In a screen designed to identify genes in the specification of epidermal cell fate in Arabidopsis primary roots we have isolated 8 new mutants that fall into 6 complementation groups corresponding to the ‘root hairless’ genes RHL1, RHL2 and RHL3 and the ‘ectopic root hair’ genes ERH1, ERH2 and ERH3. The erh2 mutant is allelic to pom1, a conditional root expansion mutant, and reveals a possible link between epidermal root hair initiation and radial cell expansion. Apart from erhl the mutants also show defects in shoot development, indicating a complex role for the affected genes. Mutant phenotypes in the patterning and shape of leaf trichomes in rh11, rhl2, rhl3 and erh3 were particularly obvious. The root hairless mutants are only partly responsive to increased ethylene concentrations, while the ectopic root hair mutants are fully responsive to reduced concentrations of ethylene, a permissive regulator of root hair initiation. This result and the analysis of double mutants suggest a complex pathway leading to root hair initiation that requires the RHL and ERH genes for correct differentiation.

The Arabidopsis primary root has a well defined and predictable structure (Dolan et al., 1993, 1994). Cells derived from rings of root meristem initials divide transversely and grow in organized files along the root axis. When they stop dividing, they continue to elongate and when they finish elongating, they differentiate. The number of cell files in each tissue layer is relatively constant. For example, there are 8 endodermal cell files surrounding the stele tissue. Outside the endodermal cells there are 8 cortical cell files surrounded in turn by about 19 epidermal cell files. The latter are divided into two types. Trichoblasts are those cells that will go on to form hairs, and these overlie the clefts between the underlying 8 cortical cell files. In between are atrichoblasts, or cells that will go on to form non-hair cells and these only contact the outer wall of one cortical cell file.

Trichoblasts can already be distinguished from atrichoblasts in the meristematic zone of wild-type roots by their denser cytoplasm. This simple and predictable pattern of cells provides a good model system to address genetically the way in which cellular pattern is established and the nature of the signalling pathways involved in cell fate specification and cell differentiation.

To understand the pathway of root epidermal cell differentiation at the genetic and molecular level a number of mutants isolated in other screens have already proved useful. Ectopic root hairs develop on the roots of the ttg mutant (Galway et al., 1994) that, in addition, lacks trichomes, anthocyanins and seed coat mucilage (Koorneef, 1981). However, Arabidopsis roots overexpressing the maize R (R-Lc) gene product, a putative TTG homologue (Lloyd et al., 1992), have no root hairs. Since non-hair cells are found in hair cell positions when TTG is constitutively expressed, this supports the idea that TTG is a positive regulator of non-hair cell fate.

Ectopic root hairs have also been found on the gl2 mutant (Masucci et al., 1996). Initially this mutant was identified by its abnormally expanded trichomes and its lack of proper seed coat mucilage (Koorneef, 1981). Expression of the GL2 gene is largely confined to the atrichoblasts in the wild-type root epidermis (Masucci et al., 1996), and the affected gene product, a homeobox containing protein (Rerie et al., 1994), is probably involved in the positive regulation of non-hair cell fate. In ttg mutants GL2 expression is greatly reduced suggesting that in the wild type TTG positively regulates GL2 (Di Cristina et al., 1996).

Another recessive mutant that displays ectopic root hairs is ctrl. The wild-type CTR1 protein was presented by Kieber et al. (1993) as a negative regulator of the ethylene response pathway that showed homology to members of the Raf family of protein kinases. After investigating ctrl roots Dolan et al. (1994) suggested that ethylene is a positive regulator of root hair formation and that the CTR1 protein acts as a negative regulator of the pathway. In subsequent studies Tanimoto et al. (1995) showed that indeed the exogenous application of the ethylene precursor 1-aminocyclopropane-1-carboxylic acid (ACC) induced ectopic hairs, mimicking the ctrl phenotype, whereas root hairs disappeared upon treatment with inhibitors of either ethylene synthesis or its perception.

