Functional characterization of transporter genes in rice node at single-cell resolution through multi-omics technologies

Abstract

Background

The rice node is a critical hub for the distribution of mineral nutrients, mediated by transporters. Manganese (Mn) is an essential micronutrient for plant growth. However, the precise cell types and the cell-type-enriched transporter genes in rice node, and the molecular mechanisms underlying the translocation and distribution of Mn in rice remain poorly understood.

Results

We characterize 11 distinct cell types using multiple cluster-enriched genes in rice node I through single-nucleus RNA sequencing (snRNA-seq), systematically profile the expression patterns of putative 1,144 transporter genes within 11 cell types, and identify six candidate transporter genes linked to the tissue-specific deposition of six elements through combining spatial ionomics in node, respectively. Furthermore, we functionally characterize OsMTP7 that is highly expressed in phloem cells in node, as well as in root stele cells and anther. OsMTP7 is localized to plasma membrane in rice and shows efflux activity for Mn. OsMTP7 knockout inhibites Mn uptake and xylem-mediated Mn translocation in root and Mn distribution in node, leading to decreased Mn concentration in various organs and root xylem sap, resulting in reduced biomass and yield. We reveal that OsMTP7 knockout alters expression of genes for multiple biological processes in spikelets using bulk RNA-seq, resulting in increased oxidative stress in anther and low fertility.

Conclusions

Our study reveals the precise cell types and the cell-type-enriched transporter genes in node, identifies candidate transporters for elements deposition in node, and demonstrates a novel and critical Mn efflux transporter mediating Mn uptake, translocation and distribution for improving growth and yield in rice.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13059-026-04046-6.

Background

Mineral elements are absorbed by roots and then translocated to shoots via xylem, driven by root pressure and leaf transpiration. Theoretically, few elements will be distributed to the developing organs, such as the newest leaf and panicle, due to their small surface area and low transpiration rate. However, these developing organs usually require and contain abundant mineral nutrients for active growth [1, 2]. In the past decade, the node, particularly the uppermost node/node I directly linked to the flag leaf and panicle, has been identified as the critical hub for the preferential and efficient distribution of mineral nutrients to developing organs in plants, including barley, maize, Arabidopsis, and especially rice [3–7].

The node is a crucial junction organ connecting internodes, leaves, and axillary buds/tillers. Rice typically has 4–5 elongated nodes on a main tiller at the reproductive stage although 13–18 nodes in total [2, 8]. Each rice node contains three primary types of axial vascular bundles (VBs), including enlarged VB (EVB), transit VB (TVB), and diffuse VB (DVB) [2, 3]. The vascular connectivity follows a specific pattern. EVB in node n (EVB_n) directly links TVB of the lower node (TVB_n-1) and leaf_n. TVB_n directly connects DVB_n-1 and EVB of the upper node (EVB_n+1). DVB_n connects to TVB_n+1. Additionally, the horizontal nodal vascular anastomosis (NVA) establishes a direct link between three VBs at the basal part of each node. Crucially, the inter-vascular transfer from EVB to DVB within each node facilitates the preferential and efficient distribution of nutrients to developing organs. The parenchyma cell bridge (PCB), located between EVB and DVB, is a key tissue enabling this inter-vascular nutrient movement. VBs and NVA comprise both xylem and phloem regions, while PCB is distinguished by densely packed plasmodesmata [2, 3]. Beyond these components, rice node contains other complex and highly organized tissues [2, 3, 8], while the precise cellular composition remains incompletely characterized.

The inter-vascular transfer in node is mediated by various transporters. To date, only some of transporters involved in the three key steps of the inter-vascular transfer have been characterized in rice. In step one, OsLsi6 for silicon (Si) [9], OsZIP3 for zinc (Zn) [10], and OsOPT7 for iron (Fe) are specifically localized in EVB xylem [11], OsPHO1;2 for phosphorus (P) is localized in xylem of EVB primarily and DVB partially [12], and OsHMA5 for copper (Cu) is primarily localized in DVB xylem [13], which mediate ion transport from xylem to PCB or phloem. In step two, OsLsi2 and OsLsi3 for Si are mainly localized in bundle sheath cells and PCB [9], respectively, and OsVMT for 2'-deoxymugineic acid is localized in PCB [14, 15], which mediate ion transport from EVB xylem to PCB or from PCB to DVB. In step three, OsPHO1;1 for P is primarily localized in DVB phloem [12, 16], and OsZIP4 and OsHMA2 for Zn are localized in phloem of DVB and EVB [17, 18], which mediate ion transport into phloem. However, the cell-type-specific expression patterns of other transporter genes within rice node remain poorly understood.

Manganese (Mn) is an essential micronutrient for plant growth and development. Mn serves as a cofactor or component in over 30 enzymes, playing an indispensable role in key biological processes such as photosynthesis, respiration, reactive oxygen species (ROS) scavenging, and hormone signaling [19, 20]. Plants require only trace amounts of Mn, but Mn²⁺ concentration in soil solutions can vary widely, from sub-micromolar to millimolar levels, depending on soil conditions, posing risks of both deficiency and toxicity [19, 20]. To adapt to these fluctuations, plants have evolved complex Mn homeostasis systems regulated by diverse transporters. For example, rice OsNRAMP5 and OsMTP9 with polarity localization, as well as OsNRAMP1, play key roles in Mn uptake in root [21–23]. OsMTP8.1 and OsMTP8.2 regulate intracellular Mn storage via transporting Mn into vacuole [24, 25]. The Golgi-localized Mn transporters OsPML3 and OsMTP11 also contribute to Mn homeostasis [26, 27]. Additionally, OsMTP9, OsYSL2, OsYSL6, OsNRAMP3, and OsNRAMP5 are involved in translocation or distribution of Mn in root, leaf or node [22, 28–31]. However, the molecular mechanisms underlying Mn efflux either in root stele cells for translocation or in node for distribution are unknown.

Single-cell RNA sequencing (scRNA-seq) is a powerful technology that has revolutionized the study of plant biology by enabling the exploration of plant-specific cell types, cellular developmental trajectories, stress responses, and the identification of key genes involved in various biological processes [32, 33]. This approach relies on either protoplast isolation or single nuclei capture, both of which yield comparable transcriptomic information [34]. Single-nucleus RNA sequencing (snRNA-seq) has been particularly advantageous for analyzing organs and tissues where protoplast isolation is challenging [33]. In rice, scRNA-seq has been successfully applied to tissues such as plumule, root, culm, leaf, shoot apex, tiller bud, panicle, and spikelet or seed [34–39], except for node. Here, we first characterized 11 distinct cell types in rice node I at the single cell resolution using snRNA-seq, systematically profiled the expression patterns of various transporter genes across these cell types, and identified candidate transporter genes for the varied deposition patterns of elements. Notably, we functionally validated the critical role of OsMTP7 in maintaining Mn homeostasis and improving grain yield in rice.

Results

SnRNA-seq analysis of rice node I

To dissect the cellular heterogeneity of rice node I at single-cell resolution, nuclei were isolated from node I samples at three developmental stages: booting (RNA3), flowering (RNA2), and filling (RNA1) stages (Fig. 1A, Additional file 1: Fig. S1A), respectively. SnRNA-seq was performed using the 10 × Genomics Chromium platform. Raw reads were mapped to the T2T reference genome of Nipponbare [40], yielding a gene expression matrix encompassing 23,143 genes across 46,669 high-quality nuclei (Additional file 2: Table S1). Stage-specific counts included 10,064 (booting stage), 20,533 (flowering stage), and 16,072 (filling stage) nuclei (Additional file 1: Fig. S1B), with median gene counts of 1,125, 1,254, and 1,143, and median UMI counts of 1,716, 1,871, and 1,764 per nucleus, respectively (Additional file 2: Table S1). For downstream comparative analyses, nuclei from all three stages were first integrated to correct for batch effects and generate a unified dataset in a shared high-dimensional expression space. The integrated data were then subjected to linear dimensionality reduction using principal component analysis (PCA), retaining the top 30 principal components and performing graph-based clustering at a resolution of 0.2. The resulting clusters were visualized using UMAP and t-SNE, which consistently revealed 11 distinct cell clusters conserved across developmental stages (Fig. 1B, Additional file 1: Fig. S2). On the other hand, the ratios of 11 cell clusters in three samples of node I were similar (Additional file 1: Fig. S1B), and the spearman correlation analysis of pseudo-bulked expression profiles from each cell cluster showed that homologous clusters from different samples were highly correlated (Additional file 1: Fig. S1C), except for cluster 10. These results indicated that the identified differentially expressed genes and cluster identities are reproducible across growth stages, and genes likely showed similar expression patterns in the three growth stages.

