Rewiring an E3 ligase enhances cold resilience and phosphate use in maize

Abstract

Cold stress restricts plant growth and inorganic phosphate (Pi) uptake, reducing yield and increasing fertilizer demand 1 , 2 , 3 . Enhancing both cold tolerance and phosphorus use efficiency (PUE) is crucial for sustainable crop productivity. Here we identify the SPX-domain-containing E3 ubiquitin ligase NITROGEN LIMITATION ADAPTATION (NLA) as a central regulator that links cold signalling to Pi homeostasis in maize (Zea mays L.). Under cold conditions, NLA promotes the degradation of the transcriptional repressor JAZ11, activating jasmonate signalling to enhance cold tolerance; however, NLA also simultaneously represses Pi uptake, through inositol polyphosphate (InsP)-dependent ubiquitination of the Pi transporter PT4. A ubiquitinome-informed genome-wide association study identified a natural PT4(K267A) (lysine-to-alanine substitution) variant that attenuates NLA-mediated degradation and increases Pi uptake in cold conditions. To overcome this nutrient–stress trade-off, we combined artificial-intelligence-guided structural modelling and ligand docking with genome editing to generate the nla Δ12 allele, which encodes an NLA variant in which binding to InsP is impaired but JAZ11 targeting is retained. The Δ12 modification selectively redirects the activity of NLA towards jasmonate signalling, resulting in improved cold resilience, higher PUE and increased yield in multi-site field trials. These findings reveal a tunable SPX regulatory module that integrates environmental and nutrient signals, and provide a molecular framework for engineering climate-resilient, nutrient-efficient crops.

Main

Maize (Zea mays L.), the most widely cultivated cereal crop globally, originates from tropical regions and is highly susceptible to cold stress 4 , 5 . During long-term adaptation, maize has evolved diverse physiological strategies to mitigate cold-induced damage, including the expansion of cold-responsive gene families and the selective retention of naturally occurring cold-tolerant alleles 6 , 7 . Although these adaptations have improved cold tolerance, they are frequently accompanied by compromised nutrient uptake and stunted growth, often leading to reduced yield potential 3 , 8 , 9 . Addressing these trade-offs requires a deeper understanding of the molecular mechanisms that coordinate stress resilience with nutrient use efficiency in maize.

Phosphorus (P) is an essential macronutrient involved in energy metabolism and developmental processes, absorbed mainly in the form of inorganic phosphate (Pi) 2 , 10 . Cold stress markedly reduces Pi availability by impairing its mobility in soil and inhibiting root activity, often causing physiological P deficiency even under P-sufficient conditions 1 , 11 . In parallel, global P resources are depleting owing to inefficient fertilizer use and the non-renewable nature of phosphate rock, raising concerns about long-term agricultural sustainability 12 , 13 , 14 , 15 . Developing climate-resilient crop varieties with enhanced PUE is thus crucial for reducing dependence on fertilizers while maintaining yield stability.

Plants have evolved intricate Pi sensing and signalling networks to optimize uptake, allocation and homeostasis 16 , 17 . Central to these systems are SPX (for SYG1, Pho81 and XPR1) domain-containing proteins, which sense intracellular Pi levels through binding to InsPs 18 , 19 . SPX proteins are classified on the basis of their domain architecture into four subfamilies: SPX, SPX-MFS (major facilitator superfamily), SPX-RING (really interesting new gene) and SPX-EXS (ERD1, XPR1 and SYG1). These mediate distinct Pi-responsive functions, including protein–protein interactions, transport and catalytic capacities, thereby enabling plants to adapt to fluctuating Pi availability 20 , 21 , 22 , 23 , 24 , 25 . Although SPX proteins have a well-known function in regulating Pi homeostasis, their role in integrating cold tolerance with PUE remains mostly unclear.

In this study, we show that the SPX-RING-type E3 ubiquitin ligase NITROGEN LIMITATION ADAPTATION (NLA) 26 functions as a central node linking cold tolerance and P acquisition in maize. It enhances cold tolerance by targeting the jasmonate (JA) signalling repressor JAZ11 for degradation, while concurrently restricting Pi uptake through destabilization of the transporter PT4. Here, using artificial intelligence (AI)-assisted structural prediction, we engineered a truncated variant, NLA Δ12 , which disrupts the binding of InsPs within the SPX domain. This variant retains the ability to target JAZ11 turnover, but leads to substantially reduced degradation of PT4, effectively dissociating cold tolerance and Pi suppression. Field trials show that NLA Δ12 enhances cold tolerance, PUE and grain yield. Our study represents an effective strategy for the development of maize varieties that are adapted to cold and nutrient-limited environments.

NLA promotes cold tolerance in maize

To identify regulators that coordinate cold tolerance and Pi homeostasis, we performed BLAST searches and phylogenetic analyses of SPX-domain-containing proteins in maize, rice and Arabidopsis, and identified 16 SPX family members in maize (Extended Data Fig. 1a,b and Supplementary Table 1). Functional screening of overexpression and CRISPR–Cas9 knockout transgenic lines highlighted NLA, which encodes an SPX-RING-type E3 ubiquitin ligase (Fig. 1a), as a key regulatory candidate. Two independent knockout alleles, nla-1 and nla-2, were generated using CRISPR–Cas9 (Extended Data Fig. 1c). Both mutants exhibited normal growth phenotypes in non-stressed conditions (Fig. 1b and Extended Data Fig. 1d), but showed pronounced cold sensitivity, with a markedly reduced uninjured leaf area compared with the wild-type inbred line LH244 (Fig. 1b,c). F 1 hybrids between the two alleles showed similar cold sensitivity (Extended Data Fig. 1e,f), confirming that NLA has an essential role in cold tolerance.

a, Predicted structural model of NLA generated using AlphaFold3. aa, amino acids. b, c, Representative chilling phenotypes (b) and quantification of uninjured leaf area (c) of wild-type (WT) (LH244) and nla-mutant seedlings after treatment at 4 °C. Scale bars, 5 cm. d, Pi uptake rate in the maize inbred lines 4019, B73 and B111 under optimal (25 °C) and cold (8 °C) conditions for the indicated times. e, Total shoot P content across 30 diverse maize inbred lines after 2 weeks of treatment at 25 °C and 8 °C. Box plots show the median (centre line), 25th and 75th percentiles (box limits), and minimum and maximum values (whiskers). f, Shoot P content in wild type and nla mutants exposed or not to a temperature of 8 °C for 20 days. g, h, Immunoblot analysis (g) and quantification (h) of NLA–GFP accumulation in seedlings exposed or not to cold for the indicated times. Actin was used as a loading control. In c, d, f, h, data are mean ± s.d. n = 3 in c, d, f, h and n = 30 in e, independent biological replicates. In c–e, h, significance was determined using two-tailed Student’s t-tests (*P < 0.05, **P <0.01, ***P < 0.001). In f, different letters denote significant differences (P < 0.05) from a Duncan’s new multiple range test. For exact P values, see Source Data.