The first mutant specifically isolated for its defect in root hair initiation was rhd6 (Masucci and Schiefelbein, 1994). The recessive rhd6 mutant displays only 15-17% of the wild-type number of root hairs. However, the wild-type phenotype can be rescued by exogenous application of an ethylene precursor or auxin. In terms of reduced root hair formation rhd6 shows phenotypic similarities to the auxin-, ethylene- and abscisic acid-resistant mutant axr2 which, however, is dominant (Wilson et al., 1990). All the growth factor-related mutants ctrl, rhd6 and axr2 show the early structural characteristics of wild-type trichoblast development indicating that their defects are related to later stages in root hair initiation (Masucci and Schiefelbein, 1996). Moreover ectopic root hairs induced by the ethylene precursor ACC appear independently of early cell differentiation events in the meristematic zone, showing that ethylene is still effective very late in root epidermal development. A preliminary model suggests that TTG and GL2 act in a position-dependent manner upstream of the ethylene-mediated processes (Masucci and Schiefelbein, 1996).

Despite recent progress many more genes remain to be identified, and since few of the known genes have been identified from a root specific screen, we conducted a visual screen of 1500 M2 families to identify mutants that completely lack root hairs or that have root hairs in ectopic positions. We isolated 5 root hairless mutants specifying 3 genes and 3 ectopic root hair mutants specifying a further 3 genes. As most of these new mutants display a pleiotropic phenotype the affected genes are assumed to play a complex regulatory role in plant development. In particular, correlations to leaf trichome development and radial cell expansion in the root have been found. The wild-type gene products identified through the recessive root hairless mutants are assumed to be positive regulators in the pathway leading to root epidermal cell differentiation, whereas the recessive ectopic root hair mutants are assumed to represent negative regulators of root hair initiation or alternatively positive regulators of non-hair cell fate. The partial complementation of the root hairless mutants with increased levels of the ethylene precursor ACC suggests that at a later developmental stage high ACC levels can overrule the block imposed on root hair initiation by the rhl mutations. However, wild-type hair-cell development appears to require functional RHL gene products. This idea is further explored by analysing double mutant plants homozygous for both a root hairless and an ectopic root hair mutation. The results are summarized in a preliminary model for root epidermal cell differentiation.

Plant material and growth conditions

The rhl1 mutant (ws background) was isolated from T-DNA transformed lines (Feldmann, 1991). rhl3-2 and erh1 (Landsberg erecta background) derive from seeds mutagenized with fast neutrons (Lehle seeds) and were isolated from a screen of 500 independent families. rhl2-1, rhl2-2, rhl3-1, erh2 and erh3 were isolated by screening the progeny of 1000 ‘Columbia’ (Col) seeds treated with 0.5% ethyl methanesulfonate (EMS). As all homozygous rhl mutants are lethal the mutated genes have to be maintained in segregating populations. All erh mutants are viable in the homozygous state. The ctr1 allele HE14 was a gift from P. Benfey’s lab, gl2 seeds were provided by J. Schiefelbein, the pom1-1 allele was made available by M.-T. Hauser and the ttg mutant and the Landsberg erecta (Ler) wild type were obtained from the Ohio seed stock centre. Plates for seedling growth including ACC and 2-aminoethoxyvinyl-glycine (AVG) treatments were prepared essentially as described by Tanimoto et al. (1995).

Genetic analysis

Each mutant was outcrossed at least once to its respective wild-type ecotype. A segregation of nearly three wild types to one mutant in the F2 generation indicated in all cases that single genes were affected and that the mutants were recessive. Pairwise crosses between mutant lines were performed to determine complementation groups. For the construction of double mutants between root hairless and ectopic root hair mutants, erh-like F2 plants (homozygous for the mutation causing ectopic root hairs) were allowed to self. Two out of three were expected to be heterozygous for the rhl mutation as well. Therefore F3 families were scored for segregating populations. Among these, approximately 25% of the plants were double mutants.

Mapping

The chromosomal location of the mutant genes was determined by calculating recombination frequencies of SSLP-(Bell and Ecker, 1994) and CAPS-markers (Konieczny and Ausubel, 1993). Each mutant was crossed to a different wild-type ecotype: rhl1, rhl2-1, erh2 and erh3 were crossed to Landsberg erecta; rhl3-2 and erhl were crossed to ‘Columbia’. Mutant F2 plants were used for the analysis. In cases where the homozygous mutant is lethal, heterozygous siblings were used, whose progeny segregate for the mutant phenotype in the F3 generation. Each marker was tested on approximately 30 homozygous or 60 heterozygous plants to determine linkage of the mutant genotype with the marker pattern. DNA was extracted from single seedlings or a few leaves according to a slightly modified Dellaporta protocol (Dellaporta et al., 1983). PCR conditions were essentially as described by Konieczny and Ausubel (1993) and Bell and Ecker (1994). DNA fragments were separated in 1.5% agarose gels or 5% polyacrylamide gels (Hauser et al., 1995).

