10-day-old whole HC6 seedlings to natural dehydration and assessed the expression levels of their precursor transcripts at multiple time points using quantitative RT-PCR. The results revealed that the precursor transcript levels of most miR164 family members initially declined and subsequently increased under drought conditions, suggesting a potential central role in the regulation of drought resistance in soybean (Figure 2). For example, MIR164a, MIR164f, MIR164g, and MIR164k exhibit rapid down-regulation, decreasing to their lowest expression at 2 h and then returning to basal levels within 6 h. This dynamic fluctuation suggests a potential role of miR164 in regulating drought resistance in soybean.

3.3. miR164 Cleaves and Degrades the mRNA of GmNAC35

To uncover the molecular mechanisms by which miR164 modulates drought stress responses in soybean, we predicted its potential target genes using the psRNATarget database. Several members of the NAC transcription factor family emerged as putative targets of miR164. Among these, GmNAC35 has previously been implicated in drought stress adaptation and other stress adaptation [34]. Analysis of publicly available degradome sequencing data from soybean roots and leaves under drought stress conditions identified GmNAC35 as a cleaved target of miR164 [27]. In soybean leaves, degradome analysis revealed that miR164 mediates cleavage at the 607th nucleotide of GmNAC35, corresponding to the 11th nucleotide of the mature miR164 sequence (5′-UGGAGAAGCAGGGCACGUGCA-3′) (Figure 3A). To validate these findings, we conducted RNA ligase-mediated 5′ rapid amplification of cDNA ends (5′ RLM-RACE) assays, which confirmed that miR164 directs cleavage of GmNAC35 mRNA at the predicted binding site, specifically between the 10th and 11th nucleotides from the 5′ end of the miR164 complementary region in GmNAC35 (Figure 3B).

To assess whether GmNAC35 is regulated by miR164, we constructed a miR164 overexpression vector driven by a tandem 35S promoter (2 × 35S) (Figure 3C). The construction was introduced into HC6 to induce hairy roots via Agrobacterium-mediated transformation. Transgenic hairy roots were generated through antibiotic selection and tissue culture-based regeneration, followed by molecular characterization. Quantitative RT-PCR analysis revealed a 3.3-fold increase in mature miR164 expression in the overexpression miR164 hairy roots compared with the empty vector (EV) control (Figure 3D). Northern blot analysis further confirmed that the accumulation of mature miR164 was increased approximately 2.2-fold in hairy roots overexpressing miR164 than the EV (Figure 3E). Among them, GmNAC35 expression was markedly reduced compared to the HC6 control (Figure 3F). These findings suggest that GmNAC35 is a direct target of miR164.

Collectively, these results provide definitive evidence that GmNAC35 is a bona fide target of miR164. To elucidate the role of GmNAC35 in the soybean drought stress response, we quantified its temporal expression dynamics in HC6 leaves following the dehydration treatment using qRT-PCR. GmNAC35 transcript levels were rapidly induced, reaching a peak at 0.5 h post-treatment, and subsequently declined (Figure 3G). Notably, when compared to the expression patterns of miR164 family members under the same conditions (Figure 2), GmNAC35 displayed a largely inverse expression profile. These findings reinforce the notion of a negative regulatory interaction between GmNAC35 and the miR164 family during dehydration stress.

3.4. The Overexpression of GmNAC35 Enhances Drought Tolerance in Transgenic Soybean Plants

To investigate the regulatory mechanism of GmNAC35, we generated transgenic soybean lines overexpressing this gene. A GmNAC35 overexpression construction, driven by the 2 × 35S promoter, was introduced into soybean (Glycine max) HC6 explants via Agrobacterium-mediated transformation. Transgenic plants were obtained through antibiotic selection and tissue culture regeneration, followed by generational segregation and molecular screening. Quantitative RT-PCR (qRT-PCR) analysis showed that GmNAC35 transcript levels were significantly elevated approximately 20-fold in two homozygous overexpression lines, GmNAC35 OE-1 and GmNAC35 OE-2, compared to HC6 (Figure 4A).

Western blotting further confirmed the accumulation of the GmNAC35–GFP fusion protein in individual plants from both overexpression lines (Figure 4B). Together, these results indicate that the generated transgenic lines provide a robust system for subsequent functional analyses.

To elucidate the role of GmNAC35 in the drought stress response of soybean, we performed soil-drought phenotyping on transgenic lines overexpressing GmNAC35 (Figure 4C). Compared with a wild-type (HC6) survival rate of 58%, 75% of GmNAC35 OE-1, and 77.7% of GmNAC35 OE-2 survived after ten days of drought stress followed by a 3-day recovery period. Compared to the HC6 control, GmNAC35-overexpressing plants exhibited a markedly higher survival rate following drought exposure (Figure 4D). Water-loss analysis showed that the detached leaves of GmNAC35 overexpressing plants lost water slower than those of HC6 (Figure 4E). Collectively, these results demonstrate that GmNAC35—acting as a downstream target of miR164—is a critical positive regulator of drought resilience in soybean.

