Introduction

Plant roots are crucial to their anchorage, absorption of nutrients and water, as well as to establishment of beneficial symbioses with the surrounding soil microorganism communities1,2. In some cases, such as in the dicot model plant Arabidopsis, the radicle is generated during embryogenesis within a seed. Following germination, the radicle elongates as the primary root (PR), and typically grows into a central taproot together with the sequential development of the associated lateral roots (LRs)3. However, monocot plants such as rice (Oryza sativa) and maize (Zea mays) develop a more complicated root system4. Apart from a specific embryonic PR and several seminal roots (SRs) in maize, most cereals possess an expanding shoot-borne root system5. The postembryonic shoot-born roots, called brace roots (BRs) and crown roots (CRs), are also able to branch developing lateral roots6. Moreover, some plants species such as chrysanthemum (Chrysanthemum morifolium), African violet (Saintpaulia spp.), strawberries (Fragaria spp.), and garlic (Allium sativum), which propagate vegetatively, firstly develop numerous adventitious roots (ARs) from the basal cuttings or stolons in their typical ecological environments3. Successively, LRs or higher-order LRs originate from the existing ARs to expand the root system in order to obtain more water and nutrients.

Adventitious roots, which are similar to lateral roots, develop post-embryonically. ARs usually arise from vegetative organs, such as the stolons, rhizomes, leaves, and stems, while LRs often originate from existing roots, such as the PRs, previous LRs, or ARs3. Despite the differences in origin, the formation and development of ARs and LRs is controlled by a suite of similar endogenous and environmental factors3,7,8. Among these common regulatory factors, auxin is the most vital regulator of both AR and LR development9,10,11,12,13. Natural auxins (e.g., indole-3-acetic acid (IAA)) and synthetic analogs (e.g., indole-3-butyric acid (IBA)) have a powerful and stimulatory effect on rooting in many plant species12,14,15. A diverse range of studies have shown that auxin is central to AR and LR development in plants, where it cross-talks with other signals (e.g., calcium signal)16,17, regulatory genes (e.g., AtMYB93, SHR, and ERF3)18,19,20, or phytohormones (e.g., ethylene and gibberellins)21,22,23. Furthermore, auxin-related biological processes, such as signal transduction, polar transport, and local biosynthesis, account for the primary underlying molecular mechanisms regulating AR or LR formation10,11,24,25,26,27.

Generally, auxin is produced in the aerial tissues (such as apical meristems) and is then distributed locally and systemically throughout whole plant via two distinct yet interconnected ways: a direct and fast flow from shoots to roots via the vascular central cylinder, and cell-to-cell active polar transport through the outer layers of the root cells28,29. However, auxin is also synthesized in the root tips, where auxin transport is characterized by dual polarities. In the roots, auxin polar transport has been described as acropetally (towards root apex) and basipetally (from apex to base)30. The auxin efflux carriers PIN-FORMED (PIN) proteins as well as auxin influx carriers AUXIN RESISTANT1/LIKE AUX1 (AUX1/LAX) proteins have been identified as the main components responsible for auxin transport31,32. The asymmetric subcellular distribution and localization of carrier proteins contributes to their polarity33, among which PIN polarity has been shown to be the primary direction-determining factor in auxin polar transport34. So far, PIN members have mostly been well studied in Arabidopsis where they exhibit unique but somewhat overlapped localization in various cell types. PIN1 localizes in the basal end of the vascular cells, facilitating the root-ward movement of auxin35. PIN2 predominantly resides basally in cortical cells, and apically in the epidermal and root cap cells34,36. PIN3 has been detected in the columella cells of the roots in an apolar manner. PIN7 localizes basally in the stele cells in the meristem and elongation zone37. Recent studies reported that PIN5, PIN6, and PIN8 located on both the endoplasmic reticulum and plasma membrane38,39,40,41,42. Additionally, PINs have been functionally identified as vital regulators of numerous auxin-related developmental processes. For instance, OsPIN1 takes part in auxin-dependent AR emergence and development in rice43. OsPIN2 regulates tiller angle, number, as well as plant height by enhancing basipetal auxin transport in rice44. In Arabidopsis, the triple mutant pin1 pin3 pin4 is defective in PR development33; PIN6 is required for LR and AR organogenesis by controlling auxin homeostasis and distribution42; and PIN8 exerts a crucial role on pollen development and functionality38,39.

Transcription factors (TFs) represent a large portion of the essential regulators of many developmental aspects in plants. With respect to root development regulation, a multitude of TFs have been reported to participate in, and influence, a diverse set of developmental stages of different root types in many plant species18,19,20,45,46. The MADS-box TFs became a point of interest into the genetic regulation of root development. In addition to developmental regulation of flower, fruit, seed, and leaf developmental regulation in plants47,48,49,50, an increasing number of MADS-box genes have been reported to be involved in root system development. For example, AtANR1 was the first MADS-box TF gene identified to regulate LR elongation under heterogeneous nitrate conditions51. The MADS-box gene XAL1/AGL12 gene was determined to be necessary for normal root development and growth via a positive control on cell cycle components52. XAL2/AGL14 plays an essential role in robust root patterning by modulating auxin polar transport53. AGL21 was found to be a positive regulator of LR initiation and growth by increasing local auxin biosynthesis in Arabidopsis27. The ANR1-like gene OsMADS25 positively regulates both PR and LR development by promoting nitrate accumulation in rice54. GmNMHC5, a MADS-box TF gene in soybean, promotes LR development in a sucrose-dependent manner55.

