A novel calmodulin-interacting Domain of Unknown Function 506 protein represses root hair elongation in Arabidopsis.

Domain of Unknown Function 506 proteins are ubiquitous in plants. The phosphorus (P) stress-inducible REPRESSOR OF EXCESSIVE ROOT HAIR GROWTH1 (AtRXR1) gene encodes the first characterized DUF506. AtRXR1 inhibits root hair elongation by interacting with RabD2c GTPase. However, functions of other P-responsive DUF506 genes are still missing. Here, we selected two additional P-inducible DUF506 genes for further investigation. The expression of both genes was induced by auxin. Under P-stress, At3g07350 gene expressed ubiquitously in seedlings, whereas At1g62420 (AtRXR3) expression was strongest in roots. AtRXR3 overexpressors and knockouts had shorter and longer root hairs, respectively. A functional AtRXR3-green fluorescent protein fusion localized to root epidermal cells. Chromatin immunoprecipitation and quantitative reverse-transcriptase-polymerase chain reaction revealed that AtRXR3 was transcriptionally activated by RSL4. Bimolecular fluorescence complementation and calmodulin (CaM)-binding assays showed that AtRXR3 interacted with CaM in the presence of Ca2+ . Moreover, cytosolic Ca2+ ([Ca2+ ]cyt ) oscillations in root hairs of rxr3 mutants exhibited elevated frequencies and dampened amplitudes compared to those of wild type. Thus, AtRXR3 is another DUF506 protein that attenuates P-limitation-induced root hair growth through mechanisms that involve RSL4 and interaction with CaM to modulate tip-focused [Ca2+ ]cyt oscillations.

INTRODUCTION

Phosphorus (P) is a structural component of nucleic acids, membrane phospholipids and energy metabolites. As such, P is essential for all life on earth. Plants assimilate P as inorganic phosphate (Pi) from the soil. However, the concentration of Pi in many unfertilized soils is low and therefore limiting for plant growth (Hammond et al., 2004; Raghothama, 1999). Plants employ various strategies to improve the acquisition, remobilization and efficient use of Pi when P is limited. The strategies include altering root morphology and secreting phosphatases and nucleases to mobilize or release Pi from soil insoluble organic sources (Lynch, 2011; Plaxton & Tran, 2011; Svistoonoff et al., 2007; Williamson et al., 2001). Increasing the density and length of root hairs (RHs) is one morphological response of roots to low P. RHs account for up to 70% of the total root surface area in many plant species. The increase in root surface area due to RHs facilitates the absorbance of soil Pi (Bates & Lynch, 2001; Bucher, 2007; Crombez et al., 2019; Lynch, 2019; Ma et al., 2003). Moreover, longer RHs improve root penetration in hard and dry soils (Choi & Cho, 2019; Haling et al., 2013), enhance plant tolerance to drought stress (Marin et al., 2020; Zhang et al., 2020) and enable roots to interact more efficiently with beneficial microbes (Brown et al., 2013).

RHs develop from specialized root epidermal cells called trichoblasts. In Arabidopsis, trichoblasts are located between two underlying cortical cells (Grierson et al., 2014). Among the many genes that function in RH formation, ROOT HAIR DEFECTIVE6 (RHD6), which encodes a bHLH transcription factor (TF), stands out (Masucci & Schiefelbein, 1994). RHD6, along with RHD6‐LIKE2 (RSL2) and RHD6‐LIKE4 (RSL4) directly bind to promoter RH elements (RHEs, TN6CA[CT]G[TA]) of various RH‐specific (RHS) genes. Binding of these bHLH TFs to RHEs modulate the expression of RHS genes and consequently RH elongation (Bhosale et al., 2018; Datta et al., 2015; Hwang et al., 2017; Kim & Dolan, 2016; Kim et al., 2017; Vijayakumar et al., 2016; Yi et al., 2010).

Phytohormones and other signalling factors, such as reactive oxygen species (ROS), calcium ions (Ca2+) and the short peptide RAPID ALKALINIZATION FACTOR1 (AtRALF1), modulate RH development (Abarca et al., 2021; Feng et al., 2017; Foreman et al., 2003; Han et al., 2020; Kapulnik et al., 2011; Mangano et al., 2017, 2018; Pitts et al., 1998; Tian et al., 2020; Vissenberg et al., 2020; Wang et al., 2017; Zhu et al., 2020). Two models explain the mechanisms of RH elongation when exposed to low external P (Bhosale et al., 2018; Song et al., 2016). One model proposes that under P‐deficiency, elevated endogenous auxin activates auxin influx carrier (AUX1)‐mediated auxin transport and induces auxin response factor19 (ARF19) expression. The activation of auxin transport and signalling then stimulates the expression of RSL2 and RSL4 to enhance RH elongation. The second model postulates that the ethylene‐activated TF ETHYLENE‐INSENSITIVE3 (EIN3) regulates P‐limitation‐induced RH growth by interacting with RHD6 or activating the expression of RSL4 target genes (Feng et al., 2017; Song et al., 2016).

P starvation significantly affects the cytosolic Ca2+ ([Ca2+]cyt) signature of Arabidopsis roots (Matthus et al., 2019). [Ca2+]cyt signatures are decoded by the Ca2+‐binding protein, calmodulin (CaM) and related proteins collectively called CaM‐like proteins (CMLs). In Arabidopsis, seven genes encode for CaMs and another 50 encode CMLs (Zielinski, 1998). CaMs associate with other enzymatic effectors such as cyclic nucleotide‐gated channels (CNGCs) to facilitate their regulatory functions (Brost et al., 2019). The NADPH oxidase, ROOT HAIR DEFECTIVE2, triggers the accumulation of ROS, which subsequently specifies the frequency and amplitude of [Ca2+]cyt oscillations, and steepness of [Ca2+]cyt gradients at the RH tip through the activation of CNGCs (Foreman et al., 2003; Tian et al., 2020). Plasma membrane‐localized CNGC14 mediates Ca2+ influx into elongating RHs (Zhang et al., 2017). Moreover, CaM7 interacts with CNGC14 in Arabidopsis to inhibit RH growth (Zeb et al., 2020). Although there is accumulating evidence for the involvement of CaM and [Ca2+]cyt in RH elongation, their role in modulating P‐starvation‐induced RH growth is unknown.

The Domain of Unknown Function 506 (DUF506) gene family is omnipresent in many plant species (Ying et al., 2022). The Arabidopsis genome has 13 DUF506 genes. The proteins encoded by these genes have three conserved domains, which are concentrated in the C‐terminus. On the other hand, the N‐terminus of DUF506 is highly diverse (Ying, 2021). The only characterized DUF506 gene so far is At3g25240 (AtRXR1). AtRXR1 is strongly induced by P‐stress in a PHR1/PHL1‐dependent manner. The PHR1/PHL1‐AtRXR1 represses RH elongation based on overexpression and mutant studies. Similar to Arabidopsis, overexpressing the Brachypodium RXR1 orthologue (Bdi2g58590) in Brachypodium distachyon leads to short RH (Ying et al., 2022).

