A Phytosulfokine signaling module activates the Nod factor receptor to control soybean nodulation Authors: Jingjing Lu, Jiahuan Chen, Kaige Jiang, Fuyu Zhang, Yonglian Peng, Yuwei Zhang, Minghui Long, Xiaoxu Dong, Wei Yang, Yongkang Gao, Jiaying Li, Xia Li, Zhijuan Wang Published: 2026 ABSTRACT AbstractThe rhizobia-legume symbiosis, essential for nitrogen fixation, is initiated through Nod Factor (NF) perception by the NF receptors (NFRs). However, the mechanisms regulating NFR activity remain poorly defined. Here, we identify a NODULE INCEPTION 1 (GmNIN1)- Subtilisin-like Protease 1.2 (GmSBT1.2) - Phytosulfokine 4 (GmPSK4) - Phytosulfokine Receptor 1 (GmPSKR1) signaling module that directly activates the GmNFR1ɑ to promote nodulation in soybean. We demonstrate that GmNIN1a specifically activates GmSBT1.2b/1.2d, which are important for nodulation. GmSBT1.2b cleaves the proprotein of the phytosulfokine peptide GmPSK4a into mature form. Knockout of GmPSK4a and its homolog GmPSK4b severely impairs nodulation. Genetic analyses confirm the functionally interdependency of GmSBT1.2b/1.2d and GmPSK4a/4b. GmPSKR1a/1b/1c, identified as GmPSK4 receptors, are beneficial to nodulation. Notably, GmPSKR1a physically interacts with GmNFR1ɑ and enhances its kinase activity. Thus, the activation of GmNFR1ɑ by GmPSKR1a is important to its role in initiating NF signaling. Our findings uncover a previously unknown regulatory mechanism that controls the core NF signaling, providing fundamental insights into the governance of symbiotic nitrogen fixation.Subject terms: Rhizobial symbiosis, Plant signalling, Plant physiology Introduction Legumes have evolved the remarkable ability to establish symbiotic relationships with soil rhizobia, leading to the formation of specialized symbiotic organs-root nodules. Within these structures, rhizobia convert atmospheric nitrogen into ammonia through symbiotic nitrogen fixation, thereby fulfilling the nitrogen requirements of the host plants1,2. This highly specific partnership is initiated by a precise molecular dialogue between the host plants and rhizobia, activating a series of complex cellular and developmental events3,4. Under nitrogen-deficient conditions, legumes release flavonoids that attract rhizobia and induce the synthesis of lipochitooligosaccharide Nod factors (NFs)5–7. NFs are subsequently perceived by LysM-type receptor kinases (NF receptors, NFRs) located on the root hair plasma membrane8,9. Activation of the intracellular kinase domain of NFRs subsequently triggers the downstream signaling, including nuclear calcium oscillations and the activation of the calcium/calmodulin-dependent protein kinase CCaMK. This signaling induces the expression of nodulation essential transcription factors the NODULATION SIGNALLING PATHWAY 1/2 (NSP1/2) and NODULE INCEPTION (NIN) to activate rhizobial infection and nodule formation10–15. Thus, perception of bacterial NFs by the root hair-localized NFRs complex constitutes the pivotal trigger for the symbiotic program. The NFR complex is composed of NFR1 and NFR5 in Lotus japonicus (L. japonicus), NFP and LYK3 in Medicago truncatula (M. truncatula), and GmNFR1ɑ and GmNFR5ɑ in soybean8,16–19. These receptors share a conserved structure, featuring an extracellular LysM domain for ligand binding, a transmembrane domain, and an intracellular kinase domain20. The kinase domain is capable of autophosphorylation and trans-phosphorylation, and is essential for NF signaling21,22. In soybean, GmNFR1ɑ phosphorylates specific tyrosine residues within the intracellular region of GmNFR5ɑ, an event necessary for initiating downstream calcium oscillations22. Furthermore, GmNFR1ɑ can directly phosphorylates downstream substrates, such as guanine nucleotide exchange factor (GmGEF2) and Regulator of G-protein Signaling (RGS), to transduce the nodulation signal, thereby activating nodulation23,24. Although NFRs play a central role in nodulation signalling, a critical gap remains in our understanding of the upstream regulation of NFR1α itself: specifically, how its kinase activity is modulated and which kinases are responsible for its phosphorylation. In plant development and immunity, the activity of receptor kinases is often regulated by post-translational modifications, protein-protein interactions, and small peptide hormones25,26. Among them, the phytosulfokine (PSK), a class of sulfated peptide growth factors, plays a central role in cell proliferation and immunity by binding to its receptor, PSKR, a leucine-rich repeat receptor kinase (LRR-RLK)27,28. The perception of PSK by PSKR1 initiates a signaling to regulate plant growth, immunity, and nitrogen metabolism29–31. Intriguingly, emerging evidence suggests a potential role for PSK signalling in legume nodulation. In L. japonicus, overexpression of LjPSK1 or AtPSKR1 increases nodule number without affecting early rhizobial infection and nodule primordia formation32, and overexpression or exogenous application of a legume-specific PSK-δ subtype also promotes nodulation in M. truncatula33. These findings identify PSK as a peptide signal involved in nodulation, yet its functional role and signaling mechanism in legumes remain largely unexplored. In Arabidopsis, PSKR1 possesses serine/threonine kinase activity, enabling it to phosphorylate the substrate upon ligand binding27,28. The phosphorylation of PSKR1 itself serves as a core molecular switch that controls its activation and downstream signaling29. This complex signaling circuitry allows PSKR1 to precisely integrate multiple stimuli34; for instance, it engages in reciprocal phosphorylation with another LRR-RLK, B. cinerea resistance-associated kinase (BRAK), thereby forming a positive regulatory circuit that synergistically balances PSK-mediated growth signals with immune responses30. PSKR1 also directly phosphorylates key metabolic enzymes, such as glutamine synthetase (GS2), thereby optimizing nitrogen metabolism to fine-tune the balance between plant growth and defense31. However, the precise molecular function, the specific genetic components involved, and crucially, whether it directly intersects with the core NF signalling pathway remain entirely elusive. The proteolytic maturation of active PSK from its precursor protein (proPSK) requires cleavage by subtilisin-like proteases (SBTs), serine proteases that act as precise “molecular scissors” for diverse substrates35,36. SBTs are structurally characterized by a signal peptide, a prodomain essential for proper folding and inhibition, a catalytic subtilisin domain containing the characteristic Ser-His-Asp catalytic triad, a protease-associated (PA) domain, and a C-terminal fibronectin III (FnIII)-like domain implicated in substrate recognition and protein-protein interactions37,38. Beyond their role in peptide hormone maturation, SBTs are functionally diverse, interpreting extracellular cues in processes from immunity to root development39–42. In symbiosis, multiple SBT genes (SbtM1, SbtM3, SbtM4, and SbtS) are induced by NFs, but SbtM1 and SbtM3 mutants primarily affect arbuscular mycorrhizal symbiosis without an apparent impact on rhizobial nodulation43. This functional specificity suggests that different SBT family members may have been recruited for distinct symbiotic programs. These findings highlight a significant and specific, yet mechanistically unsolved role for SBTs in the legume-rhizobia symbiosis, particularly regarding their precise molecular functions and key substrates in soybean. In this study, we systematically defined the GmNIN1a-GmSBT1.2-GmPSK4-GmPSKR1 module in soybean. We identified the key GmSBT1.2 isoforms controlled by GmNIN1a and their role in GmPSK4 maturation. We established the necessity of GmPSK4 and its receptor GmPSKR1 for nodulation. Most importantly, we uncovered the mechanistic core, GmPSKR1 directly interacts with and enhances the kinase activity of GmNFR1α, revealing a direct interface between peptide hormone signaling and the core symbiotic NF pathway. Our findings thus delineate a complete signalling pathway that directly activates NFR receptors, offering an additional paradigm for the control of symbiotic nodulation. Results GmSBT1.2s are highly expressed in nodules and responsive to rhizobia To identify upstream regulators of NF signaling pathway, we analyzed the root hair transcriptome upon rhizobial inoculation and identified the subtilase (SBT) gene family GmSBT1.2 as one of the most significantly induced44. This prompted us to investigate the role of these genes in nodulation. The soybean genome contains four SBT1.2 genes, which are named as GmSBT1.2a, GmSBT1.2b, GmSBT1.2c and GmSBT1.2d based on their position in chromosomes, respectively (Suppl. Fig. 1a). Of these four proteins, GmSBT1.2a and GmSBT1.2c exhibit high similarity (97.4%), while GmSBT1.2b and GmSBT1.2d display high similarity (90.5%) (Suppl. Fig. 1b). Analysis of publicly available expression data revealed that all the GmSBT1.2 genes are predominantly expressed in nodules, with GmSBT1.2b showing the highest expression level and GmSBT1.2a being highly expressed in roots (Suppl. Fig. 1c). To validate this expression pattern, we analyzed the expression of GmSBT1.2s genes via qRT-PCR using tissues harvested at 14 days after inoculation (DAI). GmSBT1.2b and GmSBT1.2d were highly expressed