We next assessed whether NLA modulates Pi acquisition in cold conditions. A nutrient depletion assay revealed that exposure to 8 °C substantially impaired Pi uptake in three maize inbred lines (4019, B73 and B111), compared with 25 °C controls (Fig. 1d). Consistently, shoot P content decreased markedly across 30 diverse inbred lines after 20 days of cold treatment at 8 °C (Fig. 1e), indicating a general suppression of Pi acquisition under cold stress. Notably, nla mutants accumulated higher total levels of P than did wild-type plants at both temperature regimes, and exhibited a mitigated decline in P levels when exposed to cold (Fig. 1f and Extended Data Fig. 1g), suggesting that NLA negatively regulates Pi uptake during cold stress.

Expression analysis revealed a dual mode of NLA regulation by cold. At the transcriptional level, NLA transcripts showed a modest but progressive induction under cold conditions (Extended Data Fig. 1h). Notably, in NLA-GFP transgenic lines driven by the Ubiquitin promoter, NLA protein abundance was markedly stabilized at 4 °C but not at 25 °C (Fig. 1g,h and Extended Data Fig. 1i), demonstrating post-translational regulation. Collectively, these findings establish NLA as a cold-responsive regulator that enhances cold tolerance while suppressing Pi acquisition under cold conditions in maize.

NLA regulates JA genes for cold response

To investigate how NLA confers cold tolerance, we first examined its enzymatic activity. In vitro ubiquitination assays showed that NLA functions as an active E3 ubiquitin ligase in the presence of Arabidopsis E1 (UBA1), human E2 (UBCH5B) and ubiquitin (Extended Data Fig. 1j).

We next performed integrative transcriptomic and ubiquitinomic profiling of wild type and nla mutants that were exposed, or not, to 4 °C conditions for 12 h (Extended Data Fig. 2a). Cold treatment triggered 2,610 differentially expressed genes (DEGs) and 3,756 differentially ubiquitinated peptides (DUPs) (Extended Data Fig. 2b–e and Supplementary Tables 2 and 3). Functional enrichment analyses revealed prominent regulation of pathways related to JA signalling, ion transport, glutathione metabolism and protein ubiquitination (Extended Data Fig. 2f,g).

Among these, 15 signalling pathways were co-regulated by NLA at both the transcriptional and the ubiquitination level (Supplementary Table 4), with JA signalling emerging as one of the most significantly enriched (Extended Data Fig. 3a,b). Given the well-established role of JA in promoting cold tolerance 27 , 28 , 29 , we intersected NLA-dependent, cold-responsive genes with known JA-responsive targets, identifying 616 co-regulated genes (Extended Data Fig. 3c and Supplementary Table 5). These included key components of JA biosynthesis and signalling 30 , 31 , such as ALLENE OXIDE SYNTHASE (AOS), LIPOXYGENASE (LOX), BENZOXAZINELESS (BX) and JASMONATE ZIM-DOMAIN (JAZ) (Extended Data Fig. 3d). Functionally, the lox8 mutants showed enhanced cold sensitivity (Extended Data Fig. 3e–g), whereas BX11-overexpressing lines exhibited improved cold tolerance (Extended Data Fig. 3h–j). These results suggest that NLA promotes cold tolerance by activating JA-responsive genes under cold stress.

NLA targets JAZ11 in JA signalling

To elucidate how NLA regulates JA signalling under cold conditions, we performed a yeast two-hybrid screen and identified 36 putative NLA-interacting proteins (Supplementary Table 6). Among these, JAZ11, a known repressor of JA signalling 32 , 33 , was selected for further analysis (Fig. 2a). The interaction between NLA and JAZ11 was validated by in vitro pull-down and co-immunoprecipitation (co-IP) assays in maize protoplasts (Fig. 2b,c). JAZ11 encodes a clade III JAZ protein that shares high sequence identity with JAZ25 but low similarity with JAZ13 in maize (Extended Data Fig. 4a). Yeast two-hybrid assays revealed that NLA interacted with both JAZ11 and JAZ25, but not with JAZ13, and that this interaction requires the N-terminal domain of JAZ11 (Extended Data Fig. 4b,c).

a, Yeast two-hybrid assay showing the interaction between NLA and JAZ11. Empty vector (EV) was used as a negative control. Yeast was grown on −LW (−Leu and −Trp) or −LWHA (−Leu, −Trp, −His and −Ade) synthetic dropout medium. Yeast cells were co-transformed with the indicated pairwise combinations of pGADT7 and pGBKT7 fusion constructs. b, c, Validation of NLA–JAZ11 interaction by in vitro pull-down (b) and in vivo co-IP assays (c). GST, glutathione S-transferase; MBP, maltose-binding protein. d, e, NLA ubiquitinates JAZ11 in vitro (d) and in maize protoplasts (e). f, g, Cell-free degradation assay (blot (f) and quantification (g)) showing that NLA promotes JAZ11 protein turnover. h, i, Immunoblot analysis (h) and quantification (i) of JAZ11–GFP protein abundance in the roots of JAZ11-GFP plants exposed or not exposed to a temperature of 4 °C for the indicated times. j, k, JAZ11 protein levels (blot (j) and quantification (k)) in the roots of JAZ11-GFP and nlaJAZ11-GFP plants after 12 h of cold treatment. l, m, Representative chilling phenotypes (l) and quantification of uninjured leaf area (m) of nla, jaz11 and nla jaz11 double mutants after 4 °C treatment. Scale bars, 5 cm. In b–e, three independent experiments were performed with similar results. In g, i, k, m, data are mean ± s.d. (n = 3 independent biological replicates). In g, i, k, significance was determined using two-tailed Student’s t-tests (*P < 0.05, **P < 0.01); NS, not significant. In m, different letters denote significant differences (P < 0.05) from a Duncan’s new multiple range test. For exact P values, see Source Data.

We next examined whether JAZ11 is a direct ubiquitination substrate of NLA. In vitro ubiquitination assays showed that NLA ubiquitinated JAZ11 in the presence of E1, E2 and ubiquitin (Fig. 2d and Extended Data Fig. 4d). In vivo ubiquitination assays, performed by co-expressing NLA–MYC and JAZ11–GFP with Flag–TUBE (a trypsin-resistant tandem ubiquitin-binding entity) 34 in maize protoplasts, revealed a strong accumulation of polyubiquitinated JAZ11 in the presence of the 26S proteasome inhibitor MG132 (Fig. 2e). Consistently, cell-free degradation assays showed that JAZ11 degraded more rapidly in wild-type extracts than in nla-mutant ones, and this degradation was abolished by treatment with MG132 (Fig. 2f,g). Moreover, increasing NLA–MYC expression progressively reduced the abundance of JAZ11–GFP in maize protoplasts (Extended Data Fig. 4e,f). To further examine the effects of NLA on JAZ11 stability, we generated transgenic lines overexpressing JAZ11-GFP (JAZ11-OE) in wild-type and nla-mutant backgrounds. Immunoblot analysis showed that the levels of JAZ11 protein declined progressively from 1 h to 12 h under cold treatment in JAZ11-GFP plants (Fig. 2h–k), whereas this reduction was markedly less pronounced in the nla-mutant background (Fig. 2j,k). Collectively, these results show that NLA mediates the cold-induced degradation of JAZ11.