Recently, it was reported that the ectopic expression of CmANR1/CmAGL44 in Arabidopsis could promote LR development56. However, the underlying mechanisms of CmANR1 modulation of LR development are largely unknown. Here, it was found that CmANR1 positively modulates AR and LR development by directly regulating CmPIN2 in chrysanthemum. The potential application of the CmANR1 gene in controlling root system development and its theoretical research value in breeding programs in chrysanthemum are discussed in this paper.

Materials and methods

Chrysanthemum and growth conditions

The 35S::CmANR1-1258 (green fluorescent protein (GFP) tag) overexpressed vector was constructed as previously described56. It was then transformed into Agrobacterium strain GV3101. The wild-type (WT) tissue-cultured chrysanthemum were kindly provided by Professor Gao (China Agricultural University). The CmANR1-transgenic chrysanthemum were obtained by Agrobacterium-mediated transformation of leaf discs57. In tissue-cultured condition, CmANR1-transgenic and WT plants were cultivated in vitro on Murashige and Skoog (MS) medium in the standardized culture room.

In hydroponic-cultured condition, the chrysanthemum were cultivated in improved Hogland nutrient solution (CaCl2 555 mg/L, MgSO4•7H2O 493 mg/L, KH2PO4 136 mg/L, FeSO4•7H2O 27.6 mg/L, EDTA-2Na 3.73 mg/L, KNO3 10 mM, H3BO4 2.86 mg/L, MnCl2•4H2O 1.82 mg/L, ZnSO4•7H2O 0.23 mg/L, H2MoO4•H2O 0.09 mg/L, CuSO4•5H2O 0.08 mg/L, pH = 5.6). A simple aeration device was used to supply oxygen in hydroponic condition in case that roots would go rotted.

Arabidopsis AR rooting assays

The Arabidopsis lines used here were the CmANR1-overexpressing (OE) lines and the “Columbia” ecotype. For the procedure of seed sterilization and the growth conditions of Arabidopsis, refer to our previous study56. After sterilization, the seeds were planted on MS medium with 1% (w/v) sucrose and 0.7% (w/v) agar. Then, the seedlings with only two cotyledons on were trimmed, leaving only the hypocotyls. The hypocotyls were transferred vertically on either MS or MS added with 0.1 μM IBA for AR rooting analysis. The AR rooting assays were performed in a dark growth incubator (23 ± 1 °C, 40% relative humidity) for a week.

Morphological characterization of roots in chrysanthemum and Arabidopsis

The 20-day-old in vitro chrysanthemum and 40-day-old hydroponic-cultured chrysanthemum were used for root morphological characterization. The relevant root data such as root total length, root volume, and root surface were analyzed by WinRHIZO software (Regent Instruments Inc., Canada). The root numbers of AR and LR were counted using Image J software (NIH, Bethesda, MD, USA) of digital images of roots. The root morphology of CmANR1-overexpressing and WT control Arabidopsis seedlings was on observed MS solid medium with 0.8% agar. Photos of the seedlings were taken after 1 week of darkness, about 10 days old. AR number and AR length was measured by hand using Image J software (NIH, Bethesda, MD, USA).

RNA-Seq data processing, de novo assembly, and annotation

The samples were the 40-day-old hydroponic-cultured CmANR1-overexpressing plants (CmANR1-OVX56, -OVX67, -OVX81) and WT plants. Three independent plants of each line consisted of the triplicate samples. Total RNAs of the samples were extracted using the RNeasy plant mini kit (New England Biolabs Inc., New England). The complementary DNA (cDNA) library was prepared as described by Grabherr et al.58 and the sequencing was performed on the Illumina HiSeq Platform (Ori-gene Inc., Beijing, China).

Raw RNA-Sequencing (RNA-Seq) reads were conducted with Cutadapt based on BMA algorithm to remove sequence artifacts such as adapter sequences on both ends, low-quality trailing (Q30), 3’-end barcode sequences, and reads with lengths less than 60 bp. The remaining valid cleaned reads were processed into de novo transcript assembly according to a previous study58. Furthermore, the resulting reads were assembled using iAssembler with a threshold of (-p) set to 9959,60.

The resulting unigenes were screened by BLAST (Basic Local Alignment Search Tool) against the GenBank non-redundant (NR), TrEMBL, Swiss-Prot, Pfam, and KOG. The unigenes with a cutoff of E-value of ≤1e−5 and ≥30% identify were needed for further functional annotation. The chrysanthemum assembled unigenes and their corresponding homologs in the UniProt database were assigned to Gene Ontology (GO) terms. Biochemical pathway prediction of the chrysanthemum transcripts were annotated and enriched by the Pathway Tools61,62.

