Molecular Characterization of Cytokinin Response Regulator ARR21 in Arabidopsis Seed Development

ABSTRACT

The role of cytokinin among plant hormones in seed development remains largely unknown. The Arabidopsis response regulator 21 (ARR21) is one of the cytokinin response regulators and a Type‐B ARR with a nuclear localization signal and a GARP motif similar to the MYB‐like DNA‐binding domain. ARR21‐sGreen fluorescent protein (GFP) signals were localized within the nucleus, and ARR21 showed the highest expression levels in developing seeds. In addition, histochemical analysis revealed ARR21 expression in the silique coats, chalazal seed coat, chalazal endosperm, and throughout the developing seed at 6 days after pollination. Two independent mutants were generated using the CRISPR/Cas9 system: arr21‐3 (51 bp in‐frame deletion) and arr21‐4 (2 bp insertion). The seed size and weight of the arr21 mutants decreased by an average of 10.7% and 37%, respectively, compared to the wild‐type (WT). In arr21 mutants, the cotyledon length of embryos and the size of seed coat cells were reduced. Seed‐specific overexpression of ARR21 in arr21‐4 restored the seed length to WT levels. This study suggests that ARR21 regulates seed size by functioning in the chalazal endosperm and embryo, thereby providing insights into the role of cytokinin in seed development.

Abbreviations

AHK

Arabidopsis histidine kinase

AHP

Arabidopsis histidine phosphotransfer protein

ARR

Arabidopsis response regulator

CDS

coding DNA sequence

CKX2

cytokinin oxidase 2

DAP

day after pollination

DsRED

discosoma red fluorescent protein

DW

distilled water

eIF4a

eukaryotic initiation factor 4

FAME

fatty acid methyl ester

GFP

green fluorescent protein

LEC2

LEAFY COTYLEDON 2

MAPKs

mitogen‐activated protein kinases

PPT

phosphinothricin

RT‐qPCR

reverse transcription ‐quantitative polymerase chain reaction

sgRNAs

single‐guide RNAs

TAG

triacylglycerol

TFL1

TERMINAL FLOWER 1

WT

wild‐type

1. Introduction

Seeds are essential for the propagation of plant descendants (Fait et al. 2006). Plant seeds can be used to benefit humans in many ways; rice, wheat, and soybeans, for example, are plant seeds that have become staple foods (Birla et al. 2017; Khalid et al. 2023; Tripathi and Khare 2016). As the climate crisis worsens, the importance of ensuring plant seed production is sustainable into the future is becoming increasingly apparent. Plant seeds are regarded as a resource for addressing food crises and as an environmentally friendly source of biofuels (Mafra et al. 2024; Muthulakshmi et al. 2021; A. K. Tiwari et al. 2023). Therefore, it is necessary to ensure a deeper understanding of seed development processes.

The mature seed structure of Arabidopsis ( Arabidopsis thaliana ) consists of an embryo, an endosperm, and a seed coat. Each component is regulated by the expression of various genes during seed development (Chaudhury et al. 2001; Haughn and Chaudhury 2005; Le et al. 2010). The chalazal endosperm, which is established during cellularization in the endosperm development process (Olsen 2004), is closely linked to the funiculus, which connects the seed to the ovule of the maternal plant (Schneitz et al. 1995; Xu et al. 2016). The chalazal endosperm provides maternal nutrients to developing seeds (Nguyen et al. 2000). Sucrose, which is essential for embryo and seed development, moves to the central endosperm via the chalazal endosperm. It then passes through the seed coat and reaches the embryo, which is surrounded by the endosperm region (Morley‐Smith et al. 2008; Robert 2019). The chalazal endosperm also influences the timing of endosperm cellularization. For example, TERMINAL FLOWER 1 (TFL1), which is expressed in the chalazal endosperm, moves to the syncytial peripheral endosperm. The seed sizes of tfl1‐1 and tfl1‐20 mutants were larger than those of the wild type (WT). In addition, the tfl1‐20 mutant showed delayed endosperm cellularization. TFL1 interacts with the endosperm‐preferred bZIP transcription factor ABSCISIC ACID INSENSITIVE 5 (ABI5), which regulates the timing of endosperm cellularization (Zhang et al. 2020). Mitogen‐activated protein kinases (MAPKs) are involved in diverse signal transduction pathways, and the expression of MAP kinase 10 (MPK10) was highest in the chalazal endosperm. MPK10 inhibits the transcriptional activity of WRKY10 and regulates endosperm development (Xi et al. 2021). This suggests that chalazal endosperms affect seed size and growth by consuming nutrients and regulating endosperm cellularization during early seed development. Ultimately, the endosperm is absorbed by the growing embryo once it completes its cellularization. In mature seeds, only the layer of endosperm surrounding the embryo remains intact.

Cytokinin is a plant hormone with a variety of functions in plants. It regulates organ formation in the shoots, flowers, and roots (Wybouw and De Rybel 2019). It also plays a role in cell division, proliferation, and expansion (Park et al. 2021; Schaller et al. 2014). Cytokinin in the seeds can regulate seed size and affect seed yield. For example, the cytokinin oxidase 2 (CKX2) gene encodes a protein that degrades active cytokinin and has been identified as a direct target gene of the IKU pathway, a major pathway that regulates seed size (Li et al. 2013; Luo et al. 2005). The reduced seed size was recovered by CKX2 overexpression, indicating that CKX2 positively regulates seed size (Li et al. 2013). In another crop, oilseed rape, ckx3,5 plants showed an increase in flower number, along with a larger inflorescence meristem diameter than the WT. Additionally, total seed weight was 20% higher in the ckx3,5 mutant than in the WT. The total number of seeds was also greater than that in the WT. However, the number of seeds per pod remains unchanged (Schwarz et al. 2020).