We next analysed the role of JAZ11 in cold tolerance using JAZ11-OE lines and CRISPR–Cas9-derived knockout mutants. JAZ11-OE plants were hypersensitive to cold, whereas jaz11 mutants exhibited enhanced cold tolerance (Extended Data Fig. 4g–l). Furthermore, nla jaz11 double mutants phenocopied the jaz11 single mutant (Fig. 2l,m), placing JAZ11 downstream of NLA in the cold-response pathway. Cold-induced expression of JA-responsive genes, including AOS2b, LOX8, LOX13 and BX10–BX12, was markedly suppressed in JAZ11-OE plants (Extended Data Fig. 4m), supporting the role of JAZ11 as a negative regulator of JA signalling and cold tolerance.

We further investigated natural variation in JAZ11 across 105 maize inbred lines, and identified a single-nucleotide polymorphism (SNP)—G to A at position 379—that was highly associated with cold tolerance (Extended Data Fig. 5a). This SNP defines 2 haplotypes: JAZ11 379A (7 lines) and JAZ11 379G (98 lines). The rare jaz11 379A allele conferred stronger cold tolerance than did jaz11 379G (P < 0.05), with the reference inbred line LH244 belonging to the latter group (Supplementary Table 7). The G-to-A SNP results in an R47K substitution within the N-terminal domain, which is crucial for interaction with NLA (Extended Data Fig. 4c). Because lysine residues often serve as ubiquitination sites 35 , we hypothesized that K47 enhances ubiquitination. Indeed, in vitro mass spectrometry confirmed the ubiquitination of JAZ11(K47) (Extended Data Fig. 5b). JAZ11(K47) also exhibited faster degradation than did JAZ11(R47) in cell-free degradation assays (Extended Data Fig. 5c,d). Furthermore, NLA promoted the degradation of JAZ11(K47) more effectively than that of JAZ11(R47) in maize protoplasts (Extended Data Fig. 5e,f). Together, these findings show that JAZ11(K47) enhances its turnover through NLA-mediated ubiquitination, thereby strengthening JA signalling and promoting cold tolerance in maize.

NLA targets PT4 for Pi uptake in cold conditions

To identify downstream targets of NLA that are involved in Pi regulation under cold stress, we performed a Gene Ontology (GO) enrichment analysis of NLA-regulated, cold-responsive proteins from our ubiquitinome dataset (Fig. 3a). This analysis revealed a significant enrichment of terms related to the transmembrane transport of nutrients. The putative substrates included several transporter proteins, including phosphate transporters (PT2, PT4, PT7 and so on), potassium transporters (for example, HAK5) and calcium transporters (for example, CAP2), with phosphate transporters emerging as the most promising candidates (Fig. 3a).

a, Enrichment map of NLA- and cold-regulated peptides, highlighting GO terms related to transmembrane transport. NLA-regulated DUPs are enriched in phosphate, potassium and calcium transport pathways; representative proteins are shown. The full list of candidates is provided in Supplementary Table 4. Each circle represents a significantly enriched annotation group. b, Manhattan plot of GWAS for shoot P content after two weeks at 8 °C. The dotted line indicates the Bonferroni-adjusted genome-wide significance threshold (−log 10 (P) = 7). Candidate genes above the threshold are listed in Supplementary Table 8. PT4 is associated with SNP chr.1.s_11,410,259. c, Local association mapping and LD of PT4 in 149 maize inbred lines; significant SNPs are in red. d, e, PT4 haplotypes (d) and corresponding shoot P content (e) after two weeks at 8 °C. In e, box plots show the median (centre line), 25th and 75th percentiles (box limits) and minimum and maximum values (whiskers). n = 53 (hap1) and 83 (hap2). f, Predicted structure of PT4 showing three amino acid polymorphisms between haplotypes, including K267; ubiquitinated lysine residues are indicated. g, Relative ubiquitination levels of identified K-GG sites on PT4 in wild-type and nla mutants after 4 °C treatment, normalized to wild type (1.0). h, Co-IP showing in vivo NLA–PT4 interaction. i, Comparison of ubiquitination between PT4(A267) and PT4(K267) proteins in protoplasts co-expressing NLA–MYC. In h, i, three independent experiments were performed with similar results. j, k, PT4 protein abundance (blot (j) and quantification (k)) in PT4 K267 -GFP transgenic plants with or without 4 °C treatment. Data are mean ± s.d. (n = 3 independent biological replicates). l, Model showing how NLA modulates cold tolerance and Pi uptake by ubiquitinating JAZ11 and PT4 under cold stress. In e, k, significance was determined by two-tailed Student’s t-tests (*P <0.05, **P < 0.01). The illustration in f was created in BioRender; Yang, S. (2026).

To assess their functional relevance, we performed a genome-wide association study (GWAS) using shoot total P content as a quantitative trait in 192 diverse maize inbred lines subjected to 8 °C treatment for two weeks (Extended Data Fig. 6a). This analysis identified ten SNPs that were significantly associated with P content under cold conditions (P < 1.0 × 10 −7 ; Fig. 3b and Extended Data Fig. 6b). Notably, a synonymous G-to-C SNP at position 603 (chr1.s_11410259, P = 9.03 × 10 −8 ) was located within the coding region of PT4, a member of the PHT1 Pi transporter family (Fig. 3b and Supplementary Table 8). Subsequent resequencing identified seven additional SNPs within PT4 that were significantly associated with P content (positions 540, 549, 550, 554, 799, 800 and 807; P < 1.0 × 10 −5 ) (Fig. 3c and Supplementary Table 9). Linkage disequilibrium (LD) analysis grouped these variants into two major haplotypes (minor allele frequency (MAF) ≥ 0.05) (Fig. 3d). Inbred lines containing haplotype 2 (hap2) accumulated significantly higher levels of shoot P under cold stress than did those containing hap1 (P < 1.0 × 10 −4 ), with the LH244 inbred line belonging to hap1 (Fig. 3e).

Protein sequence alignment revealed that hap2-specific SNPs resulted in three amino acid substitutions: L184I and A185V in the third transmembrane domain, and K267A in the cytoplasmic region (Fig. 3f). K267 was identified as a ubiquitination site in our ubiquitinome dataset, and its ubiquitination was markedly reduced in nla mutants under cold conditions (Fig. 3f,g), suggesting that K267 of PT4 is a functional target of NLA. By contrast, L184I and A185V substitutions were predicted to have minimal effects on PT4 structure or function.

Given that NLA mediates the degradation of PHT1 family transporters in Arabidopsis and rice 16 , 25 , 36 , we next examined whether maize PT4 is regulated in a similar manner. Co-IP confirmed a physical interaction between NLA and PT4 (Fig. 3h). In vivo ubiquitination assays showed that NLA catalysed PT4 ubiquitination, with the hap2 variant (PT4(A267)) showing markedly reduced ubiquitination compared with the hap1 variant (PT4(K267)) (Fig. 3i). Consistently, PT4(K267) degraded more rapidly than PT4(A267) did in cell-free degradation assays (Extended Data Fig. 6c,d), and when co-expressed with NLA in maize protoplasts (Extended Data Fig. 6e,f). Moreover, cold treatment substantially accelerated the degradation of PT4(K267) in PT4 K267 -GFP transgenic plants (Fig. 3j,k). Together, these data suggest that NLA promotes the cold-induced degradation of PT4, thereby restraining Pi uptake under cold stress (Fig. 3i).