in young nodules but low in leaf, root, and stem; GmSBT1.2a was highly expressed in root and nodule but nearly undetectable in leaf and stem; and GmSBT1.2c was highly expressed in nodule, low in stem and root, and nearly undetectable in leaf. (Suppl. Fig. 2a). Further assay revealed that all GmSBT1.2s were expressed at their highest level in nodules at 14 DAI, with GmSBT1.2b exhibiting the strongest expression (Suppl. Fig. 2b). To test whether GmSBT1.2s are responsive to rhizobia, we analyzed their expression in roots with or without rhizobial inoculation. GmSBT1.2a, GmSBT1.2c and GmSBT1.2d were induced by rhizobia at 1 and 3 hours after inoculation (HAI), and then decreased sharply at 6 HAI, maintaining a low expression level thereafter (Fig. 1a). GmSBT1.2b expression was upregulated by rhizobia at 1, 3 and 6 HAI, peaking at 3 HAI; however, it declined at 12 HAI, showing no significant difference at 24 HAI compared to the uninoculated control (Fig. 1a). Thus, GmSBT1.2b was further analyzed in more detail. To further examine the cellular expression pattern of GmSBT1.2b, we generated a construct in which the GUS reporter gene was driven by the 3-kb promoter region of GmSBT1.2b. Histochemical GUS staining revealed that GmSBT1.2b was predominantly expressed to vascular tissues in the absence of rhizobia (Suppl. Fig. 2c). Following rhizobial infection at 1, 3 and 6 HAI, GmSBT1.2b expression expanded to cortex cells, epidermis layers and root hairs (Suppl. Fig. 2c). Notably, GmSBT1.2b exhibited strong expression in young nodules but lower expression in mature nodules, with cross-sections showing predominant localization to cortex cells (Suppl. Fig. 2d). These results suggest that GmSBT1.2s, especially GmSBT1.2b, play an important role in the initial stages of nodulation in soybean. GmSBT1.2s are induced by rhizobia and GmSBT1.2b/1.2d are transcriptionally activated by GmNIN1a. a qRT-PCR analysis of GmSBT1.2s expression in roots with or without rhizobia at 0, 1, 3, 6, 12, and 24 HAI (hour-after-inoculation). Expression values are presented as fold-change relative to the 0 h sample (set to 1). b, c qRT-PCR analysis of the expression of GmNINs and GmSBT1.2s in hairy roots expressing empty vector (EV1) or GmNINs-RNAi (b) and EV2 or 35S:GmNIN1a (c) at 3 HAI (hour-after-inoculation). Expression values are presented as fold-change relative to the EV1 (b) or EV2 (c) sample (set to 1). d GUS assays showing the direct regulation of GmNIN1a to GmSBT1.2b in N. benthamiana leaf. The construct harboring proGmSBT1.2b:GUS was co-transformed with EV2 or 35S:GmNIN1a into epidermal cells of N. benthamiana leaves, and GUS assays were performed 48 h after transformation. Data are presented as means ± SD from three biological replicates. Asterisks indicate significant differences compared to EV2 control. Two-sided Student’s t-test, **p < 0.01. e ChIP-qPCR assay showing binding of GmNIN1a to the GmSBT1.2b promoter. DNA fragments were co-immunoprecipitated with anti-GFP antibody from chromatin suspensions prepared from 35S:GmNIN1a-GFP or EV2 samples. Data are presented as means ± SD of three biological repeats. Asterisks indicate significant differences relative to the EV2 control. Two-sided Student’s t-test, **p < 0.01;***p < 0.001. f EMSA showing that His-GmNIN1a binds to the NBS1 and NBS2 motifs of the GmSBT1.2b promoter. Competition for binding was performed using 100× and 200× competitive GmSBT1.2b probes. Three independent experiments were repeated with similar results. EV1 represents the empty RNAi vector; EV2 represents the empty overexpression vector driven by 35S promoter. For (a–c), GmELF1b was used as the endogenous control gene. Data are shown as means ± SD from three biological replicates. Asterisks indicate significant differences compared to the uninoculated samples at the same time (a) or compared to vector controls (b, c). Two-sided Student’s t-test, *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Source data are provided as a Source Data file. GmNIN1a transcriptionally activates GmSBT1.2b and GmSBT1.2d The rapid induction pattern of GmSBT1.2s prompted us to investigate how they are activated by rhizobia. To address this, we first analyzed the cis-elements in the GmSBT1.2s promoter regions. Notably, we identified two NIN-binding sites (NBS) in the promoters of GmSBT1.2b and GmSBT1.2d (NBS1: AAAGGT; NBS2: AAGGGT) (Suppl. Fig. 3a), suggesting that these two genes may be direct targets of GmNINs. To test this hypothesis, we examined GmSBT1.2s expression in GmNINs-RNAi hairy roots at 3 HAI. Notably, expression levels of GmSBT1.2b and GmSBT1.2d, not GmSBT1.2a or GmSBT1.2c, were significantly reduced compared to that of roots expressing empty vector (EV1) (Fig. 1b). To determine whether GmNIN1a/1b or GmNIN2a/2b is responsible for regulating GmSBT1.2b and GmSBT1.2d, we examined their expression in roots with reduced expression of GmNIN1a/1b or GmNIN2a/2b. Knockdown of either GmNIN1a/1b or GmNIN2a/2b significantly decreased the transcript levels of GmSBT1.2b and GmSBT1.2d. By contrast, the expression of GmSBT1.2a and GmSBT1.2c was either unchanged or only marginally reduced (Suppl. Fig. 3b, c). These results suggest that both GmNIN1a/1b and GmNIN2a/2b regulate GmSBT1.2b and GmSBT1.2d expression. Furthermore, expression levels of both GmSBT1.2b and GmSBT1.2d were significantly elevated in GmNIN1a overexpression (35S:GmNIN1a) hairy roots compared with empty vector (EV2) (Fig. 1c). To further validate the direct regulation of GmSBT1.2b transcription by GmNIN1a, we performed a transient co-expression assay in Nicotiana benthamiana (N. benthamiana) epidermal cells. The proGmSBT1.2b:GUS construct was co-expressed with either EV2 or 35S:GmNIN1a. A GUS activity assay demonstrated that GmSBT1.2b promoter activity was significantly enhanced when co-expressed with GmNIN1a (Fig. 1d and Suppl. Fig. 3d), thus confirming that GmNIN1a is a transcriptional activator of GmSBT1.2b. To test whether GmNIN1a directly binds to the promoters of GmSBT1.2b and GmSBT1.2d, we first conducted chromatin immunoprecipitation (ChIP) assays using soybean hairy roots expressing 35S:GmNIN1a-GFP at 14 DAI. The results revealed significant enrichment of GmNIN1a at the NBS-containing promoter regions of both GmSBT1.2b and GmSBT1.2d (Fig. 1e and Suppl. Fig. 3e). To confirm direct binding of GmNIN1a to specific promoter regions of GmSBT1.2b and GmSBT1.2d, we performed electrophoretic mobility shift assays (EMSA). We expressed and purified full-length His-tagged GmNIN1a protein (His-GmNIN1a) in Escherichia coli (E. coli). EMSA results demonstrated a specific interaction between His-GmNIN1a and DNA fragments containing NBS motifs. This binding was effectively competed by a 200-fold excess of unlabeled probe, confirming the specificity of the protein-DNA interaction (Fig. 1f and Suppl. Fig. 3f). Collectively, these results demonstrate that GmNIN1a binds directly to the promoters of GmSBT1.2b and GmSBT1.2d to upregulate their expression. GmSBT1.2b and GmSBT1.2d are positive regulators of soybean nodulation To functionally characterize the role of GmSBT1.2s genes in nodulation, we generated composite soybean plants with hairy roots independently downregulating four GmSBT1.2s and specific gene pairs (GmSBT1.2a/c and GmSBT1.2b/d) using RNA interference (RNAi) method. Knockdown of all four GmSBT1.2s genes significantly reduced nodule numbers by approximately 63% at 21 DAI compared to EV controls. Strikingly, the simultaneous downregulation of GmSBT1.2a and GmSBT1.2c did not significantly affect nodulation, whereas co-silencing of GmSBT1.2b and GmSBT1.2d led to a 51% reduction in nodule numbers (Suppl. Fig. 4a–e). These genetic assays suggest that GmSBT1.2b and GmSBT1.2d, but not GmSBT1.2a and GmSBT1.2c, play important roles in soybean nodulation. To further investigate the role of GmSBT1.2s in nodulation, we overexpressed GmSBT1.2a and GmSBT1.2b in hairy roots under the control of CaMV 35S promoter respectively (Suppl. Fig. 4f, g). GmSBT1.2b overexpression nearly doubled nodule numbers per hairy root compared to that of EV control, whereas GmSBT1.2a overexpression resulted in a more modest 29% increase (Suppl. Fig. 4h, i). Collectively, these results demonstrate that GmSBT1.2b and GmSBT1.2d play more important roles in soybean nodulation than GmSBT1.2a and GmSBT1.2c. To further confirm the importance of GmSBT1.2b in nodulation, we generated stable transgenic soybean plants carrying CRISPR-Cas9-induced mutations in GmSBT1.2b alone (Gmsbt1.2b) and in both GmSBT1.2b and GmSBT1.2d (Gmsbt1.2b sbt1.2d double mutants). There was a 5-bp insertion and a 153-bp deletion in the Gmsbt1.2b-1 and Gmsbt1.2b-2 mutant, respectively, leading to a pre-mature stop codon and partial protein of GmSBT1.2b (Suppl. Fig. 5a). For Gmsbt1.2b sbt1.2d-1, there was a 2-bp deletion in GmSBT1.2b and 1-bp deletion in GmSBT1.2d, respectively; while for Gmsbt1.2b sbt1.2d-2, there was a 1-bp insertion in both GmSBT1.2b and GmSBT1.2d (Suppl. Fig. 5a). Both Gmsbt1.2b single mutants displayed significantly fewer nodules and lower nitrogenase activity compared to wild-type (WT) plants, and the nodulation defect was more severe in the Gmsbt1.2b sbt1.2d double mutants (Fig. 2a–c), indicating that the effects of these two genes on nodulation are additive. We also generated stable transgenic lines