Fig. 1.

SnRNA-seq analysis and cell types annotation of rice node I. A Overview of snRNA-seq workflow using rice node I. Plants were grown in a ½ Kimura B solution until reproductive stage, and then nuclei were isolated from each sample containing over 20 node I samples. SnRNA-seq libraries were generated using the 10 × Genomics platform, followed by high-throughput sequencing. B UMAP visualization of 11 cell clusters from 46,669 cells in three samples of rice node I from the booting stage, flowering stage, and filling stage. Each dot denotes a single cell. Colors denote corresponding cell clusters. C Expression patterns of representative marker or enriched genes for 11 cell types. The two most enriched genes and several confirmed or reported genes in each cell cluster were shown. Dot diameter represents the proportion of a given gene expressed in one specific cell cluster, and color represents their relative expression levels in these cell clusters. Enlarged vascular bundles, EVB; diffuse vascular bundles, DVB; transit vascular bundles, TVB; parenchyma cell bridge, PCB; nodal vascular anastomosis, NVA; small vascular bundles, SVB. D Tissue localization of eight marker or enriched genes in node I. RNA in situ hybridization for Os09g0333600. Immunostaining was performed using an antibody of OsLsi6, OsHMA5, OsHMA2, OsLsi2 or OsLsi3 in WT plants, or GFP antibody in transgenic lines carrying ProOsPHO1;1:GFP or ProOsMT2c:GFP. The purple or red color shows the signal from RNA in situ hybridization or an antibody, respectively. The blue/cyan color shows signal from cell wall autofluorescence. E Schematic illustration of various tissues in rice node I based on the anatomical morphology analysis and cell types annotation of snRNA-seq

Annotation of cell types in rice node I

To assign cell identities, ​1,837 cluster-enriched genes​ were identified through differential expression analysis across clusters (Additional file 2: Table S2), and an expression heatmap of the selected top enriched genes was generated for each cluster (Additional file 1: Fig. S3). The expression patterns of the two most enriched genes and several confirmed or reported genes in each cell cluster were shown for 11 cell types annotation (Fig. 1C, Additional file 1: Fig. S4). Specifically, (1) RNA in situ hybridization of Os09g0333600, the top marker gene of Cluster 10, showed high expression in parenchyma cells (Fig. 1D). Furthermore, OsACR9 and OsRL14​, enriched in Cluster 10 (Additional file 2: Table S2), and ​OsNPF2.2,​ showing high expression in parenchyma or mesophyll cells [41–43], indicated Cluster 10 as parenchymal (Fig. 1C, E, Additional file 1: Fig. S4). (2) Cluster 3​ was characterized by EVB xylem enriched genes, OsZIP3, OsLsi6, OsOPT7, and OsPHO1;2 [9–12]. OsLsi6 localization was confirmed through immunostaining using its antibody (Fig. 1D), indicating Cluster 3​ as EVB xylem cells (Fig. 1C, E, Additional file 1: Fig. S4). (3) OsHMA5, predominantly expressed in DVB xylem confirmed by immunostaining (Fig. 1D) [13], showed a similar expression profile to the top two markers of ​Cluster 2, suggesting that ​Cluster 2​ represents DVB/TVB xylem cells (Fig. 1C, E, Additional file 1: Fig. S4). (4) Cluster 0​ was characterized by OsPHO1;1 that primarily expressed in DVB phloem, confirmed by immunostaining using GFP antibody in transgenic lines carrying ProOsPHO1;1:GFP (Fig. 1D) [12]. It indicated that ​Cluster 0​ corresponds to DVB/TVB phloem cells (Fig. 1C, E, Additional file 1: Fig. S4). (5) OsZIP4 and OsHMA2, enriched in both Clusters 0 and 9 (Additional file 2: Table S2), were highly expressed in DVB and EVB phloem [17, 18]. The tissue localization of OsHMA2 was confirmed by immunostaining (Fig. 1D), implying ​Cluster 9​ as EVB phloem cells (Fig. 1C, E, Additional file 1: Fig. S4). (6) ​Cluster 5​ was defined by OsLsi2 with high expression in bundle sheath cells [9, 15], which was confirmed using immunostaining (Fig. 1D), indicating its identity as bundle sheath cells (Fig. 1C, E, Additional file 1: Fig. S4). (7) OsLsi3 and OsVMT with high expression in PCB [9, 14], which was confirmed by immunostaining using OsLsi3 antibody (Fig. 1D), showed expression patterns similar to the top markers of ​Cluster 1, suggesting that ​Cluster 1​ represents PCB (Fig. 1C, E, Additional file 1: Fig. S4).

Besides, (8) OsMT2c and OsMT2b, previously reported to be highly expressed in DVB and EVB phloem [44], were enriched in both ​Cluster 9​ (EVB phloem) and ​Cluster 7 (Additional file 2: Table S2), with higher expression in ​Cluster 7 (Fig. 1C, Additional file 1: Fig. S4). Further study using ProOsMT2c:GFP transgenic line indicated that GFP expression was higher in NVA phloem than in phloem of DVB and EVB (Figs. 1D, 2E, Additional file 1: Fig. S5), suggesting cluster 7 as NVA phloem cells (Fig. 1C, E, Additional file 1: Fig. S4). (9) Os05g0406100 and Os04g0412500 were classified as the enriched genes of cluster 6, but primarily expressed in Cluster 3/EVB xylem (Fig. 1C, E, Additional file 1: Fig. S4). Notably, ​Cluster 6​ was closely related to ​Cluster 7 (Fig. 1B), and the enriched genes from ​Cluster 3/EVB xylem​ also exhibited high expression in ​Cluster 6 (Additional file 1: Fig. S4). These findings suggest that ​Cluster 6​ might represent xylem of NVA or small VBs (SVB) (Fig. 1C, E, Additional file 1: Fig. S4), while further investigation is required to confirm.

Fig. 2.

Effect of OsMTP1 knockout on Zn accumulation and the schematic representation of Zn transport in node. A GFP localization in the transverse section of node I in the transgenic lines carrying ProOsMTP1:GFP. Parenchyma cell bridge, PCB; bundle sheath cells, BSC. B Short-term labeling experiment of ⁶⁷Zn. ⁶⁷Zn concentration was shown in different organs of wild-type and osmtp1 mutants. Plants grown in a 1/2 Kimura B solution until the filling stage were transferred to a solution containing 0.4 μM ⁶⁷Zn, 1 μM Rb and 1 μM Sr. After 2 d, different organs were sampled and subjected to element determination by ICP-MS. Internode, Int N. C, D Zn deposition (C) and relative Zn signal intensity in PCB and bundle sheath cells (BSC) (D) in the transverse section of node I in wild-type and osmtp1 mutants at the filling stage. The accumulation of Zn was measured by LA-ICP-MS. The relative Zn signal intensity was estimated based on the gray value by Photoshop. Data are means ± SD (n = 4 in B; 6 in D). Asterisks above bars indicate significant differences compared to WT (P < 0.05, Tukey’s test). E Schematic representation of Zn transport in node. OsZIP3 mediates Zn influx from EVB xylem to PCB, OsMTP1 mediates Zn efflux from cytosol to vacuole resulting in high Zn deposition in PCB and BSC, OsHMA2 mediates Zn influx from xylem to phloem in both EVB and DVB, and OsMT2b/2c binding Zn mediates Zn translocation in phloem

Additionally, (10) OsTF1L [45], OsGT1 [46] and OsSRL2 [47], mainly expressed in epidermal cells as confirmed by in situ hybridization previously, and OsHL6, regulating trichome formation in epidermal cells [48], showed expression patterns similar to the top two markers of Cluster 8. Furthermore, the enriched genes of Cluster 8 showed the highest correlation with the reported-enriched genes of epidermal cells from rice tissues of different organs [49], suggesting Cluster 8 may represent epidermal cells (Fig. 1C, E, Additional file 1: Fig. S4). (11) OsCOMT, enriched in Cluster 4 (Additional file 2: Table S2), and OsCAD2, expressed highly in sclerenchyma cells as confirmed by Promoter:GUS transgenic lines previously [50, 51], matched the top two markers of Cluster 4. The enriched genes of Cluster 4 showed the highest correlation with the reported-enriched genes of sclerenchyma [49], indicating Cluster 4 may be sclerenchyma identity (Fig. 1C, E, Additional file 1: Fig. S4).