Cryo-scanning electron microscopy

3-day old roots, young leaves and flower bud tissue were prepared for cryo-scanning electron microscopy essentially as described by Dolan et al. (1994).

Tissue fixation and embedding

3-day old seedlings were vacuum infiltrated with fixative consisting of 2.5% glutaraldehyde in 50 mM sodium cacodylate, pH 7.2. After 1 hour prefixing the seedlings were arranged in groups of 4 or 5 on a thin layer of 1% agarose (low gelling point; Sigma) and covered with the residual agarose at 60°C. The specimens were allowed to cool on ice and were refixed overnight. The infiltration with Historesin (Leica) was performed according to the manufacturer’s instructions. To quantify ectopic root hairs thick transverse serial sections (10 pm) were cut with a Leica Autocut 2055 microtome using a 45° glass knife. Thin sections (2 pm) were examined for structural analysis of the root meristematic region. For photomicroscopy, sections were stained with toluidine blue.

Mutant isolation, root phenotype and genetic characterization of the new root hairless and ectopic root hair mutants

Primary roots of three differently mutagenized M2 populations were scored 3 days after germination for a complete absence of root hairs or a ‘bushy’ appearance, putatively caused by ectopic root hairs. The heritability of putative mutants was examined and they were then outcrossed to their respective wild-type ecotype and crossed to each other to establish complementation groups. All putative ectopic root hair mutants were crossed to ttg, ctrl and gl2, the three existing mutants known to possess root hairs in ectopic positions. Interestingly the screen identified a further allele of both ttg and ctrl which are not referred to here in any detail. Segregation data indicate that the new mutants presented here are caused by mutations in single recessive genes.

The wild-type phenotype and the mutant phenotypes are displayed in Fig. 1. On 3-day old wild-type seedlings root hairs start to emerge from above the elongation zone (Fig. 1A). They appear to be evenly distributed along the primary root with increasing length at later developmental stages. In addition there is an increased concentration of root hairs at the transition zone between root and hypocotyl, the collet region. In transverse sections through the meristematic zone the wild-type trichoblasts, situated over under-lying clefts between cortical cells, are characterized by dense cytoplasm and a low degree of vacuolization compared to the atrichoblasts, with very few exceptions (Fig. 1B). In the wild type, only trichoblasts give rise to root hairs in the hair formation zone (Table 1; Fig. 3 A). A few exceptions to this pattern were observed in the Landsberg erecta ecotype.

Table 1.

Quantification of root hairs in wildtype and mutant backgrounds

Quantification of root hairs in wildtype and mutant backgrounds
Quantification of root hairs in wildtype and mutant backgrounds
Fig. 1.

Phenotypes of 3-day old seedlings and transverse sections through their primary roots in the meristematic zone. (A,B) Wild type, (C,D) rhl1, (E,F) rhl2-2, (G,H) rhl3-1, (I,J) erhl, (K,L) erh2 and (M,N) erh3. The white scale bar in M equals 500 μm and refers to the magnification of all seedlings in this figure. The black scale bar in N equals 20 μm and represents the magnification of all transverse sections in this figure. Arrowheads in N indicate erh3 cells in trichoblast positions that do not stain as wild-type trichoblasts.

Fig. 1.

Phenotypes of 3-day old seedlings and transverse sections through their primary roots in the meristematic zone. (A,B) Wild type, (C,D) rhl1, (E,F) rhl2-2, (G,H) rhl3-1, (I,J) erhl, (K,L) erh2 and (M,N) erh3. The white scale bar in M equals 500 μm and refers to the magnification of all seedlings in this figure. The black scale bar in N equals 20 μm and represents the magnification of all transverse sections in this figure. Arrowheads in N indicate erh3 cells in trichoblast positions that do not stain as wild-type trichoblasts.

Fig. 2.

Scanning electron micrographs of 3-day old primary roots of (A) wild type and (B) rhl3-2. Scale bar equals 30 μm.

Fig. 2.

Scanning electron micrographs of 3-day old primary roots of (A) wild type and (B) rhl3-2. Scale bar equals 30 μm.