The effect of RXR1 on P‐limitation‐mediated RH growth prompted us to study other Arabidopsis DUF506 genes, particularly those whose expression was induced by low P. One Arabidopsis DUF506 gene (At1g62420, hereafter referred to as AtRXR3) was investigated in more detail. Our results reveal that, like AtRXR1, AtRXR3 is a repressor of RH elongation growth. However, the mechanism by which AtRXR3 represses RH growth differs from that of AtRXR1. While AtRXR1 represses RH growth via a small GTPase, AtRXR3 does so through RSL4 and its interaction with various cytosolic CaMs.

MATERIALS AND METHODS

Plant material, growth and treatment

Arabidopsis thaliana (ecotype Col‐0) seeds were sterilized and placed on a half‐strength Murashige and Skoog (1/2MS) solid medium according to (Ying et al., 2022). Petri dishes were stratified for 3 days at 4°C in the dark before transfer to a growth chamber (120 µmol−2 s−1 light intensity, 22°C/20°C, 16 h light cycle and 8 h dark cycle). For phosphate‐deprived (−P) treatment, experiments were performed as described previously (Morcuende et al., 2007). Briefly, Arabidopsis seedlings were germinated and grown in 1/2MS liquid medium and then transferred into P‐deficient liquid culture on Day 7. Shoot and root tissues were harvested on Day 9 or as indicated in Section 4, frozen in liquid N2 and stored in −80°C freezer. The rxr3‐1 (Salk_016908), rxr3‐2 (Salk_078603), cam7 (CS873256) and arf7 arf19 (CS24629) T‐DNA insertion mutant lines were obtained from Arabidopsis Biological Resource Center (ABRC). The rsl4 mutant was kindly provided by Dr. Elison Blancaflor (Noble Research Institute). Homozygous mutant plants were identified or validated according to the SiGnAL database protocol (http://signal.salk.edu/).

Quantitative reverse transcription‐PCR (qRT‐PCR)

Total RNA was extracted using RNeasy® plant mini kit (Qiagen) and genomic DNA was removed using Turbo DNA‐free™ kit (Invitrogen) following the manufacturer's instructions. RNA (1 µg) was converted into cDNA using SuperScript™ III Reverse Transcriptase (Invitrogen) according to the manufacturer's instruction. Quantitative PCR was performed using QuantStudio™ 7 Flex Real‐Time PCR System (Invitrogen) and KiCqStart® SYBR® Green qPCR ReadyMix™ (Millipore Sigma). The qRT‐PCR reaction procedures were conducted as described previously (Ying et al., 2022). The Arabidopsis GAPDH (At1g13440) gene was used as an internal control (Czechowski et al., 2005). All the experiments were repeated at least three times using cDNAs prepared from two different biological replicates with representative results shown in the figures. Sequences of primers used in the study are listed in Table S2.

Measurement of RH length

Seedlings were germinated vertically for 3 days on 1/2MS solid medium containing 0.4% Gelzan™ CM (Millipore Sigma) and then transferred to plates containing either 0 (P‐deficient) or 675 µM (P‐sufficient) phosphate. Gelzan™ (0.4% w/v) was selected as gelling agent to prevent unexpected phosphate contamination from different batches of agar (Jain et al., 2009). Three days after transferring the seedlings, RH images were captured using a Nikon SMZ1500 stereomicroscope. RH length was determined by measuring at least 700 RHs located between 2 and 6 mm from the tip of the primary root in 10 individual plants for each genotype. All the measurements were repeated at least three times with representative RH images shown in the figures.

Generation of transformants

To generate RXR3 overexpressing lines, the RXR3 full‐length (FL) coding region was amplified from Arabidopsis seedling cDNA using Phusion® High‐Fidelity DNA polymerase (New England Biolabs) and cloned into GATEWAY® entry vector pENTR™/SD/D‐TOPO® (Invitrogen). The sequencing‐confirmed vector was recombined with the pMDC32 destination vector (Curtis & Grossniklaus, 2003). For histochemical GUS analysis, a 1300‐bp fragment upstream from the start codon of RXR3 was amplified from Arabidopsis genomic DNA and cloned into the pBGWFS7 vector (Karimi et al., 2002). For complementation analysis of the rxr3‐1 mutant, the promoter fragment and 852‐bp coding region of the RXR3 gene was amplified and were cloned into pMU64 destination vector as described previously (Ying et al., 2022). For chromatin immunoprecipitation (ChIP) assays, the promoter fragment and FL coding region of RSL4 were amplified from Arabidopsis seedling genomic DNA or cDNA and eventually recombined into the pMU64 vector, resulting in the proRSL4::RSL4‐GFP construct. For calcium oscillation analysis, the proUBQ10::GCAMP3 construct was transformed into individual rxr3 mutant lines (Kwon et al., 2018).

All constructs were introduced into Agrobacterium tumefaciens strain GV3101 using the freeze–thaw procedure and were transformed into Arabidopsis by floral dipping (Zhang et al., 2006). Transformants were selected on 1/2MS agar (0.8% w/v) medium containing 25 µg ml−1 hygromycin (Omega Scientific).

Western blot analysis

For detection of the green fluorescent protein (GFP) fusion protein in RXR3 complementation plants, total protein was extracted as described previously (Ying et al., 2022). Equal amounts (30 µg) of each sample were separated by sodium dodecyl sulphate‐polyacrylamide gel electrophoresis (SDS‐PAGE), transferred into Immobilon‐P PVDF membrane (0.45 μm; Millipore Sigma) and probed with 1/5000‐diluted monoclonal mouse anti‐GFP horseradish peroxidase (HRP)‐conjugated antibody (Miltenyi Biotec). Chemiluminescence detection was performed with ECL™ Prime Western Blotting System (Millipore Sigma) and UVP ChemStudio imager (Analytik Jena).

Histochemical GUS staining and GFP imaging

Histochemical GUS activity assays were performed as previously described (Jefferson et al., 1987). Briefly, seedlings were incubated in a GUS staining solution containing 100 mM sodium phosphate (pH 7.0), 1 mM ethylenediaminetetraacetic acid (EDTA), 0.05% (v/v) Triton X‐100, 1 mM potassium ferricyanide/ferrocyanide and 0.5 mg ml−1 X‐glucuronide (Goldbio) at 37°C for 1–3 h. Then, samples were then cleared in a graded series of 30%, 50%, 70% and 100% (v/v) ethanol for 30 min. Images were acquired using a Nikon SMZ1500 stereomicroscope. GFP fluorescence of the proRXR3::RXR3‐GFP complementation plants was imaged with a Leica TCS SP8 confocal laser‐scanning microscope (Ex: 488 nm; Em 507 nm).

Measurement of RH tip [Ca2+]cyt oscillation

To prepare Arabidopsis seedlings for imaging, transgenic seeds expressing the intensiometric [Ca2+]cyt GCaMP3 reporter were directly plated on coverslips coated with a thin layer of ½MS‐supplemented 0.4% Gelzan™ CM following the set‐up of (Rincón‐Zachary et al., 2010). After vernalization at 4°C for 2 days, Petri dishes were transferred to a growth chamber (120 µmol−2 s−1 light intensity, 22°C/20°C, 16 h light cycle and 8 h dark cycle) and were kept vertically to enable the primary root to grow down and along the surface of the gel. Growing RHs of 5‐day‐old seedlings were imaged with a Leica TCS SP8 confocal laser‐scanning microscope (Ex: 488 nm; Em 510 nm).

