FERONIA receptor-like kinase regulates RHO GTPase signaling of root hair development

FER RLK Is a Broadly Expressed ROPGEF-Interacting RLK.

Using ROPGEF1 as bait in a yeast two-hybrid screen, we identified members of the Arabidopsis CrRLK family as potential interacting proteins for ROPGEFs (Fig. 1A and Fig. S1 B and C). Focusing on one of these, At3g51550, we used bimolecular fluorescence complementation (BiFC) assays (36) in protoplasts to confirm that this RLK also interacted with ROPGEF1 in plant cells (Fig. 1B). Given the importance of RAC/ROP signaling to polarized cell growth (15–18) and that root hairs are more accessible than pollen tubes for observation, we initially screened T-DNA insertion mutants in these RLKs for root hair defects (see SI Materials and Methods) and found that mutations in At3g51550, which turned out to be identical to FER when its sequence was elucidated (33), uniquely induced severe root hair defect (see below). Hence we adopt the FER nomenclature and the family as FER related. We also included in our studies siréne (srn), whose mutation is allelic with fer, the other originally described mutations, and inducing similar female gametophytic defects as fer (31–33).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

FER is a broadly expressed ROPGEF-interacting RLK. (A) The kinase domain of FER [FER(K)] and ROPGEF1 (GEF1) interact in yeast two-hybrid assays, complementing histidine deficiency (First and Third rows). FER(K) interacts with several other GEFs and ROPGEF1 interacts with RAC/ROP (Fig. S1 B and C) (19, 20). (B) FER and ROPGEF1 interact in BiFC assays in protoplasts. YFP image for the interacting YN–EE–GEF1 and FER–HA–YC was acquired by autoexposure (Top); the same exposure condition was used for control protoplasts expressing one of the partners (Middle and Bottom). (Scale bar, 10 μm.) (C) pFER::GUS expression pattern in 10-d-old seedlings and roots and leaf from 10-d-old plants. Arrowheads, root hairs. [Scale bars, 1 mm (seedling and inflorescence), 50 μm (root and stigma), and 500 μm (leaf).] (D) FER–GFP localization in pFER::FER-GFP transformed Arabidopsis. [Scale bars, 100 μm (root), 50 μm (trichome), and 10 μm (root, leaf epidermis, and root hairs).]

FER promoter (pFER)::GUS analysis in transgenic Arabidopsis indicates that in addition to expression in the ovule (33), FER is broadly expressed (Fig. 1C) as previously shown by transcript analyses (33, 37). pFER expressed FER–GFP was localized to the plasma membrane of diverse cell types, including emerging and growing root hairs (Fig. 1D).

FER Is Essential for Root Hair Development.

We identified two T-DNA–induced alleles, fer-4 and fer-5 (Fig. 2A and Fig. S2A). The originally described fer and srn are both in the vicinity of fer-4 (33) (Fig. 2A). RT-PCR analysis showed that fer-4, with its T-DNA insertion in the extracellular domain-coding region, is a null mutant, whereas fer-5 harbors truncated transcripts (Fig. 2B), consistent with its T-DNA insert being closed to the end of its kinase domain-coding region of the gene. fer-4 and srn showed indistinguishable and multiple phenotypes, whereas fer-5’s defects were milder and more restricted (see below).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Mutations in FER induce root hair and other vegetative phenotypes. (A) Domain map for FER. Locations of the T-DNA insertion alleles fer-4 (GABI_GK106A06) and fer-5 (Salk_ 029056c), fer, and srn (31–33) and primers used in PCR analyses are indicated. SP, signal peptide; TM, transmembrane domain; UTR, untranslated region. Mature FER is ∼807 aa long. Genomic DNA analysis for fer-4 and fer-5 is shown in Fig. S2A. (B) RT-PCR analysis for FER expression. FER(K), PCR used primers 3 and 4; FER(Ex), PCR used primers 1 and 2. (C) Root hairs from 4-d-old wild-type (WT), fer-4, fer-5 and srn, and pFER::FER-GFP complemented fer-4, -5 seedlings. (Scale bar, 100 μm.) See Fig. S2B for FER–GFP expression in the complemented fer seedlings. (D) fer mutations induce a broad spectrum of root hair defects. Top shows normal root hairs (Left) and fer mutation-induced collapsed (the most severe), burst [seen with discharged cytoplasm (arrowhead)], arrested, truncated, split, and short (the least severe) root hairs. (Scale bar, 100 μm.) The histogram shows distribution of the different classes of root hairs in WT and mutant seedlings. Each data bar represents the mean ± SD where n = 1,200 root hairs from 15–24 four-day-old seedlings. Δs highlight normal root hairs from each sample. (E) fer-induced growth suppression. (Scale bars, 10 mm (25 d) and 50 mm (45 d.) (F) fer-4 induces trichome defects. Arrows, collapsed; arrowheads, more than three branches. Approximately 30% of trichomes in fer-4 seedlings were normal compared with >90% in WT (Fig. S2C). (G) fer-4 and srn show similar reproductive phenotype. (Scale bar, 2 mm.)