Quantitative real-time (qRT)-PCR analysis

Total RNA was extracted from the roots of 40-day-old hydroponic-cultured transgenic and WT chrysanthemum using the RNA plant plus Reagent (Tiangen, Beijing, China).Then, cDNA was synthesized using the PrimeScript first-strand cDNA synthesis kit (TaKaRa, Dalian, China). Each qRT-PCR reaction (20 μl) included 1 μl cDNA template, 1 μl of both up and down primers (10 μM), 10 μl SYBR Green Ι, and 7 μl RNase-free H2O. The qRT-PCR assays were carried out according to the StepOne real-time PCR system (Applied Biosystems). All reactions were repeated three times, and a chrysanthemum Ubiquitin gene (CmUBI) served as the reference gene60. Relative gene transcript abundances were computed with the 2−ΔΔCt method56. The primers used for qPCR reactions are referred to in Supplementary Table 4.

The cloning and analysis of the promoters

Genomic DNA extracted from the leaves of “Jinba“ using the Plant Whole-genome Extraction Kit (Tiangen, Beijing, China) served as the PCR template. The promoters of four auxin-responsive genes pCmPIN2, pCmGH3.1, pCmTAA1, and pCmAB37G were cloned according to the instructions of Genome-walking Kit (TaKaRa, Dalian, China). A length of 1994 bp pCmPIN2 was a twice genome-walking result. The specific primers for cloning the promoters were designed according to cDNA sequences searched in our RNA-Seq result. Related primers are listed in Supplementary Table 4. PLACE cis-acting regulatory DNA elements analysis was completed on the following website (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/).

The expression and purification of CmANR1-His fusion protein

The open reading frame (ORF) of CmANR1 was cloned from the cDNA of CmANR1 using the paired primers with BamHI and XhoI sites (Supplementary Table 4), and then was constructed into the pET-32a vector, which had a histone (His) tag sequence. Then, the recombinant plasmid was introduced into Escherichia coli BL21 (DE3). The CmANR1 (His)-BL21 bacteria was incubated at 37 °C constant temperature shaker (200 rpm) for about 2 h. Subsequently the bacteria were treated with 3 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) for inducing the generation of the CmANR1-His fusion protein. After 4 °C centrifugation of the bacteria, the precipitate was denatured and renatured with a series of specific concentrations of urea solution63. The final fusion protein was transferred to a cobalt chelate affinity resin, which contained the immobilized His-tag. The tube was incubated at 4 °C for 2 h on the shaker. After three times separation and abstersion, the protein was collected and detected by western blot using His antibodies (Abcam, Cambridge, UK).

ChIP-qPCR and EMSA analysis

The chromatin immunoprecipitation quantitative PCR (ChIP-qPCR) tests were performed using the EpiTect ChIP OneDay kit (QIAGEN,Shanghai, China) as described in the previous study63. The primers used for ChIP-qPCR are described in Supplementary Table 4. Gel electrophoresis mobility shift assay (EMSA) was carried out following the instructions of the manufacturer in the Light Shift Chemiluminescent EMSA Kit (Thermo, Waltham, MA, USA). Concisely, the biotin-labeled probe was incubated in the 1× gel/DNA shift binding buffer containing 5 mM MgCl2, 50 mM KCl, 2.5% glycerol, and 10 mM EDTA with or without CmANR1 protein at 24 °C for 25–30 min. The unlabeled probe with specified concentrations (50×, 100×) was used for cold probe competition. Related primers are referred to in Supplementary Table 4.

In vivo firefly luciferase (Luc) imaging assay

The Luc imaging assays were carried out in Nicotiana benthamiana leaves, while the transient expression was performed as previously described64. The promoter of CmPIN2 was cloned into pGreenII 0800-LUC vector, generating the reporter CmPIN2pro::LUC. The effector (35Spro::CmANR1) was constructed by cloning the fragment of CmANR1 (ORF) into the pGreenII 62-SK vector. Then, the recombinant vectors CmPIN2pro::LUC and 35Spro::CmANR1 as well as the empty vectors pGreenII 0800-LUC (LUC) and pGreenII 62-SK (35S) were introduced into Agrobacterium strain LBA4404, respectively. The four independent Agrobacterium bacteria with similar OD600 absorbance were 1:1 pairwise mixed. The four kinds of mixed bacteria were infiltrated on the four sites of a same mature N. benthamiana leaf, respectively. A fluorescence imaging instrument (NightOWL II LB983) in conjunction with the Indigo software was used for LUC imaging and luminescence intensity quantification. Infiltrated leaves were sprayed with little luciferin (100 mM), then were put in darkness for 5–10 min before LUC imaging65. Related primers are listed in Supplementary Table 4.

Determination of total IAA in roots

About 0.2 g (fresh weight) root samples were prepared and vacuum-dried at −35 °C for about 12 h. After quick grind in liquid nitrogen, the powder of the samples was extracted in accordance with the method described by Lin et al.66. The total free IAA was detected by the high-performance liquid chromatography.