Fig. 3.

Transverse sections through root hair zone of primary Arabidopsis roots. (A) Wild type, (B) erhl, (C) erh2 and (D) erh3. Ectopic root hairs, i.e. outgrowths of epidermal cells facing only one cortical cell, are indicated with arrows. The ectopic localization of the root hairs was confirmed by analyzing subsequent sections. Scale bar in D equals 25 μm; all pictures were taken at the same magnification.

Fig. 3.

Transverse sections through root hair zone of primary Arabidopsis roots. (A) Wild type, (B) erhl, (C) erh2 and (D) erh3. Ectopic root hairs, i.e. outgrowths of epidermal cells facing only one cortical cell, are indicated with arrows. The ectopic localization of the root hairs was confirmed by analyzing subsequent sections. Scale bar in D equals 25 μm; all pictures were taken at the same magnification.

rhl1 (Fig. 1C,D) was isolated from a T-DNA mutagenized population (Feldmann, 1991) and we have shown that the mutation is caused by a T-DNA insertion into the gene (K. Schneider, unpublished results). This mutant has few hairs on its primary root and it is noticeable that the collet region is also bald. However, occasionally a few root hairs surround the emergence of lateral roots (data not shown). The primary roots of the rhl1 mutant are shorter than wild-type roots and exhibit abnormal gravitropic responses. The number of cells per tissue layer is the same as the wild type. The density of the cytoplasm and the degree of vacuolization is similar in all epidermal cells of thin sections through the meristematic zone, indicating that the morphological distinction between putative trichoblasts and atrichoblasts is much less pronounced than in the wild type. Thus in rhl1 seedlings root epidermal development is already impaired at the onset of early cell differentiation.

All other root hairless mutants: the EMS mutants rhl2-1, rhl2-2, rhl3-l and the fast neutron mutant rhl3-2 display the same phenotype with rhl3-2 having a maximum of around 10 root hairs per primary root (Fig. 1E,G). As there was no detectable difference between the phenotypes of rhl2-1 and rhl2-2 either of them was used for subsequent studies. Of the two rhl3 alleles rhl3-2 was chosen for subsequent analysis. The outer epidermal cell walls of all rhl-mutants are perfectly smooth and show no sign of root hair initiation or bulge formation at the distal end of the epidermal cell (Fig. 2).

erhl (Fig. 1I) was identified among the fast neutron treated mutants. erhl roots are shorter than wild-type roots and their appearance is very bushy. Serial sectioning along 3-day old seedling roots showed that 56% of hairs are in an ectopic position (Table 1). A typical transverse section through the root hair zone of erhl is shown in Fig. 3B. With 200 root hairs per mm erh1 has also the highest root hair density of all the mutants (Table 1). The general root architecture in erh1 appears to be the same as in the wild type, with 8 endodermal and 8 cortical cell files, although all cells seem to be radially expanded generating a root diameter that is 50% larger than in the wild type (Fig. 3A,B). In the meristematic zone some cells overlying the cleft between two cortical cells do not contain dense cytoplasm, on the other hand this feature is displayed by erh1 cells in non-hair positions indicating developmental aberrations in erh1 at very early stages (Fig. 1J). erh2 is an EMS mutant with 35% of its root hairs in ectopic positions (Figs 1K, 3C; Table 1). In the meristematic zone, erh2 roots show the wild-type pattern with 8 endodermal and 8 cortical cell files and 8 densely cytoplasmic trichoblast files in the correct positions (Fig. 1L). Radial cell expansion is already visible, but becomes more apparent in the hair zone (Fig. 3C). erh2 cortical cells have a radial dimension that is 2.5-times that of the wild type. This increased radial cell expansion, and the map position on top of chromosome I (Fig. 4), eventually suggested allelism to pom1 and reciprocal crosses proved that indeed erh2 is an allele of pom1. According to accepted procedures of Arabidopsis nomenclature erh2 was renamed and should be referred to as pom1-l2 in future. erh3 (Fig. 1M), derived from the EMS mutant screen, has 32% of its root hairs in an ectopic position and the lowest ectopic and total root hair density of the erh mutants (Fig. 3D; Table 1). erh3 frequently diverges from the constant wild-type primary root anatomy. In a sample of 12 individual roots, 3 specimens had only 7 cortical cell files (compare Fig. 3D), 5 roots possessed 9 cortical cell files and only 4 specimens displayed the wild-type pattern of 8 cortical cell files. Cytoplasmic density and the degree of vacuolization do not strictly correlate with the position of the epidermal cell over one or two cortical cells (Fig. 1N) indicating that the erh3 mutation causes defects at early stages of root epidermal cell differentiation. All cell layers of erh3 roots, and in particular the endodermis, appear to be radially expanded from the elongation zone onwards, leading to a root diameter nearly twice that of the wild type in the root hair zone (Fig. 3A,D).