For image analysis, the average fluorescence intensity was acquired by marking a rectangular region at the RH apex using the rectangular selection tool of the Leica microscope software. Fluorescence intensity values were normalized using the formula (F − F 0)/F 0, where F is the fluorescence intensity at any given time point and F 0 is the lowest fluorescence intensity point for the data set (Kwon et al., 2018). [Ca2+]cyt oscillations from the tips of at least 6 to 11 RHs from four to six seedlings of each genotype were measured.

AtRALF1 treatment

Seedlings were germinated in a regular 1/2MS liquid medium for 5 days and then transferred to a fresh liquid medium containing 1 µM synthesized AtRALF1 peptide (49 amino acids; Pepscan). After incubating in AtRALF1‐supplemented 1/2MS for 6 h, seedlings were collected for expression and fluorescence analysis. All the treatments were repeated at least two times with representative results shown in the figures.

ChIP assay

Roots of RSL4‐GFP‐complemented seedlings, which were grown vertically on 1/2MS agar (0.8% w/v) medium for 12 days, were harvested and stored in −80°C for further steps. ChIP assays were conducted as described previously with minor modifications (Haring et al., 2007; Nelson et al., 2006; Peng et al., 2017; Saleh et al., 2008). In brief, nuclear DNA was sheared by sonication (6 s × 15 pulses, 1 min break between each pulse) and cell debris was removed by centrifugation. ChIP was performed by incubating extracted DNA with a rabbit anti‐GFP polyclonal antibody (Abcam) followed by treatment with Protein A/G magnetic beads (Pierce). The beads were washed three times with high‐salt wash buffer, which contained 20 mM Tris‐HCl (pH 8.0), 500 mM NaCl, 0.2% (w/v) SDS, 0.5% (v/v) Triton X‐100, 2 mM EDTA and the Pierce™ Protease Inhibitor mini tablet. Chromatin was eluted in buffer containing 1% (w/v) SDS and 100 mM NaHCO3 and de‐cross‐linked at 65°C for 16 h. Cellular RNA and protein were removed by RNase (Qiagen) and proteinase‐K (Applied Biosystems) treatments, respectively. Purified DNA was used as a template to determine the enrichment of target genes through PCR using primers listed in Table S2.

CaM‐binding assay

For heterologous expression of RXR3, its FL and N‐terminal‐ (Δ1–22, residues 1–2, including CaM‐binding domain [CaMBD]) truncated cDNA fragment was subcloned into the pMAL™‐c5X vector (New England Biolabs) carrying an N‐terminal maltose‐binding protein (MBP) tag. For recombinant protein production, constructs were separately introduced into Escherichia coli (NEB Express Competent cells C2523H). Purification of RXR3 was performed using amylose resin (New England Biolabs) according to the manufacturer's instructions. Protein concentration was quantified using the Pierce™ rapid gold BCA protein assay kit (Thermo Fisher Scientific‎) and stored in a −80°C freezer.

CaM‐binding assays were carried out according to a previous study with minor modifications (Kato et al., 2013). Briefly, recombinant MBP‐RXR3FL or MBP‐RXR3Δ1‐22 protein (2 µg) were incubated with 25 µl CaM‐agarose (Millipore Sigma) in buffer containing 20 mM Tris (pH 7.0), 150 mM NaCl and 0.5 mM CaCl2 or 2 mM EGTA for 1 h at 4°C. The supernatant was discarded by centrifugation at 1470 g for 2 min. After washing three times with the same buffer, bound proteins were dissociated by incubation with 2× SDS‐PAGE protein sample buffer (Millipore Sigma). Samples were separated and transferred to the PVDF membrane as described above. Western blot analysis was performed using 1/5000‐diluted monoclonal mouse anti‐MBP‐HRP antibody (Miltenyi Biotec).

Bimolecular fluorescence complementation (BiFC) assays

BiFC analysis was performed as previously described (Kudla & Bock, 2016; Waadt et al., 2014). Briefly, RXR3 was fused to the N‐terminal EYFP in the pSITE‐cEYFP or pCAMBIA1305‐CFP (gift of Dr. Elison Blancaflor) vector, whereas various Arabidopsis CaM genes were fused to the C‐terminal part of EYFP in pSITE‐nEYFP. The At3g25240/RXR1 gene was also constructed into the pSITE‐cEYFP vector and used as a negative control. These different vector pairs were cotransformed into Nicotiana benthamiana through agroinfiltration. Forty‐eight hours after infiltration, YFP or CFP fluorescence was detected using a Leica TCS SP8 confocal laser‐scanning microscope (CFP, Ex: 405 nm; Em: 485 nm; YFP, Ex: 514 nm; Em: 527 nm). All the experiments were repeated at least three times with representative images shown in the figures.

Measurement of phosphate content

Phosphate content measurement was conducted as described previously (Carter & Karl, 1982) with minor modifications. Briefly, 50 mg (fresh weight) of Arabidopsis tissue was ground and resuspended in 600 µl of MilliQ water. After centrifugation at 13,000 g for 2 min, the clarified supernatant was used to quantify phosphate content. A mixture consisting of 20 µl of supernatant, 90 µl of 1 N HCl and 90 µl of malachite green dye solution (0.042%, w/v) was incubated at room temperature for 5 min. The absorbance of the mixture at 630 nm was recorded with a spectrophotometer. The collected values were calibrated by a standard curve created with various concentrations of K2HPO4. All measurements were repeated at least three times with representative results shown in the figures.

Statistical analysis

Statistical analyses and data plotting were performed with GraphPad Prism 9.0 (GraphPad Software, Inc.). Data on gene expression, RH length, shoot biomass, Pi content and amplitudes and frequencies of [Ca2+]cyt oscillation were analysed statistically by one‐way analysis of variance with Tukey's test or Student's t test.

Accession numbers

Sequence data from this article can be found in the Phytozome 12 online genomic resource (https://phytozome.jgi.doe.gov/pz/portal.html) under the following accession numbers: AtRXR1 (At3g25240), AtRabD2c (At4g17530), AtRXR3 (At1g62420), Arabidopsis DUF506 gene family members (At1g12030, At1g77145, At1g77160, At2g20670, At2g38820, At2g39650, At3g07350, At3g22970, At3g54550, At4g14620 and At4g32480), AtCAM1 (At5g37780), AtCAM2 (At2g41110), AtCAM6 (At5g21274), AtCAM7 (At3g43810), AtCNGC14 (At2g24610), AtRALF1 (At1g02900) and AtEXPA7 (At1g12560).