Statistical analysis

All samples were analyzed in at least triplicate repeats and represented as the mean ± standard deviation unless specifically labeled. Significance analysis was determined by Student’s t-test. P ≤ 0.001 meant a extremely significant difference, p ≤ 0.01 represented a significant difference, while n.s. meant no significance.

Results

CmANR1 promotes AR development in chrysanthemum under tissue culture conditions

The 35S::CmANR1-1258 (GFP) recombinant plasmid was introduced into chrysanthemum leaf discs using Agrobacterium GV3101-mediated transformation. Subsequently, several positive candidates of CmANR1-transgenic plants, which were preliminarily screened by PCR detection using the genomic DNA as the template, were further identified by qRT-PCR analysis (Supplementary Fig. S1). Among them, three independent CmANR1-transgenic lines (CmANR1-OVX56, -OVX67, -OVX81) with significantly distinct expression levels of CmANR1 were selected for further investigation (Fig. 1a). Immunoblotting assays showed that the CmANR1 protein accumulated in these three CmANR1-transgenic chrysanthemums much more than in the WT control (Fig. 1b). These three CmANR1-transgenic and WT chrysanthemums were then rooted and cultivated in vitro. Following this, the CmANR1-overexpressing plants exhibited a stimulation on AR development compared with the WT plants (Fig. 1c, d). Remarkably, the numbers and total lengths of the ARs in the CmANR1-OVXs lines were increased by 43.7–98.8% and 61.8–184.5%, respectively, in comparison to the WT plants (Fig. 1e, f). Therefore, a significant increase in both root surface and volume was showed in the transgenic plants compared to the WT plants (Supplementary Table 1). These results suggest a positive role of CmANR1 on AR development in chrysanthemum.

Fig. 1: CmANR1 plays a positive role in adventitious root development in tissue-cultured chrysanthemum.
figure 1

a Relative expression level of the CmANR1 gene in the CmANR1-transgenic and the WT chrysanthemum lines. b The level of the CmANR1-GFP fusion protein in CmANR1-transgenic plants, as determined by immunoblot analysis using an anti-GFP antibody. The anti-actin antibody was used as a loading control. c Phenotypes of CmANR1-OVXs and WT chrysanthemum (tissue culture). Scale bar = 1 cm. d The root-specific scanning pictures of CmANR1-transgenic and WT plants (tissue culture). Scale bar = 1 cm. e AR number and total root length (f) of CmANR1-transgenic and WT plants (tissue culture). Data are shown as the mean ± standard error (SE) based on three replicate. Statistical significance was determined using Student’s t-test. No significance (n.s.): p > 0.01; *p < 0.01; **p < 0.001

Full size image

Surprisingly, fewer LRs were found in both the CmANR1-transgenic and WT plants under tissue culture conditions, which seemed to be in contrast to the positive role of ANR1 on LR growth in Arabidopsis51,56. To further confirm the role of CmANR1 on AR development, AR rooting experiments were performed using CmANR1-overexpressing (CmANR1-OE3, OE6 and OE9) and WT Arabidopsis seedlings, which were obtained in our previous study56. Only the hypocotyls of those plants were placed on the MS medium and the MS medium with 0.1 µM IBA added, and cultivated vertically under dark conditions. Following this, the seedlings developed new ARs after about 1 week of growth. The total length and numbers of ARs in the CmANR1-transgenic seedlings were significantly increased compared to the WT plants, with respective increases of 8.8–56.6% and 50.4–150.2% (Supplementary Fig. S2). Additionally, the exogenous application of IBA almost abolished the developmental differences of the ARs between the transgenic and WT seedlings, suggesting that auxin may have some relationship with AR development.

CmANR1 promotes AR and LR development in chrysanthemum under hydroponic culture conditions

To better evaluate the function of CmANR1 in root system development, the WT and CmANR1-overexpressing chrysanthemums exhibiting uniform growth under tissue-cultured conditions were then cultivated hydroponically. The results showed that the CmANR1-overexpressing chrysanthemums possessed a much more extensive root system, including more ARs and LRs, compared to the WT plants after about 35–40 days of growth (Fig. 2a, b). CmANR1-overexpressing chrysanthemum exhibited a significant increase in root volume and total root length compared to the WT plants, by 0.6–1.9-fold and 0.5–1.2-fold, respectively (Fig. 2c, d). Meanwhile, the numbers of ARs and LRs in the three CmANR1-overexpressing plants were much greater than those of the WT plants, with 0.3–0.5-fold and 0.2–1.1-fold increases, respectively (Fig. 2e, f). The stronger root system of the CmANR1-transgenic plants indicated the positive effect of CmANR1 on rooting in chrysanthemum under hydroponic conditions. Furthermore, a significant stimulation on shoot height was observed in the CmANR1-overexpressing chrysanthemums compared to the WT plants, elevated by 16.1–51.9% (Supplementary Table 2). The increase in shoot height of the CmANR1-OVXs plants may be attributed to a feedback-enhanced uptake of water and nutrients by the more extensive root system.