The cytokinin signaling system, known as the “hybrid two‐component system,” involves the Arabidopsis histidine kinase (AHK), the Arabidopsis histidine phosphotransfer protein (AHP), and the Arabidopsis response regulator (ARR). Each component is biochemically linked through His‐to‐Asp phosphorylation. Cytokinin signaling occurs in the following sequence: AHK → AHP → ARR. Arabidopsis has 3 AHKs, 5 AHPs, and 23 ARRs (Aoyama and Oka 2003; Hwang and Sheen 2001; Lohrmann and Harter 2002; Stock et al. 2000; To and Kieber 2008). ARRs are divided into three types: Type‐A, Type‐B, and Type‐C. Type‐B ARRs are transcription factors (Mason et al. 2005). Type‐B ARRs, such as ARR1, ARR10, and ARR12, have been extensively studied in various contexts. Dry seeds of arr1/10/12 mutants were larger than those of the WT, and triple mutants showed more severe growth defects than double mutants in Arabidopsis. Additionally, the silique length was reduced, and a cytokinin‐insensitive phenotype was observed (Argyros et al. 2008).

Arabidopsis response regulator 21 (ARR21) is a member of the Type‐B ARRs (Hwang et al. 2002; Riechmann et al. 2000). GUS staining showed that ARR21 is expressed in the chalazal endosperm and ARR21 was an endosperm‐preferred gene (Day et al. 2008; S. Tiwari et al. 2006). Previous studies reported that the phenotypes of the arr21‐1 dSpm transposon insertion mutant and the arr21‐2 (SALK_005772) T‐DNA insertion mutant were nearly indistinguishable from the WT (Horák et al. 2003; Hill et al. 2013).

Reverse transcription polymerase chain reaction (RT‐PCR) analysis of the ARR21 C‐terminal domain overexpression transcript revealed the upregulation of not only Type‐A ARRs but also the cytokinin biosynthesis gene isopentenyltransferase 4 (IPT4) and the auxin‐responsive genes indole‐3‐acetic acid inducible 4 (IAA4) and IAA5 (Tajima et al. 2004). ARR21 expression increases in tobacco leaves when the transcription factor LEC2, induced by the senescence promoter, is overexpressed. Only ARR21 overexpression in tobacco leaves increases triacylglycerol (TAG) levels. ARR21 has been identified as a transcription factor that is indirectly regulated by LEC2 (H. U. Kim et al. 2015; I. Kim et al. 2024).

However, the role of ARR21 in seed development has not yet been investigated. To characterize the molecular function of ARR21, two mutants, arr21‐3 and arr21‐4, were generated using the CRISPR/Cas9 system. The size of dry seeds in the arr21 mutants was significantly smaller than that in the WT. Furthermore, the seed length in the complementation lines was longer than that in arr21‐4 and was restored to levels similar to that in the WT. These results demonstrate that ARR21, which is highly expressed in the chalazal endosperm and embryo of developing seeds, regulates seed length. This study provides insights into the role of Type‐B transcription factor ARR21 and the relationship between cytokinin signaling and seed development.

2. Materials and Methods

2.1. Plant Growth Conditions and Transformation Method

This study used A. thaliana (L.) ecotype Columbia 0 (Col‐0) as a control. The dry seeds were sterilized with 70% ethanol for 2 min and 0.5% NaOCl for 5 min. Subsequently, 1 mL of distilled water (DW) was added, and the washing process was repeated at least six times. The seeds were then stratified for 3 days in the dark at 4°C. After cold treatment, the seeds were plated on 1/2 Murashige and Skoog (MS) medium, consisting of 2.2g MS medium, 1% sucrose, and 7 g agar per 1 L. The 1/2 MS medium was kept in a culture chamber at 23°C under 16 h of light (100 μmol·m⁻² s⁻¹) and 8 h of darkness for 10–12 days. Next, the seedlings were transferred to soil and grown in a growth chamber at 23°C with 16 h of light (100 μmol·m⁻² s⁻¹) and 8 h of darkness. All mutant and transgenic plants used in this study were grown under the same conditions as the WT plants, following the selection of homozygous lines. All vectors constructed in the experiments described below were transformed into Agrobacterium GV3101 using the freeze–thaw method (Weigel and Glazebrook 2006). Transformation into WT plants was performed using the Agrobacterium floral dip method (Clough and Bent 1998). All primers used for vector and gene expression analyses are listed in Table S1.

2.2. Phylogenetic Tree of Type B ARRs

The protein sequences of Type‐B ARRs were obtained from The Arabidopsis Information Resource (https://www.arabidopsis.org/). A phylogenetic tree was generated using the Maximum Likelihood method using Molecular Evolutionary Genetics Analysis (MEGA) Version 11.0.13.

2.3. Subcellular Localization

cDNA was synthesized from RNA extracted from the developing seeds of Arabidopsis WT. cDNA synthesis was performed using the Prime Script II 1st Strand cDNA Synthesis Kit (Takara, Japan). The 1863 base pair (bp) ARR21 (AT5G07210) coding DNA sequence (CDS), excluding the stop codon, was amplified using primers with XmaI restriction enzyme sites at the forward and reverse primer ends. The amplified product was ligated into a pGEM‐T Easy Vector (Promega, USA) and transformed into Escherichia coli . The vector was sequenced to verify accurate cloning of the ARR21 gene sequence. The confirmed pGEMT‐ARR21 vector was digested with XmaI and cloned into the 326‐sGreen fluorescent protein (GFP) vector. Plasmid DNA was extracted using a Plasmid DNA Extraction Midi Kit (FAVORGEN, Taiwan). The tobacco plants were grown for 4 weeks in a growth chamber at 23 °C, with a light intensity of 100 μmol·m⁻² s⁻¹ under a 16‐h light/8‐h dark cycle. The protoplasts were isolated from the fourth and fifth leaves, excluding the cotyledons. The vectors, such as 326‐sGFP and 326‐sGFP‐ARR21, were transfected into the protoplasts at a concentration of 10 μg (Yoo et al. 2007). After 20 h of incubation at 25°C, the cells were observed under an inverted fluorescence microscope (Eclipse Ts2R‐FL, Nikon, Japan). The nuclei were stained with DAPI and incubated at 37°C for 3 h.