RESULTS

Expression of Arabidopsis DUF506 genes during P‐limitation

To functionally characterize AtDUF506 genes under P‐deprivation, their expression patterns from public RNA‐seq databases (i.e. GENEVESTIGATOR®) was determined. Through database searches, two additional DUF506 genes, namely, At2g20670 and At3g07350, whose expression was induced by P‐limitation, were found (Table S1). Next, transcript changes of all AtDUF506 genes in response to P‐deprivation were investigated using qRT‐PCR analysis (Figure 1). In addition to the two P‐limitation inducible DUF506 genes noted above, the expression of At1g62420/RXR3 (in root) and At4g32480 were significantly up‐regulated (i.e., fold change >2) by P‐limitation. Similar to At3g25240/RXR1, expression of At1g62420, At3g07350 and At4g32480 were induced in both shoot and root tissues, whereas At2g20670 exhibited shoot‐specific induction when exposed to P‐stress. Among the DUF506 genes, expressions of At3g25240/RXR1, At1g62420 and At3g07350 were the most P‐inducible in roots (~10‐fold up‐regulation). Next, the tissue‐specific expression pattern of At3g07350 and At1g62420 was analysed using promoter–GUS fusions. As shown in Figure S1, At3g07350 expressed weakly in vascular bundles and lateral root tips under P‐sufficient (control) condition. This gene was strongly induced in the entire seedling (e.g., cotyledon, primary/lateral roots and root tips) when seedlings were cultured in P‐deprived liquid medium for 24 h. On the other hand, GUS activity expressed under the control of the At1g62420 promoter was mainly observed in primary roots. Expression was strongest in vascular bundles, root tips and RHs of primary roots. In contrast, no GUS signals were detected in P‐stressed shoots (Figure 2A). In P‐sufficient seedlings, GUS signal was only observed in root caps. These results indicate that the levels of expression of additional Arabidopsis DUF506 genes (i.e., At1g62420, At2g20670, At3g07350 and At4g32480) respond to P‐limitation and therefore their roles during P‐stress warrant further investigation.

Figure 1

Expressions of Arabidopsis DUF506 genes in response to P‐limitation. Five‐day‐old WT seedlings were transferred to P‐sufficient or P‐deficient half‐strength MS liquid medium and then the shoots and roots were harvested separately after 24 h. Expression was measured by qRT‐PCR. The At3g54550 gene was undetectable because of its silique‐specific expression (Ying, 2021). The transcript level of each DUF506 gene in shoots (P‐sufficient condition) was set to 1. The Arabidopsis GAPDH gene was used as a reference for normalization. The data represent the mean values of three replicates ± SD. DU506, Domain of Unknown Function 506; MS, Murashige and Skoog; P, phosphorous; qRT‐PCR, quantitative reverse transcription‐PCR; R, root; S, shoot; WT, wild type.

Figure 2

At1g62420 responds to phosphate starvation. (A) GUS staining of the proAt1g62420::GUS transgenic line. Five‐day‐old transgenic seedlings were grown on P‐sufficient (a–c) or deficient (d–g) half‐strength MS solid medium for 5 days and then stained. Scale bar = 1 mm (a, b and d) or 50 µm (c, e–g). (B) Time course of At1g62420 expression determined by qRT‐PCR analysis. Five‐day‐old WT seedlings were transferred to fresh regular (P‐sufficient) or no phosphate (P‐deficient) half‐strength MS liquid medium and then the samples were harvested at the indicated time points. **p < 0.01 indicates statistical significance (P‐deficient vs. P‐sufficient) as determined by Student's t test. (C) Expression changes of At1g62420 in roots of the phr1 phl1 double mutant. Three‐day‐old seedlings were grown and treated as described in Figure 1. (D) Expression changes of At1g62420 in wild‐type roots treated with auxin (IAA). Three‐day‐old seedlings were transferred to P‐sufficient (Control) or supplemented with 1 µM IAA in half‐strength MS liquid medium. qRT‐PCR analysis was performed on roots 5 days after exposure to IAA. The data represent the mean values of three replicates ± SD in panels (B–D). MS, Murashige and Skoog; P, phosphorous; qRT‐PCR, quantitative reverse transcription‐PCR; WT, wild type.

Transcriptional characterization of P‐limitation‐inducible At1g62420 gene

Because its expression is induced strongly by P‐limitation in roots and it has low sequence similarity to RXR1, functional characterization of the At1g62420 gene was of interest. Transcript changes of the At1g62420 gene at various times after the onset of P‐starvation were measured to determine if its response was specific to low P. Under P‐sufficient conditions, the levels of At1g62420 transcript did not change, whereas in P‐stressed seedlings, At1g62420 expression was induced after 12 h and peaked at 24 h (Figure 2B). Three hours after Pi readdition (R3h), the expression level of At1g62420 returned to levels seen in P‐sufficient conditions, indicating a direct response of At1g62420 transcripts to plant P‐status.

PHOSPHATE STARVATION RESPONSE1 (PHR1) and its homolog PHR1‐LIKE (PHL1) are critical TFs in the plant P‐responses (Bustos et al., 2010; Nilsson et al., 2007; Rubio et al., 2001). Because RXR1 was transcriptionally activated by PHR1 and PHL1 (Ying et al., 2022), the regulation of At1g62420 expression by these two TFs was studied. This was first tested by searching PHR1 binding sequences (P1BS, GNATATNC) in the At1g62420 promoter region. No P1BS was found. Next, transcript changes of At1g62420 in phr1 phl1 double mutant in response to P‐limitation were examined. In root tissues of phr1 phl1 double mutant, At1g62420 exhibited identical expression patterns as wild type (Figure 2C), indicating that its transcriptional response to P‐stress was independent of PHR1 and PHL1.

During P‐limitation, the concentration of endogenous auxin accumulates and redistributes in various organs. High auxin levels induce the expression of auxin‐responsive factors (ARFs, e.g., ARF5, 7, 8 and 19), leading to altered root system architecture (RSA) (Bhosale et al., 2018; Mangano et al., 2017). Thus, expression changes of At1g62420 in response to auxin treatment were investigated. Notably, three copies of the auxin‐responsive element (i.e., AACGAC) were found in At1g62420 promoter and were located in 116, 184 and 454 bp upstream of the start codon. In roots, the auxin‐induced transcriptional changes of At1g62420 (10.7 ± 0.69‐fold up‐regulation) were similar to those of P‐stress (10.5 ± 1.38‐fold up‐regulation; Figures 1 and 2D). In addition, qRT‐PCR indicated that transcripts of At1g62420 were moderately reduced in arf7 arf19 double mutant regardless of P‐status (Figure S2). Collectively, results show that the transcription of At1g62420 in roots is regulated in a manner that might be affected by but is not exclusively dependent on ARF7/ARF19 signalling.

