Involvement of a truncated MADS-box transcription factor ZmTMM1 in root nitrate foraging.

Plants can develop root systems with distinct anatomical features and morphological plasticity to forage nutrients distributed heterogeneously in soils. Lateral root proliferation is a typical nutrient-foraging response to a local supply of nitrate, which has been investigated across many plant species. However, the underlying mechanism in maize roots remains largely unknown. Here, we report on identification of a maize truncated MIKC-type MADS-box transcription factor (ZmTMM1) lacking K- and C-domains, expressed preferentially in the lateral root branching zone and induced by the localized supply of nitrate. ZmTMM1 belongs to the AGL17-like MADS-box transcription factor family that contains orthologs of ANR1, a key regulator for root nitrate foraging in Arabidopsis. Ectopic overexpression of ZmTMM1 recovers the defective growth of lateral roots in the Arabidopsis anr1 agl21 double mutant. The local activation of glucocorticoid receptor fusion proteins for ZmTMM1 and an artificially truncated form of AtANR1 without the K- and C-domains stimulates the lateral root growth of the Arabidopsis anr1 agl21 mutant, providing evidence that ZmTMM1 encodes a functional MADS-box that modulates lateral root development. However, no phenotype was observed in ZmTMM1-RNAi transgenic maize lines, suggesting a possible genetic redundancy of ZmTMM1 with other AGL17-like genes in maize. A comparative genome analysis further suggests that a nitrate-inducible transcriptional regulation is probably conserved in both truncated and non-truncated forms of ZmTMM1-like MADS-box transcription factors found in grass species.

Introduction

Natural and agricultural soils present a large spatial and temporal diversity in nutrient distribution (Giehl and von Wirén, 2014). Plant root systems exhibit highly flexible plasticity to explore the fluctuating nutrients in soils (Giehl ). Nitrate, as the main inorganic <span class="Chemical">nitrogen source in aerobic soils, is heterogeneously distributed due to its high solubility and rapid mobility in soil solution (Lark ; Miller ). To capture the unevenly distributed nitrate, plants stimulate lateral root (LR) growth in nitrate-rich patches, which has been reported as a typical nutrient-foraging response in many plant species, including Arabidopsis, barley, rice, and maize (Drew, 1975; Zhang and Forde, 1998; Wang ; Remans ; Liu ; P. Yu ). Lines of experimental evidence suggest that nitrate per se supplied locally to roots acts as a signaling molecule to modulate LR growth (Zhang and Forde, 1998; Remans ; Ho ; Bouguyon ). In Arabidopsis, local nitrate is sensed by the nitrate transceptor NRT1.1/NPF6.3, which then triggers the signaling pathway involving the MADS-box transcription factor ANR1 to stimulate LR proliferation in a nitrate-enriched zone (Zhang and Forde, 1998; Remans ; Bouguyon ). In addition, local nitrate can inhibit NRT1.1-facilitated auxin redistribution in LR primordia, allowing auxin accumulation for LR emergence and elongation (Krouk ; Mounier ).

MADS-box transcription factors govern diverse developmental processes in plants, including floral formation, pollen maturation, fruit development, and root development (Theissen ; Ng and Yanofsky, 2001; Smaczniak ; Alvarez-Buylla ). The MIKC-type, also termed as type II MADS-box transcription factors, contain four conserved domains: the MADS-box (M-) domain, the intervening (I-) domain, the keratin-like (K-) domain, and the C-terminal (C-) domain (Alvarez-Buylla ). Many MIKC-type MADS-box genes, particularly AGL17-like genes, control root development (Becker and Theissen, 2003; Alvarez-Buylla ). In Arabidopsis, three out of four members within the AGL17-like clade, specifically AGL17, AGL21, and ANR1, are preferentially expressed in roots (Rounsley ; Zhang and Forde, 1998; Alvarez-Buylla ; Burgeff ; C. Yu ). ANR1 promotes LR elongation under local nitrate supply via enhancing LR meristem activity (Zhang and Forde, 1998; Gan ), while AGL21 is required for sustaining LR growth under N deficiency by increasing local auxin biosynthesis (L.-H. Yu ). In rice, four out of five members of the AGL17-like genes, OsMADS23, OsMADS25, OsMADS27, and OsMADS57, regulate root development (Puig ). OsMADS25 promotes growth of both the primary root (PR) and the LR, coinciding with the elevated auxin levels in the presence of nitrate, suggesting the roles of this AGL17-like gene in root nitrate foraging (C. Yu ; Zhang ). OsMADS57 also promotes seminal and adventitious root elongation, and, in parallel, increases root to shoot nitrate translocation by modulating gene expression of NRT2 nitrate transporters (Huang ). Interestingly, a monocot-exclusive miRNA, miR444, specifically targets OsMADS23, OsMADS27, and OsMADS57 for degradation, and inhibits LR growth in rice (Yan ).

Roots are shaped differently in monocot and dicot species, as they develop the overall root architecture composed of multiple root types with distinct anatomic features. The <span class="Species">maize root system is composed of multiple root types including the embryonic primary and seminal roots, and the post-embryonic crown roots initiating from below-ground nodes and brace roots from above-ground nodes (Hochholdinger ). LR development across the individual root types in maize shows distinct morphological responses to nitrate heterogeneity (P. Yu ). Specifically, local nitrate supply mainly promotes LR elongation in the primary, seminal, and crown roots, while both LR initiation and elongation are stimulated in brace roots (P. Yu , 2015, 2016). Yet, the molecular mechanisms and key regulators of nitrate-foraging responses remain unclear in maize.

Transcriptome analysis of nitrate-responsive genes implicates that many transcription factors participate in <span class="Chemical">nitrate-foraging responses in maize roots, including several MADS-boxes (Liu ; Yu ). In this study, we report functional characterization of a novel truncated MIKC-type MADS-box transcription factor gene ZmTMM1 identified as a local nitrate-responsive gene from our previous transcriptome analysis of maize roots (Liu ). Sequence analysis reveals that ZmTMM1 encodes a previously uncharacterized truncated form of an AGL17-like MADS-box lacking K- and C-domains. Functional analysis of ZmTMM1 reveals its involvement in root nitrate-foraging responses, complementing the function of Arabidopsis ANR1 and AGL21. Moreover, based on the comparative genome analysis, we uncover the local nitrate-inducible transcriptional regulation of ZmTMM1 as a common feature conserved within orthologs present in grass species.

Materials and methods

Plant material and growth conditions

Species">Maize (<span class="Species">Zea mays L.) inbred line B73 was used for gene cloning, gene expression, and root morphological analyses. Maize genotype Hi II was used to generate ZmTMM1-RNAi transgenic lines. Arabidopsis thaliana accession Columbia (Col-0) served as the wild type. The anr1 and agl21 knockout mutants were dSpm transposon insertion lines identified from the Sainsbury Laboratory Arabidopsis thaliana collection (SLAT). Homozygotes of anr1 and agl21 were screened from SLAT mutant pool N40108 by PCR (Gan ). Specifically, for the anr1 homozygous mutant, a gene-specific primer SG2 and a dSpm primer dSpm1 were used to verify the transposon insertion, while gene-specific primers SG2 and SG10 were used to confirm the homozygosity of the insertion (Gan ). For the agl21 homozygous mutant, a gene-specific primer SG4 and a dSpm primer dSpm8 were used to verify the transposon insertion, and gene-specific primers SG3 and SG6 were used to confirm the homozygosity of the insertion. These primers are listed in Supplementary Table S2 at JXB online. The double mutant anr1 agl21 (dko mutant) was generated by crossing the anr1 and agl21 single mutants. Rice (Oryza sativa L. japonica) genotype Nipponbare was used for gene expression analysis.