2.4. GUS Staining

The 1859 bp sequence from The Arabidopsis Information Resource, predicted to include the ARR21 promoter site, was amplified through PCR, targeting the ARR21 5′ UTR. The ARR21 promoter was cloned into the pDONR vector via a BP reaction using the Gateway system. The verified ARR21 promoter was cloned from the pDONR vector by the LR reaction. This vector was then used to transform WT plants. T₁ seeds showing resistance to carbenicillin (Carb) and hygromycin (Hyg) were harvested. Before progressing to the T₂ generation, only the seeds grown on 1/2 MS medium containing Hyg, showing a 3:1 segregation ratio, were transferred to the soil. In the T₃ generation, only seeds that survived on 1/2 MS medium containing Hyg were transferred to soil. Plants from the T₃ generation were used for the GUS staining assay. The GUS staining solution consisted of 1 mg·ml⁻¹ X‐Gluc, 100 mM sodium phosphate (pH 7), 10 mM EDTA, 0.1% Triton X‐100, 2 mM potassium ferricyanide, 2 mM potassium ferrocyanide, and 100 μg·ml⁻¹ chloramphenicol. The remaining volume was then filled with DW. This solution was prepared immediately before the experiment. Each sample was treated with 1 mL of GUS staining solution and incubated at 37°C overnight in the dark. The following day, the samples were washed five times with 1 mL of 70% ethanol. Seedlings, rosette leaves, flowers, and siliques were observed and photographed using a stereomicroscope (SMZ745T, Nikon, Japan). Developing seeds and embryos were observed and photographed using an HK6E3 light microscope (Nikon, Japan).

2.5. CRISPR/Cas9 Vector Cloning for arr21 Knock‐Out

To create arr21 mutants, two single‐guide RNAs (sgRNAs) targeting the third exon of ARR21 gDNA were designed using the RGEN tool (Bae et al. 2014; Park et al. 2015). The two designed sgRNAs were amplified using KOD polymerase with a BsaI site and pCBC‐DT1T2 as the template. The sgRNAs were ligated into pHEE401 using the Golden Gate system. This vector was then used to transform WT plants to generate mutations. Sequencing was performed to confirm accurate cloning of the target guide RNAs. T₁ seeds showing discosoma red fluorescent protein (DsRED) fluorescence were harvested, and sequencing PCR was performed on the rosette leaf gDNA. The primers for PCR sequencing were designed to include both sgRNAs. The deletion site was confirmed using Sanger sequencing. T₁ lines showing loss of DsRED advanced to the T₂ generation. In the T₂ generation, gDNA was extracted from rosette leaves, and Sanger sequencing was performed. Mutants with insertions or deletions were advanced to the T₃ generation. Two independent mutants were selected and named arr21‐3 and arr21‐4. All experiments used T₃ or T₄ generations.

2.6. Generation and Selection of Phaseolin Promoter‐Mediated Complement Lines

A pPhas‐gateway vector with a seed‐specific phaseolin promoter was used for complementation experiments. The verified pENTR‐ARR21 CDS was cloned into the pPhas‐Gateway vector via LR reaction. This vector was transformed into E. coli . Vector transformation of arr21‐4 plants was performed as a complementation experiment. T₁ seeds were grown on 1/2 MS medium containing Carb and phosphinothricin (PPT), and plants showing antibiotic resistance were transferred into soil. T₁ plants were genotyped using the attB1 and attB2 primers. In the T₂ generation, seeds were selected in a 3:1 ratio on 1/2 MS medium containing PPT. In the T₃ generation, the homozygous lines that survived on 1/2 MS medium containing PPT were selected. Three homozygous lines were obtained and referred to as arr21‐4/Phas1, arr21‐4/Phas2, and arr21‐4/Phas3.

2.7. Gene Expression Analysis by RT‐qPCR and RNA‐Seq

WT plants used to confirm tissue‐specific expression of ARR21 were grown on 1/2 MS medium for up to 12 days, and were then transplanted into the soil. After transplantation into the soil, samples were collected at 28 days old from the leaves, flowers, stems, roots, silique coats, and developing seeds. All fully open flowers at stage 14 were sampled (Alvarez‐Buylla et al. 2010). Developing seeds from all the siliques attached to the main stem were sampled by removing the silique coats.

RNA was extracted from vegetative tissue and developing seed using the DNA‐free RNA isolation protocols described in Onate‐Sanchez and Vicente‐Carbajosa (2008). cDNA was synthesized from 1000 ng of RNA. cDNA synthesis was carried out with the Prime Script II 1st Strand cDNA Synthesis Kit. The eukaryotic initiation factor 4A (eIF4a), which is expressed in all tissues, was used as a reference gene. cDNA (1000 ng) was diluted to 7 ng for reverse transcription‐quantitative polymerase chain reaction (RT‐qPCR), performed using the TB Green Premix EX Taq II (Takara, Japan).

Developing seeds were collected from WT and arr21‐4 siliques at the S6 stage, which is 11 days after pollination (DAP) (I. Kim et al. 2024). Total RNA was extracted as described above, and RNA‐seq libraries were prepared using the Illumina TruSeq RNA Sample Preparation Kit. Sequencing was performed on the Illumina NovaSeq platform with 101 bp paired‐end reads. Low‐quality reads and adapters were removed, and the cleaned reads were mapped to the A. thaliana reference genome (TAIR10). Differential gene expression was analyzed using the DESeq package (Love et al. 2014). Transcript abundance was quantified for all annotated Arabidopsis genes, and the complete expression dataset is provided in Table S2. A curated summary of cytokinin‐related genes, including Type‐A and Type‐B ARRs, is presented in Table S3.

2.8. Phenotypic Analysis of Plant and Seed Morphology

The number of rosette leaves was counted at the time of the first flowering. Plant height was measured 51 days after transplantation. The number of siliques per plant was at 48 days after transfer to soil. To compare the silique lengths of the WT and arr21 mutants, 10 siliques were sampled from the main stem at 9 DAP and photographed. The lengths were measured using ImageJ software (http://imagej.nih.gov/ij/), and the average length of 10 siliques was calculated for four replicates. Seeds per silique were determined by counting seeds harvested from mature siliques.