Fig. 4.

Approximate chromosomal map positions of loci involved in Arabidopsis root hair initiation.

Fig. 4.

Approximate chromosomal map positions of loci involved in Arabidopsis root hair initiation.

All mutants have been mapped using a set of molecular markers (see Materials and Methods). Recombination frequencies were as follows: for rhl1 11.8% with GAPB (CAPS) and 25% with Athgenea (SSLP); for rhl2 5% with Athctr1 (SSLP) and 24% with nga225 (SSLP); for rhl3 5% with GL1 (CAPS) and 21% with AthCHIB (SSLP); for erhl 2.3% with nga76 (SSLP) and 10% with nga129 (SSLP); for erh2 5.9% with PVV4 (CAPS) and 7.6% with nga59 (SSLP) and for erh3 3% with ADH (CAPS) and 11% with nga111 (SSLP). The map positions of the affected genes, together with 5 other known root hair initiation genes are indicated in Fig. 4.

Other aspects of the mutant phenotype

The mutants were grown to maturity after the initial screen on the third day after germination. At subsequent developmental stages we observed further deviations from the wild-type phenotype (in seven out of eight mutants) that always cosegregated with the root phenotype.

The shoots of the five different rhl mutants were hardly distinguishable, and therefore we describe rhl1 as the represen-tative of the group. 3-week old rhl1 shoots only consist of leaf rosettes as small as 1 cm in diameter which corresponds to less than half of the wild-type size (Fig. 5A). rhl1 plants never produce flowers, show necrotic symptoms after 4 weeks and die.

Fig. 5.

Shoot phenotypes of (A) 2-week old Col wild-type and rhll mutant seedlings, (B) 5-week old Col and erh3 mutant plantlets and (C) 9-week old Col and erh3 mutant plants.

Fig. 5.

Shoot phenotypes of (A) 2-week old Col wild-type and rhll mutant seedlings, (B) 5-week old Col and erh3 mutant plantlets and (C) 9-week old Col and erh3 mutant plants.

In the wild-type leaf epidermis a trichome apparatus consists of 8 socket cells and the trichome, a large twice-branched single cell (Fig. 6A, Hülskamp et al., 1994). The outer trichome wall is covered with small globular deposits. In the mature wild-type leaf one trichome apparatus is separated from the next by at least 4 epidermal cells.

Fig. 6.

Scanning electron micrographs of (A) wild-type leaf trichomes, (B) rhl1 mutant trichomes, (C) erh3 mutant trichomes (the asterisk indicates 2 adjacent erh3 trichomes). (D) Col wild-type flower bud and (E) erh3 mutant flower bud.

Fig. 6.

Scanning electron micrographs of (A) wild-type leaf trichomes, (B) rhl1 mutant trichomes, (C) erh3 mutant trichomes (the asterisk indicates 2 adjacent erh3 trichomes). (D) Col wild-type flower bud and (E) erh3 mutant flower bud.

rhl1 trichomes are mostly unbranched cellular outgrowths near the leaf margins and do not possess any globular deposits on their surface (Fig. 6B). They are still supported by 8 socket cells, but they appear to be less organized. In sharp contrast to wild-type leaves the socket cells of different rhl1 trichomes often occur directly adjacent to each other. erh2/pom1-12 shoots are comparable to wild type, but their growth is reduced by about 50% as described by Hauser et al. (1995) for other pom1 alleles, and finally erhl had no obvious shoot phenotype.

erh3 shows the most complex phenotype. The development of its very erect shoot is delayed in comparison with the wild type, but mature erh3 plants are the same size as wild-type plants (Fig. 5B,C). Interestingly the erh3 leaf epidermis also shows defects. Most obviously the trichomes are unbranched or only branch once (Fig. 6C). The initial erh3 trichome outgrowth is wider in diameter than wild-type trichomes and it has the same surface deposits as the wild type. Occasionally the socket cells of different erh3 trichomes directly abut each other (Fig. 6C). Occasional paired stomata were also seen in contrast to wild-type leaves where they are always separated by other epidermal cells (data not shown).