Maize culture

For hydroponic experiments under uniform nutrient supply, maize seeds were surface sterilized in 10% (v/v) <span class="Chemical">H2O2 for 30 min, and germinated in paper rolls (Gu ). Six days after germination, endosperms were removed and seedlings with two visible leaves were transferred to Hoagland solution (pH 5.7) containing 2 mM NH4NO3 as the N source. Nutrient solution was renewed every 2 d. For N starvation, NH4NO3 was removed from the Hoagland solution. For nitrate or ammonium resupply, 4 mM KNO3 or 2 mM (NH4)2SO4 was added, respectively. All seedlings were grown in a climatic chamber under 14 h/10 h (light/dark) and 28 °C/22 °C rhythms at a light intensity of 300 µmol m–2 s–1 and 60% humidity. For split-root experiments, a two-compartment split-root hydroponic system developed previously (Liu ) was used as described below. Maize seedlings were pre-cultured in the Hoagland solution with 0.5 mM KNO3 for 9 d, and then the primary and seminal roots were removed so that four crown roots remained on the seedlings. After 3 d of N starvation, these four crown roots were equally divided into two groups and transferred to a two-compartment split-root system containing 1 mM KNO3 or 0.5 mM (NH4)2SO4 in the +N compartment and 0.5 mM K2SO4 in the –N compartment, respectively.

Arabidopsis culture

Arabidopsis seeds were surface sterilized in 70% <span class="Chemical">ethanol with 0.05% Triton X-100. For Arabidopsis phenotyping experiments, N-free half-strength Murashige and Skoog (MS; Murashige and Skoog, 1962) mineral salts medium supplied with 1% sucrose, 2.5 mM MES (pH 5.8), and 1% Difco agar (BD Biosciences) was used as basic medium, and the N sources were added to be adjusted to final concentrations as indicated in the figures. Seedlings were cultured vertically in a growth chamber under 16 h/8 h (light/dark) diurnal cycles with temperatures during the day and night controlled at 22 °C and 20 °C and with a light intensity of 120 µmol m–2 s–1. In experiments with homogeneous N supply, seedlings were cultivated on N-free half-strength MS medium supplemented with 1 mM KNO3 as the N source for 12 d. For Arabidopsis split-root experiments, seedlings were pre-cultured on N-free half-strength MS medium supplemented with 1 mM KNO3 for 6 d, and then PRs were excised, leaving only two first-order LRs. After a recovery for 3 d, plants with two comparable LRs were transferred to the vertical split-agar plates containing 1 mM KNO3 in the +N side and 0.5 mM K2SO4 in the –N side. Root phenotypes were analyzed at 6 d after the transfer. For the local activation experiments using the glucocorticoid receptor (GR) fusion system, PRs of 9-day-old seedlings were pruned to have two LRs, and these LRs were transferred onto the vertically segmented agar plates containing 1 µM dexamethazone (DEX) in the +DEX side and 0.01% ethanol as a mock control in the –DEX side. Segmented agar plates for the local DEX treatment were supplied with 1 mM KNO3 or 0.5 mM Gln as the N source. Root phenotypes were analyzed at 8 d after the transfer.

Rice culture

For split-root experiments, rice seedlings were pre-cultured in IRRI <span class="Species">rice culture solution (Yoshida ) containing 1 mM NH4NO3 as the N source for 25 d. After 3 d of N starvation, rice plants were transferred into the two-compartment split-root system (Liu ) containing 1 mM KNO3 in the +N compartment and 0.5 mM K2SO4 in the –N compartment. Roots were harvested for gene expression analysis at 12 h after the transfer to validate the effect of the localized nitrate treatment.

Root growth measurements

<span class="Species">Arabidopsis and <span class="Species">maize roots were scanned by an Epson V700 scanner at a resolution of 300 dpi. Root lengths were measured by Image J or WinRHIZO Pro 2007 software. The lengths of the axis root and LR branching zone were measured and the visible LR root number was counted. To assay LR initiation and primordial development, Arabidopsis roots were cleared by HCG solution (chloral hydrate:water:glycerol=8:3:1) for 10 min, and LR primordia embedded in PR were counted under differential interference contrast (DIC) microscopy. Developmental stages of LR initials were classified according to Malamy and Benfey (1997).

Statistical analysis

All statistical analyses were performed by SigmaPlot 11.0 software. A paired t-test or one-way ANOVA was used to test the statistical significance, and the P-value of each statistical anaylsis is described in the figure legends.

In situ RNA hybridization

In situ RNA hybridization was performed as previously described (Gu ). Maize roots were harvested from the hydroponic system and fixed in FAA solution (<span class="Chemical">formaldehyde, ethanol, acetic acid=10%, 50%, 5%) overnight at 4 °C. Thin sections of paraffin-embedded roots of 8–12 μm thickness were generated for the hybridization. A 115 bp gene-specific region of the ZmTMM1 coding sequence, which was confirmed by quantitative real-time PCR (qRT-PCR), was selected for synthesis of digoxigenin (DIG)-labeled RNA sense and antisense probes in vitro. The hybridization reaction was conducted for 24 h at 50 °C. Anti-DIG alkaline phosphatase-conjugated antibody and the NBT (nitro-blue tetrazolium)/BCIP (5-bromo-4-chloro-3-indolyl phosphate) staining method were used to visualize the tissue-specific localization of ZmTMM1 transcript expression in root sections. Photographs were taken by Olympus BX51 microscope (Japan).

qRT-PCR

Total RNA was extracted from 50–100 mg of frozen root samples using TRIzol (Invitrogen) solution following the manufacturer’s protocol. Genomic DNA was removed by RQ1 RNase-Free DNase (Promega). Template cDNA was synthesized from 1 µg of total RNA using M-MLV reverse transcriptase (Promega) and oligodT(18) primers. qPCR was conducted by the ABI 7500 Real-Time PCR System (Applied Biosystems) using SYBR Green real-time PCR Master Mix Kit (TOYOBO). Gene expression levels were normalized relative to ZmGAPDH and <span class="Gene">ZmTUB4 in maize, AtUBQ10 in Arabidopsis, and OsACTIN in rice as internal controls, and by using the ΔΔCt method (Pfaffl, 2001). Gene-specific primers are listed in Supplementary Table S2.