overexpressing GmSBT1.2b under the control of CaMV 35S promoter in W82 background (Suppl. Fig. 5b). At 21 DAI, all three GmSBT1.2b-overexpressing lines (OE-2, OE-3, and OE-4) showed significantly increased nodule numbers and nitrogenase activity compared to WT plants (Fig. 2d–f), confirming the positive role of GmSBT1.2b in nodulation. These results demonstrate that GmSBT1.2b positively regulates both nodule formation and subsequent nitrogen fixation activity. GmSBT1.2b and GmSBT1.2d positively regulate nodulation by promoting rhizobial infection and the nodule formation. a, b Representative images (a) and quantification data (b) of nodule number in Gmsbt1.2b single and Gmsbt1.2b sbt1.2d double mutants at 21 DAI (n = 10). Scale bars = 1 cm. c Nitrogenase activity of nodules of wild type (WT), Gmsbt1.2b and Gmsbt1.2b sbt1.2d double mutants plants at 28 DAI (n = 10 for WT, Gmsbt1.2b-1 and n = 8 for Gmsbt1.2b-2 and Gmsbt1.2b sbt1.2d-1, 2). d, e Representative images (d) and quantification data (e) of nodule number in GmSBT1.2bOE transgenic plants at 21 DAI (n = 10). Scale bars = 1 cm. f Nitrogenase activity of nodules of WT, GmSBT1.2bOE transgenic plants at 28 DAI (n = 8 for WT and n = 9 for GmSBT1.2bOE-2, 3, 4). g, h Representative images (g) and quantification data (h) of infection thread number in WT, Gmsbt1.2b and Gmsbt1.2b sbt1.2d mutants at 4 DAI (n = 8). Scale bars = 100 μm. i, j Representative images (i) and quantification data (j) of nodule primordia number in WT, Gmsbt1.2b and Gmsbt1.2b sbt1.2d mutants at 7 DAI (n = 10). Scale bars = 100 μm. k, l Representative images (k) and quantification data (l) of infection thread number in WT and GmSBT1.2bOE lines at 4 DAI (n = 8). Scale bars = 100 μm. m, n Representative images (m) and quantification data (n) of nodule primordia number in WT and GmSBT1.2bOE lines at 7 DAI (n = 8). Scale bars = 100 μm. For (b, c, e, f, h, j, l, n), data are presented as means ± SD. Three independent experiments were repeated with similar results. Different letters indicate significant differences at p < 0.05 (One-way ANOVA). For (b, e), box plots display the median (centre), 25th–75th percentiles (box bounds), and minimum-maximum values (whiskers). Source data are provided as a Source Data file. Our results that GmSBT1.2b/1.2d were induced by rhizobia and reached its highest expression level at 3 HAI led us to hypothesize that GmSBT1.2b/1.2d may regulate early nodulation events. To test this hypothesis, we counted the number of infection thread and nodule primordia in Gmsbt1.2b, Gmsbt1.2b sbt1.2d mutants, and GmSBT1.2bOE lines (Fig. 2g–n). Both infection thread formation and nodule primordia were significantly impaired in Gmsbt1.2b single mutant and further diminished in the Gmsbt1.2b sbt1.2d double mutants (Fig. 2g–j). Conversely, GmSBT1.2bOE lines showed significantly higher levels of these early symbiotic events than WT controls (Fig. 2k–n). Consistent with these results, GmNIN1a expression was significantly increased in GmSBT1.2bOE roots, but decreased in the roots of both Gmsbt1.2b single and Gmsbt1.2b sbt1.2d double mutants compared to WT (Suppl. Fig. 5c, d). These results demonstrate that GmSBT1.2b and GmSBT1.2d are key regulators of rhizobial infection and nodule formation. GmSBT1.2b is responsible for the cleavage of GmPSK4a precursor The next question is how GmSBT1.2b regulates nodulation. As a subtilase family protease, GmSBT1.2b likely functions in the maturation of peptides. To determine whether the subtilase activity of GmSBT1.2b is required for its role in nodulation, we inhibited its function by overexpressing Extracellular Protease Inhibitor 10 (EPI10), a specific subtilase inhibitor45, in hairy roots. Compared to the EV control, EPI10 overexpression significantly reduced nodule number by approximately 53% at 21 DAI (Suppl. Fig. 6). This result indicates that the subtilase activity is important for nodule formation. Previous studies have established that SBTs mediate proteolytic processing of peptide precursors including PSKs35,36 and LjPSK4 plays positive roles in nodulation32. Based on these findings, we hypothesized that GmSBT1.2b regulates nodulation through cleaving PSK precursor proteins to generate active peptides. To prove this hypothesis, we first analyzed the GmPSK genes in soybean genome and found 16 GmPSKs (Suppl. Fig. 7a). Expression assay from publicly available data showed that GmPSK4a and GmPSK4b are highly expressed in nodules, with GmPSK4a additionally showing higher expression in root hairs (Suppl. Fig. 7b). qRT-PCR and GUS assays confirmed that GmPSK4a and GmPSK4b were highly expressed in both roots and nodules, particularly in young nodules (Fig. 3a–c). Further expression analyses showed that GmPSK4a and GmPSK4b were induced by rhizobia at 1, 3, and 6 HAI, with expression peaking at 6 HAI. Nevertheless, their expression decreased at 12 HAI, reaching levels similar to the uninoculated control by 24 HAI (Fig. 3d, e). GmPSK4s are highly expressed in nodules and cleaved by GmSBT1.2b. a, b Expression analysis of GmPSK4a and GmPSK4b in different tissues at 14 DAI (a) or in nodules at different stages (b). The data are shown as means ± SD from three biological replicates. Different letters indicate significant differences at p < 0.05 (One-way ANOVA). c Representative GUS staining images in nodule of proGmPSK4b:GUS at 14, 21, 28 and 36 DAI. Three independent experiments were repeated with similar results. Scale bars = 40 μm. d, e qRT-PCR analysis of GmPSK4a (d) and GmPSK4b (e) in roots with or without inoculation. Expression values are presented as fold-change relative to the 0 h sample (set to 1). The data are shown as means ± SD from three biological replicates. Asterisks indicate significant differences compared to uninoculated samples at the same time. Two-sided Student’s t-test, *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. f proGmPSK4a is processed by GmSBT1.2b in vivo. GmSBT1.2a-HA, GmSBT1.2b-HA or empty vector (HA) was coexpressed with GmPSK4a-GFP respectively in N. benthamiana leaves. The red arrow indicates cleaved product by GmSBT1.2a. g proGmPSK4a is processed by GmSBT1.2b in vitro. GmSBT1.2b-HA and empty vector (HA) were expressed in N. benthamiana leaves. The purified GmPSK4a precursor was incubated with total protein from leaves expressing 35S:GmSBT1.2b or EV and treated with or without EPI10 for 2 h. The red arrow indicates cleaved product by GmSBT1.2b. h proGmPSK4a is processed by GmSBT1.2b in vitro. Total proteins extracted from the wild-type and Gmsbt1.2b sbt1.2d double mutants were incubated with the purified His-GmPSK4a precursor protein, respectively. After 2 h of reaction, western blot was performed using anti-His antibody to detect His-GmPSK4a. Lane 4 was from a separate membrane and is used only for purity verification. For (f, g, h), the numbers marked on the right indicate the molecular weights of the fusion protein or ACTIN. Three independent experiments were repeated with similar results. Source data are provided as a Source Data file. To test whether GmPSK4a is a cleavage substrate of GmSBT1.2b, we co-expressed HA-tagged GmSBT1.2b (GmSBT1.2b-HA) and GFP-tagged GmPSK4a (GmPSK4a-GFP) fusion proteins in N. benthamiana leaves. A lower band was observed upon co-expression of GmSBT1.2b but not in the EV control (Fig. 3f). Sequencing the lower band detected peptide sequence “RTLTAHIDYIYTQK” and “TLTAHIDYIYTQK” (Suppl. Fig. 8), indicating that the lower band was the cleavage product of GmPSK4a. To determine the substrate specificity of the protease, we co-expressed GmSBT1.2a-HA with GmPSK4a-GFP and did not detect the cleavage product (Fig. 3f), indicating that GmSBT1.2b, but not GmSBT1.2a, processes GmPSK4a in plants. To confirm the cleavage of GmPSK4a by GmSBT1.2b, we expressed GmSBT1.2b-HA in N. benthamiana leaves and GmPSK4a-His in E. coli. When His-GmPSK4a was incubated with total protein extracts from N. benthamiana leaves expressing GmSBT1.2b-HA, significantly more cleavage products were observed compared to the EV control (Fig. 3g). This suggests that GmSBT1.2b directly processes GmPSK4a in vitro. To confirm this, we treated the proteins with EPI10 and found that EPI10 treatment almost completely inhibited the cleavage of GmPSK4a (Fig. 3g). Furthermore, we incubated His-GmPSK4a with total protein extracted from WT and Gmsbt1.2b sbt1.2d mutant plants. Incubation with WT protein extracts resulted in a higher accumulation of the cleavage product, as confirmed by sequencing (Suppl. Table 1), whereas this product was markedly reduced in the Gmsbt1.2b sbt1.2d mutant background (Fig. 3h). These results demonstrate that GmSBT1.2b is responsible for processing GmPSK4a. GmPSK4a and GmPSK4b positively regulate soybean nodulation To investigate the roles of GmPSK4a and GmPSK4b in soybean nodulation, we firstly examined their effect on nodulation using a hairy root transformation system by overexpressing or downregulating their expression. The number of nodules was reduced in the roots of composite plants that had downregulated GmPSK4a and GmPSK4b (Suppl. Fig. 9a–c), while increased in roots overexpressing GmPSK4a or GmPSK4b, compared to the EV controls (Suppl. Fig. 9d–g). These results suggest that GmPSK4a and GmPSK4b play a positive role in soybean nodulation. To validate the function of GmPSK4a/4b in