At1g62420/RXR3 represses RH growth

To elucidate the biological function of At1g62420, overexpression lines were generated (Figure 3A). Two independent homozygous transgenic lines (i.e., OX‐5 and OX‐12) exhibited significantly shorter RHs than those of wild type (e.g., P‐sufficient/deficient condition, 0.189 ± 0.003/0.242 ± 0.005 mm [OX‐12] vs. 0.237 ± 0.003/0.308 ± 0.004 mm [wild type]) (Figure 3C,E). Two independent T‐DNA mutants (SALK_016908 and SALK_078603) were identified from ABRC (Figure 3B). qRT‐PCR did not detect At1g62420 transcript in SALK_016908 and the transcript was significantly reduced in SALK_078603 (Figure 3A). RHs of both mutants were ~40% longer than RHs of wild type under P‐deficient conditions (e.g., P‐sufficient/deficient condition, 0.270 ± 0.005/0.433 ± 0.007 mm [SALK_016908]; 0.299 ± 0.005/0.437 ± 0.007 mm [SALK_078603]; Figures 3C,E and S3). Because At1g62420 mutants phenocopied the long RHs of rxr1 (Ying et al., 2022), SALK_016908 and SALK_078603 were hereafter named rxr3‐1 and rxr3‐2, respectively, and At1g62420 will be referred to as RXR3. Additional phenotyping of the mutants under P‐deficient conditions showed that rxr3‐1 had higher endogenous P‐content and biomass than wild type (Figures 3I,J and S4), suggesting that RXR3 functions as an RH growth repressor that affects P‐uptake and plant growth.

Figure 3

RXR3 represses Arabidopsis root hair elongation. (A) qRT‐PCR analysis of RXR3 transcript level in RXR3 overexpressers or rxr3 mutants, under P‐sufficient condition.ND, not detectable. (B) Schematic diagram of two T‐DNA insertion mutants used in the study. Exons, black boxes; UTRs, white boxes; intron, black line. (C and D) Representative images showing the effect of RXR3 on root hair elongation. Three‐day‐old seedlings were transferred to P‐sufficient or P‐deficient half‐strength MS solid medium and then grown vertically for 2 days. While the loss‐of‐function mutants (rxr3‐1 and rxr3‐2) grow longer root hairs, constitutive overexpression of RXR3 (proCaMV35S::RXR3, OX‐5 and 12) represses root hair growth (c). Root hair growth is restored to wild‐type levels in rxr3‐1 mutants complemented with RXR3 (proRXR3::RXR3‐GFP, Com#1 and #2; d). Scale bar = 1 mm. (E and F) Box plots of root hair lengths in WT, RXR3 overexpressers, rxr3 mutants and complemented lines, under P‐sufficient or P‐deficient condition. Box limits indicate 25th and 75th percentiles, horizontal line is the median and whiskers display the minimum and maximum values. Between 885 and 1026 root hairs from 10 seedlings were measured for each genotype and numbers at the bottom indicate median values (in mm) for each genotype. Statistical significance of differences was tested by one‐way ANOVA Tukey's analysis (p < 0.001). Means with different letters indicated statistical significance. (G) Immunoblot analysis of RXR3‐GFP fusion protein in complementation lines (#1 and #2) under different P‐conditions. Coomassie blue‐stained RuBisCO protein is shown as a loading control. (H) Initiating root hairs (a) and root epidermal (b, transverse view) showing the localization of RXR3‐GFP fusion protein in Arabidopsis complemented line (Comp#1) under P‐stress. Scale bar = 100 µm. (I and J) Box plots of shoot phosphate content (I) or fresh biomass (J) of 10‐day‐old wild type (blue) and rxr3‐1 mutant (orange) seedlings grown on P‐sufficient or P‐deficient half‐strength MS solid medium (n = 3). Horizontal line is the median and whiskers that display the minimum and maximum values. (I and J) **p < 0.01 or *p < 0.05 indicates statistical significance as determined by Student's t test. ANOVA, analysis of variance; GFP, green fluorescent protein; MS, Murashige and Skoog; ND, not determined; ns, no significance; P, phosphorous; qRT‐PCR, quantitative reverse transcription‐PCR; Rubisco, ribulose 1,5‐bisphosphate carboxylase/oxygenase; UTR, untranslated region; WT, wild type.

To ascertain that the observed RH traits were related to RXR3 function, an endogenous promoter‐driven RXR3‐GFP construct was introduced in rxr3‐1. GFP signal was detected in two independent complementation lines by Western blot analysis, indicating that RXR3 was correctly transcribed and translated in vivo (Figure 3G). Moreover, the RXR3‐GFP was predominately detected in root epidermal cells and RH tips (Figure 3H), which points to the direct involvement of RXR3 in an RHS process. The promoter‐driven RXR3‐GFP (complemented line #1) partially rescued the long RH phenotype of rxr3‐1 (e.g., P‐sufficient/deficient condition, 0.265 ± 0.004/0.362 ± 0.007 mm [complemented line #1] vs. 0.233 ± 0.004/0.327 ± 0.006 mm [wild type]; Figures 3D,F and S3).

Because loss‐of‐function RXR1 and RXR3 led to longer RHs, rxr1 rxr3‐1 double mutants were generated to determine if they genetically interact during RH development (Figure 4). Under P‐sufficient condition, RHs of rxr1 rxr3‐1double mutant (0.360 ± 0.005 mm) was comparable to rxr1 (0.366 ± 0.006 mm), but significantly longer than rxr3‐1 (0.309 ± 0.005 mm) and wild type (0.256 ± 0.004 mm). Under P‐deficient conditions, the double mutant (0.456 ± 0.007 mm) exhibited significantly longer RHs (p < 0.01) than either one of the single mutants, suggesting an additive function (rxr1, 0.421 ± 0.007 mm; rxr3‐1, 0.430 ± 0.006 mm). Taken together, the double mutant studies indicate that RXR3 inhibits RH growth through a pathway that is independent of RXR1.

Figure 4

Root hair phenotype of rxr1 and rxr3‐1 single mutants, and rxr1 rxr3 double mutant. (A) Representative root hair images of 5‐day‐old seedlings. Scale bar = 1 mm. (B) Box plots of root hair lengths in WT, rxr1 and rxr3‐1 single mutants and rxr1 rxr3 double mutant, under P‐sufficient or P‐deficient condition. Box limits indicate 25th and 75th percentiles, horizontal line is the median and whiskers display the minimum and maximum values. Between 906 and 1037 root hairs from 10 seedlings were measured for each genotype and numbers at the bottom indicate median values (in mm) for each genotype. Statistical significance of differences was tested by one‐way ANOVA analysis (p < 0.001) and is indicated by lower case letters. ANOVA, analysis of variance; P, phosphorous; WT, wild type.