Generation of transgenic plants

Transgenic Arabidopsis lines (dko-p35S::ZmTMM1) were generated by introducing a construct to express ZmTMM1 under the control of the <span class="Species">Cauliflower mosaic virus (CaMV) 35S promoter in the background of the dko mutant. The coding sequence of ZmTMM1 was cloned into the pMD19-T (TaKaRa) vector, and then subcloned into a plant transformation vector pPT-Hyg (Yuan ). ZmTMM1–GR, ANR1S–GR, and ANR1–GR constructs were produced by fusing GR protein to the C-terminus of ZmTMM1, ANR1S, and ANR1 protein-coding regions, and inserting these GR fusions downstream of the CaMV 35S promoter. The GR fusion constructs were subsequently introduced into the dko mutant. The promoter–green fluorescent protein (GFP)–nuclear localization sequence (NLS) constructs for ANR1 and AGL21 were made by cloning PCR-amplified 2 kb promoter regions into pENTR/D-TOPO (Invitrogen) and integrating the inserts into a Gateway compatible binary vector pBGGN (Funakoshi, Japan), which allows in situ analysis of promoter activities based on expression of GFP fused with an NLS. Arabidopsis was transformed by the floral dipping method facilitated by Agrobacterium tumefaciens GV3101 (Clough and Bent, 1998). Transgenic plants made by using pPT-Hyg and pBGGN vectors were selected by hygromycin and basta resistance, respectively, and homozygous T3 lines were used for subsequent experiments. For studying ZmTMM1 subcellular localization, GFP was fused to the C-terminus of the ZmTMM1 protein-coding region and inserted into the pBI121-GFP vector to be expressed under the control of the CaMV 35S promoter. The p35S::ZmTMM1-GFP fusion construct was transiently expressed in onion epidermal cells. To generate the ZmTMM1-RNAi construct, a 115 bp coding sequence of ZmTMM1 was inserted into an RNAi vector pTCK303 to be expressed under the control of the maize Ubiquitin 1 gene promoter (Wang ). The pUBQ::ZmTMM1-RNAi vector was introduced into A. tumefaciens EHA105 to transform maize genotype Hi II. Three independent ZmANR1-RNAi transgenic maize lines were selected by phosphinothricin (PPT) resistance and further backcrossed with wild-type B73 twice.

Phylogenetic analysis

Protein sequences encoded by the AGL17-like genes in <span class="Species">A. thaliana, Brassica rapa, Glycine max, O. sativa, Brachypodium distachyon, Sorghum bicolor, and Z. mays were obtained from the PlantGDB database. The phylogenetic tree was generated by MEGA4 software (https://www.megasoftware.net/mega4/) via the Neighbor–Joining method. Numbers shown at the nodes indicate the bootstrap values expressed as percentages obtained from 1000 trials. Motifs were identified by the MEME algorithm (http://meme-suite.org/tools/meme). Gene structure annotation was conducted by the yrGATE algorithm (http://www.plantgdb.org/prj/yrGATE/).

Accession numbers

<span class="Species">Arabidopsis: <span class="Gene">ANR1 (AT2G14210), AGL16 (AT3G57230), AGL17 (AT2G22630), and AGL21 (AT4G37940); maize: ZmTMM1 (GRMZM2G044408), ZmMADS2 (GRMZM2G316366, GRMZM2G492156), GRMZM2G055782, and GRMZM2G032905; rice: OsMADS23 (LOC_Os08g33488), OsMADS25 (LOC_Os04g23910), OsMADS27 (LOC_Os02g36924), OsMADS57 (LOC_Os02g49840), and OsMADS61 (LOC_Os04g38770, LOC_Os04g38780).

Results

Identification of a truncated MIKC-type MADS-box gene ZmTMM1 in maize

To investigate the effect of local N supply on maize root development, we developed a split-root hydroponic system (Fig. 1A; Liu ). After 5 d of local <span class="Chemical">nitrate treatments, the mean length of LRs increased up to 2.5-fold compared with roots growing in the N-free compartment, and the visible LR density increased by 10% (Fig. 1B, C). This indicated that the localized supply of nitrate promoted both LR branching and LR elongation in maize. In contrast, the localized supply of ammonium significantly stimulated LR branching but not LR elongation (Fig. 1B, C).

Fig. 1.

Development of maize LRs in response to localized ammonium or nitrate supply. Maize seedlings with four crown roots were subjected to a two-compartment split-root system which contained either 1 mM KNO3 or 0.5 mM (NH4)2SO4 in the +N compartment, and 0.5 mM K2SO4 in the –N compartment. Root growth was measured at 5 d after transfer. (A) Root growth of maize seedlings in a two-compartment split-root system. (B) Mean LR length. (C) Visible LR density. Data represent means ±SE (n=5 replicates; each replicate represents two seedlings). Asterisks indicate significant differences between the roots grown in the +N and –N compartments at *P<0.05; **P<0.01; ***P<0.001; ns, not significant (Student’s t-test).

To uncover the mechanism underlying LR proliferation in response to local <span class="Chemical">nitrate supply, a microarray experiment was conducted in our previous study for identifying nitrate-responsive genes (Liu ). A MADS-box transcription factor gene (GRMZM2G044408) was identified as a nitrate-responsive gene showing 4-fold up-regulated expression after 1 h of local nitrate treatment. By conducting 5'- and 3'-RACE, we confirmed the full-length transcript of GRMZM2G044408 that contains a 270 bp ORF, encoding a polypeptide with 89 amino acid residues, which was named truncated MIKC-type MADS-box transcription factor, ZmTMM1. According to the conservation of the DNA-binding domain, ZmTMM1 was assigned to the AGL17-like clade of MADS-box genes with 83% similarity at the protein level to Arabidopsis ANR1 (Zhang and Forde, 1998; Gan ).

Phylogenetic analysis showed that AGL17-like genes were subdivided into <span class="Disease">dicot- and monocot-specific clades (Fig. 2A). Motif analysis revealed that a typical complete MIKC-type MADS-box transcription factor contains four domains: the M-domain, I-domain, K-domain, and C-domain. All AGL17-like genes in dicots encoded the complete MIKC-type MADS-box protein structure, whereas the truncated AGL17-like-encoded MADS-box proteins existing only in monocots lacked either the K- or C-domain, or even both. Interestingly, ZmTMM1 was devoid of both the K- and C-domains, leaving the remaining portion to be present only with the conserved M-domain and I-domain (Fig. 2A).

Fig. 2.

Phylogenetic, motif, and comparative genome analysis of AGL17-like genes in plants. (A) Phylogenetic tree and motif analysis of AGL17-like MADS-box transcription factors. Protein sequences encoded by AGL17-like genes in Arabidopsis thaliana, Brassica rapa, Glycine max, Oryza sativa, Brachypodium distachyon, Sorghum bicolor, and Zea mays were used for the phylogenetic and motif analysis. ZmTMM is highlighted in red. Numbers shown at the nodes indicate the bootstrap values expressed as percentages. (B) Comparative genome analysis of AGL17-like genes in grass species. Rings in different colors represent genomes of different species: rice, gray; Brachypodium, blue; sorghum, green; maize, brown. Spots in different colors indicated on the rings represent AGL17-like genes in grass species: rice, red; Brachypodium, purple; sorghum, blue; maize, yellow. Red solid line represents intragenomic collinearity in rice. Red dashed lines represent intragenomic micro-collinearity in rice. Gray dashed lines represent interspecific collinearity across the grass species.

Phylogenetic, motif, and comparative genome analysis of AGL17-like genes in plants. (A) Phylogenetic tree and motif analysis of <span class="Gene">AGL17-like MADS-box transcription factors. Protein sequences encoded by AGL17-like genes in Arabidopsis thaliana, Brassica rapa, Glycine max, Oryza sativa, Brachypodium distachyon, Sorghum bicolor, and Zea mays were used for the phylogenetic and motif analysis. ZmTMM is highlighted in red. Numbers shown at the nodes indicate the bootstrap values expressed as percentages. (B) Comparative genome analysis of AGL17-like genes in grass species. Rings in different colors represent genomes of different species: rice, gray; Brachypodium, blue; sorghum, green; maize, brown. Spots in different colors indicated on the rings represent AGL17-like genes in grass species: rice, red; Brachypodium, purple; sorghum, blue; maize, yellow. Red solid line represents intragenomic collinearity in rice. Red dashed lines represent intragenomic micro-collinearity in rice. Gray dashed lines represent interspecific collinearity across the grass species.