nodulation, we generated Gmpsk4a single and Gmpsk4a psk4b double mutants (Gmpsk4a-1, Gmpsk4a psk4b-1, and Gmpsk4a psk4b-2) using CRISPR-Cas9 technology. The Gmpsk4a-1 and Gmpsk4a-2 mutants contained a 5-bp insertion and a 4-bp deletion within GmPSK4a exon, respectively. Consequently, both mutations generated a premature stop codon, yielding truncated proteins. In the double mutant, Gmpsk4a psk4b-1 carried a 6-bp deletion in GmPSK4a and a 4-bp deletion in GmPSK4b, while Gmpsk4a psk4b-2 contained a 6-bp deletion in GmPSK4a and a 55-bp deletion in GmPSK4b (Suppl. Fig. 10a). As expected, both nodule number and nitrogenase activity were significantly reduced in the Gmpsk4a mutants compared to the WT, with the double mutants exhibiting an even more pronounced reduction (Fig. 4a–c), suggesting a redundant role of GmPSK4a and GmPSK4b in nodulation. To confirm the function of GmPSK4a in nodulation, we generated stable transgenic plants overexpressing GmPSK4a under the control of 35S promoter in the W82 background (GmPSK4aOE-4, GmPSK4aOE-10 and GmPSK4aOE-21) (Suppl. Fig. 10b). Phenotypic analyses revealed that all three GmPSKOE lines exhibited significantly increased nodule numbers and the nitrogenase activity compared to WT (Fig. 4d–f), confirming that the level of GmPSK4a correlates positively with its function in nodulation. GmPSK4s promote nodulation by positively regulating rhizobial infection and the formation of nodule. a, b Representative images (a) and quantification data (b) of nodule number in Gmpsk4a single and Gmpsk4a psk4b double mutants at 21 DAI (n = 10). Scale bars = 1 cm. c Nitrogenase activity of nodules of WT, Gmpsk4a and Gmpsk4a psk4b double mutants plants at 28 DAI (n = 10 for WT; n = 8 for Gmpsk4a-1, 2 and Gmpsk4a psk4b-1, 2). d, e Representative images (d) and quantification data (e) of nodule number in GmPSK4aOE transgenic plants at 21 DAI (n = 10). Scale bars = 1 cm. f Nitrogenase activity of nodules of WT, GmPSK4aOE transgenic plants at 28 DAI (n = 8 for WT and n = 7 for GmPSK4aOE-4, 10, 21). g, h Representative images (g) and quantification data (h) of infection thread number in WT, Gmpsk4a and Gmpsk4a psk4b double mutants at 4 DAI (n = 8). Scale bars = 100 μm. i, j Representative images (i) and quantification data (j) of nodule primordia number in Gmpsk4a and Gmpsk4a psk4b double mutants at 7 DAI (n = 8). Scale bars = 100 μm. k, l Representative images (k) and quantification data (l) of infection thread number in WT and GmPSK4aOE transgenic lines at 4 DAI (n = 8). Scale bars = 100 μm. m, n Representative images (m) and quantification data (n) of nodule primordia number in WT and GmPSK4aOE transgenic lines at 7 DAI (n = 8).Scale bars = 100 μm. For (b, c, e, f, h, j, l, n), data are presented as means ± SD. Three independent experiments were repeated with similar results. Different letters indicate significant differences at p < 0.05 (One-way ANOVA). For (b, e), box plots display the median (centre), 25th–75th percentiles (box bounds), and minimum-maximum values (whiskers). Source data are provided as a Source Data file. To determine the role of GmPSK4a and GmPSK4b in early nodulation events, we examined infection threads and nodule primordia in the knockout mutants and stable transgenic plants. Gmpsk4a-1 showed significantly fewer infection threads and nodule primordia than WT, and Gmpsk4a psk4b exhibited even more severe defects (Fig. 4g–j). Conversely, GmPSK4aOE lines displayed increased numbers of both infection threads and nodule primordia (Fig. 4k–n). Additionally, GmNIN1a expression was reduced in Gmpsk4a and Gmpsk4a psk4b roots but elevated in GmPSK4aOE roots (Suppl. Fig. 10c, d). These results demonstrate that GmPSK4a and GmPSK4b positively regulate nodulation. GmSBT1.2b and GmPSK4a are interdependent in regulating nodulation Because GmSBT1.2b cleaves GmPSK4a to facilitate its processing, we then investigated whether GmSBT1.2b-mediated nodulation requires GmPSK4a. To do this, we overexpressed GmSBT1.2b in the hairy roots of Gmpsk4a psk4b and quantified nodule number at 21 DAI. Although GmSBT1.2b overexpression in WT increased nodule number compared to EV control, this positive effect was abolished in Gmpsk4a psk4b-2 roots (Fig. 5a, b), suggesting that GmSBT1.2b-dependent nodulation regulation requires functional GmPSK4a. We next examined the reciprocal relationship by overexpressing GmPSK4a in Gmsbt1.2b sbt1.2d mutant hairy roots. GmPSK4a overexpression significantly enhanced nodulation in WT background compared to EV control, but it failed to increase nodule number in Gmsbt1.2b sbt1.2d-1 mutant (Fig. 5c, d), indicating that GmPSK4a-mediated nodulation is dependent on GmSBT1.2b function. To confirm that the role of GmSBT1.2b and GmPSK4a in nodulation depends on the production of mature PSK peptide, we examined the effect of synthetic PSK peptide on nodulation in Gmsbt1.2b sbt1.2d and Gmpsk4a psk4b mutants. Treatment with PSK significantly increased nodule number in both mutants (Suppl. Fig. 11). These results indicate that GmPSK4a and GmSBT1.2b positively regulate nodulation through the generation of PSK peptide. Taken together, these results reveal an interdependent relationship between the GmSBT1.2b and GmPSK4a in promoting soybean nodulation. GmSBT1.2b and GmPSK4a depend on each other to regulate nodulation. a, b Representative images (a) and quantification data (b) of nodule number per hairy root expressing EV2 and 35S:GmSBT1.2b in WT or Gmpsk4a psk4b-2 mutant (n = 13 for EV2 and 35S:GmSBT1.2b in WT; n = 18 for EV2 and 35S:GmSBT1.2b in Gmpsk4a psk4b-2 mutant). Scale bars = 1 cm. c, d Representative images (c) and quantification data (d) of nodule number per hairy root expressing EV2 or 35S:GmPSK4a in WT or Gmsbt1.2b sbt1.2d-1 mutant (n = 12 for EV2 and 35S:GmPSK4a in WT; n = 18 for EV2 and n = 15 for 35S:GmPSK4a in Gmsbt1.2b sbt1.2d-1 mutant). Scale bars = 1 cm. EV2 represents the empty overexpression vector driven by 35S promoter. For (b, d), data are presented as means ± SD. Three independent experiments were repeated with similar results. Different letters indicate significant differences at p < 0.05 (One-way ANOVA). For (b, d), box plots display the median (centre), 25th–75th percentiles (box bounds), and minimum-maximum values (whiskers). Source data are provided as a Source Data file. GmPSKR1s promote nodulation in a GmPSK4s-dependent manner The next question is how GmPSK4a/4b regulate nodulation. It is clear that PSK is perceived by its receptors PSKR1 and PSKR2 in Arabidopsis28,46. To investigate whether the soybean PSKR transduces GmPSK4 signals to regulate nodulation, we analyzed the soybean orthologues of Arabidopsis PSKR1 and PSKR2 and identified three PSKR genes in soybean, named GmPSKR1a, GmPSKR1b and GmPSKR1c based on their chromosomal localization (Suppl. Fig. 12a, b). Both GmPSKR1a and GmPSKR1b were expressed at the highest level in the nodule and at a lower level in the root and stem, while GmPSKR1c was expressed at a higher in the nodule and at a lower level in the stem and was nearly undetectable in the leaf and root (Suppl. Fig. 12c, d). To determine that GmPSKR1a is the receptor of GmPSK4, we tested the binding of mature PSK peptide to LRR domain of GmPSKR1a (GmPSKR1a-LRR) using Localized Surface Plasmon Resonance (LSPR) and found that PSK peptide bind to GST-GmPSKR-LRR, but not to the GST control (Suppl. Fig. 12e, f). We then investigated the impact of GmPSKR1a, GmPSKR1b, and GmPSKR1c on nodulation using CRISPR-Cas9 or overexpressing GmPSKR1a in hairy roots. GmPSKR1s knockout significantly reduced nodule numbers, with a decrease of up to 71% compared to controls, while GmPSKR1a overexpression increased nodule number compared to vector control (Fig. 6a–e; Suppl. Fig. 13a), indicating that GmPSKR1s promote nodulation. In agreement with this, the GmPSKR1s knockout mutant exhibited significantly reduced expression of GmNFR1α, GmNINs, GmSBT1.2b/1.2d, and GmPSK4a/4b (Fig. 6f), indicating that GmPSKR1s affect both NF and PSK signal transduction. GmPSKR1s positively regulate nodulation in a GmPSK4s-dependent manner. a, b Representative images (a) and quantification data (b) of nodule number per hairy root expressing Cas9-EV and Cas9-GmPSKR1s. Data are presented as means ± SD (n = 16). Scale bars = 1 cm. c, d Representative images (c) and quantification data (d) of nodule number per hairy root expressing EV2 and 35S:GmPSKR1a. Data are presented as means ± SD (n = 11 for EV2 and n = 22 for 35S:GmPSKR1a). Scale bars = 1 cm. e qRT-PCR analysis the expression of GmPSKR1a in roots transformed with EV2 and 35S:GmPSKR1a at 21 DAI. Expression values are presented as fold-change relative to the EV2 (set to 1). f qRT-PCR analysis of the expression of GmNINs, GmSBT1.2b, GmSBT1.2d, GmPSK4s and GmNFR1α in roots transformed with Cas9-EV and Cas9-GmPSKR1s at 21 DAI. Expression values are presented as fold-change relative to the EV1 (set to 1). g, h Representative images (g) and quantification data (h) of nodule number per hairy root expressing EV2 and 35S:GmPSKR1a in WT or Gmpsk4a psk4b-2 mutant. Data are presented as means ± SD (n = 17 for EV2 and 35S:GmPSKR1a in WT; n = 14 for EV2 and 35S:GmPSKR1a in Gmpsk4a psk4b-2 mutant). Three independent experiments were repeated with similar results. Different letters indicate significant differences at p < 0.05 (One-way