Modulation of Phospholipid Signaling by GLABRA2 in Root-Hair Pattern Formation

See allHide authors and affiliations

Science  30 May 2003:
Vol. 300, Issue 5624, pp. 1427-1430
DOI: 10.1126/science.1083695


The root-hair pattern of Arabidopsis is determined through a regulatory circuit composed of transcription factor genes. The homeobox gene GLABRA2 (GL2) has been considered a key component, acting farthest downstream in this regulation. GL2 modified to include a transactivating function caused epidermal cells to develop ectopic root hairs or root hair–like structures. With this system, the phospholipase Dζ1 gene (AtPLDζ1) was identified as a direct target of GL2. Inducible expression of AtPLDζ1 promoted ectopic root-hair initiation. We conclude that GL2 exerts its regulatory effect on root-hair development through modulation of phospholipid signaling.

Cell morphogenesis and its disposition pattern are crucial for plants to ensure functional tissue and organ structures, particularly because the relative positions of plant cells do not change after proliferation. Root-hair formation provides a simple model system that allows exploration of the mechanisms that regulate both pattern formation and morphogenesis (1, 2).

In Arabidopsis, the root epidermis is composed of two types of cell files, only one of which produces root hairs (Fig. 1A) (3). Each hair cell file makes contact with two adjacent, underlying, cortical cell files, whereas each hairless cell file makes contact with a single cortical cell file. Mutations in the GLABRA2 (GL2) (4, 5), WEREWOLF (WER) (6), and TRANSPARENT TESTA GLABRA 1 (TTG1) (7) genes cause all the cell files in the root epidermis to develop root hairs. Conversely, mutations in CAPRICE (CPC) (8) reduce the frequency of root hairs in hair cell files. GL2, which encodes a homeodomain protein (9), is expressed predominantly in hairless cell files and is thought to be a negative regulator of root-hair development (4, 5). Because the level of GL2 transcription is decreased in ttg1 mutant plants (4), TTG1 has been suggested to regulate GL2 positively. WER and CPC, which encode Myb transcription factors (6, 8) and cross-regulate each other in a feedback loop (10), positively and negatively regulate the position-specific expression of GL2, respectively (6, 10). Although such a transcriptional regulatory circuit may be involved in determining root-hair pattern formation, subsequent events remain obscure. We have analyzed these downstream events by identifying genes affected by GL2, the transcription factor farthest downstream in the known feedback loop.

Fig. 1.

Phenotypes of transgenic plants expressing VP16-GL2ΔN. (A) Root hairs of a wild-type plant. (B to F) Transgenic plants carrying the GL2 promoter–driven VP16-GL2ΔN gene. (B) Root hairs. (C) Root-hair bulges. [(D) to (F)] Root hair–like structures on the surface of [(D) and (E)] hypocotyls and (F) the hypocotyl/cotyledon junction. (G to I) The leaf surfaces of (H) DEX-untreated and [(G) and (I)] DEX-treated transgenic plants carrying the GVG-inducible VP16-GL2ΔN gene. Scanning electron microscope images are shown in (E), (H), and (I). Scale bars, 100 μm [(A) to (C), (E), (H), and (I)], 1 mm [(D), (F), and (G)].

To increase the expression of GL2 target genes, we converted GL2 into a strong transcriptional activator by replacing its N-terminal acidic region with the transactivating domain of the herpes viral protein VP16. The modified GL2 gene (VP16-GL2ΔN) was expressed in transgenic plants under the control of the GL2 promoter or of the glucocorticoid-inducible gene expression system GVG (11). In 16 of 21 independent transgenic lines carrying the GL2 promoter–driven transgene, all the cell files in the root epidermis developed root hairs (Fig. 1, B and C), and hypocotyl epidermal cell files composed of relatively elongated cells developed root hair–like structures (Fig. 1, D and E). In plants with severe phenotypes, root hair–like structures were also observed on the abaxial surface of the hypocotyl/cotyledon junction (Fig. 1F). The cells that underwent these morphological changes were some of those with GL2 promoter activity (5, 12). Transgenic plants carrying the GVG-inducible VP16-GL2ΔN gene grew normally in the absence of dexamethasone (DEX), a glucocorticoid derivative (Fig. 1H). After their leaves were soaked in DEX solution, a population of leaf epidermal cells changed shape and developed protruding root hair–like structures in 10 of 13 independent lines obtained (Fig. 1, G and I).