Despite the cold-induced suppression of P accumulation, no significant correlation between shoot P content and cold tolerance was observed across the 192 maize lines examined (Extended Data Fig. 6g–i and Supplementary Table 10). Consistent with this observation, the cold tolerance of transgenic lines overexpressing PT4 under the control of the Ubi promoter was indistinguishable from that of the wild type, despite the fact that the transgenic lines accumulated significantly higher levels of P after cold treatment (Extended Data Fig. 6j–m). Notably, PT4 expression itself was transcriptionally induced by cold stress (Extended Data Fig. 6n). These results suggest that cold tolerance and P accumulation are genetically separable traits in maize.

This observation prompted us to investigate whether the regulatory roles of NLA in cold tolerance and Pi homeostasis are mechanistically distinct. Co-IP assays showed that the interaction between NLA and PT4 was enhanced in an InsP 6 -dependent manner (Extended Data Fig. 6o,p). By contrast, the NLA–JAZ11 interaction was independent of InsP 6 or the SPX domain, as revealed by co-IP and yeast two-hybrid assays (Extended Data Fig. 6q–s). Collectively, these results suggest that NLA integrates cold tolerance and Pi regulation under cold stress through a dual mode of regulation, in which its E3 ligase activity promotes the degradation of JAZ11 to enhance cold tolerance, whereas its InsP-dependent Pi-sensing function mediates PT4 turnover to restrict Pi uptake.

nla Δ12 boosts JA signalling and Pi uptake

SPX-domain-containing proteins maintain phosphate homeostasis in eukaryotes by binding to InsPs mainly through α-helices 1, 2 and 5, which form the core of the InsP-binding pocket 19 , 37 , 38 . Although the structure of NLA has not been experimentally determined, AlphaFold3 modelling 39 combined with DiffDock-L docking 40 predicts that the maize NLA SPX domain comprises six α-helices, with InsPs forming polar contacts mainly with α1, α2 and α5 (Extended Data Fig. 7a), consistent with established SPX–InsP interactions 19 , 41 . Three conserved phosphate-binding cluster (PBC) residues (Y27, K31 and K149), together with K2 and K35, form the major InsP-binding determinants (Extended Data Fig. 7a).

To selectively weaken the InsP-dependent regulation of PT4 while preserving the interaction with JAZ11, we performed site-directed mutagenesis targeting α1 and α2 of the SPX domain, on the basis of structural predictions (Extended Data Fig. 7b). In addition to the nla-1 and nla-2 alleles, three new mutants were generated: nla-3 (hereafter, nla Δ12 ), containing a 12-bp deletion removing the conserved 18 FLGQ 21 motif; nla-4, with a 3-bp deletion (ΔQ 21 ); and nla-5, with a frameshifting single-base insertion (Extended Data Fig. 7c). Phenotypic analysis showed that nla-5 resembled nla-1 and nla-2 in exhibiting strong sensitivity to cold, whereas nla-4 behaved like the wild type (Extended Data Fig. 7d). Notably, nla Δ12 plants exhibited enhanced cold tolerance together with substantially increased shoot P accumulation under cold stress (Fig. 4a–c). These plants were therefore selected for further analysis.

a, b, Representative chilling phenotypes (a) and quantification of uninjured leaf area (b) of wild-type plants and nla-1 and nla Δ12 mutants after 4 °C treatment. Scale bars, 5 cm. c, Shoot P content in wild-type plants and nla-1 and nla Δ12 mutants after two weeks at 8 °C. d, Structural model of the NLA–InsP 8 complex (left), with residues corresponding to the 12-bp deletion in nla Δ12 highlighted in green; right, enlarged view of the InsP-binding pocket, showing helix arrangement and InsP 8 molecule. e, f, Binding affinities (e) and quantification (f) of InsPs binding to His-tagged NLA, NLA Δ12 and NLA PBC proteins, measured by MST. g–j, Interactions of NLA–MYC or NLA Δ12 –MYC with JAZ11 (g, h) and PT4 (i, j) in maize protoplasts. In b, c, f, h, j, data are mean ± s.d. (n = 3 independent biological replicates). In b, c, f, different letters denote significant differences (P < 0.05) from a Duncan’s new multiple range test. For exact P values, see Source Data. In h, j, significance was determined by two-tailed Student’s t-tests.

The deleted 18 FLGQ 21 motif lies within the loop bridging α1 and α2 (Fig. 4d), a region predicted to stabilize the spatial arrangement of helices that is essential for InsP binding 37 . Microscale thermophoresis (MST) analysis showed that wild-type NLA bound to both InsP 6 and InsP 8 , with particularly high affinity for InsP 8 (dissociation constant (K d ) = 0.0014 mM). By contrast, the NLA Δ12 variant exhibited an approximately 50-fold reduction in InsP 8 binding (K d = 0.071 mM) (Fig. 4e,f). As expected, the NLA PBC mutant showed even weaker InsP 8 binding (Fig. 4e,f). Co-IP assays further revealed that NLA Δ12 retained normal binding to JAZ11 but showed markedly reduced interaction with PT4 (Fig. 4g–j), a pattern mirrored by the NLA PBC variant (Extended Data Fig. 8a–d).

To determine whether the impaired PT4 interaction affects NLA subcellular distribution, we examined the localization of NLA and NLA Δ12 in transgenic plants (Extended Data Fig. 8e). Wild-type NLA localized to both the plasma membrane and the cytoplasm, whereas NLA Δ12 showed a reduction in the membrane-associated fraction and an increase in cytosolic accumulation (Extended Data Fig. 8f). Given that PT4 is localized at the plasma membrane, whereas JAZ11 resides in the cytoplasm and nucleus (Extended Data Fig. 8g,h), this redistribution suggests that the disrupted PT4 binding releases NLA Δ12 from the plasma membrane, thereby increasing the cytosolic pool of NLA Δ12 that is available to interact with JAZ11 and amplify JA-mediated cold responses. Consistent with this observation, JA-responsive genes, such as AOS2, LOX13 and BX10–BX12, showed higher expression in nla Δ12 -mutant plants than in wild-type ones, contrasting with the downregulation that was observed in nla-1 (Extended Data Fig. 8i). Collectively, these findings indicate that deletion of the InsP-binding motif in NLA compromises NLA-mediated Pi sensing and PT4 regulation, and simultaneously enhances the activation of JA signalling.