ANOVA). Scale bars = 1 cm. EV2 represents the empty overexpression vector driven by 35S promoter. For (b, d, e, f), the data are shown as means ± SD from three biological replicates. Asterisks indicate significant differences relative to the EV controls. Two-sided Student’s t-test, **p < 0.01; ***p < 0.001; ****p < 0.0001. For (b, d, h), box plots display the median (centre), 25th–75th percentiles (box bounds), and minimum-maximum values (whiskers). For (h), three independent experiments were repeated with similar results. Different letters indicate significant differences at p < 0.05 (One-way ANOVA). Source data are provided as a Source Data file. To determine whether the positive effect of GmPSKR1 on nodule formation depends on its kinase activity, we generated a mutant kinase by replacing a specific lysine residue in the ATP-binding region with glutamate34 (mGmPSKR1a) (Suppl. Fig. 13b, c). We expressed this kinase-dead version in hairy root and found mGmPSKR1a had no effect on nodulation (Suppl. Fig. 13d-f), suggesting positive effect of GmPSKR1a on nodulation depends on its kinase activity. Interestingly, this positive effect of GmPSKR1a on nodulation was compromised in Gmpsk4a psk4b-2 mutant (Fig. 6g, h). This result indicates that the positive role of GmPSKR1a in nodulation is dependent on GmPSK4s to some extent. GmPSKR1a interacts with GmNFR1ɑ and activates its kinase activity Given the positive role of GmPSK4a/4b in the early infection stage and its protein characteristics, we hypothesized that GmPSKR1s is highly likely to interact with NF receptor. To test this, we first investigated whether GmPSKR1a and GmNFR1ɑ are co-expressed in root hair. We expressed proGmPSKR1a:GUS or proGmNFR1α:GUS in hairy root and detected their expression at 1, 3, and 5 DAI. Both GmPSKR1a and GmNFR1ɑ were shown to be expressed in root hair (Suppl. Fig. 14a). We then tested whether these two proteins are co-localized. A subcellular localization assay showed that GmNFR1ɑ-mCherry and GmPSKR1a-GFP co-localized at the plasma membrane (Suppl. Fig. 14b). We then tested the interaction between GmPSKR1a and GmNFR1ɑ using Y2H and bimolecular fluorescence complementation (BiFC) assays. The results showed that GmPSKR1a interacts with the C-terminal end of GmNFR1ɑ (Fig. 7a, b). This interaction between GmPSKR1a and GmNFR1ɑ was further confirmed by co-immunoprecipitation (Co-IP) experiments (Fig. 7c). To assess whether the PSK peptide affects the GmPSKR1a-GmNFR1ɑ interaction, we examined their binding in the presence of PSK. Application of PSK increased the YFP signal in the BiFC assay for GmPSKR1a-YC and GmNFR1ɑ-C-YN (Fig. 7b). Consistently, PSK treatment enhanced the amount of GmNFR1ɑ-C-Flag immunoprecipitated by GmPSKR1a-MYC in Co-IP assays. These results demonstrate that PSK promotes the interaction between GmPSKR1a and GmNFR1ɑ. GmPSKR1a interacts with GmNFR1α and activates its kinase activity. a Yeast two-hybrid assay to verify the interaction between the full-length GmPSKR1a and cytoplasmic kinase domain GmNFR1α (GmNFR1α-C). Three independent experiments were repeated with similar results. b BiFC assay to detect the interaction between GmPSKR1a and the GmNFR1α-C with or without PSK treatment. Three independent experiments were repeated with similar results. GmNARK-YN and GmNARK-YC were used as negative controls. Scale bars = 100 μm. c Co-IP assay of the interaction of GmPSKR1a-MYC with GmNFR1α-C-Flag in N. benthamiana leaves with or without PSK treatment. GmSOS2-MYC were used as a negative control. Three independent experiments were repeated with similar results. d GST-GmNFR1α-C and His-GmPSKR1a-C proteins are mutually phosphorylated. In vitro phosphorylation reaction was performed using purified proteins and 32P-labeled ATP and detected by autoradiography. Coomassie Brilliant Blue to display the expression level of proteins. Three independent experiments were repeated with similar results. e The phosphorylation activity of GST-GmNFR1α-C toward GmGEF2a was enhanced in the presence of His-GmPSKR1a-C. Three independent experiments were repeated with similar results. MBP, Maltose-Binding Protein. Source data are provided as a Source Data file. Next, we expressed and purified the cytoplasmic kinase domain of GmPSKR1a-His (GmPSKR1a-C-His) and GmNFR1ɑ-GST (GmNFR1ɑ-C-GST) in E. coli. In vitro kinase assay showed that both GmPSKR1a and GmNFR1ɑ exhibit autophosphorylation activity. Notably, the phosphorylation level of GmNFR1ɑ increased with rising concentrations of GmPSKR1a (Fig. 7d; Suppl. Fig. 15). Conversely, the phosphorylation level of GmPSKR1a was also enhanced in the presence of GmNFR1ɑ (Fig. 7d). These results indicate that GmPSKR1a and GmNFR1ɑ reciprocally activate each other’s kinase activity. To test whether the activation of GmPSKR1a affects GmNFR1ɑ phosphorylation on its substrate, we examined the phosphorylation of GmGEF2a23, a known substrate of GmNFR1ɑ, in the absence or presence of GmPSKR1a. The results showed that GmNFR1ɑ phosphorylates GmGEF2a, with the phosphorylation level increasing significantly in the presence of GmPSKR1a (Fig. 7e). This result indicates that the activation of GmNFR1ɑ by GmPSKR1a enhances its ability to phosphorylate GmGEF2a. GmPSKR1a-mediated activation of GmNFR1ɑ contributes to nodulation To test whether the activation of GmNFR1ɑ by GmPSKR1a affects its function in nodulation, we first characterized the phosphorylation site of GmNFR1ɑ by GmPSKR1a by LC/MS. We conducted GmNFR1ɑ phosphorylation reaction with or without GmPSKR1a in vitro. Mass spectrometry analysis identified 13 sites (Ser-300, 349, 464, 516, 534, and Thr-312, 360, 426, 456, 465, 468, 513 and Tyr-421) in GmNFR1ɑ alone and 8 sites (Ser-464, 516, 534, 537, 609, and Thr-465, 513, 603) in the combination of GmNFR1ɑ and GmPSKR1a (Suppl. Fig. 16, Table 2). We predicted that the additional three sites (Ser-537, Thr-603 and Ser-609) and one site (Ser-534) showing increased phosphorylation intensity on GmNFR1ɑ are phosphorylated by GmPSKR1a. To prove this, we generated mutated GmNFR1ɑ (mGmNFR1ɑ) with all four Ser and Thr sites to alanine and then tested the kinase activity of mGmNFR1ɑ with or without GmPSKR1a. A in vitro assay showed that GmPSKR1a enhanced the phosphorylation level of GmNFR1ɑ, but the enhancement was decreased in mGmNFR1ɑ (Fig. 8a), indicating that the mutation of phosphorylation sites impaired the activity of GmNFR1ɑ by GmPSKR1a. Consistent with this, the phosphorylation level of GmPSKR1a by GmNFR1ɑ was decreased in the mGmNFR1ɑ (Fig. 8a). The activation of GmPSKR1a to GmNFR1α is critical for GmNFR1α function in nodulation. a Mutation of GmNFR1α reduced the reciprocal phosphorylation between GST-GmNFR1α-C and His-GmPSKR1a-C. Three independent experiments were repeated with similar results. b, c qRT-PCR analysis of the expression of GmNFR1α variants in roots transformed with EV2, 35S:GmNFR1α and 35S:mGmNFR1α in Bragg and nod49 mutant plants. Expression values are presented as fold-change relative to the EV2 (set to 1). d, e Representative images (d) and quantification data (e) of nodule number per hairy root expressing EV2, 35S:GmNFR1α and 35S:mGmNFR1α in Bragg and nod49 mutant plants. Data are presented as means ± SD (n = 15 for EV2, 35S:GmNFR1α and 35S:mGmNFR1α in Bragg; n = 14 for EV2, 35S:GmNFR1α and 35S:mGmNFR1α in nod49 mutant). Three independent experiments were repeated with similar results. Different letters indicate significant differences at p < 0.05 (One-way ANOVA). Scale bars = 1 cm. EV2 represents the empty overexpression vector driven by 35S promoter. For (e), box plots display the median (centre), 25th–75th percentiles (box bounds), and minimum-maximum values (whiskers). Source data are provided as a Source Data file. To determine the functional significance of GmNFR1ɑ phosphorylation by GmPSKR1a, we expressed wild-type GmNFR1ɑ and its phosphorylation-sites mutant (mGmNFR1ɑ) in the hairy root of GmNFR1ɑ mutant nod49 and WT (Bragg)18. In the Bragg background, overexpression of GmNFR1ɑ led to an approximately 2-fold increase in nodule number, whereas overexpression of mGmNFR1ɑ only increased nodule number by less than 0.5-fold compared with vector control; in the nod49 background, overexpression of GmNFR1ɑ fully restored the nodulation levels to WT that overexpressing GmNFR1ɑ, while overexpression of mGmNFR1ɑ led to the nodulation levels to WT that expressing vector control (Fig. 8b–e), suggesting that persistent activity of GmNFR1ɑ requires GmPSKR1a. These data demonstrate that GmPSKR1a-mediated phosphorylation of GmNFR1ɑ is vital for nodulation. Discussion The central role of NFRs in initiating symbiotic signaling pathway has been firmly established; the mechanisms that fine-tune their kinase activity remains elusive. Our work elucidates a complete previously unrecognized signalling module, in which the GmNIN1s-induced phytosulfokine pathway directly activates the Nod Factor receptor GmNFR1ɑ to regulate nodulation. We demonstrate that the receptor kinase GmPSKR1, upon perception of the mature GmPSK4 peptide processed by GmSBT1.2, physically interacts with and potentiates the kinase activity of GmNFR1ɑ. This crosstalk between two distinct receptor systems provides a distinct regulatory paradigm for the control of symbiotic signalling, suggesting that plants employ endogenous peptide hormones