Gene expression profile of transporters in rice node I

To clarify the expression profile of transporter genes in rice node, 1,144 putative transporter genes were obtained from the ARAMEMNON database (https://aramemnon.botanik.uni-koeln.de/index.ep). Eight hundred eighty-seven transporter genes showed varied expression levels in the 11 cell types of node I (Additional file 2: Table S3), indicating that they exhibited differential roles in various cell types. While 256 transporter genes were not detected in the snRNA-seq (Additional file 2: Table S3), suggesting probable minor roles in node I.

Identification of candidate transporter genes for ions deposition in rice node

Elements showed varied deposition patterns in node I through spatial ionomics analysis using laser ablation-ICP-MS (LA-ICP-MS) (Additional file 1: Fig. S6A). Fe was highly deposited in PCB and parenchyma cells (Additional file 1: Fig. S6A), which has been reported to be associated with OsVIT2 encoding a tonoplast-localized transporter for Fe accumulation in vacuole [52]. OsVIT2 showed high expression in PCB [52], consistent with snRNA-seq data (Additional file 1: Fig. S6B, C). OsVIT2 also showed some expression in parenchyma cells and sclerenchyma cells, with high correlation with Fe deposition in various tissues (Additional file 1: Fig. S6E).

In contrast, Zn was specifically and highly accumulated around PCB (Fig. 2C, Additional file 1: Fig. S6A). The tonoplast-localized OsMTP1, sequestrating Zn into vacuole [53], showed high gene expression in PCB and bundle sheath cells (Fig. 1C, Additional file 1: Fig. S4, S6B). There was a good correlation (R² = 0.6231) between OsMTP1 expression and Zn deposition in various tissues in node I (Additional file 1: Fig. S6E), while its exact role in Zn deposition in node was unclear. The ProOsMTP1:GFP transgenic lines were used to confirm its high expression in PCB and bundle sheath cells (Fig. 2A). Furthermore, osmtp1 mutants showed significantly decreased ⁶⁷Zn concentration in nodes (Fig. 2B), especially in PCB and bundle sheath cells (Fig. 2C, D), but significantly high Zn concentration in leaves, internodes, and panicles compared to WT (Fig. 2B). By contrast, OsMTP1 knockout did not affect the accumulation of Rb and Sr (Additional file 1: Fig. S7), serving as the negative control. These results indicated that OsMTP1 plays key roles in high deposition of Zn in PCB and bundle sheath cells, as well as for Zn distribution from nodes to other organs, together with OsZIP3, OsHMA2, OsMT2b, and OsMT2c (Fig. 2E).

OsSPX-MFS3, OsHMA4, OsMTP8.1 and OsABCC1 are the key tonoplast-localized transporters for sequestrating P, Cu, Mn and As into vacuole [24, 54–57], respectively. OsSPX-MFS3 was primarily expressed in PCB, and OsHMA4 was highly expressed in DVB/TVB xylem, NVA and PCB (Additional file 1: Fig. S6B, C). OsMTP8.1 showed high levels in cells of sclerenchyma, parenchyma and epidermis (Additional file 1: Fig. S6B, C), which was confirmed by immunostaining (Additional file 1: Fig. S6D). OsABCC1 was mainly expressed in EVB phloem (Additional file 1: Fig. S6B, C), consistent with the previous study [55]. There was a good positive correlation with R² > 0.5 between the four genes expression and element accumulation in various tissues of node I (Additional file 1: Fig. S6E), respectively. These results suggested that the tonoplast-localized transporters for vacuolar ions accumulation play key roles in element deposition within rice node.

Identification and expression pattern of OsMTP7

The cation diffusion facilitator (CDF)/metal tolerance protein (MTP) family plays important roles in Mn or Zn transport in plants [19, 58]. Ten MTP members were found in rice (Additional file 1: Fig. S8, S9A, Additional file 2: Table S3). Except for the high expression of OsMTP1 in PCB (Cluster 1) as its enriched gene (Figs. 1C, 2A, Additional file 1: Fig. S4), OsMTP9 and OsMTP7 were the enriched genes of bundle sheath cells (Cluster 5) and phloem of DVB/TVB (Cluster 0) in node I (Fig. 1C, Additional file 1: Fig. S4), respectively. The high expression of OsMTP9 in bundle sheath cells and PCB was confirmed by immunostaining (Additional file 1: Fig. S10). OsMTP1 and OsMTP9 mediated Zn or Mn transport, which play key roles in Zn and Mn homeostasis in rice (Fig. 2) [22, 53], respectively, while the role of OsMTP7 remains unknown.

The cloned open reading frame (ORF) of OsMTP7 contained 1,419 bp, encoding a peptide with 472 amino acids containing four putative transmembrane domains (Additional file 1: Fig. S9B, C). OsMTP7 was highly expressed in the basal stems and roots at the vegetative stage (Fig. 3A), and in the flowers (stamen + pistil), rachis, and node I at the flowering stage (Fig. 3B). Furthermore, OsMTP7 showed a five times higher level in the stamen than in the pistil (Fig. 3C). The expression of OsMTP7 in roots was hardly affected by the deficiency or excess of Mn, Zn, Cu, or Cd (Additional file 1: Fig. S11A).

Fig. 3.

Gene expression pattern and subcellular localization of OsMTP7. A-C Relative expression of OsMTP7 in various organs of rice at the vegetative stage (A) and flowering stage (B and C). Tubulin was used as an internal standard. The expression level was determined using quantitative RT-PCR. Expression level relative to roots (A and B) or stamen (C) is shown. Data are means ± SD (n = 3). Different letters above bars indicate significant differences (P < 0.05, Tukey’s test). D-G Tissue-specific expression of OsMTP7. Immunostaining with a GFP antibody was performed in the node I (D), main root tip (E), basal root zone (F), and the GFP fluorescence was detected in the anther (G) in the transgenic lines carrying ProOsMTP7:GFP. Red color indicates the GFP antibody-specific signal and blue color indicates cell wall autofluorescence (D-F). Phloem area of enlarged vascular bundle (PE), phloem area of diffuse vascular bundle (PD), xylem area of enlarged vascular bundle (XE), xylem area of diffuse vascular bundle (XD), parenchyma cell bridge (PCB), epidermis (Ep), root hair (RH), and endodermis (En) are shown. Bars, 100 μm. H-K Subcellular localization of OsMTP7. Construct of 35S:GFP-OsMTP7 was transiently expressed in the protoplasts prepared from rice shoots. FM4-64 was used as a plasma membrane-localized marker (J). Bright field (H), GFP image (I), FM4-64 image (J), and merged image (K) are shown. Bars, 5 μm

To confirm the tissue expression of OsMTP7, the transgenic rice lines expressing GFP driven by OsMTP7 promoter were employed for immunostaining with an antibody against GFP. Consistent with the snRNA-seq data with enrichment in clusters 0 and 9 (Fig. 1C, Additional file 1: Fig. S4), a strong GFP signal was detected in the phloem region of DVB and EVB in node I (Fig. 3D), which was further confirmed by qRT-PCR using four types of tissues of node I separated by laser microdissection (LMD) (Additional file 1: Fig. S11B). Additionally, GFP signal was highly detected in all cells except the epidermis in the root tip (Fig. 3E). Some GFP signal was found in the exodermis, sclerenchyma, and cortex in the basal root, but the strong GFP signal was observed in the endodermis and stele cells (Fig. 3F). Besides, a strong GFP fluorescence was detected in the pollen and vascular tissues of anther (Fig. 3G). In contrast, no GFP signal was detected in the root, node I, and anther of the WT plants (Additional file 1: Fig. S11C-E). These results implicated that OsMTP7 was constitutively expressed in rice and may play roles in root, phloem of node, and stamen.