Next, we examined how the truncated MADS-box proteins would have been generated in monocot species. An intron exists in the coding sequence of AGL17-like MADS-box genes, being inserted into the gap between the I- and K- domains (Parenicová et al., 2003; Supplementary Fig. S1; Supplementary Table S1). <span class="Gene">AGL17-like genes in Arabidopsis contain such introns with a length of 1137–2133 bp, and rice genes, except for OsMADS25, with a length of 2755–5184 bp, respectively (Supplementary Fig. S1). Notably, due to the insertion of the introns, OsMADS61 was incorrectly annotated into two tandem genes, Os04g38770 and Os04g38780, and ZmMADS2 was mis-annotated into two tandem genes, GRMZM2G316366 and GRMZM2G492156 (Supplementary Table S1; Schreiber ; Puig ). Besides being present and spliced as an intron of a gene encoding the complete MIKC-type MADS-box protein, the existence of such an extremely large size (>9–10 kb) of a gap in a genomic region may cause termination of transcription, producing a truncated MADS-box gene transcript containing only the M- and I-domains, such as BRADI5G12440 in Brachypodium, Sb07g021110 in sorghum, and GRMZM2G055782 and GRMZM2G032905 in maize. The second halves separately annotated downstream in the genome could represent genes encoding the K- and C-domains (Supplementary Table S1). In addition, due to the genome duplication and the chromosome fusion that reconstructed the genome of grass species (Bolot ), some MADS-box genes might even have lost these second halves of fragments on the genome, such as ZmTMM1 in maize, and Sb04g024010 and Sb06g019040 in sorghum (Supplementary Table S1). Taken together, the comparative genome analysis provided evidence that ZmTMM1 encodes a novel truncated AGL17-like gene product exclusively identified in grass species.

ZmTMM1 is expressed in root xylem parenchyma cells and outer layer cortical cells

Similar to other AGL17-like genes (Burgeff ; Becker and Theissen, 2003; Puig ), ZmTMM1 was preferentially expressed in <span class="Species">maize roots (Supplementary Fig. S2). At the tissue level, in situ RNA hybridization showed that ZmTMM1 transcript was not detected in either crown root tips or LR tips (Fig. 3A, D). However, along the longitudinal section of maize crown roots, ZmTMM1 transcript was detected in the central cylinder of the root maturation zone (Fig. 3B). The cross-section further revealed that ZmTMM1 transcript is mainly localized in xylem parenchyma cells of mature roots (Fig. 3C). In the LR branching zone, ZmTMM1 accumulated in both xylem parenchyma cells and outer layer cortical cells surrounding the emerging LR (Fig. 3D). In contrast, the sense probes for ZmTMM1 did not generate a detectable signal (Fig. 3E–H). Additionally, by transiently expressing ZmTMM1–GFP fusion protein in onion epidermal cells, ZmTMM1 was shown to localize in the nucleus, which is considered prerequisite for this protein being a transcription factor (Fig. 3I).

Fig. 3.

Tissue-specific and subcellular localization of ZmTMM1. (A–H) In situ RNA hybridization of ZmTMM1 in maize roots. Maize seedlings were cultivated under 4 mM KNO3 for 12 d, and crown roots were harvested. Antisense probes (A–D) and sense probes (E–H) were used to determine the localization of ZmTMM1 mRNA. Longitudinal sections of crown root tips (A and E); longitudinal sections of crown roots in the maturation zone (B and F); cross-sections of crown roots in the maturation zone (C and G); cross-sections of crown roots with emerging LRs (D and H). ep, epidermis; ex, exodermis; co, cortex; en, endodermis; pe, pericycle; ph, phloem; xy, xylem. Scale bars=100 µm. (I and J) Subcellular localization of ZmTMM1::GFP fusion protein (I) and GFP control (J) in onion epidermal cell. Scale bars=100 µm.

Tissue-specific and subcellular localization of ZmTMM1. (A–H) In situ RNA hybridization of ZmTMM1 in maize roots. <span class="Species">Maize seedlings were cultivated under 4 mM KNO3 for 12 d, and crown roots were harvested. Antisense probes (A–D) and sense probes (E–H) were used to determine the localization of ZmTMM1 mRNA. Longitudinal sections of crown root tips (A and E); longitudinal sections of crown roots in the maturation zone (B and F); cross-sections of crown roots in the maturation zone (C and G); cross-sections of crown roots with emerging LRs (D and H). ep, epidermis; ex, exodermis; co, cortex; en, endodermis; pe, pericycle; ph, phloem; xy, xylem. Scale bars=100 µm. (I and J) Subcellular localization of ZmTMM1::GFP fusion protein (I) and GFP control (J) in onion epidermal cell. Scale bars=100 µm.

ZmTMM1 is up-regulated by local nitrate in the LR branching zone

Next, we investigated the response of ZmTMM1 at the transcriptional level to different N availabilities. First, the expression of ZmTMM1 was found to be repressed by N starvation (Fig. 4A). In response to N resupply, ZmTMM1 was specifically up-regulated following addition of nitrate, showing an increase starting after 1 h and reaching maximum levels at 6–12 h by 10-fold in the time course (Fig. 4B). In contrast, the ZmTMM1 transcript level was unaffected by <span class="Chemical">ammonium resupply. In a split-root system, ZmTMM1 was specifically up-regulated by the localized nitrate supply rather than ammonium (Fig. 4C), while the root N status and transcript levels of an N-responsive marker gene ZmGS1.1 (Guo ) increased in response to both the localized supply of nitrate and ammonium (Supplementary Fig. S3). Given that the root transcriptional response is tuned by a local nitrate signal in combination with systemic N demand (Ruffel ), we further used a split-root system, in which the roots were subjected to either homogeneous or heterogeneous nitrate supply or no N, to test whether ZmTMM1 is under the control of local or systemic N signals. The maize seedlings subjected to the split-root culture consistently showed that ZmTMM1 transcript levels in portions of the split roots locally fed with nitrate and the whole root system homogeneously supplied with nitrate were comparable and significantly higher than in the roots either locally or homogeneously submerged in the no N nutrient solution, suggesting that ZmTMM1 expression is induced by local nitrate signal rather than by the systemic N status of plants (Fig. 4D). In addition, this local increase in ZmTMM1 transcripts by nitrate mainly occurred in the LR branching zones (Fig. 4E).

Fig. 4.

Regulation of ZmTMM1 in maize roots in response to different N availability. (A) ZmTMM1 transcript levels changing in response to N deprivation. Maize seedlings were pre-cultured with 2 mM NH4NO3, and then transferred to N-free solution. (B) Time course analysis of changes in ZmTMM1 transcript levels in response to the homogenous resupply of nitrate or ammonium. After N starvation for 4 d, maize seedlings were resupplied with either 4 mM KNO3 or 2 mM (NH4)2SO4. (C) ZmTMM1 transcript levels in response to local supply of nitrate or ammonium in a split-root system. Maize seedlings with four crown roots were cultivated in a split-root system supplemented with either 1 mM KNO3 or 0.5 mM (NH4)2SO4 in the +N compartment, and 0.5 mM K2SO4 in the –N compartment. (D) ZmTMM1 is regulated by local nitrate rather than systemic N signal. CK-KNO3, roots in both compartments of the split-root system homogeneously treated with nitrate (1 mM KNO3); Sp-KNO3, roots in the +N compartment (1 mM KNO3) of the split-root system; Sp-K2SO4, roots in the –N compartment (0.5 mM K2SO4) of the split-root system; CK-K2SO4, roots in both compartments of the split-root system subjected to N deprivation (0.5 mM K2SO4). Relative expression levels of ZmTMM1 after 8 h of treatment are shown. (E) Ratio of ZmTMM1 transcript levels between the root segments in the +N and –N compartments of the split-root system. Maize crown roots were harvested from the split-root system after 12 h of treatment and divided into four regions: A, root tip; B, LR initiation region; C, LR elongation region; D, mature LR region. Roots were harvested for gene expression analysis at the indicated time points. The relative transcript level of ZmTMM1 was determined by qPCR and normalized by maize Tubulin 4 (AJ420856). Data represent means ±SD (n=3 replicates; each replicate represents a single seedling). Different letters represent significant differences among means at P<0.05 (Tukey’s test). Asterisks indicate significant differences between the roots grown in the +N and –N compartments at *P<0.05; **P<0.01; ***P<0.001; ns, not significant (Student’s t-test).