to gatekeep the perception of rhizobial signals. In this study, we investigated the functional specialization within the GmSBT1.2 gene family in soybean nodulation. Although all four GmSBT1.2 members are transcriptionally induced by rhizobial infection, only GmSBT1.2b and GmSBT1.2d were found to be direct transcriptional targets of GmNINs, a master regulator of nodulation. Consistently, knockdown of GmSBT1.2b and GmSBT1.2d significantly reduced nodule numbers, whereas downregulation of GmSBT1.2a and GmSBT1.2c had no detectable effect on nodulation. These observations indicate that despite their sequence similarity and co-induction during symbiosis, the four GmSBT1.2 members have undergone functional divergence. The biological significance of the rhizobia-induced expression of GmSBT1.2a and GmSBT1.2c remains unclear. It is possible that these genes play roles in other aspects of plant-microbe interactions, such as defense modulation or later stages of nodule development, which were not captured by our focused assays. Alternatively, they may function under specific environmental conditions or in different tissue contexts. Detailed spatiotemporal expression analysis and functional characterization under varied conditions will be required to address these possibilities. We further demonstrated that GmSBT1.2b promotes nodulation by processing the signaling peptide GmPSK4. In vitro biochemical assays showed that GmSBT1.2b, but not GmSBT1.2a, efficiently cleaves the GmPSK4a precursor to generate the mature peptide. Genetic evidence supported this specificity: the nodulation-promoting effect of full-length GmPSK4a was largely abolished in the Gmsbt1.2b sbt1.2d double mutant, whereas mature PSK4 retained activity. These results establish GmPSK4a as a physiological substrate of GmSBT1.2b and place this protease-peptide module downstream of the GmNINs signaling cascade. Given the ~83% amino acid sequence similarity between GmSBT1.2b and GmSBT1.2a, their differential substrate specificity toward GmPSK4a is particularly intriguing, raising questions about how substrate recognition is achieved among highly similar proteases. Structural and mutational analyses of the substrate-binding pocket may reveal key residues governing specificity, and additional cofactors or spatial compartmentalization may also contribute. Although other proteases could potentially cleave GmPSK4a in different contexts, the strong genetic interaction-where loss of GmSBT1.2b and GmSBT1.2d nearly abolishes the nodulation response to GmPSK4a-indicates that these two proteases act as the primary processors of GmPSK4a during symbiosis. Future identification of additional substrates and interacting partners will help delineate the broader regulatory network centered on this protease family. A previous study in L. japonicus have shown that phytosulfokine peptide LjPSK4 acts as a positive regulator of nodulation32. However, the precise molecular mechanism by which this peptide hormone exerts its function and how its signal is integrated into the established NF signaling cascade remained an open question. Our findings in soybean now bridge this gap by delineating a complete signaling axis. The pleiotropic functions of PSK are known to be mediated by its receptor, PSKR. In this study, we identify the cognate receptors GmPSKR1a/1b/1c as indispensable components for nodulation. These GmPSKR1a/1b/1c are expressed during nodulation, and alteration of these receptor genes results in defective nodulation (Fig. 6a–f). Thus, our work defines the signaling pathway from the matured phytosulfokine peptides to their plasma membrane receptors during soybean nodulation. Most significantly, our study reveals the precise point of its integration with and promotes the NF pathway. Firstly, GmPSKR1a and GmNFR1ɑ co-express in root hair and co-localize at the plasma membrane (Suppl. Fig. 14); Secondly, GmPSKR1a physically interacts with GmNFR1ɑ via their cytoplasmic kinase domains, importantly, PSK peptide enhances their interaction (Fig. 7a–c); Thirdly, GmPSKR1a enhances the kinase activity of GmNFR1ɑ, boosting both its autophosphorylation and its transphosphorylation of the substrate GmGEF2 (Fig. 7d, e); Finally, we determine the phosphorylation sites (Ser-537, Thr-603, Ser-609 and Ser-534) on GmNFR1ɑ targeted by GmPSKR1a (Fig. 8a; Suppl. Fig. 16). We observed that mutating these four sites to non-phosphorylatable forms did not completely abolish the function of GmNFR1ɑ. There are two possible explanations for this phenomenon. First, there may be additional phosphorylation sites on GmNFR1ɑ targeted by GmPSKR1a. This is supported by the observation that GmNFR1ɑ could still be phosphorylated by GmPSKR1a after mutation of these four sites; if all sites phosphorylated by GmPSKR1a were mutated, a more severe inhibitory effect would be expected. Second, the basal function of GmNFR1ɑ upon NF activation may not require phosphorylation by GmPSKR1a. Phosphorylation by GmPSKR1a likely maintains GmNFR1ɑ in a persistently activated state. Therefore, when GmPSKR1-mediated phosphorylation is abolished, GmNFR1ɑ can still initiate nodulation, but its sustained activity is compromised, resulting in significantly fewer nodules. This is consistent with the observation that mutation of GmPSKR1s leads to reduced nodule number rather than complete nodulation failure. Collectively, our findings position the GmPSK4-GmPSKR1 module as a central hub that integrates developmental and environmental cues. Since nodulation is an energetically costly process, it is tightly regulated by factors such as nitrogen availability, phytohormones, and stress conditions-many of which are known to regulate PSK expression. By modulating GmNFR1α activity, this module enables the plant to fine-tune symbiotic responses according to prevailing physiological and environmental conditions. Thus, while the core Nod factor receptor is essential for initiating symbiosis, this additional layer provides the plasticity required for adaptive regulation. Intriguingly, we also found that GmNFR1ɑ reciprocally enhances GmPSKR1a kinase activity, suggesting a potential mechanism by which nodulation may feedback-regulate immunity or environmental adaptation through GmPSKR1a. In summary, our results delineate a coherent positive feedback loop in NF signal pathway. Upon NF perception by GmNFR1α and its co-receptors, early signaling events lead to the transcriptional activation of GmNINs, which then acts as a master regulator to initiate the expression of downstream genes, including GmSBT1.2b/1.2d. GmSBT1.2b/1.2d processes GmPSK4 precursors to produce mature PSK peptides. These peptides are perceived by the GmPSKR1 receptor, which in turn phosphorylates and further activates GmNFR1α. This activated GmNFR1α can then enhance NF signaling, leading to sustained or amplified GmNINs expression and consequently a second round of GmSBT1.2b/1.2d activation. Thus, the additional phosphorylation sites on GmNFR1α targeted by GmPSKR1a likely function to propagate this second wave of signaling, creating a positive feedback loop that reinforces and amplifies the nodulation response (Fig. 9). We propose that this feedforward mechanism is important for reinforcing the plant cells’ commitment to symbiosis and for the synchronized development of the nodule. Therefore, our study provides a mechanistic explanation for the earlier observations in L. japonicus and elevates the understanding of PSK in symbiosis from a positive regulator to an integral component of the NFR activation complex. Future work should explore whether this crosstalk paradigm is conserved across legume species and how the PSK-NFR module might be engineered to optimize symbiotic efficiency in agricultural contexts. A proposed working model for the phytosulfokine signaling module activates the Nod factor receptor to control soybean nodulation. Under inoculation conditions, the expression of GmSBT1.2s and GmPSK4s is induced by rhizobia. GmNIN1a directly binds to and activates the expression of GmSBT1.2b/1.2d, which subsequently cleaves GmPSK4a to generate the mature PSK peptide. Upon binding to its receptor GmPSKR1a, the mature PSK peptide facilitates the interaction between GmPSKR1a and GmNFR1α, thereby enhancing the phosphorylation activity of GmNFR1α toward GmGEF2a, ultimately regulating soybean nodulation. Methods Plant growth and rhizobial inoculation Soybean Glycine max (L.) Merrill cv. Williams 82 (W82) plants were used for gene cloning, gene expression and functional analysis. Soybean seedlings were grown in a growth room (16 h light/8 h dark, 25 °C, 50% relative humidity). The plants were transplanted to pots containing vermiculite. The plants were grown for 7 days to allow rooting and then were inoculated with 30 mL B. diaefficiens strain USDA110 suspended in distilled water (OD600 = 0.08). Roots, stems, leaves, and nodules were harvested and assayed at the indicated time points. Nodule phenotypes were evaluated at 21 DAI. Vector construction For the proGmSBT1.2b:GUS and proGmPSK4a:GUS constructs, the promoter of GmSBT1.2b (2906 bp upstream of the ATG translational start codon) and GmPSK4a (3010 bp upstream of the ATG translational start codon) were amplified and cloned into pCAMBIA1391 vector between Hind III and BamH I sites. For the 35S:GmSBT1.2a-GFP, 35S:GmSBT1.2b-GFP, 