Subcellular localization of OsMTP7

The subcellular localization of OsMTP7 was analyzed by transiently expressing OsMTP7-GFP and GFP-OsMTP7 in rice protoplasts. The signal of OsMTP7-GFP and GFP-OsMTP7 were localized to the cell periphery of the protoplast, and were highly co-localized with the plasma membrane markers including FM4-64 (Fig. 3H-K, Additional file 1: Fig. S12E-H) and AtPIP2a-mCherry (Additional file 1: Fig. S12M-T) [59]. In contrast, the signal of GFP alone was mainly localized to the cytosol and nuclei and hardly co-localized with that of FM4-64 and AtPIP2a-mCherry (Additional file 1: Fig. S12A-D, I-L). The results indicated that OsMTP7 was localized to plasma membrane in rice.

Effect of OsMTP7 knockout on Mn homeostasis

To investigate the physiological role of OsMTP7 in rice, three independent knockout lines of OsMTP7, osmtp7-1 (one G deletion at target 1), osmtp7-2 (one T insertion at target 2), and osmtp7-3 (35 bp fragment deletion at target 2), were generated using CRISPR/Cas9 technology (Additional file 1: Fig. S13A). Knockout of OsMTP7 hardly affected panicle number, but significantly inhibited plant height and dramatically reduced the dry weight of straw by 38%, panicles by 68%, and grains by 53%, as well as the fertility by 60% compared to WT at the repining stage grown in soil (Fig. 4A, Additional file 1: Fig. S13B-K). Remarkably, three osmtp7 mutants showed greatly lower Mn concentrations in the nodes, leaf sheaths, leaf blades, and total straw by 39% and brown rice by 17% (Fig. 4B, C, Additional file 1: Fig. S14C, D), but comparable concentrations of Fe, Cu, and Zn in the straw and brown rice compared to WT through ionomics analysis using ICP-MS (Additional file 1: Fig. S14). Furthermore, the distribution of Mn was significantly higher in the nodes and leaf blades and dramatically lower in the grains of osmtp7 mutants compared to WT (Fig. 4D).

Fig. 4.

Effect of OsMTP7 knockout on plant growth and Mn homeostasis. A-D Plant growth and Mn homeostasis at the ripening stage. Dry weight of straw and panicle (A), Mn concentration in various organs (node, leaf sheath, leaf blade, and inter node) of straw (B) and grain (C), and Mn distribution in various organs (D) in wild-type and osmtp7 mutants. Plants were grown in the potted soil and sampled at the ripening stage. Element concentration was determined using ICP-MS. E-N Plant growth and Mn homeostasis at the vegetative stage. Phenotype of plants and various organs (E), dry weight (F, K), Mn concentration (G, L), and Mn distribution (H, M) in various organs (roots, basal stems, leaf 2–8, or leaves), Mn uptake (I, N), and Mn concentration in the xylem sap (J) in wild-type and osmtp7 mutants. Plants were grown in a ½ Kimura B solution containing normal Mn with 0.5 μM for 6 weeks (E-J) or increased Mn with 5 μM for 4 weeks (K-N). Bars = 20 cm in (E). Data are means ± SD (n = 5 in A-D; 4 in F-J; 3 in K-N). Asterisks above bars indicate significant differences compared to WT (P < 0.05, Tukey’s test)

To reveal how OsMTP7 is involved in plant growth and Mn accumulation, hydroponic experiments were conducted. Under normal 0.5 μM Mn condition, significantly shorter leaf 8 (youngest), as well as dramatically lower dry weight of root, basal stems, and leaf 8 were observed in osmtp7 mutants compared to WT (Fig. 4E, F). Ionomics analysis demonstrated that OsMTP7 knockout significantly decreased Mn concentration in all examined organs compared to WT (Fig. 4G), which was consistent with the inhibited total Mn uptake via root by 13% and decreased Mn concentration in xylem sap by 31% (Fig. 4I, J). Furthermore, OsMTP7 knockout increased Mn distribution in the old leaves 4–6 but dramatically decreased Mn allocation in the youngest leaf 8 compared to WT (Fig. 4H). Besides, OsMTP7 knockout hardly affected Fe concentration in the examined organs, but partially decreased Cu or Zn concentrations in some leaves and partially increased Zn concentration in the roots (Additional file 1: Fig. S15A-C). The uptake of Fe, Cu, or Zn was comparable among the genotypes (Additional file 1: Fig. S15D-F).

To get rid of the effect of other Mn transporters in osmtp7 mutants, the expression of OsNRAMP5, OsNRAMP1, OsMTP9, OsNRAMP3, and OsYSL2, involved in Mn uptake and translocation [21–23, 28, 30], and OsYSL6, OsMTP8.1, and OsMTP8.2, involved in Mn detoxification [24, 25, 29], were analyzed. OsMTP7 knockout did not alter the expression of the eight genes in the roots (Additional file 1: Fig. S16), implying the direct involvement of OsMTP7 in Mn homeostasis in rice.

Furthermore, under a high 5 μM Mn condition, lower dry weight of roots and leaves (Fig. 4K) and lower Mn uptake (Fig. 4N) were also found in osmtp7 mutants compared to WT. While significantly increased Mn concentration and distribution in roots and basal stems, and significantly decreased Mn concentration and distribution in leaves were detected in osmtp7 mutants compared to WT (Fig. 4L, M). These results indicated that rice OsMTP7 plays important roles in Mn uptake via root, Mn translocation from root to shoot through xylem especially in high Mn condition, and Mn distribution from nodes and old leaves to new leaf and grain via nodes, which is crucial for plant growth and yield.

Role of OsMTP7 in fertility

Due to the extremely decreased fertility in osmtp7 mutants and the high expression of OsMTP7 in stamen (Fig. 3B, C, Additional file 1: Fig. S13F, I), the anther growth and pollen viability were compared among the genotypes. osmtp7 mutants showed partially distorted anthers (Additional file 1: Fig. S17A) and dramatically lower pollen number and pollen viability (Fig. 5A-C) compared to WT. To understand the mechanism underlying the low fertility of osmtp7 mutants, bulk RNA-seq analysis was conducted using the spikelets of WT and osmtp7 mutants. 322 upregulated genes and 472 downregulated genes were identified in both osmtp7 lines compared to WT (Additional file 1: Fig. S17B, C). Gene ontology (GO) analysis revealed enrichment in processes such as oxidoreductase activity, nutrition metabolism, hormone metabolism, mitochondrion activity, etc. (Fig. 5D). Four selected groups of genes, involved in plant growth and development, the synthesis and metabolic processes of cell wall, hormone and reactive oxygen, were shown (Fig. 5E). The expression of 17 reported or putative genes for the above four functions were further confirmed through qRT-PCR (Fig. 5F) [60–71].

Fig. 5.