erh3 is the only mutant that displays a distinctive flower phenotype. Wildtype flower buds have an oval shape and the sepals cover the organs of the interior whorls until anther maturity (Fig. 6D). In erh3 plants the pistil that is characterized by its short and fat shape, already protrudes through the anther, petal and sepal whorls at a very early bud stage (Fig. 6E). Occasionally we observed five sepals and five petals on erh3 flowers instead of the four organs per whorl found in the wild type (data not shown). erh3 siliques are only half of the wild-type length (Fig. 5C). erh3 seeds are larger than wild-type seeds, display a darker pigmentation and their shape is round as opposed to the oval shape of wild-type seeds (data not shown). However, no reduction in fertility was observed.

Ethylene responses

Tanimoto et al. (1995) showed that ethylene stimulates ectopic root hair initiation in a concentration-dependent manner. In the present study we investigated the effects of the ethylene precursor ACC on the root hairless mutants and the response of the ectopic root hair mutants to the ethylene biosynthesis inhibitor AVG. The data represent the results of two independent series of experiments.

The results of the ACC treatment on the wild type and the root hairless mutants is summarized in Fig. 7. Adding 1 μM ACC to the medium has little effect on the wildtype, but upon treatment with 10 or even 100 μM ACC the hypocotyl swells and root elongation is inhibited. The density of root hairs appears to be greatly increased.

Fig. 7.

ACC effects on seedlings. 3-day old seedlings of (A-D) wildtype, (E-H) rhl1, (I-L) rhl2-2 and (M-P) rhl3-2 grown on medium supplied with (A,E,I,M) no ACC, (B,F,J,N) 1 μM ACC, (C,G,K,O) 10 μM ACC and (D,H,L,P) 100 μM ACC. Scale bar in P equals 400 μm.

Fig. 7.

ACC effects on seedlings. 3-day old seedlings of (A-D) wildtype, (E-H) rhl1, (I-L) rhl2-2 and (M-P) rhl3-2 grown on medium supplied with (A,E,I,M) no ACC, (B,F,J,N) 1 μM ACC, (C,G,K,O) 10 μM ACC and (D,H,L,P) 100 μM ACC. Scale bar in P equals 400 μm.

The root hairless mutants display only parts of the complex response observed in the wildtype. Again little effect is seen upon treatment with 1 μM ACC, whereas higher concentrations promote a slight swelling of the hypocotyl and inhibition of root elongation in rhl1, rhl2 and rhl3. There are more root hairs than in the untreated mutants, but the initiation is still very limited. Even with 100 μM ACC, rhl1 and rhl2-2 only produce a maximum of 5-6 root hairs that emerge from below the bald collet region (Fig. 7H,L). rhl3-2, which is already characterized by a slightly higher root hair density in the untreated state, produces correspondingly more root hairs although the collet region still appears bald (Fig. 7M-P).

The phenotypes of AVG-treated wildtype and ectopic root hair mutant seedlings are presented in Fig. 8. In the wild-type root treated with 5 μM AVG root hair initiation below the collet region is inhibited while with 20 μM AVG very few root hairs appear below the collet region.

Fig. 8.

AVG effects on seedlings. 3-day old seedlings of (A-C) wildtype, (D-F) erhl, (G-I) erh2 and (J-L) erh3 grown on medium supplied with (A,D,G,J) no AVG, (B,E,H,K) 5 μM AVG and (C,F,I,L) 20 μM AVG. Scale bar in L equals 400 μ m.

Fig. 8.

AVG effects on seedlings. 3-day old seedlings of (A-C) wildtype, (D-F) erhl, (G-I) erh2 and (J-L) erh3 grown on medium supplied with (A,D,G,J) no AVG, (B,E,H,K) 5 μM AVG and (C,F,I,L) 20 μM AVG. Scale bar in L equals 400 μ m.

AVG also inhibits root hair initiation on erhl seedlings (Fig. 8D-F). Although erh2/pom1-l2 and erh3 are both impaired in root hair initiation in response to AVG (Fig. 8G-L), the treatment also appears to cause extreme radial expansion in the root epidermis leading to characteristic bulging in epidermal cells of both mutants.