Mature seeds were thoroughly dried and observed under a stereomicroscope (SMZ745T, Nikon, Japan), with 30 seeds per sample and five replicates. This experiment was performed with at least three independent replicates. The width was measured at the smallest point, and the length was measured at the largest point. The width and length were analyzed in images using ImageJ software. The dry seed phenotypes of the arr21‐4/Phas lines were assessed using T₃ seeds, and the width and length of 150 seeds were plotted. The size was defined as the product of the width and length of all seed images.

2.9. Scanning Electron Microscopy

The T₃ Arabidopsis seeds were coated with Pt to a thickness of 20 nm. These were observed using field‐emission scanning electron microscopy (FESEM). The perimeter of the seed coat cells was measured by averaging the data from 10 cells in five different seeds. Seed coat cells was determined by counting cells in the corresponding area of five seeds.

2.10. Visualization of Mucilage

Ruthenium red mucilage staining was performed as described previously (Voiniciuc et al. 2015). After removing the staining solution, DW was added again without any physical shock, and the samples were observed under an HK6E3 light microscope (Nikon, Japan). Seed area and mucilage‐containing seed area were measured using HK Basic. Mucilage area was defined as the difference between the mucilage‐containing seed area and the total seed area. The data were normalized by dividing the values by the WT average (set to 1). The relative values of arr21‐3 and arr21‐4 were then calculated by dividing by the WT average and are presented in Figure S2.

2.11. Fatty Acid Analysis

Lipid extraction was performed to quantify the total lipid content of Arabidopsis seeds. The weight of 100 seeds was measured. 500 μL of toluene and 500 μL of lipid extraction solution containing 15:0 were added to the seeds, and the mixture was heated to 85°C. The mixture was boiled for 2 h and then cooled to 22°C. After cooling, 1 mL of 0.9% NaCl and 1 mL of hexane were added, and the mixture was shaken before being centrifuged. The upper phase was separated, and the process was repeated twice by adding 1 mL of hexane, shaking, and centrifugation. Three milliliters of hexane were purged with nitrogen gas. Then, 200 μL of hexane was added, and the solution was transferred into the insert of a GC vial. The analyses were performed using a GC‐2030 instrument (Shimadzu, Japan). A DB‐23 column (30 m × 0.25 mm, 0.25 μm film, Agilent, USA) was used.

2.12. Embryo Dissection

The dried seeds were soaked in DW for approximately 3–4 h. The seed coat and embryo were separated by gently pressing them onto a glass slide. The differential interference contrast mode of an Eclipse Ts2R‐FL inverted fluorescence microscope (Nikon, Japan) was used to image intact embryos. The embryos were placed on glass slides and decolorized using Plant Biology reagent (Visikol, USA). Observations were performed using an HK6E3 optical microscope (Nikon, Japan).

2.13. Statistical Analysis

All data presented were obtained from at least three biological replicates. Statistical analyses were performed using one‐way analysis of variance (ANOVA) in GraphPad Prism 9. The smallest value was designated as “a,” and statistically significant differences were indicated by different letters (b, c, d). Data sharing in the same letter indicate no statistically significant differences between them.

3. Results

3.1. ARR21 Is a Transcription Factor Associated With the Cytokinin Signaling Pathway

Arabidopsis response regulator 21 (ARR21) is a Type‐B ARRs (Hwang et al. 2002; Riechmann et al. 2000). Type‐B ARRs are transcription factors that express cytokinin‐responsive genes. Their signaling progresses from AHK to AHP to ARR (Figure 1A). ARR21 is composed of 621 amino acid residues and contains three major domains (Figure 1B). Among these domains, there is a response regulatory domain required for cytokinin signaling and a receiver domain that can be phosphorylated by AHP or other ARR proteins, enabling the expression of cytokinin‐responsive genes (Kakimoto 2003; Stock et al. 2000). Additionally, an MYB‐like DNA‐binding domain, similar to the GARP motif, has been identified (Hosoda et al. 2002). GARP is defined as follows: G: GOLDEN2 in Zea mays , AR: ARR‐B in A. thaliana , and P: Psr1 in Chlamydomonas reinhardtii . The GARP subfamily is a plant‐specific transcription factor that plays diverse roles in nutrient (nitrogen and phosphorus) sensing, chloroplast development, circadian rhythm regulation, and hormone signaling (Safi et al. 2017). ARR21 contains a nuclear localization signal domain. To confirm that ARR21 functions as a transcription factor, its subcellular localization was determined by GFP fusion analysis. The GFP signal of 35S:ARR21 CDS‐sGFP was observed in the nuclear region of tobacco protoplasts stained with DAPI (Figure 1C). There are 23 known cytokinin ARRs (Schaller et al. 2008). A phylogenetic tree was constructed to examine the homology between the protein sequences using the maximum likelihood method (Figure 1D). ARR21 is closely related to the Type‐B ARR13.

FIGURE 1.

Cytokinin responsive transcription factor ARR21. (A) Schematic structure of the ARR21 protein. NLS, nuclear localization signal; GARP, G represents GOLDEN2 in Zea mays , AR represents ARR‐B in Arabidopsis thaliana , and P represents Psr1 in Chlamydomonas reinhardtii . (B) Green fluorescence of Nicotiana benthamiana protoplast expressing 35S:ARR21 CDS‐sGFP is observed in the nucleus stained with DAPI. Scale bar = 20 μm. (C) Phylogenetic tree of Arabidopsis Type A, B, and C ARRs. ARR21 is marked in a box. (D) Cytokinin signaling pathway. AHK, Arabidopsis histidine kinase; AHP, Arabidopsis histidine protein; D, aspartate, receiver; P, phosphate group. The green arrows indicate phosphorylation, whereas the red arrows represent the translocation of AHP from the cytoplasm to the nucleus.