Consistent with the high expression level of FER in root hairs (37) (Fig. 1C), fer-4, fer-5, and srn showed readily observable root hair defects that ranged from collapsed, burst, to short (Fig. 2 C and D). Although the level of normal root hairs was similarly low, the majority of fer-4 and srn root hairs collapsed shortly after emergence, whereas many of the fer-5 root hairs were shorter than wild type (WT) or burst after some growth was achieved (Fig. 2 C and D). fer-5 was negligibly affected in growth and development, whereas fer-4 showed significantly reduced stature, developed severe trichome defects (Fig. 2 E and F and Fig. S2C), and reproductively mimicked fer and srn (Fig.2G). The root hair defects in fer-4, -5 were complemented by pFER::FER-GFP (Fig. 2C and Fig. S2B), as were other growth and reproduction phenotypes in fer-4 (e.g., Fig. 2G). Together these observations imply that FER’s role in growth and development is considerably broader than regulating female gametophytic functions (31–33). Although kinase-deleted and extracellular domain-located mutations in some RLKs act in a dominant negative manner (38, 39), the fact that the T-DNA insert is at the end of the kinase domain and that heterozygous fer-5 plants were nondistinguishable from wild-type plants suggest fer-5 is more likely to have lost some aspects of FER’s multiple functions that rely on an intact C-terminal region, whereas the null mutants fer-4, srn, and fer have lost all of FER’s functions.

FER Functions in a RAC/ROP Signaling Pathway and Mediates Auxin-Regulated Root Hair Development.

It is well established that auxin plays an important role in regulating root hair development (e.g., refs. 40, 41) and we showed earlier that RAC/ROPs mediate auxin-signaled gene derepression via 26S proteasome-mediated repressor proteolysis (9, 10). We therefore asked whether FER functions in a RAC/ROP signaling pathway and also acts as a regulator for auxin-regulated root hair development. We observed that when grown in the presence of auxin, differences between WT and fer root hairs were further amplified because root hair elongation in WT seedlings was stimulated by auxin, whereas fer-4, fer-5, and srn root hairs failed to respond and remained short (Fig. 3 A–C and Fig. S2D). pFER::FER-GFP complemented this phenotype, restoring auxin-stimulated root hair growth in fer seedlings (Fig. 3 A and B).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

fer mutants show reduced auxin and RAC/ROP signaling capacity. (A) Four-day-old WT, mutant, and pFER::FER-GFP complemented mutant seedlings grown under standard (0 nM NAA) or auxin-supplemented conditions. Auxin-treated WT and fer-4, -5 mutants are shown magnified on the Right. (Scale bar, 2 mm.) (B) Root hair length comparison between WT, fer-5, and complemented fer-5 under standard and auxin-supplement growth conditions. Data bars represent average of the mean root hair lengths from triplicate samples ± SE. For each replicate, 300 root hairs from at least nine plants were measured. (C) Root hair length comparison between WT and fer-4 (Fig. S2D shows data for srn). (D and E) Overexpression of GFP–ROP2 rescued fer-5–induced root hair morphological defects (D), restored root hair elongation (E, compare white and black bars), and auxin-stimulated root hair elongation (E, compare 0, 50, and 100 nM NAA data bars). See Fig. S3 for genotyping of these rescued plants. (Scale bar, 100 μm.) (F) Pulldown of active RAC/ROPs. ICR1–MBP was used for fer-4; PBD–GST (8, 9) was used for fer-5. Anti-NtRac1 antibody (9) was used for detection. The fer-4 blot was visualized by chemiluminescence, the fer-5 blot by alkaline phosphatase reaction. Quantification of data is shown in Fig. S2E. (G–J) ROP2–MBP pulldown of pFER-expressed FER–GFP from microsomal proteins from transformed seedlings (G), of FER–HA expressed in transiently transfected mesophyll (H) and root (I) protoplasts, and of GEF4–HA expressed in mesophyll protoplasts (J). Lower panels show input bait proteins on Ponçeau-stained blots. Anti-GFP (G) and anti-HA (H–J) were used for detection. *band in G was unrelated to the experiment. (K) Sketch showing guanine nucleotide-regulated FER–RAC/ROP interaction as suggested by results shown in G–J. Presumably, in vivo, signal activation-induced changes in the interacting ROPGEF and RAC/ROP would weaken the GEF–RAC/ROP interaction, recycling the ROPGEF, and releasing activated RAC/ROP to interact with effectors for downstream signaling. Data for WT, fer-5, fer5 + FER–GFP and fer5 + GFP–ROP2 shown in B and E were collected in the same experiments thus the data for WT and fer-5 are used in both panels. Mutant root hair data are overestimations because only hairs with measurable lengths were included in the data set. Brackets and the numbers 1 and 2 in C denote statistical comparisons (1, significant; 2, insignificant) as described in Materials and Methods; these differences are representative for data shown in B and E.