Regulation of ZmTMM1 in Species">maize roots in response to different N availability. (A) ZmTMM1 transcript levels changing in response to N deprivation. <span class="Species">Maize seedlings were pre-cultured with 2 mM NH4NO3, and then transferred to N-free solution. (B) Time course analysis of changes in ZmTMM1 transcript levels in response to the homogenous resupply of nitrate or ammonium. After N starvation for 4 d, maize seedlings were resupplied with either 4 mM KNO3 or 2 mM (NH4)2SO4. (C) ZmTMM1 transcript levels in response to local supply of nitrate or ammonium in a split-root system. Maize seedlings with four crown roots were cultivated in a split-root system supplemented with either 1 mM KNO3 or 0.5 mM (NH4)2SO4 in the +N compartment, and 0.5 mM K2SO4 in the –N compartment. (D) ZmTMM1 is regulated by local nitrate rather than systemic N signal. CK-KNO3, roots in both compartments of the split-root system homogeneously treated with nitrate (1 mM KNO3); Sp-KNO3, roots in the +N compartment (1 mM KNO3) of the split-root system; Sp-K2SO4, roots in the –N compartment (0.5 mM K2SO4) of the split-root system; CK-K2SO4, roots in both compartments of the split-root system subjected to N deprivation (0.5 mM K2SO4). Relative expression levels of ZmTMM1 after 8 h of treatment are shown. (E) Ratio of ZmTMM1 transcript levels between the root segments in the +N and –N compartments of the split-root system. Maize crown roots were harvested from the split-root system after 12 h of treatment and divided into four regions: A, root tip; B, LR initiation region; C, LR elongation region; D, mature LR region. Roots were harvested for gene expression analysis at the indicated time points. The relative transcript level of ZmTMM1 was determined by qPCR and normalized by maize Tubulin 4 (AJ420856). Data represent means ±SD (n=3 replicates; each replicate represents a single seedling). Different letters represent significant differences among means at P<0.05 (Tukey’s test). Asterisks indicate significant differences between the roots grown in the +N and –N compartments at *P<0.05; **P<0.01; ***P<0.001; ns, not significant (Student’s t-test).

Ectopic expression of ZmTMM1 modulates LR development in transgenic Arabidopsis

Given the overlapping gene expression patterns and redundant functions of ANR1 and <span class="Gene">AGL21 in LR development (Supplementary Fig. S4; Zhang and Forde, 1998; Gan ; L.H. Yu ), an Arabidopsis double mutant (dko) of ANR1 and AGL21 was generated to serve as a genetic background for functional characterization of ZmTMM1 (Supplementary Fig. S5A). The dko (anr1 agl21) mutant showed a reduction in total LR length compared with the wild type, as explained by the decrease in the visible LR number rather than the average LR length (Supplementary Fig. S5B). Analysis of LR development by stages further indicated that a larger proportion of LR primordia were arrested in the dko mutant at earlier stages, leading to a relative decrease in the reduced visible LR density, while the number of LR initiation events monitored as the total density of LR and LR primordia was unaffected (Fig. 5E–G). Thus, the growth impairment of LR primordia was suggested to be the major defect in LR development identified in the dko mutant.

Fig. 5.

Ectopic expression of ZmTMM1 restores LR development in the dko mutant. ZmTMM1 was ectopically expressed in the Arabidopsis dko mutant under the control of the CaMV 35S promoter. Arabidopsis plants were cultivated on N-free half-strength MS agar medium supplemented with 1 mM KNO3 as the N source, and root phenotypes were analyzed 12 d after germination. (A) LR length. (B) Visible LR number. (C) Average LR length. (D) LR elongation rate during the day 10–day 12 period presented as the mean rate of the two longest LRs. (E) Total LR density. (F) Visible LR density. (G) Relative frequency of LR primordia and emerged LRs. LR primordia and LRs at specific developmental stages were classified into four groups: stage I–II; stage III–VII; emerged LR <0.5 mm; and visible LR >0.5 mm. Bars represent means ±SE (n=16 replicates; each replicate represents a single seedling). Different letters represent significant differences among means at P<0.05; ns, not significant (Tukey’s test).

Ectopic expression of ZmTMM1 restores LR development in the dko mutant. ZmTMM1 was ectopically expressed in the <span class="Species">Arabidopsis dko mutant under the control of the CaMV 35S promoter. Arabidopsis plants were cultivated on N-free half-strength MS agar medium supplemented with 1 mM KNO3 as the N source, and root phenotypes were analyzed 12 d after germination. (A) LR length. (B) Visible LR number. (C) Average LR length. (D) LR elongation rate during the day 10–day 12 period presented as the mean rate of the two longest LRs. (E) Total LR density. (F) Visible LR density. (G) Relative frequency of LR primordia and emerged LRs. LR primordia and LRs at specific developmental stages were classified into four groups: stage I–II; stage III–VII; emerged LR <0.5 mm; and visible LR >0.5 mm. Bars represent means ±SE (n=16 replicates; each replicate represents a single seedling). Different letters represent significant differences among means at P<0.05; ns, not significant (Tukey’s test).

ZmTMM1 was then ectopically overexpressed in the Species">Arabidopsis dko mutant under the control of the <span class="Species">CaMV 35S promoter, and three independent transgenic lines (dko-p35S::ZmTMM1) were selected for root morphological analysis (Supplementary Fig. S6). Under a condition with uniform nitrate supply, overexpression of ZmTMM1 recovered the LR growth of the dko mutant. Specifically, it restored the LR emergence in the dko mutant, increasing the density of visible LRs, while not altering the LR elongation rate and LR initiation (Fig. 5A–G). These results indicate that the truncated MADS-box gene ZmTMM1 is able to complement the functions of ANR1 and AGL21 promoting LR development in Arabidopsis.

Local activation of ZmTMM1 promotes LR elongation in a nitrate-dependent manner

The nuclear targeting of GR fusion protein is controlled in the presence of DEX, which allows local activation of the function of transcription factor–GR fusion proteins following <span class="Chemical">DEX treatment (Brockmann ; Gan ). To mimic the stimulatory effect of local nitrate supply on LR growth, a ZmTMM1–GR fusion gene construct was generated and introduced into the dko background (Supplementary Fig. S7B, C). In the split-root system, ZmTMM1–GR fusion protein expressed under the control of the CaMV 35S promoter was locally activated in the DEX-containing segment (Supplementary Fig. S7D). When nitrate was supplied as the N source, the local activation of ZmTMM1–GR fusion protein promoted LR growth in the transgenic plants, increasing both the LR number and the average LR length significantly, as presented by the +DEX/–DEX ratios (Fig. 6A–D; Supplementary Fig. S7D). When Gln was supplied as an alternative N source, however, DEX-induced activation of ZmTMM1–GR fusion protein only increased the LR number but not the average LR length (Fig. 6E–H; Supplementary Fig. S7D). These results indicate that the local activation of ZmTMM1 protein in transgenic Arabidopsis promotes LR growth, stimulating LR elongation in a nitrate-dependent manner and enhancing LR branching in a nitrate-independent manner.