Double mutants

Double mutants between root hairless and ectopic root hair mutants were generated to determine the interactions between genes involved in root hair initiation.

All aspects of the phenotype of the double mutants between rhl1 and erh2/pom1-l2 were identical to the rhl1 single mutant, and therefore rhl1 is formally epistatic to erh2/pom1-l2.

Seedlings homozygous for both rhl1 and ttg are completely bald on their primary roots. The shoot morphology is essentially that of rhl1 single mutants, but doubly mutant seedlings appear to be very pale due to the lack of anthocyanins in the hypocotyl and the cotyledons. However, with respect to the phenotype of the root epidermis rhl1 is clearly epistatic to ttg.

The double mutants between rhl1 and erh3 are characterized by hairless roots that are, however, increased in diameter in the same way as erh3. It is a complex phenotype, but the epidermal cell fate seems to be determined by the rhl1 mutation, i.e. rhl1 is epistatic to erh3.

Double mutants between rhl1 and ctrl displayed a phenotype that was reminiscent of rhl1 treated with 10-100 μM ACC. The root hair density was only slightly increased on primary roots whereas the collet region remained bald as in rhl1 single mutants. The length of the root was dramatically reduced as seen in ctrl single mutants. The shoot phenotype was essentially the same as in rhl1 single mutants.

Plants homozygous for the rhl2-2 and ctrl mutations also showed a few more hairs on short primary roots below the collet region, but otherwise looked like rhl2-2 single mutants.

The same result was obtained for the rhl3-2 ctrl double mutant that displayed the same phenotype as the rhl3-2 single mutant treated with 10-100 μM ACC.

In all cases functional RHL-gene products are required for the ctrl root phenotype. However, for root hair formation below the collet the rhl-mutants are not completely epistatic to ctrl, indicating a more complex or indirect relationship between the affected genes. The few root hairs of rhl ctrl double mutants indicate that the rhl mutants appear to be equipped with a functional machinery to produce root hairs. This is consistent with the idea that the RHL genes act prior to cell fate specification in the meristem (Fig. 1D,F,H) whereas CTR1 and ethylene play a role very late in the root hair initiation process irrespective of the predetermined cell fate (Masucci and Schiefelbein, 1996).

In our screen for mutants without root hairs or with root hairs in ectopic positions we have defined 6 genes that have not previously been associated with root epidermal cell differentiation. We characterize several aspects of their mutant phenotype and discuss possible functions for the affected gene products by relating both the mutants to each other and to other previously described steps in the pathway to root hair formation.

Most genes involved in root hair initiation affect shoot developmental processes as well

A new class of root hairless mutants is described here for the first time specifying the 3 genes RHL1, RHL2 and RHL3. The mutants rhl1, rhl2-1, rhl2-2, rhl3-1 and rhl3-2 all show the same pleiotropic phenotype characterized by the complete absence of root hairs, abnormal patterning of the leaf epidermis and dwarfism. This might indicate functional redundancy of the three genes.

The formation of root hairs is regulated differently on the primary root and at the transition from the primary root to the hypocotyl region (Dolan et al., 1993), but the root hairless genes RHL1, RHL2 and RHL3 appear to be required in both pathways. The only other mutant displaying this phenotype is axr3-1, the strongest allele of an auxin and ethylene resistant mutant, again with a very complex phenotype (Leyser et al., 1996). However, a comparison of map positions excludes allelism between axr3 and the root hairless mutants presented here.

Apart from the rhl-mutants erh3, an ectopic root hair mutant, shows alterations in the shoot as well as the root epidermis. Despite the opposite effects of the erh3 and the rhl mutations on the root epidermal phenotype, their effects on leaf development are similar, suggesting different roles for the corresponding gene products in roots and shoots. With gl2 and ttg, two more loci are known to be involved in general epidermal development (Masucci et al., 1996, Galway et al., 1994) indicating a complex network of genes controlling the development of specialized epidermal cells all over the plant.

erh3 is the only known ectopic root hair mutant so far that shows a significant floral phenotype as well (Fig. 6E). Occasionally aberrant numbers of floral organs suggest a defect in primordial positioning. Despite differences between wildtype and erh3, siliques and seeds, the erh3 mutant is fully fertile indicating that there is no problem with gametophyte development.