3.2. ARR21 Expression Is Highest in Developing Seeds

RT‐qPCR was performed to assess the expression of ARR21 in different Arabidopsis tissues. When the expression of ARR21 in the rosette leaves was set to 1, it was up to 88‐fold higher in the developing seeds (Figure 2A). Histochemical localization of ARR21 was confirmed by GUS staining of transgenic ARR21promoter:GUS plants. The shoot apical meristem, 7‐day‐old seedlings (Figure 2B), and rosette leaves were stained (Figure 2C). Additionally, GUS staining was observed in the petals, pistils, and receptacles of flowers (Figure 2D) and silique coats (Figure 2E). Furthermore, the chalazal seed coats of developing seeds at 3 and 4 DAP were stained (Figure 2F,G). At 6 DAP, the expression of ARR21 was observed throughout developing seeds (Figure 2H). During embryonic development, GUS staining was first detected in the cotyledons at the torpedo stage (Figure 2I). At the bent torpedo stage, staining was more intense in both the cotyledons and near the root apical meristem (Figure 2J). GUS staining was also observed in the mature embryos (Figure 2K). These results suggest that ARR21 is expressed in the chalazal seed coat, chalazal endosperm, and embryo  during the early to mid‐stages of seed development.

FIGURE 2.

Tissue‐specific and histochemical expression profile of the ARR21 gene. (A) Expression analysis of ARR21 using RT‐qPCR. All tissue samples were from 28‐day‐old wild‐type plants after transferring to the soil. DS, developing seed; FL, flower; LE, rosette leaf; RO, root; SC, silique coat; ST, stem. Error bars indicate the SD, n = 3. (B–K) Histochemical localization in Arabidopsis transformed with pMDC163‐ARR21 promoter: GUS. (B) 7‐day‐old seedlings grown in 1/2 MS medium. (C) Rosette leaf. (D) Flower. (E) Siliques. (B,D) Scale bar = 1 cm. (C,E) Scale bar = 2 cm. (F) Developing seed at 3 DAP, (G) 4 DAP, and (H) 6 DAP. (I) Late torpedo. (J) Bent torpedo. (K) Mature embryo. (F–H) Scale bar = 50 μm. (I–K) Scale bar = 100 μm.

3.3. arr21 Mutants Were Generated by CRISPR/Cas9, and Their Seed Size Is Smaller Than That of the WT

To determine the function of ARR21, arr21 mutants were constructed using CRISPR/Cas9. Two sgRNAs targeting the exon3 of ARR21 were designed to induce functional knockout mutations. The vector containing the cloned sgRNA was transformed into the WT. Sixteen T₁ generation lines were selected for the analysis. These lines were advanced to the T₂ generation, where seeds lacking DsRED were identified, and further propagated to the T₃ generation. ARR21‐specific primers were designed to assess indel variation in the ARR21 gene (Figure 3A). Indel mutant candidates were analyzed via Sanger sequencing, and two mutants with indel sequences distinct from those of the WT were identified. The arr21‐3 mutant contained a 51 bp deletion, whereas the arr21‐4 mutant contained a 2 bp insertion in coding region of third exon (Figure 3B). When the protein sequences of arr21‐3 and arr21‐4 were predicted, arr21‐3 showed a 17 – amino acid in‐frame deletion, whereas arr21‐4 was prematurely terminated (Figure 3C,D).

FIGURE 3.

CRISPR/Cas9‐mediated generation and analysis of arr21 mutants. (A) sgRNAs targeting exon3 of ARR21 gDNA were designed. The arrows are forward and reverse primers for sequencing. (B) Sanger sequencing chromatograms. (C) Predictable ARR21 protein sequence. (D) The schematic structure of the predictable ARR21 protein. RT‐qPCR primers used in (E) is depicted on the WT structure. (E) RT‐qPCR analysis targeting a region with sequence variation of WT, arr21‐3, and arr21‐4 at 9 DAP developing seeds. Error bars indicate the SD, n = 5.

RNA was extracted from the developing seeds of Arabidopsis plants at 9 DAP to determine how the expression of ARR21 changes at the transcript level. RT‐qPCR analysis was performed using an ARR21‐specific mismatch primer designed to discriminate between WT and arr21 mutant transcripts (Hayashi et al. 2004). The normal ARR21 transcript was detected in the WT, but not in the arr21‐3 and arr21‐4 mutant lines, indicating a loss of normal transcript expression in these mutants (Figure  3E).

Previous studies have shown that plant growth is inhibited and silique length is reduced in Type‐B ARR triple mutants, arr1/10/12 (Argyros et al. 2008; Reyes‐Olalde et al. 2017). Therefore, differences in phenotype are expected between the WT and the arr21 mutants. First, the phenotypes of the mutants were examined, and no differences were observed in the germination rate (Figure S1A). After transferring the plants to soil, 7‐day‐old and 20‐day‐old plants showed characteristics similar to those of WT plants (Figure S1B,C). No significant differences were observed in the number of rosette leaves during the first flowering stage (Figure S1D). Plant height and number of siliques per plant were similar to those of the WT plants (Figure S1E,F). Silique length and number of seeds per silique also showed no significant variation (Figure S1G–I).

When examining the inside of the 4‐DAP siliques, a reduction in seed size was observed (Figure 4A), prompting an investigation into the dry seed phenotype. The dry seed size of the arr21 mutants was smaller than that of the WT plants (Figure 4B). The average width of the WT was 0.3 mm, whereas the average widths of arr21‐3 and arr21‐4 were 0.28 and 0.26 mm, respectively (Figure 4C). The average length of the WT was 0.48 mm, whereas the average lengths of arr21‐3 and arr21‐4 were 0.46 and 0.45 mm, respectively (Figure 4D). The average seed size of WT was 0.14 mm², whereas the average seed size of arr21‐3 was 0.13 mm², and that of arr21‐4 was 0.12 mm². The seeds of arr21 mutants were shorter in width and length than the WT seeds; thus, the seed size of arr21 mutants was smaller than that of the WT (Figure 4E). The 100‐seed weights of arr21‐3 and arr21‐4 were much lower than those of the WT. The average weight of 100 seeds was 2.2 mg, whereas the average weights of arr21‐3 and arr21‐4 were 1.8 and 1.6 mg. The seed weight of arr21 mutants was reduced by up to 37% (Figure 4F). These results suggested that ARR21 regulates seed size.