Fig. 6.

Local activation of ZmTMM1–GR fusion protein modulates LR growth. ZmTMM1–GR fusion protein expressed under the control of the CaMV 35S promoter was locally activated by DEX supply in the split-root system. Arabidopsis seedlings harboring two LRs with similar length were transferred to vertically split agar plates containing 1 μM DEX in the +DEX side and no DEX in the –DEX side. The root phenotype was measured 8 d after transfer. (A–D) Split agar plates containing 1 mM KNO3 as the N source; (E–H) split agar plates containing 0.5 mM Gln as the N source. +DEX/–DEX ratio of first-order LR length (A and E); +DEX/–DEX ratio of total second-order LR length (B and F); +DEX/–DEX ratio of visible second-order LR number (C and G); +DEX/–DEX ratio of average second-order LR length (D and H). Bars represent means ±SE (n=7–12 replicates; each replicate represents a single seedling). Asterisks indicate significant differences between the LRs on DEX-treated (+DEX) and non-treated (–DEX) segments of agar plates at *P<0.05; **P<0.01; ***P<0.001; ns, not significant (Student’s t-test).

Local activation of ZmTMM1–GR fusion protein modulates LR growth. ZmTMM1–GR fusion protein expressed under the control of the <span class="Species">CaMV 35S promoter was locally activated by DEX supply in the split-root system. Arabidopsis seedlings harboring two LRs with similar length were transferred to vertically split agar plates containing 1 μM DEX in the +DEX side and no DEX in the –DEX side. The root phenotype was measured 8 d after transfer. (A–D) Split agar plates containing 1 mM KNO3 as the N source; (E–H) split agar plates containing 0.5 mM Gln as the N source. +DEX/–DEX ratio of first-order LR length (A and E); +DEX/–DEX ratio of total second-order LR length (B and F); +DEX/–DEX ratio of visible second-order LR number (C and G); +DEX/–DEX ratio of average second-order LR length (D and H). Bars represent means ±SE (n=7–12 replicates; each replicate represents a single seedling). Asterisks indicate significant differences between the LRs on DEX-treated (+DEX) and non-treated (–DEX) segments of agar plates at *P<0.05; **P<0.01; ***P<0.001; ns, not significant (Student’s t-test).

Accordingly, as ZmTMM1 is a truncated form of a MADS-box protein, the K- and C-domains were suggested to be dispensable for LR development. To verify this hypothesis, an artificially truncated form of <span class="Species">Arabidopsis ANR1 protein (ANR1S; 93 amino acid residues), mimicking the structure of ZmTMM1, was constructed by the deletion of the K- and C-domains (Supplementary Fig. S7A, B). Then, driven by the CaMV 35S promoter, the ANR1S–GR fusion and ANR1–GR as a positive control were expressed in the dko mutant (Supplementary Fig. S7B, C). It was shown that local activation of the truncated protein ANR1S and the intact protein ANR1 in the DEX-containing patches stimulated LR growth (Fig. 7). As ZmTMM1 did, both the locally activated ANR1S and ANR1 led to an increase in both the average LR length and the LR number in the presence of nitrate (Fig. 7A–D); in the absence of nitrate, however, there was only an increase in the LR number (Fig. 7E–H). These results supported that, despite the K- and C-domain truncations, ZmTMM1 can fully function as a MADS-box protein modulating LR development.

Fig. 7.

Truncated protein ANR1S functions as the intact ANR1 protein regulating LR development. ANR1–GR and ANR1S–GR fusion proteins expressed under the control of the CaMV 35S promoter were locally activated by DEX in the split-root system. Arabidopsis seedlings harboring two LRs of similar length were transferred to vertically split agar plates containing 1 μM DEX in the +DEX side and no DEX in the –DEX side. The root phenotype was measured 8 d after transfer. (A–D) Split agar plates containing 1 mM KNO3 as the N source; (E–H) split agar plates containing 0.5 mM Gln as the N source. +DEX/–DEX ratio of first-order LR length (A and E); +DEX/–DEX ratio of total second-order LR length (B and F); +DEX/–DEX ratio of visible second-order LR number (C and G); +DEX/–DEX ratio of average second-order LR length (D and H). Bars represent means ±SE (n=7–12 replicates; each replicate represents a single seedling). Asterisks indicate significant differences between the LRs on DEX-treated (+DEX) and non-treated (–DEX) segments of agar plates at *P<0.05; **P<0.01; ***P<0.001; ns, not significant (Student’s t-test).

Truncated protein ANR1S functions as the intact <span class="Gene">ANR1 protein regulating LR development. ANR1–GR and ANR1S–GR fusion proteins expressed under the control of the CaMV 35S promoter were locally activated by DEX in the split-root system. Arabidopsis seedlings harboring two LRs of similar length were transferred to vertically split agar plates containing 1 μM DEX in the +DEX side and no DEX in the –DEX side. The root phenotype was measured 8 d after transfer. (A–D) Split agar plates containing 1 mM KNO3 as the N source; (E–H) split agar plates containing 0.5 mM Gln as the N source. +DEX/–DEX ratio of first-order LR length (A and E); +DEX/–DEX ratio of total second-order LR length (B and F); +DEX/–DEX ratio of visible second-order LR number (C and G); +DEX/–DEX ratio of average second-order LR length (D and H). Bars represent means ±SE (n=7–12 replicates; each replicate represents a single seedling). Asterisks indicate significant differences between the LRs on DEX-treated (+DEX) and non-treated (–DEX) segments of agar plates at *P<0.05; **P<0.01; ***P<0.001; ns, not significant (Student’s t-test).

ZmTMM1-RNAi transgenic maize shows no discernible phenotype compared with wild-type plants

To investigate the function of ZmTMM1 in maize, we knocked down ZmTMM1 transcripts by RNAi in transgenic <span class="Species">maize. In three independent ZmTMM1-RNAi transgenic maize lines that we generated, the ZmTMM1 transcript abundance decreased specifically, while the levels of its homologous gene transcripts (e.g. ZmMADS2 and GRMZM2G055782) remained unchanged relative to the wild type (Supplementary Fig. S8B). In the ZmTMM1-RNAi lines, the induction of ZmTMM1 transcript expression by local nitrate supply was reduced to 10–50% the levels of the wild-type plants (Supplementary Fig. S8B), but their LRs still responded to nitrate and grew to a similar extent as the wild type (Supplementary Fig. S8A, C). Consistent with these observations, the response of LR proliferation to local nitrate supply was also unaffected in the Arabidopsis dko mutant in the split-root system (Supplementary Fig. S5C). These results may imply the functional redundancies among the AGL17-like genes or other compensatory mechanisms complementing the loss of function of MADS-box proteins for root nitrate foraging.