nla Δ12 improves cold tolerance and yield

The identification of nla Δ12 as a variant that separates cold-induced JA activation from Pi suppression prompted us to evaluate its agronomic performance under field conditions. Multi-site trials were performed at three climatically distinct locations: Gongzhuling (43° 11′ N, 124° 02′ E), Sanya (18° 09′ N, 108° 56′ E) and Zhuozhou (39° 21′ N, 115° 44′E) (Fig. 5a). At Zhuozhou (2024) and Sanya (2023), grain yield per ear and per hectare showed no significant differences among wild-type, nla and nla Δ12 plants (Fig. 5b,c). By contrast, at the high-latitude site of Gongzhuling in 2023, where a short episode of chilling (temperature below 15 °C) occurred during germination (Fig. 5a), nla Δ12 plants exhibited higher grain yields than the wild type (Fig. 5b,c). During grain filling, nla Δ12 plants accumulated more P in ear leaves and kernels than wild-type plants did, but less than nla-1 mutants did (Fig. 5d), consistent with the reduced InsP binding and weakened PT4 association of NLA Δ12 (Fig. 4g–h). These observations suggest that the superior yield performance of nla Δ12 plants reflects contributions from both enhanced cold tolerance and improved PUE.

a, Temperature profiles for three locations: Gongzhuling (43° 11′ N, 124° 02′ E; sowing took place on 8 May 2023), Sanya (18° 09′ N, 108° 56′ E; sowing on 1 November 2023), and Zhuozhou (39° 21′ N, 115° 44′ E; sowing on 30 April 2024). All soils contained more than 20 mg kg −1 available P. b, c, Grain yield per plant (b) and per hectare (c) of wild-type, nla and nla Δ12 plants grown at the three locations. All soils contained more than 20 mg kg −1 available P. d, P content in ear leaves (top) and seeds (bottom) of wild-type, nla and nla Δ12 plants grown in Gongzhuling. e, Temperature profiles and growth stages for normal sowing (NS), early sowing (ES) and late sowing (LS) in field trials. Date of early sowing: 15 April 2024. f, g, Plant survival rate (f) and harvest rate (g) of wild-type and nla Δ12 plants under NS, ES and LS conditions. h, Representative ear morphology of wild-type and nla Δ12 plants grown under NS, ES and LS conditions. i, j, Grain yield per plant (i) and per hectare (j) of wild-type and nla Δ12 plants under NS, ES and LS conditions. In b–d, f, g, i, j, data are mean ± s.d. (n = 3 independent biological replicates). In b, c, significance was determined by two-tailed Student’s t-tests. In d, f, g, i, j, different letters denote significant differences (P < 0.05) from a Duncan’s new multiple range test. For exact P values, see Source Data.

To further assess the temperature adaptability of nla Δ12 plants, early, normal and late sowing trials were performed in 2024 at Gongzhuling, a site characterized by frequent chilling during seedling establishment and reproductive development (Fig. 5e). Early sowing exposed seeds and seedlings to long-term cold stress (around 50 nights below 15 °C) (Fig 5e), leading to substantial cold-induced mortality and a reduced final yield despite normal grain filling in surviving plants (Fig. 5f–j). Late sowing imposed chilling during reproduction, severely impairing ear development and, in severe cases, eliminating yield entirely (Fig. 5g–j). Across both stress regimes, nla Δ12 plants showed smaller losses in yield than the wild type did (Fig. 5g–j). The nla Δ12 allele increased early-sowing survival and enhanced per-ear yield across all sowing conditions, with the strongest benefits seen under late sowing conditions (15.7% per plant and 13.5% per hectare), followed by early sowing (11.4% and 9.5%) and normal conditions (8.9% and 8.9%) (Fig. 5i,j). These results indicate that nla Δ12 maintains yield under cold stress by simultaneously reinforcing JA-mediated cold resilience and sustaining P acquisition.

To evaluate hybrid performance, nla Δ12 and wild-type (LH244 background containing the PT4 K267 haplotype) lines were crossed with PH4CV (favourable PT4 A267 haplotype), an elite parent of the commercial hybrid Xianyu 335. Across all sowing regimes in Gongzhuling, F 1 hybrids carrying nla Δ12 consistently outperformed wild-type hybrids in both grain yield and yield stability (Extended Data Fig. 9a,b). In the segregating F 2 population (Extended Data Fig. 9c,d), homozygous PT4 A267 nla Δ12 plants accumulated the highest shoot P levels among all genotypes (Extended Data Fig. 9e), whereas PT4 A267 nla Δ12 and PT4 K267 nla Δ12 exhibited comparable cold tolerance (Extended Data Fig. 9f). These findings show that PT4 A267 and nla Δ12 act additively to enhance P acquisition, but that cold-tolerance enhancement is derived mainly from nla Δ12 .

Together, these results show that NLA Δ12 enhances PUE, stabilizes yield and improves field performance under cold-prone conditions, establishing it as a promising genetic resource for breeding high-performance maize that is adapted to temperate climates.

Discussion

Enhancing stress resilience while maintaining nutrient efficiency remains a key challenge for crop improvement under increasing climate variability. Here we identify the SPX-domain E3 ubiquitin ligase NLA as a central regulator that connects cold tolerance with Pi homeostasis in maize. NLA integrates these processes through a dual mechanism in which its E3 ligase activity promotes the degradation of JAZ11 to activate JA signalling and enhance cold tolerance, whereas its InsP-dependent Pi-sensing function mediates PT4 turnover to restrict Pi uptake. Ubiquitinome profiling further expands this framework by revealing additional substrates of NLA that are involved in nutrient transport, such as sucrose transporters (SUCs), potassium transporters (HAKs) and sugar exporters (SWEETs), suggesting that NLA has broader roles in nutrient allocation and source–sink regulation. Future studies should examine natural variation in nitrate transporters (NRTs) and other NLA substrates to uncover further targets for crop improvement.

Using AlphaFold-based structural modelling and CRISPR–Cas9 editing, we selectively disrupted InsP sensing while retaining E3 ligase activity. The engineered nla Δ12 allele reveals how NLA partitions its activity between nutrient and stress pathways. The 12-bp deletion weakens InsP binding and PT4 association, reducing membrane sequestration and increasing the cytosolic pool of NLA, thereby amplifying JA activation under cold stress. Physiological and population data indicate that the enhanced cold tolerance of nla Δ12 results mainly from reinforced JA signalling, rather than increased Pi accumulation.

SPX-domain proteins act as canonical InsP-dependent Pi sensors across plants 42 , 43 , 44 , 45 . In Arabidopsis, the stability of NLA depends on Pi availability 25 , 46 , whereas in maize it accumulates under cold stress, indicating temperature-dependent modulation of SPX control. Similar Pi-independent responses in SPX4 in rice 43 suggest that SPX proteins integrate diverse environmental cues. The nla Δ12 allele exemplifies precise reprogramming of this conserved module. Selective perturbation of InsP sensing redirects the activity of NLA towards stress signalling without compromising Pi uptake. Given the evolutionary conservation of SPX domains 19 , this strategy could be broadly applicable to reconfiguring nutrient–stress coordination in cereals.

Field trials across temperature regimes reveal that NLA Δ12 improves yield stability at both the seedling and the grain-filling stage. Cold episodes often suppress Pi uptake, causing a P blockade that is typical of temperate climates 1 . The nla Δ12 allele enables plants to overcome this limitation, mainly by sustaining Pi acquisition while reinforcing JA-mediated cold protection. This dual advantage explains the consistent yield gains observed under cold-prone conditions. The enhanced PUE supports biomass and productivity, whereas increased JA signalling strengthens cellular resilience. Thus, NLA Δ12 redefines the balance between cold resilience and nutrient efficiency, achieving both simultaneously rather than trading one for the other.