RXR3 is directly regulated by RSL4

RSL4 is a critical TF that specifies RH initiation and elongation by modulating the expression of several RH‐related genes (Yi et al., 2010). Therefore, it was hypothesized that RXR3 expression could be regulated by RSL4. RSL4 binds to RHEs in the promoter regions of RH‐expressed genes (Hwang et al., 2017). Four RHEs were identified in the RXR3 promoter region (Figure 5A) and the P‐limitation‐inducible expression of RXR3 was not observed in rsl4 (Figure 5B). These findings suggest that RSL4 might directly bind to RHEs of RXR3 to regulate its expression. In doing so, RSL4 and RXR3 could synergistically control RH growth under P‐limitation. To test the hypothesis that RSL4 binds to RHEs in RXR3 promoter, RSL4 expression was examined in response to P‐limitation or auxin. As shown in Figure S5, RSL4 transcripts in roots were induced (~2‐fold) by both treatments. Next, an endogenous promoter‐driven RSL4‐GFP was expressed in rsl4. The RH phenotype of the complemented rsl4 mutant was similar to the one of wild type, indicating that the RSL4‐GFP fusion protein rescued the loss‐of‐function RH phenotype of rsl4 (Figure S6). ChIP analysis was then conducted to determine whether RSL4 binds to the RHEs in RXR3 promoter. EXPANSIN A7 (EXPA7) is a well‐known RSL4‐regulated gene (Yi et al., 2010); therefore, it was used as a positive control in the ChIP assays (Figure S7). As shown in Figure 5C, RSL4 interacted with probes 2 and 3, which contained RHE motifs of RXR3 (Figure 5A). No interaction was observed between RSL4 and probe 1, which did not have RXR3 RHEs. These results suggest that RSL4 directly binds to RHEs within the RXR3 promoter.

Figure 5

RSL4 binds to the RXR3 promoter. (A) Schematic diagram of the RXR3 promoter region showing the relative positions of RHEs. The RHEs (TN6CA[CT]G[TA]) are marked by grey rectangles, and relative positions and sizes of the different PCR‐amplified fragments are indicated by black lines under the RHE(s). (B) qRT‐PCR analysis of RXR3 expression in rsl4 mutants and complementation lines. Three‐day‐old Arabidopsis seedlings were transferred to fresh P‐sufficient or deficient half‐strength MS liquid medium. qRT‐PCR assays were done on roots after 24 h. The data represent the mean values of three replicates ± SD. Statistical significance of differences was tested by one‐way ANOVA analysis (p < 0.01) and is indicated by lower case letters. (C) ChIP analysis to detect the association between RSL4 and the RHEs within the RXR3 promoter in the RSL4‐GFP transgenic line. The ChIP signals with (anti‐GFP) and without (No Ab) addition of anti‐GFP are indicated. Ab, antibodies; ANOVA, analysis of variance; ChIP, chromatin immunoprecipitation; GFP, green fluorescent protein; qRT‐PCR, quantitative reverse transcription‐PCR; RHE, root hair element.

Recently, Zhu et al. (2020) proposed that the peptide RALF1 and the leucine‐rich receptor‐like kinase, FERONIA (FER), play central roles in RH growth. One mechanism by which RALF1 governs FER‐mediated RH growth is through translational regulation of RSL4 by FER‐phosphorylated eIF4E1. Although RALF1 was insensitive to P‐limitation (Table S1), it was hypothesized that RALF1 treatment might simulate P‐starvation to activate RSL4, and consequently lead to RXR3 transcript accumulation. To test this hypothesis, 5‐day‐old RXR3 promoter‐GUS or rxr3‐1 complemented seedlings were incubated with 1/2MS liquid medium supplemented with 1 µM RALF1 for 6 h. As shown in Figure 6A,B, RXR3 exhibited the same expression pattern in the root epidermis in response to RALF1 as that of P‐stress (Figure 3H). Furthermore, RXR3 transcript in wild type and RSL4 complemented line was induced by RALF1 treatment, but not in rsl4 mutants (Figure 6C). These results support the hypothesis that RXR3 expression is transcriptionally regulated by RSL4 through pathways that intersect with RALF1.

Figure 6

RSL4 is required for RXR3 expression. Five‐day‐old seedlings were transferred to control (P sufficient), or supplemented with 1 µM RALF1 half‐strength MS liquid medium. After incubation for 6 h, root tissues were harvested for analysis. (A) GUS staining of proRXR3::GUS roots after RALF1 treatment. Scale bar = 100 µm. (B) Initiating root hairs (a) and root epidermal (b, transverse view) localization of RXR3‐GFP fusion protein in Arabidopsis complemented line (Comp #1) after RALF1 treatment. Scale bar = 100 µm. (C) qRT‐PCR analysis of RXR3 expression in rsl4 mutants and its complementation line, after RALF1 treatment. The data represent the mean values of three replicates ± SD. Statistical significance of differences was tested by one‐way ANOVA analysis (p < 0.01) and is indicated by lower case letters. ANOVA, analysis of variance; MS, Murashige and Skoog; P, phosphorous.

RXR3 interacts with CaMs

In contrast to RXR1 (Ying et al., 2022), RXR3 has an N‐terminal CaMBD between amino acids 11 and 22 (motif 1–12, [FILVW]xxxxxxxxxx[FILVW]), in addition to the DUF506 domains that define the protein family. This observation led to the hypothesis that RXR3 might interact with CaMs. This was tested by heterogeneously expressing and purifying MBP‐tagged recombinant RXR3 protein (MBP‐RXR3FL), and N‐terminally truncated mutant (MBP‐RXR3Δ1–22, lack of CaMBD; Figures 7A and S8). These recombinant proteins were separately incubated with CaM‐agarose in the presence or absence of Ca2+. The bound protein was dissociated, subjected to SDS‐PAGE and detected by immunoblotting with antibodies against the MBP tag. As shown in Figure 7B, RXR3FL bound to CaM‐agarose in the presence of Ca2+, but not in the presence of Ca2+ chelator, EGTA. By contrast, RXR3Δ1–22 did not bind to CaM‐agarose, in the presence of Ca2+. These results indicate that the N‐terminal CaMBD of RXR3 is required and sufficient for binding to the Ca2+–CaM complex.

Figure 7

RXR3 interacts with calmodulins (CaMs). (A) Diagram of full‐length (FL) RXR3 protein tagged with maltose‐binding protein (MBP, black boxes) and its truncated mutant (Δ1–22), which lacks the CaM‐binding domain (CaMBD, red box). Green boxes indicate the three conserved motifs present in the plant DUF506 family (Ying et al., 2022). (B) Immunoblot analysis of recombinant MBP‐RXR3FL and MBP‐RXR3Δ1–22 protein binding with CaM‐agarose in the presence or absence of Ca2+. Recombinant protein (2 µg each) was incubated with CaM‐agarose in the presence of 0.5 mM Ca2+ or 2 mM EGTA. Proteins bound to the CaM‐agarose were separated by SDS‐PAGE gel and detected by immunoblotting with anti‐MBP tag antibody. Coomassie blue‐stained recombinant protein (10% input) is shown as the loading control. (C) BiFC analysis to detect the interaction between RXR3 and CaMs in Nicotiana benthamiana. FL RXR3 or RXR1 (as negative control) and various CaM genes were cloned into pSITE‐cEYFP or pSITE‐nEYFP vectors. Constructs were cotransformed into N. benthamiana leaves via syringe infiltration. All the measurements were repeated at least three times with representative results shown. Scale bar = 20 µm. BiFC, bimolecular fluorescence complementation; [Ca2+]cyt, cytosolic Ca2+; CFP, cyan fluorescent protein; DUF506, Domain of Unknown Function 506; SDS‐PAGE, sodium dodecyl sulphate–polyacrylamide gel electrophoresis; YFP, yellow fluorescence protein.