Induced expression by local nitrate is a common feature of ZmTMM1-like genes in cereals

In grass species, comparative genome analysis provides a deeper insight into the paralogous and orthologous relationships among the members of large gene families, uncovering their intraspecific and interspecific collinearity (Bolot ). In rice, among the five members of MADS-box genes in the <span class="Gene">AGL17-like clade, OsMADS27 and OsMADS61 showed an intragenomic collinearity that is caused by the intragenome duplication between chromosome 2 and chromosome 4, and OsMADS23 revealed an intragenomic micro-collinearity with both due to a segmental duplication event. In contrast, OsMADS25 and OsMADS57 are likely to be the homologs that evolved independently (Fig. 2B; Supplementary Dataset S1). In addition, the interspecific collinearity of AGL17-like MADS-box genes was dissected in Brachypodium, sorghum, and maize by using the rice genome as a reference (Fig. 2B; Supplementary Dataset S1). Notably, GRMZM2G344991 and GRMZM2G052045 in maize lost the DNA-binding domain, thus they were not recognized as AGL17-like transcription factor genes. Comparative genome analysis revealed that ZmTMM1-like genes, comprising ZmTMM1, Sb04g028960, BRADI3G57017, and OsMADS57, independently evolved in grass species, implying that they might exhibit unique functional features.

Within the ZmTMM1-like clade, BRADI3G57017 and OsMADS57 are the complete MIKC-type MADS-box genes, while Sb04g028960 lacks the K-domain and ZmTMM1 is a truncated form devoid of both the K- and C-domains (Fig. 2A), suggesting that the absence of K- and C-domains is not a common feature of ZmTMM1-like genes in grass species. By evaluating transcript expression levels in the split-root system, OsMADS57 and ZmTMM1 were shown to be specifically up-regulated by local <span class="Chemical">nitrate supply among all the AGL17-like genes in rice and maize (Fig. 8A, B). Thus, we propose the local nitrate-inducible expression as a conserved feature of the ZmTMM1-like genes in grass species. In contrast, all the AGL17-like genes in Arabidopsis were not regulated by local nitrate supply at the transcriptional level (Supplementary Fig. S9), suggesting that nitrate regulates AGL17-like genes via distinct mechanisms between dicots and monocots.

Fig. 8.

Local nitrate-inducible transcriptional regulation is a conserved feature of ZmTMM1-like genes. (A) Expression of maize AGL17-like genes in response to local nitrate supply. Maize seedlings were cultivated in a split-root system containing 1 mM KNO3 in the +N compartment and 0.5 mM K2SO4 in the –N compartment. After 12 h of local nitrate treatment in the split-root system, the transcript levels of target genes in roots were determined by qPCR. Values are normalized relative to maize Tubulin 4 (AJ420856). (B) Expression of rice AGL17-like genes in response to local nitrate supply. Rice seedlings were cultivated in a split-root system as described for maize. After 12 h of local nitrate treatment, the transcript levels of target genes in roots were determined by qPCR. Values are normalized by OsACTIN (LOC_Os03g50890). Data represent means ±SD (n=3 replicates; each replicate represents a single seedling). Asterisks indicate significant differences between the gene expression in roots in the +N and –N compartments at: *P<0.05; **P<0.01; ns, not significant (Student’s t-test). (C) A model scheme describing the function and regulation of the truncated MADS-box gene ZmTMM1. In Arabidopsis, local nitrate activates ANR1 via a post-translational mechanism to stimulate LR proliferation; while in maize, local nitrate induces the expression of ZmTMM1 transcripts to promote LR development.

Local nitrate-inducible transcriptional regulation is a conserved feature of ZmTMM1-like genes. (A) Expression of <span class="Species">maize AGL17-like genes in response to local nitrate supply. Maize seedlings were cultivated in a split-root system containing 1 mM KNO3 in the +N compartment and 0.5 mM K2SO4 in the –N compartment. After 12 h of local nitrate treatment in the split-root system, the transcript levels of target genes in roots were determined by qPCR. Values are normalized relative to maize Tubulin 4 (AJ420856). (B) Expression of rice AGL17-like genes in response to local nitrate supply. Rice seedlings were cultivated in a split-root system as described for maize. After 12 h of local nitrate treatment, the transcript levels of target genes in roots were determined by qPCR. Values are normalized by OsACTIN (LOC_Os03g50890). Data represent means ±SD (n=3 replicates; each replicate represents a single seedling). Asterisks indicate significant differences between the gene expression in roots in the +N and –N compartments at: *P<0.05; **P<0.01; ns, not significant (Student’s t-test). (C) A model scheme describing the function and regulation of the truncated MADS-box gene ZmTMM1. In Arabidopsis, local nitrate activates ANR1 via a post-translational mechanism to stimulate LR proliferation; while in maize, local nitrate induces the expression of ZmTMM1 transcripts to promote LR development.

Discussion

Root system architecture shows complex morphology and plasticity to adapt to nutrient heterogeneity (Giehl and von Wirén, 2014). Identification of key regulatory components modulating root morphological responses to nutrient heterogeneity is of remarkable ecological and agronomic relevance for improving nutrient use efficiency in cropping systems (Rogers and Benfey, 2015). In this study, we identified a truncated <span class="Gene">MIKC-type MADS-box transcription factor gene ZmTMM1 involved in root <span class="Chemical">nitrate-foraging responses.

ZmTMM1 is a unique truncated MADS-box gene modulating LR development

Genome duplications and chromosome fusions frequently occur in grass species during evolution, which reconstruct the genome and lead to functional diversification among the paralogs and orthologs (Bolot ). The AGL17-like gene products in <span class="Species">dicots exhibit the complete protein structure of the classic MIKC-type MADS-box genes. In contrast, the truncated AGL17-like MADS-box genes are found exclusively in grass species (Fig. 2A). It is suggested that these truncated forms have evolved as a result of the insertion of an ancestral intron between the I- and K-domains (Supplementary Table S1; Supplementary Fig. S1). This intron can be removed during the mRNA splicing process, as in the case of AGL17-like genes in Arabidopsis and rice (Supplementary Fig. S1). However, if the size of the intron was too large for mRNA splicing, then the first half of the coding sequence only with the M- and I-domains causes a truncated MADS-box protein to be produced, although the remaining second half of the coding sequence may still exist downstream on the genome, such as those identified for BRADI5G12440, Sb07g021110, GRMZM2G055782, and GRMZM2G032905 (Supplementary Table S1). In addition, not via mRNA splicing, the truncated forms can also be directly transcribed from the MADS-box genes, in which the genomic regions corresponding to the introns and the following coding sequences are completely lost presumably during the genome reconstruction, as exemplified by ZmTMM1, Sb04g024010, and Sb06g019040 (Supplementary Table S1). It should be noted that all AGL17-like gene products in rice still exhibit the complete protein structure of MIKC-type MADS-box genes (Fig. 2A; Supplementary Fig. S1), indicating that the deletion of K- and C-domains might have occurred after the subdivergence of rice with other grass species.

For MIKC-type MADS-box transcription factors, the M-domain is responsible for DNA binding, the K-domain facilitates protein–protein interaction, and the C-terminal domain participates in transcriptional regulation (Kaufmann ). The C-terminal deletion of two <span class="Species">Arabidopsis MADS-box genes, SOC1 and AG, produces two truncated non-functional isoforms that act as artificial interfering peptides to inactivate the original SOC1 and AG by protein–protein interaction (Seo ). In some other cases, however, the C-terminal domain seems to be dispensable for the functions, such as rice OsMADS57 in control of tillering (Guo ) and Arabidopsis AP3 and PI in flowering regulation (Piwarzyk ; Su ). Thus, the MADS-box domain and I-domain appear to form the minimum unit responsible for DNA binding, which is essential for the specificity of MICK-type MADS-box transcription factors (Kaufmann ). In this study, ZmTMM1 containing only the M- and I-domains has been shown to be a functional MADS-box transcription factor (Fig. 2A). ZmTMM1–GFP fusion protein is located in the nucleus, and both the overexpression of ZmTMM1 and the DEX-mediated local activation of ZmTMM1–GR fusion protein recovered LR development of the Arabidopsis dko mutant (Figs 5, 6). The GR fusion of an artificially truncated ANR1S protein, mimicking the protein structure of ZmTMM1, also promotes the LR growth similar to the way in which the intact ANR1 protein does in the dko background (Fig. 7), implicating that K- and C-domains are not essential for the function of AGL17-like genes to regulate LR development.