Fig. 2: CmANR1 promotes both adventitious root and lateral root development in hydroponic-cultured chrysanthemum.
figure 2

a Phenotypes of CmANR1-transgenic and WT chrysanthemums under hydroponic culture conditions. Scale bar = 1 cm. b The root-specific magnified pictures of CmANR1-OVXs and WT plants (hydroponic culture). Scale bar = 1 cm. c–f The root volume (c), total root length (d), AR number (e), and LR number (f) of CmANR1-transgenic and WT plants. The data represent the means ± SE of three independent experiments. Statistical significance was determined using Student’s t-test. n.s.: p > 0.01; *p < 0.01; **p < 0.001

Full size image

Transcriptome sequencing of CmANR1-overexpressing and WT chrysanthemum roots

To reveal the underlying mechanism of CmANR1 in controlling root system development, strand-specific RNA sequencing libraries from the roots of the WT and CmANR1-overexpressing chrysanthemums were constructed. De novo assembly of the valid cleaned reads produced 51,481 unigenes with a mean length of 653 bp and a longest length of 8281 bp. The length distribution of the assembled unigenes is exhibited in Supplementary Table 3. Then, we annotated the assembled unique transcripts by BLAST against several protein databases. A total of 24,292 (47.2%), 36,870 (71.7%), 37,115 (72.2%), 28,442 (55.3%), 30,328 (59.0%), 30,392 (59.1%), and 12,050 (23.4%) unique transcripts obtained significant hits (identity ≥30%, E-value ≤ 1e−5) in the Swiss-Prot, TrEMBL, GenBank NR, Pfam, eukaryotic orthologous groups (KOG), GO (http://www.geneontology.org/), and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases, respectively. Remarkably, “Signal transduction mechanisms” was the second-most abundant group in these KOG functional categories after “Posttranslational modification, protein turnover, chaperones”, irrespective of the poorly characterized ones (Supplementary Fig. S3a). In contrast, the GO terms “metabolic process” in the biological process category, “cell” in the cellular category, and “binding” in the molecular function category were the most enriched in these three categories, respectively (Supplementary Fig. S3b). In the KEGG pathway classification, “Carbohydrate metabolism” in the metabolism category was the most abundant group (Supplementary Fig. S3c).

Subsequently, FPKM (Fragments Per Kilobase of transcript per Million fragments mapped) was used to evaluate the expressional abundances of the assembled unigenes based on the transcriptome sequencing data. The volcano plot showed the relationship between the significance of the p value and fold change of all the differentially expressed unigenes (DEGs) (Fig. 3a). Simultaneously, the MA Value Plot presented the distribution and differences in DEGs in the WT and CmANR1-overexpressing chrysanthemums (Supplementary Fig. S4a). In addition, a heatmap of the DEGs provided a visual illustration of the expressional differences between the WT and CmANR1-overexpressing chrysanthemums, with high expression levels shown in red and low expression levels shown in green (Supplementary Fig. S4b). A total of 7612 DEGs were identified in the CmANR1-overexpressing chrysanthemums compared to the WT plants (Fig. 3b; Supplementary Appendix S1). Among them, 5698 (11.1%) were up-regulated and 1914 (3.7%) were down-regulated (Fig. 3b, c; Supplementary Appendix S1). Subsequently, these up-regulated and down-regulated genes that were involved in the main biological functions, as well as the biochemical metabolic and signal transduction pathways in GO and KEGG enrichment, were analyzed (Supplementary Appendix S2 and S3). GO terms involving “catalytic activity” and “binding” in the molecular function category and “metabolic process”, “response to stimulus”, and “signaling” in the biological process category were highly abundant in the GO enrichment annotation of the up-regulated unigenes (Fig. 3d). Meanwhile, “Plant hormone signal transduction” was abundant in the KEGG-enrich analysis and enriched bubble diagram of the up-regulated unigenes (Fig. 3e). These information suggested that CmANR1 might participate in the regulation of plant hormone signaling processes. However, DEGs being classified to “Plant-pathogen interaction” term in KEGG-enrich analysis accounted for the most abundant, which attracted our attention and would be discussed later.

Fig. 3: Annotation enrichment of DEGs in the roots of CmANR1-OVXs in contrast with WT chrysanthemum.
figure 3

a The volcano plot of the DEGs. A small dot in the picture represents a unigene; the x-axis represents the log2(fold change) of a gene expression difference between WT and CmANR1-OVXs; the y-axis represents the negative log Kow of the p value of the gene expression change. Unigenes with significant differential expression are indicated with a red dot. b The numbers of up-regulated and down-regulated unigenes in the RNA-Seq results. c Pie chart representing the percentage of up- and down-regulated unigenes and indistinguishable unigenes in RNA-Seq data. d GO enrichment analysis of the DEGs. The x-coordinate is the GO classification, which is the secondary function of GO; the left vertical coordinate is the percentage of the DEGs; and the right shows the corresponding numbers. e KEGG bubble chart of the DEGs. The bubble size represents the number of DEGs, and the bubble color represents the p value. The rich-factor equals the number of DEGs/the number of background genes in a certain signaling pathway