To establish that FER indeed acted upstream of RAC/ROPs, we examined whether increasing RAC/ROP signaling capacity by overexpressing GFP–ROP2 may counteract the root hair defects in fer-4 and fer-5. Indeed fer-5 root hairs were restored to normal, including ability to respond to auxin stimulation (Fig. 3 D and E and see also Fig. S3), consistent with reduced RAC/ROP signaling as underlying the fer-induced root hair defects under normal growth conditions and decreased sensitivity to auxin stimulation of root hair growth. However, a rescued fer-4 was not recovered. Possibly, the higher level of RAC/ROP overexpression required to counteract fer-4’s deficiency could have prevented transgenic plant regeneration. Alternatively, and more pertinent to FER’s multiple functionality as suggested by the multiple phenotypes induced by fer-4 and other null or knockdown mutations (29–33), it is possible that fer-5 is defective only in FER’s RAC/ROP-interacting capacity, whereas fer-4 lacks all of FER’s functions, including those potentially requiring an intact extracellular domain.

To substantiate the FER and RAC/ROP signaling linkage on a biochemical level, we used a functional pulldown assay that specifically targets activated RAC/ROPs (5, 8, 9). Indeed, the level of activated RAC/ROPs was reduced in fer-4, -5 seedlings relative to WT (Fig. 3F and Fig. S2E) consistent with FER being necessary for RAC/ROP activation thus working as an upstream regulator for these small GTPases. Moreover, ROP2 maltose-binding protein (MBP) efficiently pulled down pFER-expressed FER–GFP from seedling microsomal proteins or when FER–HA was overexpressed in mesophyll or root protoplasts (Fig. 3 G–I), suggesting complex formation via their mutual interaction with ROPGEFs, several of which are known to interact with FER (Fig. S1B).

It is known that most ROPGEFs preferentially bind to GDP-bound RAC/ROPs versus the GTP-bound activated form (19, 20), presumably upon nucleotide exchange stimulated by upstream signaling, the activated GTPases would be released to interact with downstream effectors. Therefore the presence of a signaling complex involving an upstream regulator, a ROPGEF and a RAC/ROP should exist but would likely be transient in vivo (see Fig. 3K). To substantiate that the observed pulldown of FER by RAC/ROPs (Fig. 3 G and H Left) was functionally relevant, we included either GDP or GTP in the pulldown reaction to determine whether these guanine nucleotides differentially affect the pulldown efficiency. GDP and GTP would shift RAC/ROPs to the inactive or activated state, respectively, mimicking the action of these guanine nucleotides under physiological conditions. Indeed, pulldown of FER–HA was substantially enhanced by GDP but reduced by GTP (Fig. 3 H Right and I), consistent with the observation that most ROPGEFs (e.g., ROPGEF4, Fig. 3J) are preferentially pulled down by GDP-bound RAC/ROPs but not as efficiently by the GTP-bound form (19, 20). Together with observations that ROPGEFs interact with FER (Fig. 1 A and B and Fig. S1B) and with RAC/ROPs (19, 20; Fig. S1C), results from these pulldown assays are consistent with a dynamic signaling complex involving interactions between FER, a ROPGEF, and a RAC/ROP, mediating nucleotide exchange and activating the small GTPases (Fig. 3J).

FER Regulates RAC/ROP-Mediated NADPH Oxidase-Dependent ROS Accumulation in Root and Root Hairs.