Population and ubiquitination analyses revealed natural allelic diversity in substrates of NLA, such as JAZ11 and PT4, which modulates protein stability and stress responses. Combining the favourable PT4 A267 allele with nla Δ12 further improved the PUE, illustrating the synergistic potential of integrating natural variation with targeted molecular design. Moreover, the enrichment of favourable JAZ11 alleles in teosinte (the wild ancestor of maize)—in contrast with the scarcity of these alleles in modern maize (Extended Data Fig. 9g–i and Supplementary Tables 7, 11 and 12)—highlights wild relatives as valuable reservoirs of adaptive diversity 47 , 48 , 49 .

In summary, our study uncovers a regulatory mechanism through which NLA integrates cold resilience with Pi homeostasis, and shows that SPX-mediated signalling can be modularly rewired (Extended Data Fig. 10). These findings establish a molecular framework for engineering climate-resilient crops with enhanced nutrient efficiency and yield stability.

Methods

Plant materials and growth conditions

Transgenic maize (Zea mays L.) lines overexpressing JAZ11, PT4, BX11 or GFP-tagged NLA, JAZ11 or PT4 were generated in the LH244 inbred background following a previously described protocol 50 . GFP-tagged NLA Δ12 plants were generated in an nla-mutant background. The nla, jaz11, nla jaz11 and lox8 loss-of-function lines were also generated in the LH244 inbred line background through CRISPR–Cas9-mediated genome editing, as described previously 51 . At least two independent transgenic lines were selected, and homozygous plants were used for subsequent analyses (Supplementary Table 13).

A diverse panel of maize lines with natural variation, comprising tropical and temperate inbred lines, was used for the GWAS, as previously described 52 . DNA samples from 1,008 maize landraces were used to analyse the global geographical distribution of maize 53 . For allele frequency analysis, 115 teosinte accessions (including 88 Zea mays ssp. mexicana and 27 Zea mays ssp. parviglumis) were examined.

To generate the nla Δ12 × PH4CV F 2 segregating population, nla Δ12 (LH244 background containing the PT4 K267 haplotype) was crossed with the inbred line PH4CV, which is homozygous for the favourable PT4 A267 haplotype. From the resulting F 2 population, 500 individuals were genotyped using KASP markers (Supplementary Table 14) specific to the NLA and PT4 loci to identified homozygous genotypes for subsequent phenotypic evaluation.

For controlled experiments, maize seeds were sown in plastic pots (30 cm × 20 cm × 15 cm; length × width × depth) and cultivated in growth chambers or greenhouses at 25 °C under a 16-h light–8-h dark photoperiod, with a light intensity of 200 μmol m −2 s −1 and 60% relative humidity. The soil nutrient composition was: total nitrogen, 6.98 g kg −1 ; total P, 0.093%; and total potassium, 17.3 g kg −1 . Cold stress was applied to 14-day-old seedlings after thorough watering.

Plasmid construction and plant transformation

For overexpression constructs, the coding sequences (CDSs) of JAZ11 and BX11 were PCR-amplified, treated with Taq polymerase to add 3′ A overhangs and ligated into the XcmI-linearized pCXUN vector 54 . Transgene expression was driven by the Ubi promoter.

To generate the NLA-GFP, NLA Δ12 -GFP, JAZ11-GFP and PT4-GFP constructs, the CDSs of the respective genes were ligated into the XcmI-linearized CAUC3078-CeGFP vector (a derivative vector from pCXUN containing a GFP tag adjacent to the XcmI site), also under the Ubi promoter.

To generate Super::NLA-MYC, Super::JAZ11-GFP, Super::PT4-GFP and their mutant variants, the CDSs of NLA, JAZ11 and PT4 were cloned into the Super1300-MYC and Super1300-GFP (KpnI) vectors using the In-Fusion PCR cloning system (CloneSmarter). Site-directed mutagenesis was then used to generate the Super::NLA Δ12 -MYC, Super::NLA PBC -MYC and Super::JAZ11 K47 -GFP constructs.

For recombinant protein expression in Escherichia coli, the CDSs of NLA and JAZ11 were inserted into pGEX4T-1 (EcoRI) and pMALc5x (BamHI) using the In-Fusion PCR cloning system. Site-directed mutagenesis was used to generate the GST-JAZ11 K47 construct. The MBP-His-NLA construct was generated by cloning the NLA CDS into a modified pMALc2 vector containing a 6×His tag adjacent to MBP.

For yeast two-hybrid assays, full-length or truncated CDSs of JAZ11, JAZ25, JAZ13 and NLA were cloned into pGBKT7 (EcoRI) and pGADT7 (EcoRI) vectors, respectively.

All constructs were introduced into Agrobacterium tumefaciens strain EHA105 for maize transformation. Immature zygotic embryos (12 days after pollination) from LH244 were used for Agrobacterium-mediated transformation to generate transgenic lines (Supplementary Table 13). All primers used are listed in Supplementary Table 14.

For CRISPR–Cas9-mediated genome editing of NLA, JAZ11 and LOX8, guide RNAs were designed using Cas-OFFinder ( ). Off-target prediction was performed with 20-nucleotide query sequences (gN19 format), allowing ≤4 mismatches and considering NRG PAM sequences (excluded from the query length). All candidate targets satisfied the 5′ guanine (g) requirement. Guide RNA expression cassettes were cloned into pBUE411 as described 51 . After Agrobacterium-mediated maize transformation and Sanger sequencing validation, edited lines were selected for phenotypic evaluation for cold tolerance, PUE and yield.

Cold stress treatment

Seedlings grown at 25 °C for 14 days were transferred to 4 °C for 4–5 days and then recovered at 25 °C for 1–2 days. Representative images were captured after two days of recovery. Leaf injury was quantified by measuring the area of the second leaf using ImageJ (v.1.8.0) 55 . Relative uninjured leaf area was calculated as: (1 − B/A) × 100%, where A is the total area of the second leaf and B is the visibly injured area.

For P content analysis under cold stress, total shoot P was measured from 14-day-old seedlings after either 2 weeks (for GWAS samples) or 20 days (for transgenic plants) of treatment at 8 °C.

Pi depletion assay

For Pi depletion assays, two seedlings per treatment were transferred to a flask containing 500 ml of depletion solution (750 μM K 2 SO 4 , 650 μM MgSO 4 , 2 mM Ca (NO 3 ) 2 , 100 μM KCI, 250 μM KH 2 PO 4 , 1 μM MnSO 4 , 1 μM ZnSO 4 , 0.1 μM CuSO 4 , 0.05 μM (NH4) 6 Mo 7 O 24 , 1 μM H 3 BO 3, 100 μM Fe-EDTA and 100 μM FeSO 4 ). Seedlings were maintained at either 25 °C or 8 °C, and flasks were gently agitated every 30 min. At the indicated time points, 500 μl aliquots were collected, and Pi concentrations were determined using the molybdenum blue colorimetric method 56 .