Full size image

Expressional profiling of the genes involved in root system development in chrysanthemum

Four groups of potential candidate unigenes from the DEGs associated with root system development, including auxin-responsive group, calcium (Ca2+) signaling-related group, ethylene-related group, and cell cycle group, were selected for further investigation. The corresponding ID numbers and log2(fold changes) values of these unigenes are listed in a supplementary file (Supplementary Appendix S4). The heatmaps display the average absolute expression values after log2 transformation between WT and CmANR1-overexpressing chrysanthemums (Fig. 4). Among them, the expressions of some auxin transport unigenes (PIN2, AUX1, AB37G, and AB11B) and a range of auxin-responsive protein genes (SAU20, SAU36, RHM1, and XTH20) showed a significant increase in the CmANR1-overexpressing chrysanthemum compared to the WT plants in the auxin-responsive group (Fig. 4a). Ca2+ signaling-related unigenes, such as the calcium-binding protein genes CMLs (CML45, CML23, and CML50) and calcium-dependent protein kinase genes CDPKs (CDPK6, CDPK10, CDPK30, and CDPK33), were also found to be significantly up-regulated in the CmANR1-overexpressing chrysanthemums compared to the WT plants (Fig. 4b). The ethylene biosynthesis genes ACS7 and ethylene-responsive TF genes ERFs (ERF3, ERF78, and ERF109) were significantly up-regulated in the CmANR1-overexpressing chrysanthemums compared with the WT plants (Fig. 4c). In addition, the expressions of cell cycle-related genes (CYCB1-4, CYCD3-2, ALISs, MLH1, and MKK4) were also significantly altered in the CmANR1-overexpressing chrysanthemums compared to the WT plants (Fig. 4d).

Fig. 4: Heatmaps of four groups of candidate unigenes associated with root system development in the DEGs.
figure 4

a Heatmap of auxin-responsive unigenes. b Heatmap of Ca2+ signaling-related unigenes. c Heatmap of ethylene-related unigenes. d Heatmap of cell cycle-related unigenes. Differences in expression levels are represented by color gradients. Red and orange strips indicate highly to moderately up-regulated unigenes, while dark blue to light blue strips represent highly to moderately down-regulated unigenes. The ID numbers of the four groups of unigenes are provided in Supplementary Appendix S4. The red unigenes in the four groups were selected for qRT-PCR

Full size image

To further validate our RNA-Seq results, 18 unigenes that were marked in red in Fig. 4 from the four groups were selected for qRT-PCR verification. As a result, most of these unigenes were significantly increased in the CmANR1-overexpressing chrysanthemums compared to the WT plants (Fig. 5a, b). Additionally, we found that the exact fold changes of the detected unigenes varied between the qPCR analysis and RNA-Seq expression data (Fig. 5c), but the high correlation (R2 = 0.882) correlated with a simple linear regression equation y = 0.96x + 0.467 demonstrated that a good consistency existed between the two experimental methods (Fig. 5d).

Fig. 5: Verification of the RNA-Seq results by qRT-PCR.
figure 5

a The relative transcript abundance of some IAA-responsive unigenes that were selected based on their differential expression in the roots of WT vs. CmANR1-transgenetic plants. b The relative expression level of the unigenes in Ca2+ signaling-related, ethylene-related, and cell cycle-related groups in the roots of WT and CmANR1-transgenetic plants. c Comparison of the expression level of unigenes between the RNA-Seq and qRT-PCR. d Scatter diagram of the log ratios (log2 FC) of the unigenes. The qRT-PCR data were normalized to the internal control CmUBI. Note that in a and b, data are shown as the mean ± SE based on three or more replicate. Statistical significance was determined using Student’s t-test. n.s.: p > 0.01; *p < 0.01; **p < 0.001

Full size image

CmANR1 facilitates auxin polar transport by direct transcriptional activation of CmPIN2

Accumulating evidence suggests that auxin plays a central role in both AR and LR development30. To determine the causality between auxin and root system development, we measured the endogenous free IAA content in the roots of the WT and CmANR1-transgenic plants. As observed with the up-regulated expressions of auxin polar genes, such as PIN2, PIN3, and AUX1 (Fig. 5a), the free IAA level in the roots of CmANR1-transgenic chrysanthemum was elevated by 23.4–89.4% compared to the WT plants (Fig. 6a), suggesting that auxin was the underlying cause of the more robust root system in the CmANR1-transgenic plants.