3.5. Transcriptomic Analysis of the GmNAC35 Overexpressing Lines

In this study, transcriptome sequencing was conducted on two-week-old GmNAC35 overexpressing line (GmNAC35 OE-1) and the wild-type HC6. Principal component analysis (PCA) revealed distinct separation among the treatments, with biological replicates within the same treatment forming well-defined clusters (Figure S1A). Moreover, the Group_RPKM_correlation analysis confirmed high reproducibility among biological replicates and revealed significant differences between comparison groups (Figure S2). Compared with the HC6, we identified 3943 up-regulated and 3395 down-regulated genes in GmNAC35 OE-1 (Figure 5A, Table S2). Moreover, a heat map, showing gene expression level with a statistically significant difference in GmNAC35 OE-1 compared to HC6, was generated (Figure S1B).

To investigate the biological functions of GmNAC35-regulated differentially expressed genes (DEGs), we conducted Gene Ontology (GO) enrichment analysis (Figure 5B). The analysis revealed that these DEGs exhibited multidimensional functional roles in drought stress responses. At the biological process level, DEGs were significantly enriched in categories such as “response to stress” and “organic substance biosynthetic process”, suggesting that drought stress triggers broad-spectrum defense mechanisms in plants, accompanied by the enhanced biosynthesis of osmoregulatory compounds (e.g., proline and soluble sugars) to maintain cellular osmotic homeostasis. Cellular component analysis indicated that the products of these DEGs were primarily localized to “intracellular organelles” and “non-membrane-bounded organelles,” implying that drought stress modulates the metabolic activity of organelles such as chloroplasts and mitochondria to coordinate energy production and reactive oxygen species (ROS) detoxification. At the molecular function level, DEGs were significantly enriched in terms such as “ion binding” and “nucleotide binding,” indicating that drought signaling may drive transcriptional reprogramming through calcium ion (Ca²⁺) signaling pathways and ATP-dependent enzymatic processes. Together, these findings demonstrate that GmNAC35 promotes drought tolerance in soybean by orchestrating biosynthetic pathways, organelle function, and signal transduction networks, thereby integrating multiple protective mechanisms, including osmoregulation, redox homeostasis, and energy metabolism (Figure 5B).

To elucidate the key metabolic pathways enriched among the downstream differentially expressed genes (DEGs), KEGG functional annotation analysis was conducted, revealing a coordinated regulatory network underlying the drought stress response (Figure 5C). The analysis showed that DEGs were significantly enriched in the “plant hormone signal transduction” pathway, particularly in genes associated with abscisic acid (ABA) signaling, suggesting that drought stress activates defense responses through the hormonal signaling system. Significant enrichment was also observed in the “phenylpropanoid biosynthesis” and “cutin, suberin, and wax biosynthesis” pathways, implying that plants mitigate water loss by enhancing cell wall remodeling and cuticle formation. Notably, the enrichment of the “MAPK signaling pathway—plant” highlighted the molecular mechanisms of drought signal transduction, while the upregulation of genes involved in flavonoid and other antioxidant metabolite biosynthesis within the “secondary metabolite biosynthesis” pathway indicated that plants counteract drought-induced oxidative damage by strengthening their antioxidant defenses. Collectively, these findings outline a multi-layered drought adaptation strategy, spanning signal perception (hormone signaling), signal transduction (MAPK cascade), and physiological response (osmotic regulation and antioxidant defense), and provide pathway-level insights into the role of the GmNAC35 regulatory network in drought resistance (Figure 5C).

Based on the GO and KEGG enrichment analyses, we further analyzed the DEG related to drought (Table S3). Differential expression analysis of drought-responsive genes identified distinct up- and down-regulated subsets, with up-regulated genes including 5 ABA signaling pathway-associated genes, 19 involved in phenylpropanoid biosynthesis, 8 related to cutin, suberine, and wax biosynthesis, 36 MAPK signaling pathway components, 36 transcription factors, 5 ROS-responsive genes, 8 associated with alanine, aspartate and glutamate metabolism, and 29 linked to starch and sucrose metabolism. Down-regulated genes comprised 27 phenylpropanoid biosynthesis-related genes, 8 involved in cutin, suberine and wax biosynthesis, 4 associated with alanine, aspartate and glutamate metabolism, and 16 starch and sucrose metabolism-related genes. Then we selected six drought-related genes for experimental validation: one WRKY-family drought-responsive transcription factor (Glyma.17g035400), one NAC-family drought-responsive transcription factor (Glyma.04g167200), one MYB-family drought-responsive transcription factor (Glyma.05g234600), one ABA biosynthesis-related gene (Glyma.06g085000) and two AP2-family drought-responsive transcription factors (Glyma.03g191800 and Glyma.09g199800). qRT-PCR analysis confirmed that the expression patterns of these genes in GmNAC35 OE-1 transgenic plants matched the RNA-seq data (Figure 6).