P content measurement

For quantification of the total P content of transgenic plants, tissues (excluding kernels) were dried at 80 °C to a constant weight and finely ground, and approximately 0.2 g of each sample was used per analysis. Samples were carbonized at 300 °C for 10 h and washed at 575 °C for 12–14 h. The ash was extracted with 10 ml of 0.1 M HCl for at least 8 h, and absorbance was measured at 410 nm. For kernel samples, ashing was extended to 24–28 h. The total P content was determined as previously described 57 .

For GWAS samples, P content was determined according to the Chinese agricultural industry standard NY/T 2017-2011 (Convinced-Test, Nanjing). In brief, 0.2 g of dried material was digested with concentrated sulfuric acid at 250 °C, followed by oxidation at 400 °C, with sequential additions of hydrogen peroxide until the solution became clear. The digestate was diluted to 50 ml, and an aliquot (containing 0.04–1.0 μg P) was mixed with 2,4-dinitrophenol indicator and 2.5 ml ammonium molybdate–ascorbic acid chromogenic reagent. After 30 min of colour development, the absorbance was measured at 700 nm.

Protein detection by immunoblotting

For immunoblot analysis, roots from 14-day-old Ubi::NLA-GFP, Ubi::JAZ11-GFP and Ubi::PT4-GFP seedlings were collected after treatment at either 25 °C or 4 °C. Total proteins were extracted using 2% SDS extraction buffer supplemented with 1 mM dithiothreitol (DTT) and 50 μM MG132 (Sigma-Aldrich, C2211). Proteins were separated by SDS–PAGE and subjected to immunoblotting with anti-GFP antibodies (Transgen, HT801-01, 1:5,000 dilution), or anti-PEPC (Agrisera, AS09458, 1:5,000 dilution).

Ubiquitinome and transcriptome analysis

For ubiquitinome profiling, 14-day-old wild-type (LH244) and nla-1 mutant seedlings were treated at 25 °C or 4 °C for 12 h, with MG132 (25 µM) added 15 min before treatment. Total proteins were extracted from seedling roots and digested with trypsin overnight at 37 °C, following a method described previously 58 , 59 . The resulting peptides were used for proteomic analysis and enrichment of ubiquitinated peptides using an anti-ubiquitin antibody-based affinity purification system (PTMScan Ubiquitin Remnant Motif (K-ε-GG) kit; Cell Signaling Technology). Enriched samples were analysed by liquid chromatography–tandem mass spectrometry (LC–MS/MS), spectrometry data were processed in Xcalibur v.2.2.48 (Thermo Fisher Scientific) and raw MS data were analysed using Spectronaut 16.0 (Biognosys) searched against the UniProt Z. mays database (Uniprot_Maize_2018-07, 30,208 entries). Differentially ubiquitinated peptides and sites (DUPs) were defined by fold change > 1.5 and adjusted P value (Q) < 0.05 across three biological replicates. The ratio of site-specific protein ubiquitination (r) was calculated by dividing the intensity of ubiquitinated peptide by the expression level or abundance of the corresponding protein.

For transcriptome profiling, total RNA was extracted from 14-day-old wild-type (LH244) and nla-1-mutant seedlings treated at 25 °C or 4 °C for 12 h. cDNA libraries were prepared and sequenced by Novogene. Raw data (raw reads) in fastq format were subjected to quality control by FastQC (v.0.11.9) and the clean reads were aligned to the reference maize genome (B73 RefGen_v4, AGPv4) using the HISAT2 program (v.2.0.4) with default parameters. Gene read counts were determined by FeatureCounts (v.2.0.1). Differential expression analysis of genes (DEGs) was performed using the R package DESeq2 (v.1.30.0). P values were adjusted according to the Benjamini–Hochberg procedure. DEGs were defined by fold change > 1.5 and Q < 0.05 across three biological replicates.

GO enrichment analyses of DEGs and DUPs were conducted using the Gene Ontology Consortium database ( ) and DAVID 60 ( ), using the hypergeometric test with a significance cut-off of P < 0.05. Cluster analyses were performed using the R package pheatmap (v.1.0.12; ). Heat maps were generated using TBtools (v.2.315) 61 , and GO networks were constructed using the EnrichmentMap (v.3.5.0) plug-in in Cytoscape (v.3.10.3) 62 , 63 .

Protein–protein interaction assays

For yeast two-hybrid assays, CDSs were cloned into pGADT7 and pGBKT7 vectors and co-transformed into yeast strain AH109. Empty vectors were used as negative controls. Protein–protein interactions were assessed on SD/−Ade/−His/−Leu/−Trp dropout medium (Clontech).

For pull-down assays, GST or GST–JAZ11 fusion proteins were immobilized on glutathione sepharose beads (GE HealthCare, 17513202) and incubated with MBP–NLA fusion proteins. After extensive washing, bound proteins were detected by immunoblotting with anti-MBP (AbM59007-3-PU) and anti-GST (AbM59001-2H5-PU) antibodies (Beijing Protein Innovation; both 1:5,000 dilution).

For co-IP assays, maize protoplasts were transiently co-transfected with NLA–MYC and either GFP–JAZ11 or GFP constructs. After 16 h of expression, total proteins were extracted using IP buffer (50 mM Tris-HCI (pH 7.5), 150 mM NaCl, 10% glycerol, 1% NP-40, 1 mM DTT and 1× protease inhibitor cocktail (Roche, 04693132001)). Extracts were incubated with GFP-agarose beads (Chromotek, GTA-20) at 4 °C for 2 h and washed three times with IP buffer. Immunoprecipitates were detected by immunoblotting with anti-MYC (TransGen Biotech, HT101-02, 1:5,000 dilution) and anti-GFP antibodies (TransGen Biotech, HT801-01, 1:5,000 dilution). Source gels and blots are provided in Supplementary Fig. 1.

JAZ11 antibody production

A synthetic peptide corresponding to residues 155–169 of JAZ11 (C-VPVISLPAAQADAKK-NH 2 ) was used to immunize rabbits (Shanghai Youke Biotechnology, 4516-2-R1). The resulting polyclonal antibody was used at 1:1,000 dilution for immunoblotting.

In vitro and in vivo ubiquitination assays

For in vitro ubiquitination assays, recombinant proteins, including E1 (His–AtUBA1), E2 (His–UBCH5B), E3 (MBP–His–NLA) and the substrate (MBP–JAZ11), were expressed in E. coli BL21 (CodonPlus) RIL cells and purified as previously described 64 . MBP–JAZ11 was purified using amylose resin and cleaved with Factor Xa protease (NEB, P8010V) at a 1:100 molar ratio for 16 h at 4 °C. Ubiquitination reactions were performed at 30 °C for 2 h in a buffer containing 50 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 5 mM ATP and 2 mM DTT. Reaction products were analysed by immunoblotting with anti-His, anti-JAZ11, and anti-ubiquitin antibodies (Cell Signaling Technology, SC8017; 1:1,000 dilution).

In vivo ubiquitination assays were done as previously described 65 . NLA–MYC, JAZ11–GFP, UBQ and Flag–TUBE constructs were co-expressed in maize protoplasts for 16 h. Total proteins were extracted using IP buffer and incubated with anti-Flag agarose beads at 4 °C for 1 h, followed by three washes with IP buffer. Polyubiquitinated JAZ11 was detected by immunoblotting with anti-GFP antibodies.