Fig. 6: CmANR1 binds directly to the CmPIN2 promoter.
figure 6

a Total free IAA content in the roots of 20-day-old WT and CmANR1-transgenic chrysanthemums. b The relative enrichment folds of the fragment containing the CArG-box motif by ChIP-qPCR in the promoters of four auxin-responsive unigenes. c CmANR1 directly binds to the CArG motif presented in pCmPIN2 in vitro, as indicated by EMSA. Competition for CmANR1 binding was performed with 10×, 50×, and 100× unlabeled probes (wt) or 50× and 100× CArG-box-mutated probes (mut). The symbol “+” indicates presence and “−” indicates absence. d The Luc imaging assay indicates that CmANR1 activates the expression of CmPIN2. Representative images of Nicotiana benthamiana leaves 72 h after infiltration are shown. Statistical significance was determined using Student’s t-test. n.s.: p > 0.01; *p < 0.01; **p < 0.001

Full size image

MADS-box TFs bind to specific DNA sequences with an overall consensus of CC(A/T)6GG, called CArG-box motifs67. To further verify whether CmANR1 directly activates the expression of auxin-responsive genes, the promoters of CmPIN2, CmGH3.1, CmTAA1, and CmAB37G were cloned by the Genome-Walking method. Plant cis-acting regulatory DNA elements (PLACE) analysis found that there was one CArG-box motif in the promoters (Supplementary Fig. S5). Subsequently, the ChIP-qPCR assays were carried out using CmANR1-transgenic and WT chrysanthemum. The results demonstrated that the CArG-box motif of CmPIN2 was significantly recruited by CmANR1, while others were not (Fig. 6b). The results provided in vivo evidence for the binding of CmANR1 to the CmPIN2 promoter on the CArG-box motif site.

To further validate the binding of CmANR1 to the CArG-box recognition site in the CmPIN2 promoter in vitro, EMSA analysis was conducted with an oligo-probe containing the CArG-box motif using the purified CmANR1-His fusion protein. The result showed that specific DNA-CmANR1 protein complexes were detected when the CArG-box motif-containing sequence was used as the labeled oligo-probe. When increasing the amounts of the unlabeled competitive probe with the same sequence, we found that the binding complexes were reduced. However, the competition was not existed in the mutated version. The specificity of the competition verified the physical interaction between the CmANR1 protein and CmPIN2 promoter, which required the specific CArG-box cis-element (Fig. 6c).

To examine whether CmANR1 directly activates CmPIN2, an in vivo firefly Luc imaging assay was performed. Constructs containing 35Spro::CmANR1 (pGreenII 62-SK) and the CmPIN2pro::Luc (pGreenII 0800-Luc), as well as Luc (pGreenII 0800-Luc) and 35Spro (pGreenII 62-SK) were respectively co-infiltrated into tobacco leaves to express these fusion proteins transiently. The results demonstrated that a strong luminescence signal was detected in the 35Spro::CmANR1/CmPIN2pro::Luc coexpression region, but no or very weak luminescence signal was detected in the negative controls (Fig. 6d). These results indicate that CmANR1 directly activates CmPIN2 transcription.

Taken together, our results suggest that CmANR1 activates the transcription of the CmPIN2 gene by direct binding to the CArG-box motif in its promoter.

Discussion

MADS-box TF genes have been extensively identified as the important regulators of flowering time, floral organ identity, and flower development68,69,70,71,72. In contrast, fewer MADS-box TF genes have been reported to regulate root development27,55,56. With regards to the molecular regulatory mechanisms on root development, MADS-box TF genes are seldom an immediate regulator. In fact, these MADS-box genes indispensably exert their effects on root development through cross-talks with other signals, such as the cross-talk between OsMADS25 and the nitrate signal54, and GmNMHC5 with the sucrose signal in controlling root development55. Remarkably, the functional mechanism of a large part of MADS-box genes on root development is mainly auxin dependent, such as ANR1, AGL21, and XAL2/AGL1427,52,53. Interestingly, our previous study on CmANR1 demonstrated that a nitrate signaling pathway as well as auxin-related processes interacted under the integration of the CmANR1 gene, giving rise to the proliferation of LR growth in Arabidopsis56.

A great deal of evidence suggests that auxin is central to both LR development9,13 and AR formation11,12. In this study, the free IAA content was highly elevated in the roots of CmANR1-OVXs plants compared to the WT control (Fig. 6a), which provided a reasonable explanation for the better developed ARs and LRs in CmANR1-transgenic chrysanthemums. It is well known that PINs and AUX/LAXs are two dominating groups of auxin efflux/influx carrier genes that are essential for auxin polar transport in plants32. Moreover, PIN genes have been reported to provide rate-limiting functions in auxin movement29,34. In our study, the expressions of the auxin transport genes PIN2, PIN3, and AUX1 were significantly up-regulated in the roots of the CmANR1-transgenic plants compared to the WT control (Fig. 5a). Although auxin is also synthesized in the root tips, the several included auxin synthesis unigenes (e.g., YUC1, AAO-Like, AAO2, NIT2-Like, and NIT4) in the RNA-Seq results showed no obvious expressional differences in the roots between WT and CmANR1-OVXs. Given the role of polarity in auxin transport, the increased auxin in the roots of the CmANR1-transgenic plants was more likely due to shoot-to-root auxin transport under the mediation of auxin transport carrier genes, but not as a direct result of auxin biosynthesis in the roots. Furthermore, CmANR1 could directly activate CmPIN2 transcription (Fig. 6), confirming the possibility of root-ward auxin transport at least partially by CmPIN2 in CmANR1-transgenic chrysanthemum.