To validate the interaction between RXR3 and CaMs in vivo, BiFC assays were conducted. In Arabidopsis, there are four isoforms of CaM (e.g., CaM1/4, CaM2/3/5, CaM6 and CaM7) that derive from seven CaM encoding genes (Bender & Snedden, 2013; Zielinski, 1998). Four of the seven CaM genes (i.e., CaM1, CaM2, CaM6 and CaM7) were selected for the BiFC assays with RXR3. Strong cytosolic yellow fluorescence was observed in RXR3‐CaM1, RXR3‐CaM2, RXR3‐CaM6 and RXR3‐CaM7 pairs, but not the RXR1‐CaM7 (negative control) pair (Figures 7C and S9). Overall, in vitro binding and in vivo BiFC assays demonstrate that RXR3 physically interacts with Ca2+–CaMs.

RXR3 influences RH tip [Ca2+]cyt oscillation

CaMs function as versatile Ca2+‐binding proteins that regulate the activity of numerous effectors (e.g., CNGCs) in Ca2+ signalling. Multiple CNGCs (e.g., CNGC6/9/14) generate and maintain RH tip‐focused [Ca2+]cyt oscillations (Brost et al., 2019). It was reported that CaM7 associates with CNGC14 to regulate RH polar growth presumably by controlling Ca2+ influx (Zeb et al., 2020). Therefore, we hypothesized that the interaction of RXR3 and CaM7 could inhibit or disrupt the CaM7–CNGC14 association. If so, RXR3 exerts its inhibitory function by regulating the affinity of CaM7 to CNGC14. When RXR3‐CFP was coexpressed with CaM7‐nEYFP and CNGC14‐cEYFP in tobacco, yellow fluorescence was not affected (Figure 8A,B). RXR3 not only colocalized with the CNCG14–CaM7 complex in the cytosol but also was found in the nucleus (arrow, Figure 8B). These results indicate that RXR3 did not interfere with the interaction between CaM7 and CNGC14, and might conduct other functions in the nucleus. Furthermore, the GCaMP3 construct was expressed in wild type, rxr3‐1 and rxr3‐2 to determine whether RXR3 influences RH apical [Ca2+]cyt oscillations (Figures 8C–E, S10 and S11 and Supporting Information Video S1). Both rxr3 mutants (rxr3‐1, 0.0517 ± 0.0027 Hz; rxr3‐2, 0.0542 ± 0.0020 Hz) exhibited ~20% higher frequencies of [Ca2+]cyt oscillation than wild type (0.0399 ± 0.0009 Hz). Moreover, the amplitude of [Ca2+]cyt oscillation was dampened (~40%) in rxr3‐1. These results suggest that RXR3 represses RH growth possibly by regulating [Ca2+]cyt oscillations. However, it cannot be ruled out that the observed disruptions in [Ca2+]cyt oscillations in rxr3 are secondary effects caused by altered RH growth.

Figure 8

RXR3 affects root hair apical [Ca2+]cyt oscillation. (A) BiFC analysis reveals an interaction between CNGC14 and CaM7 in Nicotiana benthamiana. (B) BiFC analysis shows that interaction between CNGC14 and CaM7 is independent of RXR3. Arrow, nucleus. Scale bar = 20 µm (A and B). Number in the top right corner of (B) indicates Pearson's coefficient that is calculated by ImageJ/JACoP Plug‐in. (C) Representative normalized root hair [Ca2+] oscillograms of wild type (WT) and rxr3‐1 mutant. The data were processed by baseline simulation analysis. (D and E) Box plot of [Ca2+]cyt oscillation amplitudes (D) or frequencies (E) in root hairs of rxr3 mutants and WT over 600 s. Box limits indicate 25th and 75th percentiles, horizontal line is the median and whiskers display the minimum and maximum values. Each semitransparent dot represents individual measurements from 6 to 11 root hairs per group from 4 to 6 plants. **p < 0.01 and *p < 0.05 indicates statistical significance as determined by one‐way ANOVA analysis. ANOVA, analysis of variance; BiFC, bimolecular fluorescence complementation; [Ca2+]cyt, cytosolic Ca2+​​​​; CaM, calmodulin; CFP, cyan fluorescent protein; ns, no significance; YFP, yellow fluorescence protein.

Overexpression of CaM7 represses RH growth by inhibiting CNGC14 activity (Zeb et al., 2020), which phenocopies RXR3 overexpressers. This led to the hypothesis that cam7 mutant might also phenocopy the long RHs of rxr3. Similar to rxr3, RHs of cam7 were longer than those of wild type (Figure S12). Additionally, to investigate whether RXR3 and CaM7 function in a similar pathway, RXR3 was overexpressed in cam7. RXR3 Overexpression did not reverse the long RH phenotype of cam7 (Figure S12). Based on these results, CaM7 might function upstream of RXR3 to affect RH growth and apical [Ca2+]cyt oscillation.

DISCUSSION

Many proteins with domains of unknown function (DUF) were identified and annotated, but remain to be functionally characterized (Ying, 2021). Recently, we demonstrated that Arabidopsis P‐limitation‐inducible DUF506 protein, AT3G25240/RXR1, negatively regulates RH growth by interacting with the small GTPase RabD2c (Ying et al., 2022). Here, we report four additional P‐limitation‐inducible DUF506 genes in Arabidopsis and show that one of these DUF506‐containing proteins, RXR3, is another repressor of RH elongation growth.

Because their expression is strongly induced by low P, we investigated two AtDUF506 genes in more depth. At3g07350, which is the segmental duplication of RXR1 (Ying, 2021), is an uncharacterized and putative RH specific gene with two RHE in its promoter region (Won et al., 2009). Here, GUS staining revealed that under P‐stress, At3g07350 expression is highly induced in roots, most notably in root tips and RHs (Figure S1). Moreover, the expression analysis revealed that At3g07350 in throot was induced by auxin (~3‐fold upregulation; Figure S13), suggesting that it is involved in auxin‐dependent regulatory pathways. Collectively, the data indicate that At3g07350 functions through an auxin‐mediated P‐signalling pathway, independent of RXR1.

To unravel the biological function of RXR3 gene, we performed genetic, biochemical and cytological experiments (Figures 2, 3, 4, 5, 6, 7, 8). Because the RXR3 expression pattern based on GUS and qRT‐PCR analysis mirrored the one of RXR1, we hypothesized that RXR3 might also function as a redundant RH growth repressor. Consistent with this hypothesis was the finding that overexpression of RXR3 reduced RH lengths, whereas loss‐of‐function mutants exhibited longer RHs than wild type (Figure 3). The shorter and longer RHs of RXR3 overexpressors and rxr3, respectively, are reminiscent of how RXR1 affects RH growth. Although RXR1 and RXR3 have similar repressor functions on RH growth, the results of this study revealed that the mechanism by which RXR3 influences RH development is distinct from that of RXR1. For instance, RXR1 responds to P‐stress by a >1000‐fold upregulation of its expression through PHR1/PHL1‐dependent and auxin‐independent pathways (Ying et al., 2022). By contrast, RXR3 influenced RH growth through PHR1/PHL1‐independent pathways and was induced by auxin (Figure 2). Furthermore, under P‐deficient conditions, the rxr1 rxr3‐1 double mutant exhibited longer RHs than both single mutants (Figure 4). Taken together, the data indicate functional and regulatory divergence of RXR1 and RXR3.