Protein stability assays

For JAZ11 cell-free degradation, total proteins were extracted from 10-day-old wild-type and nla-1 mutant seedlings in native buffer (50 mM Tris-HCl (pH 8.0), 10 mM EDTA (pH 8.0), 0.5 M sucrose, 1 mM MgCl 2 and 5 mM DTT). Recombinant GST–JAZ11 protein was incubated with the extracts at 25 °C in the presence of 1 μM ATP. HSP82 served as a loading control. To test for proteasome-dependent degradation, 50 μM MG132 was added to selected reactions. For PT4 cell-free degradation, PT4(K267)–GFP or PT4(A267)–GFP was extracted from instantaneously expressed protoplasts and reacted with the His–NLA protein under the same conditions as described above.

The stabilization of JAZ11 and PT4 variants in protoplasts was performed as previously described 64 . JAZ11 and PT4 variants tagged with GFP (30 μg) were co-transfected into maize protoplasts along with increasing amounts of NLA–MYC or empty MYC vector (maintaining constant total plasmid DNA). After 16 h of incubation at 22 °C, total proteins were extracted and analysed by immunoblotting with anti-GFP and anti-MYC antibodies (TransGen, HT101-01, 1:5,000 dilution).

GWAS for shoot P content under cold treatment

A panel of 192 diverse maize inbred lines was grown in a greenhouse at 25 °C for 14 days, followed by 8 °C treatment for 14 days. Shoot P content was measured and used for GWAS based on 11.5 million genome-wide SNPs 66 . SNPs with MAF > 0.05 and missing rate < 10% were retained. GWAS was performed using EMMAX (0~beta.20100307), incorporating kinship and population structure as covariates. Independent SNPs were selected using PLINK (v.1.19) 67 (--indep-pairwise 50 20 0.2). The significance threshold for shoot P content was set as P = 0.05/n, where n represents the number of independent effective SNPs. A Bonferroni-correction genome-wide significance threshold of P < 1 × 10 −7 was applied. Manhattan and Q–Q plots were generated from the EMMAX results using the CMplot (v.4.5.0) package in R.

For candidate gene association analysis of PT4, 149 inbred lines were resequenced. A total of 34 SNPs within the PT4 region were analysed using TASSEL (v.5.0) 68 (MLM; MAF ≥ 0.05). LD and haplotype blocks were visualized using LDBlockShow (v.1.40) 69 . DNA sequences were aligned using MEGA (v.5.0) 70 .

MST

Variants of full-length NLA, NLA Δ12 and NLA PBC were cloned into a modified pET-28a-SUMO vector carrying an N-terminal SUMO fusion protein with 6×His–Strep dual tags 71 . Recombinant proteins were expressed in E. coli SHuffle T7 cells (Weidi Biotechnology, EC2031) and purified using Ni-NTA affinity chromatography (Novagen), followed by size-exclusion chromatography, as previously described 72 .

MST was performed as previously described 19 . In brief, 500 nM purified His-tagged NLA variants were labelled with 50 nM RED-Tris-NTA dye (NanoTemper, MO-L018) and incubated with serial dilutions of InsP 8 or InsP 6 ligands. Thermophoresis was measured using a Monolith NT.115 instrument (NanoTemper), and dissociation constants (K d ) were calculated by fitting binding curves to a single-site binding model with MO.Affinity software (v.2.3).

Structural predictions

Structural predictions of proteins and complexes were performed using AlphaFold3 (ref. 39 ). The interactions of InsPs with NLA were predicted using DiffDock-L docking 40 . All structure figures were prepared in PyMOL v.3.1.0 (ref. 73 ).

Quantitative PCR with reverse transcription (RT–qPCR)

Total RNA was extracted from leaves of 14-day-old maize seedlings using TRIzol reagent (Invitrogen) and reverse-transcribed with M-MLV reverse transcriptase (Promega). Quantitative PCR (qPCR) was performed with SYBR Green (Takara) on a StepOnePlus Real-Time PCR System (Applied Biosystems). UBI served as an internal control. Primer sequences are listed in Supplementary Table 14. Each experiment was done with at least three independent biological replicates.

Co-expressed gene identification and selection analysis

Co-expressed genes of NLA, JAZ11 and PT4 were identified using ATTED-II ( ) with a co-expression threshold of z-score ≥ 5. Nucleotide diversity (π) in the region surrounding the target genes (Supplementary Table 15) was analysed in maize and teosinte using vcftools (v.0.1.15). Estimates of π were generated with a 1,000-bp sliding window advanced in 300-bp steps. Maize genotype data were sourced from the HapMap3 resource 74 , which is publicly accessible through the Panzea database (www.panzea.org). Nucleotide diversity (π) values were calculated across 5-kb genomic regions flanking NLA, JAZ11, PT4 and their co-expressed genes to assess genetic variation and selection pressure.

Field trials

Field trials were performed at three locations representing distinct soil and climatic conditions: Gongzhuling (43° 11′ N, 124° 02′ E), Zhuozhou (39° 21′ N, 115° 44′ E) and Sanya (18° 09′ N, 108° 56′ E). In Gongzhuling, sowing was performed under early, normal and late sowing regimes from mid-April to mid-May. In Zhuozhou, planting was done in early May, whereas in Sanya, trials were sown in mid-November.

For field management, in Gongzhuling, which has chernozem soil, a compound NPK fertilizer was applied at a rate of 0.8 t ha −1 after deep tillage that incorporated the previous year’s maize stover. In Zhuozhou, which has sandy loam soil, a basal application of 3.75 t ha −1 organic fertilizer and 0.8 t ha −1 compound NPK fertilizer was applied before sowing, followed by a 0.6 t ha −1 top-dressing at the jointing stage. In Sanya, which has sandy soil, 3.75 t ha −1 of organic fertilizer and 0.8 t ha −1 of compound fertilizer were applied as basal nutrients. Additional NPK compound fertilizer was top-dressed at both the V6–V8 and the V12–VT stages (0.375 t ha −1 per application, totalling 0.8 t ha −1 ). All sites had high available P (AP) levels (>20 mg/kg) 75 , 76 .

A split-plot design with three replicates was implemented at each location. Each subplot consisted of three rows (2 m in length, spaced 50 cm row spacing) with 11 plants per row (20 cm plant spacing). In Gongzhuling, plant survival and harvest rates were recorded under both normal and late sowing conditions. At maturity, all ears from the middle row of each subplot were harvested as representative samples. Grain yield was determined using three independent biological replicates.

Quantification and statistical analysis

Sample sizes were determined on the basis of previous studies, as specified in each experimental section. Unless otherwise indicated, n represents the number of independent biological replicates. Statistical analyses were performed using GraphPad Prism v.8.3.0 and IBM SPSS Statistics v.26. Data are presented as mean ± s.d., and statistical significance was assessed using two-tailed Student’s t-test or Duncan’s new multiple range test (P < 0.05). No specific randomization or blinding methods were used, and no data were excluded from analysis.