AGL17-like MADS-box genes confer a conserved function in <span class="Chemical">LR development, as indicated by roles of ANR1 and AGL21 in Arabidopsis and OsMADS25 and OsMADS57 in rice (Zhang and Forde, 1998; Gan ; L.H. Yu ; C. Yu ; Zhang ; Huang ). Among them, ANR1, OsMADS25, and OsMADS57 are involved in root nitrate foraging. Ectopic overexpression of ZmTMM1 in the Arabidopsis dko mutant revealed that ZmTMM1 stimulates LR growth, as is reflected, respectively, by the nitrate-dependent and -independent increase in the LR length and the LR number (Figs 6, 7; Gan ). A monocot-exclusive miRNA, miR444, additionally regulates AGL17-like genes (OsMADS23, OsMADS27, and OsMADS57) in rice by targeting the coding sequence in the K-domain, which leads to post-transcriptional degradation of the coding mRNA (Guo ; Yan ). The lack of a K- and C-domain in ZmTMM1 and other ZmTMM1-like truncated MADS-box genes in grass species thus may imply a strategy evolved to escape from the negative control mediated by miR444. The functional conservation of AGL17-like genes in LR development also implies the existence of genetic redundancy among the homologs. Indeed, the Arabidopsis anr1 single mutant and the anr1 agl21 double mutant showed local nitrate-induced LR proliferation to a similar extent as the wild type (Gan ; Supplementary Fig. S5C). Similarly, in maize, the response of LR growth to local nitrate was not affected in the ZmTMM1-RNAi lines (Supplementary Fig. S8), suggesting a possible functional compensation because of the presence of the remaining AGL17-like genes such as ZmMADS2 and GRMZM2G055782. Since these two genes did not respond to local nitrate supply at the transcript expression levels (Fig. 8A), their post-translational regulation can be postulated as a mechanism that could have compensatorily stimulated LR proliferation in nitrate-fed root segments in the maize ZmTMM1-RNAi lines (Supplementary Fig. S8) as ANR1 does in Arabidopsis (Fig. 8C). Further studies on multiple knockout or overexpression lines may be necessary to fully understand the function of AGL17-like genes in maize.

Local nitrate induction is a conserved feature of ZmTMM1-like genes in cereals

Root-preferential expression is an ancestral feature of the AGL17-like genes (Becker and Theissen, 2003). In <span class="Species">Arabidopsis, three of the AGL17-like members (ANR1, AGL21, and AGL17) are mainly expressed in root tips despite their differential cell type specificity of gene expression. For example, ANR1 and AGL21 are preferentially expressed in the LR primordia and central cylinder of mature roots, while AGL17 is specifically detected in the LR cap and the epidermis (Supplementary Fig. S4; Burgeff ; Remans ; L.H. Yu ). In rice, the expression of OsMADS23, OsMADS25, OsMADS27, and OsMADS57 is detected in the root central cylinder (Puig ); in particular, OsMADS57 expression is shown in the xylem parenchyma cells (Huang ). ZmTMM1 transcripts are also found in the xylem parenchyma cells of mature roots and the cortical cells surrounding the LR base (Fig. 3). Thus, during evolution, both the intact and truncated forms of AGL17-like genes, either in dicots or in monocots, may have retained the root-preferential expression pattern from their ancestor, supporting their conserved roles in root nutrient foraging.

However, AGL17-like genes in dicot and monocot species show distinct responses to N availability. In Arabidopsis, ANR1, AGL21, and AGL16 are up-regulated by N deprivation and repressed by N resupply, while AGL17 is unaffected by the changes in N availability (Gan ). In the split-root system, the expression of all these Arabidopsis AGL17-like genes is not transcriptionally regulated by local supply of nitrate (Supplementary Fig. S9). To stimulate LR proliferation in a nitrate-rich zone, ANR1 is proposed to be subject to a post-translational regulation in the presence of nitrate (Gan ). In contrast, in rice roots, OsMADS25, OsMADS27, and OsMADS57 are transcriptionally induced in response to nitrate supply (Puig ; C. Yu ; Huang ). It should be noted that previous studies reporting the expression patterns of AGL17-like genes in Arabidopsis or rice were conducted under conditions of homogenous nitrate supply, not allowing differentiation of their regulation via local or systemic nitrate signaling pathways (Gan ; Puig ; Li ; C. Yu ; Huang ). Therefore, in the present study, we performed split-root experiments to investigate the response of AGL17-like genes to local N signaling specifically (Figs 4C, D, 8A, B; Supplementary Fig. S9). Unlike other AGL17-like genes, ZmTMM1-like genes in the monocot-specific subclade are transcriptionally regulated by local nitrate rather than a systemic N signal. This expression pattern suggests that ZmTMM1-like genes are the homologs that could have retained the transcriptional control mechanisms relevant to responses to local nitrate supply even after the divergence of dicot and monocot species. It may be speculated that the two key features of ZmTMM1, truncated protein and transcriptional regulation by local nitrate, are likely to have evolved independently.

Taken together, we propose a model whereby in the dicot plant <span class="Species">Arabidopsis, local supply of nitrate activates ANR1 at the post-translational level to stimulate LR proliferation; however, in grass species such as maize, local supply of nitrate directly induces the expression of ZmTMM1 at the transcriptional level to promote LR development in nitrate-fed root segments (Fig. 8C). The post-translational control may also exist in ZmTMM1, although it is a truncated MADS-box gene. This study provides new insights into diverse functions and regulations of MICK-type MADS-box genes across different plant species, and opens up an avenue toward understanding of developmental control mechanisms involved in root nitrate foraging.

Supplementary data

Supplementary data are available at JXB online.

Fig. S1. The exon–intron structure of AGL17-like genes in <span class="Species">Arabidopsis and rice.

Fig. S2. Root-preferential expression of ZmTMM1 in <span class="Species">maize.

Fig. S3. Expression of an N-responsive marker gene <span class="Gene">ZmGS1.1 and total N concentration in <span class="Species">maize roots in response to local N supply in a split-root system.

Fig. S4. Localization of AGL21 and <span class="Gene">ANR1 promoter activities in Arabidopsis roots.

Fig. S5. <span class="Chemical">LR growth phenotype of the <span class="Species">Arabidopsis dko mutant.

Fig. S6. Ectopic expression of ZmTMM1 in the dko mutant.

Fig. S7. Construction and analysis of ZmTMM1–, <span class="Gene">ANR1S–, and <span class="Gene">ANR1–GR fusion transgenic lines.

Fig. S8. Phenotypic analysis of ZmTMM1-RNAi transgenic <span class="Species">maize under local <span class="Chemical">nitrate supply.

Fig. S9. Expression of Arabidopsis <span class="Gene">AGL17-like genes in response to local nitrate supply.

Table S1. Gene structure of truncated <span class="Gene">AGL17-like genes and orthologs in <span class="Species">monocots identified from comparative genome analysis.

Table S2. Primers used in this study.

Dataset S1. Comparative genome analysis of <span class="Gene">AGL17-like genes in <span class="Species">monocots.

Click here for additional data file.

Click here for additional data file.