We histochemically examined these transgenic plants for promoter activity of the expansin 7 gene (At-EXP7), which is expressed specifically in root-hair cells (13), using a β-glucuronidase (GUS) reporter gene. At-EXP7 promoter activity was found in cells that bore ectopic root hairs and root hair–like structures (fig. S1), suggesting that those cells had undergone the same developmental processes as root-hair cells. Neither root hair–like structures nor ectopic At-EXP7 promoter activity was observed when a gene that encoded the authentic GL2 protein was inducibly expressed by the GVG system (14).

We conducted a subtraction experiment using cDNAs prepared from DEX-treated plants that carried the GVG-inducible VP16-GL2ΔN and the GVG-inducible GL2 as tester and driver, respectively. In this experiment, we expected to detect GL2 target genes promoting root-hair development, because VP16-GL2ΔN and GL2 were thought to promote and repress root-hair development, respectively. From expression patterns of cloned subtracted cDNAs, we identified a clone that was specifically expressed in DEX-treated plants that carried the inducible VP16-GL2ΔN (Fig. 2A). The clone encoded a phospholipase D (PLD) named AtPLDζ1 (15).

Fig. 2.

GL2-target-gene analyses of AtPLDζ1. (A) RNA fractions prepared from DEX-untreated (–) or DEX-treated (+) transgenic plants that carried the GVG-inducible VP16-GL2ΔN or GL2 genes were subjected to Northern analysis with a probe specific for AtPLDζ1. The same RNA fractions were electrophoresed on an agarose gel and stained with ethidium bromide. (B) A gel mobility shift analysis was performed, with the truncated GL2 protein (amino acids 1 to 235) that contained the HD-ZIP domain and upstream DNA fragments of AtPLDζ1. The DNA regions used were probes 1 (–1079 to –855), 2 (–878 to –654), 3 (–653 to –351), 4 (–350 to –141), and 5 (–140 to –1); the negative numbers in parentheses indicate the number of base pairs upstream from the AtPLDζ1 initiation codon. Each DNA fragment (0.1 pmol) was end-labeled, incubated without (–) or with (+) the truncated protein (1.8 pmol), and fractionated on a polyacrylamide gel. (C) DNase I footprints of the top and bottom strands of probe 3 were generated with the truncated GL2 protein. The amount of protein added to the DNA fragment (0.02 pmol) was 0.0, 0.4, 1.8, and 5.4 pmol for lanes 1, 2, 3, and 4, respectively. Lanes A, G, C, and T are reference sequence ladders. The regions protected from DNase I are indicated by bars. (D) Protected regions in the DNase I footprint analysis are indicated by bars along the double-stranded sequences. The shaded and open boxes indicate a pseudopalindromic sequence and a sequence similar to the L1-box [5'-TAAATG(C/T)A-3'] (16), respectively.

We examined the physical interaction between the AtPLDζ1 promoter and the GL2 DNA-binding domain in vitro. In a gel mobility shift analysis, a truncated GL2 protein containing the homeodomain-leucine-zipper (HD-ZIP) region was specifically bound to a 303-base pair (bp)–long DNA fragment, 350 bp upstream from the AtPLDζ1 initiation codon (Fig. 2B). Deoxyribonuclease I (DNase I) footprinting analysis revealed that the protein protected an 18-bp sequence in the fragment from DNase I (Fig. 2C). This sequence has two noticeable features (Fig. 2D): a pseudopalindromic structure and a sequence similar to that of the L1-box (16). The existence of the pseudopalindromic structure is consistent with the fact that the leucine-zipper of GL2 can mediate homodimerization (4). The L1-box is reported to be bound to ATML1, a homeodomain protein belonging to the HD-ZIP IV subfamily along with GL2 (16). These results indicate that GL2 recognizes promoter sequences of, and hence directly regulates, the AtPLDζ1 gene.