In addition, calcium, as an important signaling messenger, was found to be related with the regulation of AR or LR development via cross-talks with auxin. Ca2+ and CDPK act as the downstream messengers during auxin-induced AR formation in cucumber16. AtCIPK6 significantly affects LR formation by positively regulating root basipetal and shoot-to-root auxin transport73. OsCBL1 is required for LR development in rice by mediation of auxin biosynthesis17. In line with previous studies, a multitude of Ca2+-signaling-related genes, including CML6, CML45, CML50, CIPK6, CDPK10, and CDPK33, as well as PBP1, were significantly up-regulated in the roots of CmANR1-overexpressing chrysanthemum compared to WT plants (Figs. 4b, 5b), which indicated possible cross-talk of calcium signaling with auxin-related physiological processes in regulation of root development under the mediation of CmANR1. Moreover, the complex interaction of auxin with ethylene in root development has been well documented21,22,74. Auxin can stimulate ethylene biosynthesis by increasing ACS transcription75 and positively regulating the ethylene-mediated inhibition of root growth76. Ethylene has a positive role in the regulation of auxin synthesis and promotes basipetal auxin polar transport in the roots, resulting in increased auxin as well as a subsequent triggering of ethylene-mediated root growth inhibition77,78,79. Notably, in a present study, the expression levels of numerous ethylene-related unigenes were markedly varied in the roots of CmANR1-transgenic plants compared with the WT plants (Fig. 4c), such as ACS7, ERF13, and ERF109. However, the major components in the ethylene signaling pathway, including ETR1, ETR2, EIN2, EIN3, and EIL1, showed no obvious expressional differences between the WT and CmANR1-OVX, suggesting the absence of the ethylene signaling pathway in this context. There seems to be a paradox between the ethylene signaling and RNA-Seq results here. A possible explanation is that the ERFs (ERF13 and ERF109) have other functions, or otherwise are not the main determinants involved in the ethylene signaling pathway. In fact, ERF109, which is highly jasmonic acid (JA)-responsive, plays an important role in mediating the connection of JA signaling with auxin biosynthesis during LR formation in Arabidopsis80. Therefore, ethylene biosynthesis rather than signaling may take part in regulating root development in chrysanthemum. Further researches are needed to prove this hypothesis.

In our transcriptome sequencing results, thousands of unigenes were significantly differentially expressed in the root samples of CmANR1-OVXs compared to the WT control (Fig. 3). The annotations on these DEGs were highly informative. In addition to the anticipated DEGs, there were some unexpected results that required further assessment. For example, the “Plant-pathogen interaction” term accounted for the most abundant group in the KEGG enrichment analysis on the up-regulated DEGs (Fig. 3e). In terms of pathogen defense in plants, the phytohormone JA naturally becomes a primary consideration. As a biotic and abiotic stress-related hormone, JA is essential for immunity and development in plants. In addition, methyl JA was proved to repress root growth in Arabidopsis81. Moreover, a previous study on AR formation in Arabidopsis reported that the auxin-inducible Gretchen Hagen3 (GH3) genes, GH3.3, GH3.5, and GH3.6, could lower JA content in the roots by down-regulating JA biosynthesis and enhancing JA conjugation. GH3 genes were found to fine-tune AR formation by a combination of auxin and JA regulatory pathways11. Inspired by the above studies, we discovered that the auxin-responsive GH3 unigenes (GH3.1, GH3.5, and GH3.6) were significantly up-regulated in the roots of CmANR1-OVXs plants compared to the WT control. Conversely, three JA-amido synthetase (JAR1) unigenes were down-regulated in our RNA-Seq results. Therefore, based on this observation, JA signaling may possibly be involved in AR development by cross-talks with auxin under the integration of CmANR1 in chrysanthemum.

In conclusion, we have provided a summary of a working model of the MADS-box TF gene CmANR1 on the regulation of both AR and LR development in chrysanthemum (Fig. 7). When there was a higher concentration of nitrate in the surviving environment of chrysanthemum roots, NRT1.1, as the nitrate sensor on the root cell plasma membrane, sensed and transferred the external nitrate signal to ANR1 in the nucleus82,83. Then, the nuclear-localized ANR1 was rapidly enriched in response to the nitrate signal, forming more ANR1/ANR1 homodimers56. Subsequently, the TF gene ANR1 up-regulated the expressions of several auxin polar transport genes, such as PINs and AUX1, thereby facilitating root-ward auxin transport. Alternatively, in our previous study, local auxin biosynthesis was elevated concurrently56. The resulting increased auxin in the roots promoted both AR and LR development in chrysanthemum. In contrast, the increased auxin in the roots may feedback-regulate the expression of a member of the same clade, AGL21, which was shown to interact with CmANR156, and AGL21 exerted similar effects on LR development as CmANR127,56. The formation of ANR1/AGL21 heterodimers may then collectively regulate LR development in chrysanthemum. Finally, a robust root system developed in parallel with thriving shoot development and growth in chrysanthemum (Supplementary Table 2).

Fig. 7
figure 7

The working model of CmANR1 in root development in chrysanthemum: CmANR1 promotes both AR and LR development by transcriptional activation of CmPIN2 in the roots

Full size image