The root hair defects in fer mutants closely resemble those seen in rhd2 (Fig. S4). Together with the FER, RAC/ROP signaling linkage (Fig. 3) and the RAC/ROP linkage to NADPH oxidase-dependent ROS-mediated root hair development (15–18), it seemed highly plausible that FER acted upstream of the RAC/ROP-regulated pathway. We therefore examined whether fer mutants were compromised in RAC/ROP-regulated ROS accumulation.

Roots and root hairs are known to show auxin-regulated diphenylene iodonium (DPI)-sensitive NADPH oxidase-dependent ROS accumulation and growth (e.g., refs. 13, 16, 17, 41, 42; Fig. 4 A and B; Fig. S5A). When treated with H₂DCF–DA, a commonly used ROS-detection stain, fer root and root hairs showed that they accumulated significantly lower levels of ROS relative to WT (Fig. 4 C and D–G, 0 nM NAA) and they were also nonresponsive to auxin-stimulated ROS accumulation (Fig. 4 D–G, NAA-treated samples). Like the root hair morphological defects, the reduced ROS phenotype in fer was complemented by pFER::FER-GFP (Fig. 4 H and I) and countered by overexpressing ROP2–GFP (Fig. 5 A and B and see also Fig. S5B and Fig. S6 A and B). The corollary experiment of overexpressing RAC/ROPs, ROPGEF1, or FER also resulted in increased ROS accumulation in transgenic seedling roots (Fig. 5 C and D and Fig. S6 C and D), further supporting that FER, ROPGEFs, and RAC/ROPs act in a common pathway that controls NADPH oxidase-dependent ROS production.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

FER regulates NADPH dependent- and auxin-regulated ROS production in root and root hairs. (A and B) ROS accumulation and auxin-stimulated ROS production in Arabidopsis primary root. DPI inhibition reflects NADPH oxidase-dependent ROS production (see e.g., refs. 16, 17, 41). Seedlings were treated with H₂DCF–DA to monitor ROS levels (see SI Materials and Methods). The rectangle in A indicates a representative region of interest (ROI) where average root ROS intensity was quantified for this and other samples; comparative data are shown in B. All images were acquired using the control (−DPI, −NAA) image acquisition condition. Each data bar in the histogram represents the mean ± SD of ROS intensity measured from nine roots in one representative experiment. All other experiments involving root ROS intensity comparison (D–I; Fig. 5) follow the same sampling and analysis methods. (C) ROS in WT, fer-4 and fer-5 root and root hairs. The WT ROS image was acquired by autoexposure; all other images were acquired using the WT exposure conditions. (Scale bar, 100 μm.) (D–I) ROS in the primary roots of WT, fer-4, -5 (D and E), srn (F and G), and pFER::FER-GFP complemented fer-4, -5 seedlings [H and I; under the ROS imaging conditions, FER–GFP signal was negligible (see Fig. S5B)] in normal and auxin-treated growth conditions. (E, G, and I) ROS level quantification at representative ROIs. (Scale bars, 100 μm.) 1, significant; 2, insignificant differences. Statistical differences shown in G are representative for data shown in E and I, except for fer-5’s weak but significant response to NAA (E, P = 0.002376), reflecting its weaker phenotype.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

FER and RAC/ROPs function in a common pathway regulating ROS production. (A and B) GFP–ROP2 restores ROS accumulation in fer-5 roots and root hairs. Image acquisition and quantitative analyses were as described in Fig. 4. ROS images (A) were acquired using exposure conditions for a H₂DCF–DA-treated WT control under which GFP–ROP2 signal was negligible (fourth data bar in B; see also Fig. S5B). (Scale bar, 200 μm.) Note also the root hairs in the GFP–ROP2-expressing fer-4 seedlings. (B) Quantified comparative ROS levels. Root hair ROS was also quantified using dihydroethidium for detection (54) (Fig. S6 A and B). (C and D) Overexpression of CaMV35S-driven FER–HA (Fig.S6C), ROPGEF1 or NtRAC1(CA) (9) augments ROS production in roots and root hairs. Insets in C show DIC images. Quantified comparative ROS levels are shown in D. These transgenic plants did not show readily noticeable phenotypes other than mild root hair depolarization (e.g., as seen in GEF1ox seedling shown here), reflecting increased Rac/Rop signaling; their root hair lengths were within the range seen in control seedlings. (Scale bar in C, 200 μm.) 1, significant; 2, insignificant differences.