Effect of OsMTP7 knockout on fertility at the flowering stage. A-C Pollen viability determination. Pollen staining by 1% KI–I₂ (A) and pollen number in each same view field (B), and pollen viability (C) in wild-type and osmtp7 mutants. Blue arrowhead indicates normal pollen (black color), and red arrowhead indicates abnormal pollen (yellow color). D-F Bulk RNA-seq analysis of spikelets before flowering. Gene ontology (GO) analysis of differentially expressed genes (DEGs), hierarchical clustering and expression heatmap of selected DEGs (E), and DEGs validation by qRT-PCR for 17 genes (F) in wild-type and osmtp7 mutants (D). G-H O²⁻ staining (G) and relative O²⁻ level (H) in anther of wild-type and osmtp7 mutants. The spikelets were sampled before flowering to stain O²⁻ using nitroblue tetrazolium (NBT). The relative O²⁻ level in anther was estimated based on the gray value by Photoshop. Plants were grown in the potted soil and sampled at the flowering stage. Bars = 100 μm in (A); 2 mm in (G). Data are means ± SD (n = 5 in B and C; 3 in F; 9 in H). Asterisks above bars indicate significant differences compared to WT (P < 0.05, Tukey’s test)

Since the genes for oxidoreductase activity were affected (Fig. 5D, E), Mn functions in Mn-SOD for reactive oxygen species (ROS) scavenging, and the fine ROS homeostasis is crucial for fertility [72–74], the superoxide anion (O²⁻) level in the spikelets was determined. A significantly higher level of O²⁻ in the anthers was observed in osmtp7 mutants than in the WT (Fig. 5G, H). These results indicated that knockout of OsMTP7 disturbed genes expression involved in synthesis and metabolic processes for growth and development in spikelets, leading to increased ROS level and oxidative stress in anther, resulting in low fertility and yield loss.

Transport activity test of OsMTP7 for elements in yeast

To investigate the transport activity of OsMTP7 for elements, OsMTP7 was expressed in yeast. OsMTP9, the plasma membrane-localized Mn efflux transporter [22], was employed as the positive control. The yeast cells carrying pYES2 empty vector, pYES2-OsMTP7, or pYES2-OsMTP9 showed similar growth on the plate of control or glucose (Fig. 6A, B). While on the plates containing raffinose and high Mn, the yeast cells carrying OsMTP9 showed better growth than that carrying pYES2 (Fig. 6A), consistent with the previous report [22]. Unexpectedly, the yeast cells carrying OsMTP7 showed increased sensitivity to high Mn compared to that carrying pYES2 (Fig. 6A). Since the mislocalization of the transporter in yeast could affect the yeast growth phenotype under metal stress [54, 75], the subcellular localization of OsMTP7 in yeast was investigated. GFP-OsMTP7 was found to be localized to the inner membrane in yeast (Additional file 1: Fig. S18A), different from the plasma membrane localization of OsMTP7 in rice (Fig. 3I, Additional file 1: Fig. S12). To direct OsMTP7 to yeast plasma membrane, OsMTP7 was fused with the soluble C-terminal domain of Ist2p (Ist2^C), a dominant driving protein to plasma membrane (Additional file 1: Fig. S18B) [76]. The yeast cells carrying OsMTP7-Ist2^C showed increased tolerance to high Mn compared to that carrying pYES2 under raffinose-induced condition (Fig. 6B). On the other hand, expression of OsMTP7 or OsMTP7-Ist2^C in yeast hardly affected the tolerance to Cu, Fe and Zn (Additional file 1: Fig. S19), consistent with no difference in their concentration between osmtp7 mutants and WT (Additional file 1: Fig. S14). These results indicated that OsMTP7 probably shows efflux transport activity for Mn, but not Cu, Fe and Zn.

Fig. 6.

Transport activity of OsMTP7 in yeast and the schematic representation of Mn transporters in rice. A, B Transport activity of OsMTP7 for Mn in yeast. The yeast carrying pYES2 empty vector, pYES2-OsMTP7, pYES2-OsMTP9, or pYES2-OsMTP7-lst2^c was cultured on a SD-U medium plate with glucose or raffinose containing 20, 25 or 30 mM MnCl₂ for 2 or 3 d. C Schematic representation of Mn transporters in root, node and anther of rice. OsMTP7 is expressed highly in endodermis and stele cells of root, phloem cells in node, and pollen and vascular tissue in anther. OsMTP7 is a plasma membrane-localized transporter for Mn efflux. OsMTP7 is involved in Mn uptake, translocation from root to shoot through xylem, and phloem loading for distribution in node, which is important for plant growth and fertility in rice. Besides, OsNRAMP1/3/5 mediate Mn influx and OsMTP9 mediates Mn efflux in root and node

Discussion

Rice nodes contain complex tissues and well-organized vascular systems. Several key tissues, such as EVB, TVB, DVB, SVB, PCB and NVA, have been identified based on the anatomical morphology analysis [2, 3, 8]. We here used snRNA-seq to identify 11 distinct cell clusters at the single-cell resolution depending on ​1,837 cluster-enriched genes​ in node I, and further annotated each cell type depending on the marker or enriched genes (Fig. 1, Additional file 1: Fig. S1-4, Additional file 2: Table S2). Although a lack of true biological replicates for each growth stage, the marker or enriched gene expression patterns from snRNA-seq were highly consistent with that of the reported or newly confirmed genes in node (Figs. 1, 2, 3, Additional file 1: Fig. S5-6, S10-11), confirming the high reliability of snRNA-seq data. While other sub-cell clusters could be further divided in the future through additional biological replicates and in-depth data mining, such as cluster 0/phloem of DVB/TVB, cluster 2/xylem of DVB/TVB, and cluster 4/sclerenchyma cells (Fig. 1), which would further promote the studies of the cellular developmental trajectories [32, 33]. Importantly, rice node acts as a hub for distributing mineral elements depending on various transporters [2, 3], while the expression pattern and the role of most of them in node are uncovered. We here systematically profiled the expression patterns of 1,144 putative transporter genes across 11 cell types (Additional file 2:Table S3), although 256 of them were not detected, which will serve as an important database for identifying key genes in node.

Rice nodes not only accumulate high concentration of elements [44, 55, 77], but also show specific deposition patterns for differential elements (Additional file 1: Fig. S6) [78, 79], which is important for the storage and reutilization of elements. We found that the expression patterns of OsMTP1, OsVIT2, OsSPX-MFS3, OsHMA4, OsMTP8.1 and OsABCC1, which encoding the tonoplast-localized transporter for vacuolar ions accumulation [24, 52–57], showed strong correlation with the specific deposition patterns of Zn, Fe, P, Cu, Mn and As in differential tissues in node I (Fig. 2, Additional file 1: Fig. S6), respectively. Knockout of OsMTP1 or OsVIT2 confirmed its key role in Zn or Fe deposition around PCB (Fig. 2) [52], respectively, indicating the important role of transporters for vacuolar ions accumulation in the deposition patterns of elements in rice node. On the other hand, knockout of OsMTP1 or OsVIT2 greatly increased Zn or Fe concentration in grain because of the decreased Zn or Fe accumulation in nodes (Fig. 2) [52, 53], suggesting that our data will provide candidate genes for rice breeding with higher mineral nutrients and lower toxic elements.