FIGURE 4.

Seed size and 100‐seed weight of the arr21 mutants. (A) Macrophotograph of developing seeds at 4‐DAP siliques, Scale bar = 0.5 mm. (B) Image of WT dry seeds and arr21 dry seeds. Scale bar = 0.5 mm. (C) Seed width. (D) Seed length. (E) Seed size was measured by multiplying the seed width and length. (C–E) The experiments were independently replicated three times with similar results. Error bars represent the range from min to max, n = 5. (F) 100‐seed weight. Error bars represent the range from min to max, n = 8. One‐way analysis of variance.

3.4. arr21 Mutants Showed Reduced Embryo and Seed Coat Sizes

As the seed size of the arr21 mutants was reduced, the embryo and seed coat may also differ from those of the WT. Embryos and seed coats of dry seeds from arr21‐3 and arr21‐4 mutants were examined to investigate potential differences. First, intact embryos were observed in dry seeds (Figure 5A–C). WT and arr21 mutants showed similar cotyledon widths in embryos (Figure 5G). In contrast, a reduction in cotyledon length was observed in the arr21 mutants. The average cotyledon length of the WT embryo was 0.38 mm, whereas it was 0.35 mm for arr21‐3 and 0.32 mm for arr21‐4 (Figure 5H). Decolorized embryonic cells were observed under higher magnification (Figure 5D–F). The average cotyledon cell area in the WT embryos was 90.2 mm². In both arr21 mutants, the cotyledon cell area was reduced. The average cotyledon cell area of arr21‐3 was 77.8 mm², and that of arr21‐4 was 79.6 mm² (Figure 5I).

FIGURE 5.

Changes in embryo cotyledon and cell size of arr21 mutants. (A–C) Image of dry seed embryos observed under a phase‐contrast microscope, Scale bar = 200 μm. (D–F) Photograph of cotyledon cells in decolorized dry embryos, Scale bar = 20 μm. (G) Embryo cotyledon width. (H) Embryo cotyledon length. The shorter part of the cotyledon in the dry seed embryo is the width, and the longer part is the length. Error bars represent the range from min to max, n = 50. (I) Embryo cell size. Error bars represent the range from min to max, n = 30. One‐way analysis of variance and Tukey's multiple comparison tests were used for statistical processing.

To further examine defects in the seed coat, dry seeds were photographed using scanning electron microscopy (Figure 6A–F). The average perimeter of the WT seed coat cells was 127.4 μm. The average perimeter of arr21‐3 and arr21‐4 mutants was 101.9 μm and 105.8 μm, respectively, indicating decreases of 20% and 17% (Figure 6G). In contrast, the number of cells per unit area increased in the arr21 mutants. The average number of cells per unit area was 24 in the WT seed coats. The average number of cells per unit area was 52 in the arr21 mutants (Figure 6H). These results indicate that ARR21 may regulate cell proliferation or expansion in the embryo cotyledon and seed coat.

FIGURE 6.

Alterations in seed coat cell perimeter and number of arr21 mutants. (A–H) Analysis of dry seeds by scanning electron microscopy (SEM). (A, D) Surface morphology of dry seeds from WT, (B, E) arr21‐3, and (C, F) arr21‐4. (A–C) Scale bar = 100 μm. (D–F) Scale bar = 20 μm. (G) Cell perimeter of the seed coat. The results were obtained by averaging the cell perimeter values of 10 cells from each of five different seeds. (H) Cell number of the seed coat. Error bars represent the range from Min to Max, n = 5. One‐way analysis of variance and Tukey's multiple comparison tests were used for statistical processing.

Mucilage is a floating substance found in seeds that is crucial for seed germination, seed dispersion, and protection of seeds from harmful pathogens that can potentially cause damage (Tsai et al. 2021). Because the arr21 mutants exhibited seed coat changes, ruthenium red staining was conducted to investigate whether there was a difference in mucilage levels between the WT and arr21 mutants (Figure S2A). Consistent with previous results, reduced seed area was observed in the arr21 mutants relative to the WT (Figure S2B). The results showed that the mucilage area decreased in arr21‐3 and arr21‐4 (Figure S2C). However, when the mucilage area was normalized to the seed area, the value was similar to or slightly smaller than that of the WT (Figure S2D). Therefore, ARR21 did not markedly affect the mucilage levels.

3.5. The Changes in Fatty Acid and TAG Were Not Prominent in the arr21 Mutants

When LEC2, under the control of a senescence‐inducible promoter, was transiently expressed in tobacco leaves, the amount of TAG and the expression of ARR21 increased (H. U. Kim et al. 2015). In a follow‐up study, ARR21 was selected as a candidate downstream transcription factor of LEC2. When ARR21 was transiently expressed in tobacco leaves, TAG levels increased. However, it was later determined that ARR21 was not directly regulated by LEC2 (Kim et al. 2024). Based on this study, gas chromatography was used to examine potential changes to determine whether there were changes in TAG and fatty acid composition in arr21 mutants. The fatty acid composition of arr21 mutants showed no remarkable variation (Figure S3A). In addition, the total fatty acid methyl ester (FAME) content in the arr21 mutants was the same as that in the WT (Figure S3B). However, the total FAME content per seed in arr21 mutants was lower than that in the WT seeds (Figure S3C). This reduction could be explained by the decreased seed size observed in arr21 mutants, which would naturally result in a lower absolute amount of lipids. These results implied that ARR21 did not significantly influence seed lipid metabolism.