RSL4 expression is up‐regulated by P‐limitation and its overexpression promotes RH growth (Figure S5) (Bhosale et al., 2018; Datta et al., 2015; Yi et al., 2010). Here, we demonstrate that RSL4 binds to RHE motifs in the RXR3 promoter, which results in the accumulation of RXR3 transcript (Figure 5). Zhu et al. (2020) propose an autocrine signalling pathway, which includes RALF1‐FER‐RSL4, to synergistically regulate RH size. We found that RALF1 activates RSL4, indicating that the accumulation of RXR3 transcript is directed by RSL4 TF (Figure 6). Given that auxin induces RXR3 expression (Figures 2D and S2), we conclude that RXR3 participates in auxin‐dependent and RSL4‐mediated signalling pathways, to manage P‐limitation‐induced RH elongation (Figure 9).

Figure 9

Proposed working model illustrating the different pathways by which RXR1 and RXR3 repress root hair elongation. Under P‐stress, RXR3 is transcriptionally activated by auxin through RSL4 TF, whereas RXR1 participates in the PHR1/PHL1‐dependent pathway. RXR1 interacts with RabD2c GTPase to inhibit RH elongation (Ying et al., 2022). In contrast, RXR3 binds to Ca2+‐activated CaMs to regulate tip‐focused [Ca2+]cyt oscillations and eventually RH growth. Blunt end, negative action; arrow, positive action; solid line, experimentally supported; broken line, hypothetical. [Ca2+]cyt, cytosolic Ca2+​​​​; CaM, calmodulin; RH, root hairs; TF, transcription factor.

Ca2+ is a second messenger that transduces signals by associating with various Ca2+ binding proteins (e.g., CaM and CDPK). Ca2+ binds to EF‐hand motifs of Ca2+ binding proteins to modulate many biological processes in plants (Hepler, 2005; Himschoot et al., 2015; Thor, 2019; Yuan et al., 2017). Accumulating evidence suggests that tip‐focused oscillations of intracellular Ca2+ concentration ([Ca2+]cyt) are critical for sustained RH growth (Bibikova et al., 1997; W. Feng et al., 2018; Konrad et al., 2011; Monshausen et al., 2008). CNGC14 was reported to stabilize the integrity of growing RHs, support cell expansion and interact with CaM7 to inhibit RH tip growth (Brost et al., 2019; Zeb et al., 2020;  Zhang et al., 2017). CaMBDs were identified in most Arabidopsis DUF506 members (S. Ying, 2021). Protein microarray analysis demonstrated that DUF506 protein At1g77145 binds to several CaM or CML proteins (Popescu et al., 2007). Here, BiFC and CaM‐binding assays revealed that RXR3 is also a CaM‐interacting DUF506 protein (Figure 7). Overexpression of RXR3 (CaMBD‐truncated mutant) in rxr3‐2 did not restore its long RH phenotype (Figure S14), indicating that the CaMBD is critical for inhibiting RH growth. Moreover, the association between CaM7 and CNGC14 was not noticeably affected by RXR3 (Figure 8A,B). The long RH phenotype of cam7 was not reversed by constitutively overexpressing the RXR3 gene (Figure S12), demonstrating that CaM7 is required for the inhibitory role of RXR3 on RH growth. It is notable that the transcript and protein level of RXR3 were significantly induced by P‐stress (Figures 2 and 3G). Thus, whether the P‐limitation‐induced accumulation of RXR3 protein affects the CaM–CNGC14 interaction requires further investigation.

In addition to CNGC14, other CNGCs are important regulators of RH or pollen tube tip growth (Brost et al., 2019; Chang et al., 2007; Gao et al., 2016; Pan et al., 2019; Tan et al., 2020; Tunc‐Ozdemir et al., 2013; Zhang et al., 2017). Some CNGCs spontaneously form heteromeric complexes, such as CNGC9/14 and CNGC2/4, to conduct their regulatory roles (Brost et al., 2019; Chin et al., 2013; Yu et al., 2019). In contrast to CaM7‐arrested CNGC14, the activity of CNGCs is positively regulated by CaM binding (e.g., CaM1 to CNGC12) (Dietrich et al., 2020; Jarratt‐Barnham et al., 2021). Such studies shed light on the diverse regulatory mechanisms of CNGC. From the current work and previous reports, at least two of DUF506 proteins (i.e., At1g77145 and RXR3) interact with CaMs. Future studies will investigate how these interactions affect CNGC activity or intrinsic Ca2+ dynamics.

To assess the influence of RXR3 on RH tip‐focused Ca2+ dynamics, the GCaMP3 construct was introduced into two individual rxr3 mutant lines, and the [Ca2+]cyt oscillations in RH apex were recorded (Figures 8C–E, S10 and S11 and Supporting Information Video S1). The altered Ca2+ oscillations in tips of growing RHs of rxr3 mutant, such as dampened amplitude and elevated frequency, indicated that RXR3 disturbed RH tip‐focused [Ca2+]cyt gradients. The RH phenotype and tip‐focused Ca2+ oscillations of rxr3 mutants are opposite to those of the eru mutant. The ERU gene encodes an RH‐specific plasma membrane‐localized receptor‐like kinase and is regulated by ARF7 and ARF19 (Schoenaers et al., 2018). A recent study reveals that ERU regulates RH growth through RH tip [Ca2+]cyt oscillations (Kwon et al., 2018). Our current findings support the notion that RXR3 negatively regulates RH growth, albeit the mechanisms affecting RH tip Ca2+ oscillations are still unclear. Note that GCaMP3 is a nonratiometric Ca2+sensor and is therefore less suitable to derive relative differences (e.g., amplitude) in [Ca2+]cyt. Alternative ratiometric Ca2+ biosensors, such as the Yellow Cameleon (YC) 3.6, could present a more accurate picture of the differences between Ca2+ oscillations in RH tips of wild type and rxr3. It is also notable that the frequency of [Ca2+]cyt oscillation in RH tips of rxr1 is higher than the one of wild type (Ying et al., 2022). Therefore, the possibility that RabD2c and RXR3 interact cannot be excluded, because of the omnipresence of the conserved domain 3 in the DUF506 family. On the other hand, our results substantiate previous findings that elements outside the conserved domain of DUF506 proteins are essential for their regulatory functions (Ying et al., 2022).

In summary, this study supports a model in which RXR3 has inhibitory functions that are essential for the maintenance of P‐limitation‐induced RH growth (Figure 9). It remains to be determined whether and/or how RXR3 connects with the other tip growth‐focused CNGCs. Taken together with previous studies, our data suggest that RXR1 and RXR3 function as novel components of the P‐limitation‐inducible RH growth regulatory network.

CONFLICTS OF INTEREST

The authors declare no conflicts of interest.

Supporting information.

Click here for additional data file.

Supporting information.

Click here for additional data file.