CDF/MTP members in rice and Arabidopsis were divided into two subgroups. OsMTP7 belonged to the clade together with OsMTP8.1, OsMTP8.2, AtMTP8, OsMTP9, OsMTP11, and AtMTP11 (Additional file 1: Fig. S9A), which have been reported to mediate Mn efflux from cytosol [22, 24, 26, 80–82]. OsMTP7 was localized to plasma membrane in rice (Fig. 3H-K, Additional file 1: Fig. S12), and the localization of OsMTP7 to plasma membrane in yeast resulted in increased tolerance to high Mn (Fig. 6B, Additional file 1: Fig. S18B), similar to the role of OsMTP9 in yeast (Fig. 6A) [22], suggesting its probable efflux ability for Mn (Fig. 6C). Besides, OsMTP7 likely showed no transport activity for Cu, Fe and Zn (Additional file 1: Fig. S14, S19). In rice roots, OsNRAMP5 and OsMTP9 were specifically expressed in exodermis and endodermis cells (Fig. 6C) [21, 22], and OsNRAMP1 was expressed in the cells except for stele cells (Fig. 6C) [23]. In contrast, OsMTP7 was likely a novel Mn efflux transporter gene that was highly expressed in stele cells (Figs. 3F, 6C). Additionally, OsMTP7 was also expressed in almost all cells in main root tip and partially in other cells in root basal zone (Figs. 3E, F, 6C). The polarly localized OsNRAMP5 and OsMTP9 have been suggested to compose the effective Mn uptake system in rice with the supplement of OsNRAMP1 (Fig. 6C) [21–23]. Knockout of OsMTP7 did not alter the expression levels of OsNRAMP5, OsNRAMP1, OsMTP9, and other Mn transporter genes in roots (Additional file 1: Fig. S16), hardly affected the Fe, Cu and Zn accumulation (Additional file 1: Fig. S14), but decreased Mn uptake, Mn concentration in xylem sap and Mn translocation from roots to leaves (Fig. 4), resulting in greatly decreased Mn concentration in roots, leaves, straw, and grain (Fig. 4, Additional file 1: Fig. S14C, D). These results indicated that OsMTP7 plays important roles in the uptake, radial transport, and root-to-shoot translocation of Mn in roots (Fig. 6C).

OsMTP7 was highly expressed in phloem region of both EVB and DVB in rice node (Figs. 1C, 3D). Knockout of OsMTP7 decreased Mn distribution from basal stems to leaves and from old leaves to the newest leaf at the vegetative stage, and from nodes and leaves to grains at the reproductive stage (Fig. 4D, H). The effect of OsMTP7 on Mn distribution was similar as OsNRAMP3 on Mn distribution (Fig. 6C, Additional file 1: Fig. S20) [30], OsHMA2 and OsMT2b/2c on Zn distribution [17, 44], and OsYSL16 on Cu distribution [83] through phloem in node, indicating that OsMTP7 was involved in mediating Mn distribution in nodes by loading Mn to the phloem of EVB and DVB (Fig. 6C). Additionally, OsMTP9, encoding efflux transporter for Mn in root exodermis and endodermis cells [22], was detected to be expressed primally in PCB and bundle sheath cells in node I (Fig. 1C, Additional file 1: Fig. S10), suggesting that OsMTP9 may be involved in Mn efflux from bundle sheath cells and PCB to DVB for the inter-vascular transfer of Mn (Fig. 6C), though further research is required to confirm this hypothesis.

Knockout of OsNRAMP5, OsNRAMP1, OsMTP9 or OsNRAMP3 resulted in inhibited growth or low fertility [21–23, 30]. Similarly, OsMTP7 knockout reduced growth, fertility and production, which could be attributed to Mn deficiency in the roots, leaves, straw, and grain (Fig. 4, Additional file 1: Fig. S13-14). Although the exact reasons for the inhibition in mutants of OsNRAMP1/3/5 and OsMTP7/9 might vary, decreased PSII quantum yield was found in osmtp9 mutants [22], indicating that Mn deficiency-induced photosynthesis inhibition may be the cause, because Mn acts as a cofactor in PSII supercomplex. Besides, the decreased fertility in osmtp7 mutants was presumably a combination of the retarded anther development and inhibited pollen viability (Fig. 5A-C, Additional file 1: Fig. S17A), where OsMTP7 was highly expressed (Fig. 3B, C, G). Further research through bulk RNA-seq using spikelets revealed that OsMTP7 knockout altered the expression of genes involved in multiple processes, which were important for plant growth and development, the synthesis and metabolic processes of cell wall, hormone and reactive oxygen (Fig. 5D-F) [60–71]. This altered gene expression profile was consistent with the defects in the biosynthesis and deposition of cell wall polysaccharides and ROS homeostasis resulted from disturbed Mn homeostasis in the previous studies [27, 84–86]. Moreover, increased ROS level and oxidative stress in the anther of osmtp7 mutants were confirmed (Fig. 5G, H), which may be attributed to the inhibited Mn-SOD ability that is important for ROS scavenging [72].

Conclusions

This study identifies 11 distinct cell clusters at the single-cell resolution in rice node I using snRNA-seq, and annotates each cell type through multiple marker or enriched genes. This study provides the expression profile of various transporter genes in the 11 cell clusters, and candidate transporter genes for the varied deposition patterns of multiple elements in rice node I, which will serve as an important database for identifying key genes in node. Furthermore, this study identifies OsMTP7 as a novel Mn efflux transporter gene that is highly expressed in node phloem and root stele cells. OsMTP7 mediates Mn uptake, translocation and distribution for improving growth and yield in rice.

Methods

Plant materials and hydroponic conditions

Wild-type (WT) rice (Oryza sativa cv. Nipponbare), the transgenic lines carrying ProOsPHO1;1:GFP, ProOsMT2c:GFP, ProOsMTP1:GFP, or ProOsMTP7:GFP, and three independent knockout lines of OsMTP1 or OsMTP7 generated as described below, were used in this study. Seeds were soaked in water in the dark at 30 °C for 2–3 d for germination and were then placed on a plastic net floating on a 0.5 mM CaCl₂ solution (pH 5.6). After 5 d, the seedlings were transferred to ½ Kimura B solution (pH 5.6) in a plastic pot. The nutrient solution was changed every 3 d. Plants were grown in a controlled greenhouse at 25–35 °C with natural sunlight in Kurashiki, Japan, or Kunming, China.

The ½ Kimura B solution contained macronutrients (mM) including (NH₄)₂SO₄ (0.18), MgSO₄⋅7H₂O (0.27), KNO₃ (0.09), Ca(NO₃)₂⋅4H₂O (0.18), and KH₂PO₄ (0.09), and the micronutrients (μM) including MnCl₂⋅4H₂O (0.5), H₃BO₃ (3.0), (NH₄)₆Mo₇O₂₄⋅4H₂O (1.0), ZnSO₄⋅7H₂O (0.4), CuSO₄⋅5H₂O (0.2), and Fe-EDTA (20) [87].

Preparation of node I samples and nucleus isolation

Segments of Node I were precisely excised using a sterile razor blade from over 20 individual rice (cv. Nipponbare) plants grown in ½ Kimura B solution at the booting stage (RNA3), flowering stage (RNA2), and filling stage (RNA1), respectively. The harvested node I samples were immediately immersed in ice-cold water and maintained on ice to preserve cellular integrity before subsequent nuclear isolation and fluorescence-activated nucleus sorting, as previously described [88–90]. Briefly, plant tissue samples were collected and cut into small pieces (< 0.5 cm) and then homogenized using a glass Dounce tissue grinder (Sigma, cat. no. D8938) in an ice-cold nucleus lysis buffer. The tissue was homogenized with pestle A for 25 strokes, followed by 25 strokes with pestle B in 2 mL of the lysis buffer. The homogenate was incubated on ice for 5 min, after which an additional 3 mL of cold EZ lysis buffer was added, and the sample was further incubated on ice for another 5 min to ensure complete nuclear release. The sample was then centrifuged at 500 g for 5 min at 4 °C to pellet the nucleus. The resulting nucleus pellet was washed with 5 mL of ice-cold EZ lysis buffer and incubated on ice for an additional 5 min. After centrifugation, the pellet was resuspended in 5 mL of nucleus suspension buffer (NSB, containing 1 × PBS, 0.01% BSA, and 0.1% RNase inhibitor (Clontech, cat. no. 2313 A). The suspension was filtered through a 35 µm cell strainer (Corning-Falcon, cat. no. 352235) to remove any debris. Nuclei were then counted using a hemocytometer. Finally, the suspension was adjusted to a concentration of 1,000 nuclei/µL for loading onto a 10 × Genomics platform.