3.6. Seed‐Specific ARR21 Expression in arr21‐2 Recovered Seed Length

To examine whether ARR21 affects seed size, a vector containing the ARR21 CDS fused to a seed‐specific phaseolin promoter was agrotransformed into arr21‐4 for complementation (Figure S4A). Genotyping was performed on the rosette leaves of 10 lines in the T₁ generation to confirm the presence of transgenes (Figure S4B). In the selected independent T₃ lines, the length, width, and size of 150 seeds were compared with those of the WT.

In the complementation experiments, arr21‐4/Phas1, arr21‐4/Phas2, and arr21‐4/Phas3 were selected and examined (Figure 7A). The average seed width of arr21‐4/Phas lines was comparable to or slightly larger than that of arr21‐4 (Figure 7B). However, the seed length of arr21‐4/Phas2 and arr21‐3/Phas3 was restored to 0.48mm and 0.46 mm, respectively, which are similar to that of the WT (Figure 7C). The seed size of the complementation lines was larger than that of arr21‐4 and almost recovered to WT levels (Figure 7D). Thus, ARR21 regulates seed size by influencing seed length.

FIGURE 7.

Seed length recovery in arr21‐4/Phas complementation lines. (A) Image of WT and complementation lines dry seeds. Scale bar = 0.5 mm. (B) Seed width. (C) Seed length. (D) Seed size was measured by multiplying the seed width and length. 150 dry T₃ seeds were used in this experiment. Error bars represent the range from Min to Max. One‐way analysis of variance and Tukey's multiple comparison tests were used for statistical processing.

4. Discussion

Plant seeds consist of an embryo, endosperm, and seed coat. Various genes expressed in these components interact to regulate seed development. Seed size is an important characteristic that has been primarily studied among the various seed traits. This study indicated that ARR21, a cytokinin response regulator, influences seed size by regulating seed length.

ARR21 possesses a nuclear localization signal and a GARP motif (MYB‐like binding domain), which are characteristics of Type‐B, a transcription factor within the response regulator family (Gupta and Rashotte 2012) (Figure 1A,B). Subcellular localization of ARR21‐sGFP was observed, and GFP signals overlapping with the nuclear staining solution, DAPI, were detected, supporting the findings of a previous study in which ARR21 was identified as a transcription factor (Figure 1C). RT‐qPCR was conducted to analyze the tissue‐specific expression of ARR21 in various Arabidopsis tissues, revealing that ARR21 was highly expressed in developing seeds (Figure 2A). A previous study reported that its expression was highest during Stages 5 and 6 of seed development (Kim et al. 2024). Furthermore, GUS staining was observed in the chalazal seed coat, chalazal endosperm (Le et al. 2010), and embryo of the developing seeds (Figure 2F–K). These results imply that the transcription factor ARR21 may regulate genes in the chalazal endosperms and embryos of developing seeds. Moreover, GUS staining in the shoot apical meristem, pistil, and receptacle indicated that ARR21 may function in vegetative and floral, similar to Type‐B ARRs (Meng et al. 2017; Zhu et al. 2022) (Figure 2B–E). To study the function of ARR21, the CRISPR/Cas9 system was used to generate arr21 mutants, specifically arr21‐3 and arr21‐4 lines (Figure 3). These lines represent distinct alleles from the previously reported dSpm transposon insertion arr21‐1 and T‐DNA insertion arr21‐2 (Horák et al. 2003; Hill et al. 2013). The arr21‐3 and arr21‐4 mutants did not exhibit any significant phenotypes in terms of plant growth (Figure S1). Specifically, the seed size of arr21‐3 and arr21‐4 mutants was reduced (Figure 4), but there were no differences in silique length or seeds per silique (Figure S1H,I). Unlike our mutants, arr21‐1 showed no change in seed size, suggesting that arr21‐1 may be a knockdown mutant with a transposon inserted into the third intron, allowing trace amounts of normal transcripts during RNA splicing (Horák et al. 2003). In contrast, arr1/10/12 mutants, which are other Type‐B ARRs, exhibited shortened silique length and decreased inflorescence number (Reyes‐Olalde et al. 2017). Therefore, ARR1, ARR10, and ARR12, rather than ARR21, function more crucially during the reproduction stage.

The major phenotype observed in arr21 mutants was a reduction in seed size. Seed length, width, size, and 100‐seed weight were significantly lower in the mutants than in the WT plants (Figure 4). In contrast, in triple ahk2/3/4 and quintuple ahp1/2/3/4/5 mutants, which are presumed to be upstream of ARR21, an increase in embryo and seed size, particularly in seed length, was observed (Hutchison et al. 2006; Riefler et al. 2006). Similarly, Type‐B ARRs, such as the triple arr1/10/12 mutant, increased the embryo size and seed length (Argyros et al. 2008; Ishida et al. 2008; Mason et al. 2004). However, there are no detailed studies on the regulation and function of these genes during seed development. ARR1, ARR10, and ARR12 belong to subfamily I, whereas ARR13 and ARR21 belong to subfamily II of Type B ARRs (Mason et al. 2004). Moreover, there is no evidence suggesting that subfamily II primarily functions through the cytokinin two‐component signaling system (Gupta and Rashotte 2012). Interestingly, the seed length of complementation lines where ARR1 promoter: ARR21 was expressed in the arr1/12 mutant was restored to WT levels (Hill et al. 2013). These results demonstrate that ARR21 is closely associated with seed length. In other words, ARR21, unlike other Type‐B ARRs, is specifically expressed in the chalazal endosperm, embryo, seed coat, and ultimatley the entire developing seeds. This suggests that it may regulate the genes required for supplying nutrients essential for embryo and endosperm growth. Thus, when the ARR21 gene is mutated using the CRISPR/Cas9 system, the seed size of arr21 mutants is smaller than that of the WT. This result may be due to insufficient nutrient supply to the embryo and endosperm, which could also have affected the timing of endosperm cellularization.