35S:GmPSK4a-GFP and 35S:GmPSKR1a-GFP constructs, the coding sequences were amplified and cloned into pCAMBIA1300 vector between Xba I and BamH I sites. For the 35S:GmSBT1.2b-HA and 35S:GmPSK4a-HA constructs, the coding sequences were amplified and cloned into pTF101 vector using Mlu I and Spe I sites. For the UBI:GmPSKR1a-4×MYC construct, the coding sequence were amplified and cloned into the UBI:4×MYC vector using Xba I and Kpn I sites. For the UBI:GmPSKR1a-C-4×MYC and UBI:GmNFR1α−4×MYC constructs, the coding sequences were amplified and cloned into the UBI:4×MYC vector using Stu I site. RNA interference (RNAi) technology was used to knockdown GmSBT1.2a/1.2c, GmSBT1.2b/1.2d, GmSBT1.2a/1.2b/1.2c/1.2d, GmPSK4a/4b and GmPSKR1a/1b/1c, which has been widely used in downregulating mRNA activity. Soybean hairy root transformation and B. diazoefficiens inoculation assays For the hairy root transformation, the constructs were transformed into Agrobacterium rhizogenes K599 for soybean hairy root transformation to get composite plants as previously described in ref. 47. For nodulation assays, transgenic composite plants were transplanted to pots containing vermiculite. The plants were grown for 7 days to allow rooting and then were inoculated with 30 mL B. diaefficiens strain USDA110 suspended in distilled water (OD600 = 0.08). Roots and nodules were harvested and assayed at indicated time points. Generation of transgenic plants For CRISPR-Cas9 vector construction, the pYLCRISPR/Cas9-DB vector was used as previously reported, and small guide RNAs were designed using CRISPR-P2.0 software (http://crispr.hzau.edu.cn/CRISPR2/)48. Two sgRNAs targeting both GmSBT1.2b and GmSBT1.2d and two sgRNAs targeting GmPSK4a and GmPSK4b were selected. These sgRNAs were individually integrated into different single-guide DNA expression cassettes and then introduced into the pYLCRISPR/Cas9-DB. CRISPR-Cas9-edited mutants were generated by the Agrobacterium-mediated cotyledon-node transformation system following previously described approaches49,50. Briefly, seeds were kept in a vacuum desiccator with chlorine gas for 14 h and then placed in sterile water overnight. The seed coat was gently removed, and two cotyledon-node explants were generated by making a horizontal slice through the hypocotyl and a vertical slice between the cotyledons. Explants were immersed in the inoculum of Agrobacterium tumefaciens EHA101 for 8 h with slow horizontal shaking, and then dried on filter paper and transferred to co-cultivation plates for 4 days at 25 °C in darkness. Following co-cultivation, explants were cultured on shoot-initiation medium, with the hypocotyl and cotyledon nodes embedded in the medium under a 12 h light/12 h dark photoperiod at 25 °C and subcultured biweekly to fresh medium. Explants with buds were transferred to shoot-elongation medium and subcultured biweekly in the same conditions until shoots elongated. Elongated shoots (>3 cm) were transferred to rooting medium and the rooted putative transgenic plants were then transplanted to vermiculite for further growth in a growth room. DNA extraction and validation of transgenes Plant samples harvested from the composite plants were used for DNA extraction using the CTAB method as described previously in ref. 51. Briefly, the hairy roots were ground in the isolation buffer (2% hexadecyltrimethylammonium bromide [CTAB], 1.4 M NaCl, 0.2% 2-mercaptoethanol, 20 mM EDTA, 100 mM Tris-HCl, pH8.0) and incubated at 60 °C for 30 min, DNA was then extracted with chloroform-isoamyl alcohol (24:1), after centrifugation (12,000 × g for 10 min), the aqueous phase was transferred to clean tubes and the DNA was precipitated by ethanol. The stable-transgenic mutant Gmsbt1.2b sbt1.2d and Gmpsk4a psk4b plants were first screened by leaf painting with 0.8% w/v glufosinate solution, and glufosinate-resistant plants were confirmed by PCR detection of the BAR gene. PCR-positive plants were sequenced with gene-specific primers (Suppl. Data 1). Transgenic hairy roots transformed with 35S:GmSBT1.2a-GFP, 35S:GmSBT1.2b-GFP, 35S:GmPSK4a-GFP, 35S:GmPSKR1a-GFP, UBI:GmNFR1α−4×MYC or the empty vector (with GFP reporter in another cassette) were screened using a portable fluorescence lamp to visualize GFP green fluorescence, and verified by gene expression. RNA extraction and quantitative PCR analysis Total plant RNAs were extracted using TRIpure Reagent (Aidlab Biotechnologies Ltd., Beijing, China). Genomic DNAs were removed by gDNA Wiper Mix (Vazyme Biotech Co., Ltd., Nanjing, China). cDNAs were produced using a FastQuant RT Kit (Vazyme Biotech Co., Ltd.). qRT-PCR analyses were performed using SuperReal PreMix Plus (Vazyme Biotech Co., Ltd.). The primers used in qRT-PCR are listed in Suppl. Data 1. Histochemical analysis The infected hairy roots and nodules of soybean composite plants expressing proGmSBT1.2b:GUS and proGmPSK4a:GUS were stained with X-Gluc solution [50 mM NaH2PO4, 50 mM Na2HPO4, 0.5 mM K4(Fe(CN)6), 0.5 mM K3(Fe(CN)6), 1 mg mL−1 X-Gluc and 0.1% (v/v) Triton X-100] to test β-glucuronidase activity at the specified time points. Samples were incubated in staining buffer at 37 °C for 4 h for roots and 7 h for nodules, and then washed three times with 75% (v/v) ethanol. GUS assay in transcriptional regulation The 35S:GmNIN1a-GFP and 35S:GFP constructs were individually co-expressed with proGmSBT1.2b:GUS in N. benthamiana leaves by infiltrating bacterial suspensions (OD600 = 0.3) in different combinations. Two days later, the leaves were stained with X-Gluc solution buffer at 37 °C for 4 to 6 h, and then washed three times with 75% (v/v) ethanol. Quantitative GUS activity was measured by monitoring the conversion of 4-methylumbelliferyl-β-D-glucuronide to 4-methylumbelliferone (4-MU) as described previously in ref. 52. Fluorescence was detected using a SpectraMax i3x Multi-Mode Microplate Reader with excitation at 365 nm and emission at 455 nm. Protein concentrations of the tissue extracts were determined using the Quick Start Bradford 1× Dye Reagent Protein Assay (Bio-Rad, CA, USA) following the manufacturer’s instructions. GUS activity was expressed as picomoles of 4-MU produced per minute per milligram of protein. Nitrogenase activity Nitrogenase activity of nodules was determined using the acetylenereduction-activity assay as previously described in ref. 53. One nodulated root was placed inside a sealed 40 mL vial. Of the air inside the vial, 2 mL was removed and then replaced with 2 mL pure acetylene gas under water. Vials were placed upside down and incubated at 28 °C for 2 h. The acetylene-reduction reaction was stopped by incubating vials at 4 °C for 5 min. One hundred microlitres of the gas in each vial was used for the measurement of ethylene content using a GC-4000A gas chromatograph (Dongxi). Nitrogenase activity was calculated by the amount of ethylene reduced per plant per unit of time. Synthesis of PSK peptide The PSK peptide used in this experiment was synthesized by Shanghai Sangon Biotech Co., Ltd. The small peptide was dissolved in sterile water to prepare a 1 M stock solution and stored at −80 °C for later use. The peptide sequence is YSO3IYSO3TQ, with a purity of over 95%. Yeast two-hybrid assay The full-length coding sequences of GmPSKR1a and the C-terminal coding sequences of GmNFR1α were amplified and the PCR products were cloned into the entry vector pDONR207 by BP reactions. pDONR207 harboring these genes were then ligated into pGBKT7 (BD) or pGADT7 (AD) by LR reactions. Yeast-two hybrid (Y2H) assays were done according to the Matchmaker GAL4 Two-Hybrid System (Clontech, Mountain View, CA). To verify the interactions between GmPSKR1a and GmNFR1α, the above constructs were co-transformed into Saccharomyces cerevisiae AH109. Clones were grown on selective medium (lacking Trp, Leu, and His) to test protein-protein interaction. Images were taken 3 d after incubation at 30 °C. BiFC assay The coding sequences of GmPSKR1a and GmNFR1α were cloned into the pEARLYGAYE202-YC and pEARLYGAYE201-YN through the Gateway reaction system (Invitrogen, Carlsbad, CA), respectively. The constructs were transformed into Agrobacterium tumefaciens strain GV3101. Transformed strains were collected and resuspended in infiltration buffer (10 mM MgCl2, 200 µM acetosyringone, pH 5.6) and incubated for 2–3 h at 30 °C before infiltration. Suspensions (OD600 = 0.2) were infiltrated into N. benthamiana leaves in different combinations. Two days after infiltration, the leaves were treated with 10 μM PSK for 30 min, followed by the visualization of YFP fluorescence was detected using a Leica confocal laser scanning microscope (Leica Microsystems). Co-IP assay The GmPSKR1a-MYC construct was co-expressed with GmNFR1α-C-Flag construct in N. benthamiana leaves by infiltrating bacterial suspensions (OD600 = 0.3) in different combinations. Two days later, the leaves were injected with 10 μM PSK or 10 mM MgCl2 30 min before sampling. Tobacco total proteins were extracted in extraction buffer (50 mM Tris-MES, pH 8.0, 0.5 M Sucrose,1 mM MgCl2, 10 mM EDTA, 1% NP40, 1 mM PMSF and 1× protease inhibitor cocktail) and then were incubated with 8 µL anti-MYC agarose beads (SIGMA, A7470) for 2 h at 4 °C. The beads were washed five times with extraction buffer, and the immunoprecipitated proteins were eluted with SDS