SnRNA-seq library preparation and sequencing

SnRNA-seq libraries were prepared using the Chromium Next GEM Single Cell 3' Reagent Kits v3.1 (10 × Genomics) on a Chromium Controller. Nuclei were isolated from node I and resuspended in PBS containing 0.04% BSA. Nuclei from each developmental stage were separately loaded onto the Chromium Next GEM Chip G to generate single-cell gel beads in emulsion (GEMs), according to the manufacturer's instructions. The captured nuclei were subsequently lysed, and the released RNA underwent reverse transcription to produce barcoded, full-length cDNA. Libraries were constructed following the protocol provided by the reagent kit. The quality of the libraries was assessed using the Qubit 4.0 Fluorometer and the Agilent 2100 Bioanalyzer. Sequencing was performed on the Illumina NovaSeq 6000 platform, achieving a sequencing depth of at least 50,000 reads per nucleus, with paired-end 150 bp (PE150) reads.

Data processing and dimensionality reduction for snRNA-seq of 10 × genomics

High-throughput single-nucleus transcriptome sequencing (snRNA-seq) of each rice node I sample was performed using the 10 × Genomics platform. Raw sequencing data were aligned to the rice (cv. Nipponbare) reference genome (T2T) (http://www.ricesuperpir.com/web/nip) using the STAR aligner [91], and adapter sequences and low-quality reads were filtered out using FASTP to ensure data quality. Cell barcodes were identified, and unique molecular identifiers (UMIs) were counted using UMI-TOOLS. Following alignment, Cell Ranger v7.0 was used for cell-calling to identify barcodes containing valid nuclei, which were further filtered to remove low-quality nuclei and potential doublets. Specifically, nuclei were retained if: (i) the number of detected genes was between 500 and 7,000, (ii) the total UMI count per nucleus was ≥ 100. And genes expressed in at least ten nuclei were retained for downstream analysis.

Subsequently, differential expression analysis, clustering, and cell-type identification were performed in Seurat (v4.0.1). Gene expression values were normalized with the LogNormalize method, and the top 2,000 highly variable genes (HVGs) were selected. These HVGs were subjected to standardization before principal component analysis (PCA). The top 30 principal components (PCs) were used for nonlinear dimensionality reduction and visualization by t-distributed stochastic neighbor embedding (t-SNE) and uniform manifold approximation and projection (UMAP) [92, 93]. Clustering was performed using the shared nearest neighbor (SNN) modularity optimization algorithm.

Cluster-specific marker genes were identified using the FindAllMarkers function in Seurat. A gene was considered a marker if it was expressed in at least 25% of nuclei within the cluster, with |log₂ fold change|> 0.4 and an adjusted p-value (q) ≤ 0.01.

RNA in situ hybridization

Sense and antisense RNA probes targeting the parenchyma cell marker gene Os09g0333600 were synthesized in vitro using the Digoxigenin (DIG) RNA Labeling Kit (Roche), with the primer sequences listed in Additional file 2: Table S5, and applied to tissue sections for RNA in situ hybridization. Tissue samples consisted of node I collected at the flowering stage of rice (cv. Nipponbare). Paraffin-embedded Sects. (10 µm thick) and whole-mount preparations were subjected to RNA in situ hybridization according to previously published protocols [47, 94]. Hybridization was carried out at 50 °C for 16 h. After washing, the slides were incubated with an anti-digoxigenin antibody (Roche), the signal of which was then detected in the nitro-blue tetrazolium/5-bromo-4- chloro-30-indolyphosphate stock solution (NBT/BCIP solution; Roche).

Tissue-specific gene expression or protein localization determination

The transgenic lines carrying ProOsMT2c:GFP containing 2,040 bp fragment upstream of the ATG of OsMT2c, ProOsMTP1:GFP containing 4,566 bp fragment upstream of the ATG of OsMTP1, or ProOsPHO1;1:GFP containing 2.0 kb fragment upstream of the ATG of OsPHO1;1 were constructed and used in previous studies [12, 44, 53]. To construct ProOsMTP7:GFP transgenic lines, genomic DNA of rice (cv. Nipponbare) was extracted using the phenol/chloroform method. The 2,514 bp fragment upstream of the ATG of OsMTP7 was amplified by PCR from genomic DNA and then cloned into pPZP2H-lac carrying Green Fluorescent Protein (GFP) and the terminator of the nopaline synthase gene [95]. This construct was introduced into Agrobacterium tumefaciens (strain EHA101) for transformation into rice callus (cv. Nipponbare) [96]. The roots from plants grown in the hydroponic solution, and the node I and anthers from plants grown in soil were sampled and cut into slices using a vibrating slicer.

Immunostaining was performed using an antibody against GFP (A11122; Molecular Probes, MA, USA), OsLsi6, OsHMA5, OsHMA2, OsLsi2, OsLsi3, OsMTP8.1 or OsMTP9 as described previously [9, 13, 17, 22, 24]. The fluorescence of the secondary antibody (Alexa Fluor 555 goat anti-rabbit IgG, Molecular Probes) was recorded using a laser scanning microscope (Leica TCS SP8 X). GFP fluorescence in the anthers or node I was observed using a laser scanning microscope. Primers used are listed in Additional file 2: Table S5.

Analysis of plant growth and mineral accumulation

In the hydroponic experiments, rice seedlings of WT, osmtp1 or osmtp7 mutants, generated as described below, were cultured in ½ Kimura B solution containing normal Mn with 0.5 μM for 6 weeks or increased Mn with 5 μM for 4 weeks, or grown in ½ Kimura B solution until the filling stage and then transferred to a solution containing 0.4 μM ⁶⁷Zn, 1 μM Rb and 1 μM Sr for 2 d. Various organs, including roots, basal stems, leaves, nodes, internodes, rachis, husk or brown rice, were sampled. To collect xylem sap, rice seedlings were grown in ½ Kimura B solution, the shoot (2 cm above the root) was removed with a razor, and the xylem sap was collected for 1 h using a micropipette.

In the potted soil experiments, rice seedlings were grown on the potted soil (3.5 L) in a closed glasshouse at 25–35 °C under natural sunlight. After ripening, the plant height and panicle number were recorded, and the shoot and panicle were photographed. The weights of the straw, panicles, and grains and the fertility were determined after drying in a 40 °C oven for several days, as described previously [97].

Concentrations of elements in each sample were determined using inductively coupled plasma MS (ICP-MS) after digestion as described in the following section. Uptake of element = elemental content in plant/root dry weight.

Spatial ionomics analysis through LA-ICP-MS

Node I was sampled at filling stage and element deposition was determined (two replicates) using laser ablation-ICP-MS (LA, NWR213, New Wave Research; ICP-MS 7700X, Agilent Technologies, CA, USA) based on previously described procedures [44, 79].

Clone of OsMTP7

Total RNA was extracted from rice (cv. Nipponbare) using a kit (Qiagen, http://www.qiagen.com; or Omega Bio-Tek, Norcross, GA, USA), and 500 ng of RNA was converted to cDNA using ReverTra Ace qPCR RT Master Mix (SuperScript II, Invitrogen). The full-length coding region of OsMTP7 (Os04g0298200 or LOC_Os04g23180) was amplified from cDNA using PCR and primers (Additional file 2: Table S5) that were designed based on the Rice Annotation Project database (RAP, http://rapdb.dna.affrc.go.jp/). The amplified fragment was cloned into a pGEM-T vector (Promega, https://www.promega.com/), and the sequence was confirmed using an ABI Prism 3130 sequence analyzer (Applied Biosystems, http://www.appliedbiosystems.com/).