To obtain in vivo evidence that GL2 regulates AtPLDζ1 in root epidermal cells, we examined the promoter activity of an AtPLDζ1 upstream region that encompassed the GL2-binding sequence with a GUS histochemical analysis. Transgenic plants carrying the reporter gene showed GUS activity in various loci, including meristematic and vascular tissues (14). In developing wild-type root epidermis, GUS activity was higher in hair cell files than in hairless cell files (Fig. 3A). On the other hand, under the gl2 mutant genetic background, both cell files showed the same level of GUS activity (Fig. 3B). A mutant reporter gene, of which the promoter contains substitutions at four base pairs in the sequence bound to the GL2 DNA-binding domain, conferred the same level of GUS activity to both cell files in wild-type root epidermis (Fig. 3C). These expression patterns indicated that GL2 represses the target gene in the development of hairless cells through the recognition sequence.

Fig. 3.

Analyses of AtPLDζ1 expression in transgenic plants. (A to C) Activity of the [(A) and (B)] wild-type and (C) mutant AtPLDζ1 promoters in the root epidermis was analyzed histochemically with a GUS reporter gene under the [(A) and (C)] wild-type and (B) gl2-5 mutant (21) genetic backgrounds. The arrowheads in (A) indicate root-hair cell files. (D to G) The intracellular localization of AtPLDζ1-GFP fusion protein, which was expressed under the control of the AtPLDζ1 promoter. [(D), (E), and (G)] Fluorescence from the fusion protein was observed with confocal laser scanning microscopy. The arrowheads indicate relatively strong fluorescence signals around (D) bulges and (G) the root-hair apex. (F) A transparent light image of the root in (D). Scale bars: 100 μm [(A) to (D) and (F)], 10 μm [(E) and (G)].

To investigate the function of AtPLDζ1 in root-hair development, we examined its tentative intracellular localization using a fusion protein of AtPLDζ1 and green fluorescent protein (GFP). In the transgenic root epidermis, fluorescence was observed mainly in the cell cortical region and showed a pattern suggesting the protein was localized in vesicles (Fig. 3, D to G). At early stages of root-hair development, the fusion protein was localized preferentially around bulges (Fig. 3, D and E). During growth of root hairs, relatively strong fluorescence was often observed around the root-hair apex (Fig. 3G). Next, we inducibly expressed AtPLDζ1 using the GVG system. In the absence of DEX, the transgenic plants grew normally and had the same root surface structure as wild-type plants (Fig. 4B). After soaking the root surface in DEX solution, various abnormalities in root-hair development were observed in all nine transgenic lines obtained. In hair cells at stages after bulge formation, root hairs were frequently branched and swollen (Fig. 4, A, D, and E). At early stages, the DEX treatment induced bulges in all cell files (Fig. 4C) and sometimes numerous bulges on a single cell (Fig. 4F). These abnormalities were not observed in DEX-treated transgenic plants that carried a GVG-inducible luciferase gene (14). Thus AtPLDζ1 is likely involved in both initiation and maintenance of root-hair morphogenesis. Branched root hairs with shapes similar to that in Fig. 4E were frequently observed in gl2 mutant roots (Fig. 4, G and H); 6 days after being sown on vertically standing agar medium, 16 of 20 mutant seedlings showed the phenotype. However, none of 20 wild-type seedlings did so under the same conditions, suggesting that AtPLDζ1 is negatively regulated by GL2 in hair cells as well as hairless cells. This is consistent with the fact that GL2 is expressed at a low level in hair cells (10).

Fig. 4.