As the seed size of arr21 mutants is reduced, the embryos and seed coats were also differentially observed. The length of the embryo cotyledon was shortened in arr21 mutants, and the cell size decreased (Figure 5). The seed coat perimeter of arr21 mutants was markedly diminished compared to that of the WT, and the cell number per area increased (Figure 6). Cytokinin is involved in cell proliferation and expansion (Park et al. 2021; Schaller et al. 2014). However, the specific role of cytokinin in seeds has not yet been determined. ARR21 is likely involved in cytokinin‐related cell proliferation or cell expansion in developing embryos and seed coats. To verify whether ARR21 influenced seed size, ARR21 was overexpressed specifically in the seeds of the arr21‐4. The results showed that, although there was no difference in seed width between the mutant and transgenic lines, the seed lengths of arr21‐4/Phas2 and arr21‐4/Phas3 recovered to WT levels (Figure 7). In conclusion, ARR21 regulates seed size by affecting cell proliferation or expansion in the embryo and seed coat.

RNA‐seq analysis of developing seeds showed that ARR21 expression was nearly absent in the arr21‐4 mutant, consistent with it being a loss‐of‐function allele (Figure S5A,B). This was accompanied by a modest reduction in the expression of several cytokinin‐responsive genes. Notably, Type‐A ARRs such as ARR4, ARR8, and ARR16, which act as negative regulators and are transcriptionally induced by Type‐B ARRs (Tajima et al. 2004), were consistently downregulated (Figure S5B and Table S3). These results support the conclusion that ARR21, a Type‐B ARR, contributes to the transcriptional activation of cytokinin signaling components during seed development (Argyros et al. 2008).

In summary, the unknown transcription factor ARR21 expressed in developing seeds performs functions distinct from those of other Type‐B ARRs during seed development (Figure 8). This study suggests that ARR21 plays a role in seed development, particularly in the chalazal endosperm and embryo, and may regulate cell proliferation or expansion associated with seed length.

FIGURE 8.

ARR21 may regulate seed size in the developing seeds. In arr21 mutants, the cotyledon length was reduced, the dry seed size decreased, and the perimeter of seed coat cells was smaller than in WT. Therefore, ARR21 may influence seed size. ARR21 may function through cytokinin signaling in response to AHK and AHP, although the exact mechanism has not been fully elucidated. ARR21 is expressed in the chalazal seed coat, chalazal endosperm, and embryo during seed development, suggesting that ARR21 plays distinct roles in these tissues. Additionally, it may also influence seed coat cell development.

Author Contributions

Sujeong Jeong: data curation, formal analysis, investigation, methodology, visualization, writing – original draft, writing – review and editing. Inyoung Kim: writing – review and editing. Hyun Uk Kim: conceptualization, resources, funding acquisition, project administration, supervision, writing – original draft, writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Peer Review

The peer review history for this article is available in the Supporting Information for this article.

Supporting information

Data S1 Peer review.

Figure S1 Plant phenotype characteristics of arr21 mutant. (A) Germination rate of arr21 mutants. Values are means ± SDs, n = 3. (B) The seedling stage of WT and arr21 mutants at 7 days old after transfer to soil. (C) The vegetative stage of WT and arr21 mutants at 20 days old after transfer to soil. (D) Rosette leaf number at first flowering (n = 7). (E) Plant height at 51 days old after transfer to soil (n = 9). (F) Siliques per plant: The number of siliques per plant was counted in 48‐day‐old plants after transferring to soil (n = 4). (G) Macrophotograph of siliques at 9 days after pollination (DAP). Scale bar = 2 cm. (H) Silique length: average length of 10 siliques on the main stem at 9 DAP (n = 4). (I) Seeds per silique: the number of seeds in dried siliques (n = 10). The bold dashed line in the violin plot indicates the median. One‐way analysis of variance and Tukey’s multiple comparison tests were used for statistical processing.

Figure S2. Mucilage extrusion assay of arr21 dry seeds. (A) R.R. staining images of WT and arr21 mutant dry seeds for mucilage detection. Scale bar = 100 μm. (B) Seed area. (C) Mucilage area of WT and arr21 mutants. The graphs in (B) and (C) show absolute values. (D) The mucilage area was divided by the seed area for normalization. Error bars indicate the SD, n = 25. One‐way analysis of variance and Tukey’s multiple comparison tests were used for statistical processing.

Figure S3. Fatty acid composition and fatty acid methyl esters (FAME) analysis of dry seeds from arr21 mutant. (A) Fatty acid composition of the WT and arr21 mutants. (B) Total FAME of WT, arr21‐3, and arr214. (C) Total FAME per seed. Error bars indicate the SD, n = 5. One‐way analysis of variance and Tukey’s multiple comparison tests were used for statistical processing.

Figure S4. Genotyping for vector confirmation in T₁ leaves of arr21‐4/Phas lines. (A) The vector map diagram of pPhas‐gate. blpR = the phosphinothricin acetyltransferase gene, LB = left border, RB = right border. (B) Genotyping results of T₁ plants. PCR and gel electrophoresis confirmed the arr21‐4/Phas to verify the presence of the vector. attB1 and attB2 primers were used for genotyping. The star indicates lines with observed seed weight and length.

Figure S5. Expression of cytokinin signal transduction pathway genes based on RNA‐seq analysis of developing seeds in WT and arr21‐4 (A) Relative expression of AHKs, AHPs, and ARRs genes in developing seeds of WT. (B) Expression changes of AHKs, AHPs, and ARRs in arr21‐4 seeds compared to WT. Expression changes are shown as log‐transformed values (log₂); a value of 1 corresponds to a twofold change.

Table S1 List of primers used in this study.

Table S2 Differential gene expression analysis by RNA‐seq in S6‐stage developing seeds of WT and arr21‐4.

Table S3. Cytokinin signaling gene expression in WT and arr21‐4 S6‐stage developing seeds.

Jeong, S. , Kim I., and Kim H.. 2025. “Molecular Characterization of Cytokinin Response Regulator ARR21 in Arabidopsis Seed Development.” Plant Direct 9, no. 7: e70091. 10.1002/pld3.70091.

Data Availability Statement

All data supporting the findings of this study are available in the paper and supplementary materials published online.