4. Discussion and Conclusions

GmNAC35-overexpressing soybean lines exhibited enhanced drought tolerance, as evidenced by increased survival rates under soil-drought conditions and reduced water loss from detached leaves, suggesting a positive role for GmNAC35 in promoting water retention and stress resilience. This observation implies that GmNAC35 may regulate traits such as stomatal aperture, cuticular structure, or the expression of water transport-related genes, thereby mitigating dehydration under drought conditions. When integrated with prior evidence from miRNA target prediction and gene expression analyses, these results reinforce that GmNAC35 acts as a bona fide downstream target of miR164 and plays a critical role in the drought response. The miR164–NAC regulatory module is a well-established component in plant development and stress adaptation; our findings provide direct functional evidence for its involvement in soybean drought tolerance. From a molecular perspective, GmNAC35, as a member of the NAC transcription factor family, likely exerts its function by regulating the expression of downstream stress-responsive genes, including those encoding dehydration-related proteins, antioxidant enzymes, and components of the abscisic acid (ABA) signaling pathway [35]. Future studies employing transcriptomic profiling (RNA-seq) and chromatin immunoprecipitation followed by sequencing (ChIP-seq) will be instrumental in elucidating the regulatory landscape and direct targets of GmNAC35.

To explore the biological functions of the DEGs regulated by GmNAC35, we performed GO enrichment analysis. This analysis highlighted the multifaceted roles of these DEGs in drought stress responses. At the biological process level, DEGs were significantly enriched in categories such as “response to stress” and “organic substance biosynthetic process” [36]. These findings suggest that drought stress triggers a broad-spectrum defense response in plants, accompanied by the upregulation of biosynthetic pathways for osmoregulatory compounds, such as proline and soluble sugars [37]. These compounds are essential for maintaining cellular osmotic homeostasis during water-deficit conditions. Cellular component analysis further revealed that the products of these DEGs were predominantly localized to “intracellular organelles” and “non-membrane-bounded organelles,” indicating that drought stress induces alterations in organellar metabolism, particularly in chloroplasts and mitochondria. This suggests a coordinated regulation of energy production and reactive oxygen species (ROS) detoxification, essential processes for maintaining cellular integrity under drought stress [38].

At the molecular function level, DEGs were significantly enriched in “ion binding” and “nucleotide binding” terms, implying that drought stress may initiate transcriptional reprogramming via calcium ion (Ca²⁺) signaling and ATP-dependent enzymatic processes [34,39]. These results underscore the involvement of key signaling pathways in mediating the plant’s response to drought stress. Taken together, these findings suggest that miR164 and GmNAC35 play a pivotal role in promoting drought tolerance in soybean by regulating a comprehensive network of biosynthetic pathways, organelle function, and signal transduction mechanisms. This orchestrated regulation integrates multiple protective responses, including osmoregulation, redox homeostasis, and energy metabolism, which are essential for mitigating the adverse effects of drought stress.

To further elucidate the key metabolic pathways enriched among the downstream DEGs, we conducted KEGG functional annotation analysis, which revealed a coordinated regulatory network underlying the drought stress response. The analysis revealed significant enrichment of DEGs in the “plant hormone signal transduction” pathway, particularly in genes associated with abscisic acid (ABA) signaling [40,41]. This suggests that drought stress activates defense mechanisms via hormonal signaling, with ABA playing a central role in mediating stress responses [42]. Additionally, the “phenylpropanoid biosynthesis” and “cutin, suberin, and wax biosynthesis” pathways were significantly enriched, indicating that plants reduce water loss through enhanced cell wall remodeling and cuticle formation. Notably, the “MAPK signaling pathway—plant” was also enriched, highlighting the molecular mechanisms involved in drought signal transduction. The upregulation of genes involved in the biosynthesis of flavonoids and other antioxidant metabolites within the “secondary metabolite biosynthesis” pathway suggests that plants counteract drought-induced oxidative damage by strengthening their antioxidant defenses.

Collectively, these findings outline a multi-layered drought adaptation strategy that spans signal perception (hormone signaling), signal transduction (MAPK cascade), and physiological responses (osmotic regulation and antioxidant defense) [43,44]. These results provide valuable pathway-level insights into the regulatory role of the miR164-GmNAC35 network in enhancing drought resistance in soybean.

Our study highlights the critical role of the miR164k-GmNAC35 module in soybean drought response, presenting potential molecular targets for breeding applications. However, the conversion of these findings into practical applications requires further validation. Specifically, field trials across diverse environments are required to confirm the module’s impact on drought tolerance and yield stability. Testing in elite cultivars is essential, as genetic background may influence outcomes. Additionally, agronomic assessments must be conducted to ensure that there are no unintended trade-offs in other important traits. Future research could employ CRISPR-Cas9 for precise gene editing, identify favorable allelic variants through association mapping, and explore interactions with other stress pathways to optimize drought resilience.