Abnormalities in root hairs caused by increased and decreased levels of AtPLDζ1 expression, a gl2-mutant defect, and inhibition of PLD activity. (A to F) Root hairs of (B) DEX-untreated and [(A) and (C) to (F)] DEX-treated transgenic plants carrying the GVG-inducible AtPLDζ1 gene. [(A), (D), and (E)] Branched root hairs. [(C) and (F)] Root-hair bulges. Arrowheads in (B) and (C) indicate root-hair bulges. (G and H) Root hairs of gl2-5. Arrowheads in (G) indicate branched root hairs. (I to L) Root hairs of (J) E2-untreated and [(I), (K), and (L)] E2-treated transgenic plants that XVE-inducibly express RNA interference for AtPLDζ1. (I) A globularly expanded root hair. [(K) nd (L)] Root hairs developing from random positions. Arrowheads in [(J) to (L)] indicate the distal ends of root-hair cells. (M and N) Wild-type Arabidopsis germinating on the agar medium that contains (M) 2-butanol or (N) 1-butanol. Scale bars, 10 μm [(A) to (F) and (H) to (L)], 100 μm [(G), (M), and (N)].

To analyze the effects of decreased AtPLDζ1 function, we inducibly expressed double-stranded RNA that interferes with the AtPLDζ1 expression in transgenic Arabidopsis, using the XVE system, an estrogen-inducible transcription system (17). When the transgenic plants were treated with an estrogen, 17β-estradiol (E2), both transgenic lines that we examined showed decreased AtPLDζ1 transcript levels (fig. S2). The lines had normally developing root hairs, as in wild-type plants, in the absence of E2 (Fig. 4J). In the presence of E2, however, their root hairs developed from random positions on the outer surface of hair cells (Fig. 4, K and L), not from the normal position (i.e., near the distal end) and were often expanded and globular (Fig. 4I). These abnormalities were not observed in plants that inducibly expressed AtPLDζ1 or in E2-treated transgenic plants that carried the XVE system only (14). Although complete loss of AtPLDζ1 function was expected to inhibit root-hair initiation, the AtPLDζ1 gene was not completely repressed in this system (fig. S2). However, partial repression of AtPLDζ1 decreased the specificity of tip-growth position and activity for tip-growth maintenance.

Another experiment examined the involvement of PLD activity in root-hair development with 1-butanol as a specific inhibitor of PLD and 2-butanol as a negative control (18). On an agar medium containing 0.06% 2-butanol, wild-type plants grew normally and developed root hairs around the transit region just after germination (Fig. 4M). In the presence of the same concentration of 1-butanol, germination occurred, but epidermal cells around the transit region did not develop root hairs (Fig. 4N). This result suggests the involvement of PLD function in root-hair development. Root elongation and leaf development were inhibited by 1-butanol (14), possibly because PLD activity is required for normal cell proliferation.

Thus, expression of GL2 target genes promoted root-hair development. We conclude that GL2 regulates root-hair pattern formation by repressing genes for root-hair development in hairless cells (fig. S3). Because PLD, the product of a GL2 target gene, affects cell morphogenesis (19, 20), GL2 very likely acts as an intermediary between the transcriptional regulatory circuit that involves WER, CPC, and GL2 and the morphogenic effects of PLD (fig. S3). PLD hydrolyzes phosphatidylcholine to generate phosphatidic acid and choline (19, 20). Arabidopsis encodes two types of PLDs: a C2 domain–containing type that is unique to plants and a PX-PH domain–containing type generally found in eukaryotes (15). AtPLDζ1 is of the latter type. PLD in animal and yeast cells regulates membrane-trafficking events to and from the plasma membrane (19, 20). By analogy, AtPLDζ1 might regulate vesicle trafficking and exocytosis in root-hair tip growth. The localization pattern of the GFP-fusion protein supports this postulate. As another possibility, AtPLDζ1 might also regulate root-hair morphogenesis through the translocation of the membrane proteins involved in signal transduction, including those for phytohormones.

Supporting Online Material

Materials and Methods

Figs. S1 to S3

References and Notes

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article