Gene expression analysis

For analysis of OsMTP7 expression in differential organs, various organs (roots, basal stems, leaf sheaths, leaf blades, internodes, nodes, rachises, husks, stamens, or pistils) were sampled from rice (cv. Nipponbare) grown on soil at the vegetative stage (4-w-old) and flowering stage (14-w-old). For the metal response analysis of OsMTP7 expression in the roots, rice seedlings (cv. Nipponbare) were exposed to ½ Kimura B solution with (CK) or without Mn, Zn, or Cu, or containing 200 μM MnCl₂, 100 μM ZnSO₄, 2 μM CuSO₄, or 1 μM CdSO₄. The roots were sampled for RNA extraction after 1 week. Tissue specific expression in node I was investigated using samples prepared by laser microdissection (Applied Biosystems Arcturus Laser Capture Microdissection System, Life Technologies, Carlsbad, CA, USA) as described previously [55]. For expression analysis of 8 genes involved in Mn homeostasis, the roots were sampled from plants grown in ½ Kimura B solution for 1 month. For expression analysis of 17 genes in spikelets, the spikelets were sampled from plants grown in soil before flowering. Total RNA and cDNA were prepared as described previously. Tubulin was used as the internal standard. Gene expression levels were determined using a kit (Thunderbird SYBR qPCR Mix, Toyobo, Osaka, Japan; or SuperReal PreMix Plus SYBR Green, Tiangen) and qRT-PCR (CFX96 Touch, Bio-Rad; or QuantStudio7 Flex, Applied Biosystems, Foster City, CA, USA). Primers used here were listed in Additional file 2: Table S5.

Subcellular localization

The open reading frame (ORF) of OsMTP7 was cloned and ligated to the 5ʹ or 3ʹ end of GFP under CaMV35S promoter in pUC18 (Takara, Kyoto, Japan) as described previously [53, 54]. FM4-64 or AtPIP2a-mCherry was employed as a plasma membrane-localized marker [59]. The constructs were introduced into rice leaf protoplasts using the polyethylene glycol method [98]. The fluorescence signal was recorded using a laser scanning microscope (OLYMPVS, FV3000). Primers used here were listed in Additional file 2: Table S5.

Generation of rice mutants

The three osmtp1 mutant lines were constructed through CRISPR/Cas9, which are loss-of-function mutants used in previous studies [53]. Two target sites of OsMTP7 were designed using CRISPR-P v2.0 (http://crispr.hzau.edu.cn/cgi-bin/CRISPR2/CRISPR). The sequences were synthesized and cloned into the Bbs I-linearized pU6gRNA vector, respectively [99]. OsU3-gYSA in pZDgRNA-Cas9ver.2_HPT was replaced by the above synthetic gRNA construct using Asc I and Pac I restriction sites [99]. The final constructs were transformed into calluses (cv Nipponbare) through Agrobacterium transformation [96]. A genomic fragment was amplified from the transgenic rice plants using PCR and primer pairs flanking the designed target sites, and the amplified fragment was sequenced directly to identify the genotype of the mutants. T3 generations of three independent osmtp7 mutants were used for the analysis as followed. Primer sequences were listed in Additional file 2: Table S5.

Analysis of flowers and pollen viability

Before flowering, fresh rice spikelets were collected and half the husk was removed for photographing with a microscope. For pollen viability measurement, the pollen grains from equal anthers of the genotypes were collected and stained with 1% KI-I₂ solution for 5 min, and then photographed under an optical microscope [44]. Black or yellow color pollen was recorded as normal or abnormal pollen, respectively. Pollen viability ratio = pollen stained with black color/total pollen on the screen × 100.

Bulk RNA-seq analysis

The spikelets were sampled before flowering from wild-type rice and two osmtp7 mutants grown in soil. RNA was extracted and assessed using RNA Nano 6000 Assay Kit of Bioanalyzer 2100 system (Agilent Technologies, CA, USA). Qualified RNA was used to construct cDNA library for RNA-Seq analysis on the Illumina HiSeq™ X TEN platform (Novogene Bioinformatics Technology Co., Ltd. -Tianjin, Beijing, China). The feature Counts v1.5.0-p3 was used to count read numbers mapped to each gene. The FPKM of each gene was calculated based on gene length and the read count mapped to it.

Histochemical assays of superoxide anion (O²⁻) in anther

Before flowering, fresh rice spikelets were collected and stained with the solution (pH 7.8, 120 mM K₂HPO₄, 9 mM KH₂PO₄, 0.6 mM nitro blue tetrazolium, NBT) after removing half of the husk. After 1 h, the spikelets were washed with 95% alcohol and photographed under an optical microscope [73]. The relative O²⁻ level in the anthers was estimated based on the gray value using Photoshop.

Transport activity assay of OsMTP7 in yeast

The OsMTP7, OsMTP9, GFP-OsMTP7, or GFP-OsMTP7-Ist2^C was amplified by PCR using the primers listed in Additional file 2: Table S5 and then cloned into the pYES2 vector. The pYES2 empty vector, pYES2-OsMTP7, pYES2-OsMTP9, pYES2-GFP-OsMTP7 or pYES2-GFP-OsMTP7-Ist2^C was transformed into yeast strain △ycf1 for Mn and BY4741 for Cu, Fe and Zn. The yeast was grown in SD(-Uracil) solid medium containing 0.67% (w/v) yeast nitrogen base without amino acids (Difco), 2% (w/v) glucose, 0.2% (w/v) amino acids without uracil, and 2% (w/v) agar for selection. For the plate experiment, the yeast was cultured on SD-U medium plates with glucose or raffinose containing various MnCl₂, FeSO₄, CuSO₄ or ZnSO₄ for 2–4 d. The GFP fluorescence signal of GFP in yeast was recorded using a laser scanning microscope (OLYMPVS, FV3000).

Ionomics analysis through ICP-MS

The harvested samples were dried at 70 °C for 3 d or 50 °C for 7 d, and subjected to digestion using concentrated HNO₃ (61%, UN-2031, Kanto, Japan; or 65–68%, CAS:7697–37-2, Jing Rui, China) at up to 120 °C. HNO₃ without plant samples was used as a blank control. The concentrations of elements in the digested solution were determined using inductively coupled plasma MS (ICP-MS 7700X, Agilent Technologies, CA, USA; or NexION 1000G, PerkinElmer, USA), as described previously [53].

Statistical analysis

Statistical tests were performed using Tukey’s test. The significance of differences was defined as P < 0.05.

Authors’ contributions

G.J.L. conceived and designed the experiments. H.L., H.J., M.N., F.D., W.C., O.Y., J.C., X.L., L.S., S.Z., N.Y., F.H., J.F.M., and G.J.L. performed the experiments and took part in data analysis and discussion. N.Y., and J.F.M. provided critical suggestion to the experiments. G.J.L. wrote the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by grants from the National Natural Science Foundation of China (32260450 and 32372055 to G.J.L.; 32370285 to F.D.), the Yunnan Fundamental Research Projects (202301BF070001-018, 202201AW070003 and 202201AU070013 to G.J.L.), the Japan Society for the Promotion of Science (JSPS; KAKENHI grant no. 21H05034 to J.F.M.), the National Key Research and Development Program of China (2023YFD23020001 to F.H.), New Cornerstone Science Foundation (NIC202341 to F.H.), the Hubei Hongshan Laboratory (2021hskf004 to F.D.), and Yunnan University and Yunnan Education Department (KC-252513720 to H.L.).

Data availability

The datasets generated and analyzed during the current study are available in the following repositories: The raw single-nucleus RNA-seq (snRNA-seq) data have been deposited in the NCBI Sequence Read Archive (SRA) under BioProject PRJNA1331938 [100], and the processed gene expression matrix has been deposited in the NCBI Gene Expression Omnibus (GEO) under accession GSE308757 [101]. The analysis code used in this study is publicly available at GitHub: https://github.com/JHL0410/scRNA-seq-code [102]. The source code is released under the GNU General Public License v3.0 (GPL-3.0) and has been archived in Zenodo [103]. Our reference dataset for the correlation analysis is available at http://ibi.zju.edu.cn/plantscrnadb/#/ [49]. The version of the code used for this manuscript is available at: 10.5281/zenodo.18954644 [103]. All microscopy data used for quantification on Figshare [104]. The raw bulk RNA-seq data have been deposited in the Genome Sequence Archive (GSA) under accession number CRA030039 [105, 106].

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.