loading buffer with boiling for 5 min for western blotting. Samples were analyzed by western blot with anti-flag (ABclonal, AE005, 1:2500) and anti-MYC (SIGMA, C3956, 1:2000). ChIP-qPCR assay Two grams of transgenic roots expressing 35S:GmNIN1a-GFP and 35S:GFP were used for ChIP assays. Roots were cross-linked with 1% formaldehyde at 4 °C for 10 min and neutralized with 0.125 M glycine. Roots were ground to fine power in liquid nitrogen and the nuclei were isolated. Immunoprecipitation (IP) was performed using anti-GFP antibody (ab290, Abcam). Chromatin precipitated with lgG antibody (ab171870, Abcam) served as the negative control, while chromatin isolated prior to IP was used as the input control. qPCR was conducted to analyze the precipitated chromatin. The specific primers used in this experiment are listed in Supplementary Data 1. In vivo cleavage assay and IP-MS/MS analysis The 35S:GmSBT1.2a-HA, 35S:GmSBT1.2b-HA and pTF101 constructs were individually co-expressed with 35S:GmPSK4a-GFP in N. benthamiana leaves. Tobacco total proteins were extracted in extraction buffer (50 mM Tris-MES, pH 8.0, 0.5 M Sucrose, 1 mM MgCl2, 10 mM EDTA, 1% NP40, 1 mM PMSF and 1× protease inhibitor cocktail). The samples were loaded on 15% SDS-PAGE and separated at 150 V in Invitrogen Novex vertical gel tanks and were analyzed by western blot with anti-GFP (Abcam, ab290, 1:2500) and anti-HA (SIGMA, SAB4300603, 1:2500). The GmPSK4a cleavage sites were determined by IP-MS/MS. GmSBT1.2b-HA and GmPSK4a-GFP were co-expressed in N. benthamiana and the total protein extract was incubated with GFP affinity beads for 3 h at 4 °C. The purified proteins were then separated by SDS-PAGE, and the specific cleaved products were excised and analyzed by LC-MS/MS to identify the cleavage site (n = 1). The samples were separated using a nano-flow HPLC system (nanoElute). After chromatographic separation, mass spectrometry analysis was performed on a timsTOF Pro mass spectrometer. All raw mass spectrometry files were analyzed with MaxQuant (v1.6.14) using the Andromeda search engine against the Glycine max UniProtKB reference proteome (uniprotkb_Glycine_max_85158_20260128). Search parameters included: trypsin digestion with a maximum of two missed cleavages; fixed carbamidomethylation of cysteine (+57.0215 Da); variable modifications of methionine oxidation (+ 15.9949 Da) and protein N-terminal acetylation (+42.0106 Da); 20 ppm precursor mass tolerance and 0.1 Da fragment mass tolerance; 1% FDR for both peptides and proteins. iBAQ quantification was enabled, with all other parameters set to default values. The raw data and search parameters have been submitted to the ProteomeXchange Consortium (dataset identifier: PXD076049). In vitro cleavage assay and LC-MS/MS analysis The coding sequences of GmPSK4a and EPI10 were cloned into the His and pMal-c2X (MBP) vector. The GmPSK4a fused with His and EPI10 fused with MBP were expressed in E. coli strain BL21 cells at 37 °C for 3 h and induced by 1 mM Isopropyl β-D-1-thiogalactopyranoside (IPTG). GmPSK4a-His and MBP-EPI10 were purified using High Affinity Ni-NTA Resin (GenScript) or amylose resin (New England Biolabs), respectively, according to the manufacturer’s instructions. The in vitro cleavage assay was performed as described previously in ref. 54. Fifty µg GmPSK4a-His in a total volume of 100 µl 25 mM Na2HPO4/NaH2PO4, pH 6.5, 1 mM CaCl2, 10 µM PepstatinA, were digested during 2 h at 25 °C with 5 µg GmSBT1.2b-HA or pTF101 or GmSBT1.2b-HA and EPI10. Samples were detected by western blotting with anti-His (Sigma, SAB4301134, 1: 2,500), anti-HA (Sigma, SAB4300603, 1: 2,500), and anti-MBP (ABclonal, AE016, 1: 3,000) antibodies. The cleaved products were excised and analyzed by LC-MS/MS to identify the cleavage site (n = 1). The software used for analysis and settings employed for peptide identification were same as in vivo, and the raw data and search parameters have been submitted to the ProteomeXchange Consortium (dataset identifier: PXD076048). In vitro phosphorylation assays and LC-MS/MS analysis The in vitro phosphorylation assay was performed as previously described in ref. 55 with slight modifications. Recombinant GST-GmNFR1α-C (1 µg) and His-GmPSKR1a-C (3 µg) proteins were purified by using PierceTM Glutathione Agarose (Thermo Fisher Scientific) and High Affinity Ni-NTA Resin (GenScript). Recombinant GmNFR1α-C and GmPSKR1a-C were incubated in the reaction buffer (50 mM HEPES, pH 7.4, 10 mM MgCl2, 2.5 mM MnCl2, 25 μM ATP, 1 mM DTT, and 1 µCi of [γ-32P] ATP) at 28 °C for 30 min. The reactions were terminated by adding 5 × SDS loading buffer followed by boiling for 10 min. Proteins were separated by 12% SDS-PAGE, and phosphorylation was detected by autoradiography using FUJI FLA5100 (Fujifilm, Tokyo, Japan). The SDS-PAGE gels were stained with Coomassie blue as loading controls. In vitro phosphorylation sites were identified by LC-MS/MS analysis as previously described in ref. 31. The target protein bands were excised from a Coomassie brilliant blue-stained SDS-PAGE gel for the control sample (His with GST-GmNFR1α-C) and the experimental sample (His-GmPSKR1a-C with GST-GmNFR1α-C). The excised bands were subjected to in-gel digestion with trypsin, and the resulting peptides were analyzed by LC-MS (n = 1). The samples were separated using a nano-flow HPLC system (nanoElute). After chromatographic separation, mass spectrometry analysis was performed on a timsTOF Pro mass spectrometer. All raw mass spectrometry files were analyzed with MaxQuant (v1.6.14) using the Andromeda search engine against the Glycine max UniProt reference proteome (uniprot_Glycine_max_85143_20240207). Search parameters included: trypsin digestion with a maximum of two missed cleavages; fixed carbamidomethylation of cysteine (+57.0215 Da); variable modifications of methionine oxidation (+15.9949 Da) and serine/threonine/tyrosine phosphorylation (+ 79.9663 Da); 20 ppm precursor mass tolerance and 0.1 Da fragment mass tolerance; 1% FDR for both peptides and proteins; and a minimum phosphorylation site localization probability of 0.75. iBAQ quantification was enabled, with all other parameters set to default values. The above raw mass spectrometry data and search parameters have been submitted to the ProteomeXchange Consortium (dataset identifier: PXD076050). Statistical analysis Data was analyzed using GraphPad Prism 8 (GraphPad Software, Inc., La Jolla, CA). Statistical significance was determined by ANOVA with Tukey’s post hoc test or the two-tailed Student’s t-test. The type of statistical significance test was described in the figure legends. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary information Description of Additional Supplementary Files Supplementary Data 1 Reporting Summary Transparent Peer Review file Source data Source Data Acknowledgements We thank Prof. Ertao Wang from Institute of Plant Physiology and Ecology, Shanghai Institutes for Biological Sciences, for providing the GmNFR1ɑ-C-GST. Author contributions Z.W. and X.L. designed the experiments. J.J.L., J.C., K.J., F.Z., Y.P., Y.Z., M.L., X.D, W.Y., Y.G., and J.Y.L. performed the experiments and analyzed the data; J.J.L., Z.W., and X.L. wrote the manuscript. Peer review Peer review information Nature Communications thanks Takuya Suzaki and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Funding This work was supported by the National Natural Science Foundation of China (32090062, 32472087, 32330078), the National Key Research and Development Program of China (2023YFD1200604), Natural Science Foundation of Hubei Province Project (2025AFB582) and the Fundamental Research Funds for the Central Universities (X2662024ZKPY003). Data availability All data are available in the main text and/or supplementary materials. The source data for Figs. 1–8, Supplementary Figs. 2–6, 9–13, 15 are provided as a Source Data file. The raw and processed data of mass spectrometric identification of GmPSK4a-derived peptides in vivo (accession code: PXD076049) and in vitro (accession code: PXD076048) have been deposited to the ProteomeXchange Consortium via the iProX partner repository. The raw and processed data of GmNFR1α phosphorylation sites have also been uploaded to iProX (accession code: PXD076050). Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Jingjing Lu, Jiahuan Chen. Contributor Information Xia Li, Email: xli@mail.hzau.edu.cn. Zhijuan Wang, Email: wangzj@mail.hzau.edu.cn. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-74586-9. References Associated Data Supplementary Materials Supplementary information Description of Additional Supplementary Files Supplementary Data 1 Reporting Summary Transparent Peer Review file Source Data Data Availability Statement All data are available in the main text and/or supplementary materials. The source data for Figs. 1–8, Supplementary Figs. 2–6, 9–13, 15 are provided as a Source Data file. The raw and processed data of mass spectrometric identification of GmPSK4a-derived peptides in vivo (accession code: PXD076049) and in vitro (accession code: PXD076048) have been deposited to the ProteomeXchange Consortium via the iProX partner repository. The raw and processed data of GmNFR1α phosphorylation sites have also been uploaded to iProX (accession code: PXD076050). Source data are provided with this paper.