As there is only one EMS allele of the erh3 mutation whether or not the severity of the displayed defects is due to a null mutation cannot be evaluated. However, the mutant phenotype is consistent with a role for the ERH3 gene product in lateral inhibition mechanisms involved in specifying the position and fate of cell types or organ primordia (Bünning, 1965).

The fact that our screen based on root phenotypes mostly identified mutants with pleiotropic defects indicates that root epidermal cell differentiation is interlinked with other processes and is likely to require a number of gene products that are also involved in other processes. There appear to be only a few root specific mutants and erh1 might be such a candidate. However, the pathway has not yet been screened to saturation.

Root hair initiation and radial cell expansion

A correlation between radial cell expansion and epidermal cell differentiation is revealed by the erh2/pom1-12 mutant phenotype (Hauser et al., 1995). The conditional root expansion phenotype of pom1 is dependent on a high concentration of sucrose in the growth medium (Hauser and Benfey, 1993). The growth medium used in our studies contains 3% sucrose and falls within this range. Under these conditions 3 day old erh2 seedlings show early stages of cell differentiation comparable to the wildtype in the meristematic zone that does not yet display the fully expanded phenotype (Fig. 1A,L). This indicates that ectopic root hairs in erh2 are likely to be initiated at late developmental stages correlated with cell expansion. The rhl1-like double mutant between rhl1 and erh2 reveals that both cell expansion and root hair formation in erh2 are dependent on RHL1 gene function. The radial expansion of the cortex in pom1 is prevented at sucrose concentrations below 0.5% (Hauser and Benfey, 1993). Preliminary results indicate that erh2/pom1-12 seedlings grown on medium with 0.1% sucrose hardly show any ectopic root hairs (data not shown), thus suggesting that ectopic root hairs as well as the cortex cell expansion phenomenon are conditional on the growth rate of erh2/pom1-12 seedlings.

A possible link between cell expansion and root hair initiation also becomes apparent by comparing transverse sections of ctrl roots and roots treated with the ethylene precursor ACC (Dolan et al., 1994; Tanimoto et al., 1995). The diameter of both specimens appears to be about 1.25-fold larger than the wildtype. The degree of expansion seems to be uniform in all cell layers. In both cases cell expansion is probably due to ethylene action that also promotes ectopic root hair formation, and the different ethylene pathways may or may not be linked. Masucci and Schiefelbein (1996) suggested that ethylene and auxin action do not change cell fate in the primary root epidermis, but alter cell expansion or morphogenesis at a late developmental stage to trigger hair formation. However, in erhl, erh2/pom1-12 and erh3 primary roots still appear to be expanded after blocking ethylene synthesis with AVG (Fig. 8), there even seem to be epidermal bulges on erh2/pom1-12 and erh3 roots whereas all root hairs below the collet have disappeared. This result indicates that radial cell expansion in erh1, erh2/pom1-12 and erh3 is likely to be caused by something other than ethylene synthesis.

Interestingly the diameters of erh1 and erh3 roots are also significantly wider than the diameter of wildtype roots (Fig. 3). In the case of erh1 the radial expansion seems to affect all cell layers to a similar degree whereas in erh3 the endodermal cells are particularly expanded. Preliminary results indicate that cell expansion of erh1 mutants is also conditional on growth rate although erh1 roots still produce ectopic root hairs when grown on 0.1% sucrose. Growing erh3 mutants on 0.1% sucrose, however, appears to affect neither cell expansion or ectopic root hair initiation. Both results are consistant with the idea that ERH1 and ERH3 act as early as at the stage of cell fate specification and therefore the production of ectopic root hairs is not dependent on growth rate. Moreover, cell expansion of erh3 roots is independent of growth rate altogether.

The strongly expanded bald root phenotype of the rhl1 erh3 double mutant also argues for the independence of root hair initiation and cell expansion. It reveals that in the absence of functional RHL1 gene product cell expansion per se does not promote root hair initiation. Still, there remains the possibility that some genes may regulate radial cell expansion and – dependent on it or not – the competence to undergo ethylene dependent hair formation at a late stage of root epidermal development.

A preliminary model for root hair formation

The variety of mutants impaired in correct root hair initiation and the different relationships between the corresponding gene products suggests a complex and not strictly hierarchical pathway regulated at different levels leading to root hair formation. To achieve an order of events in that process the mutants were grouped according to the developmental stages in which their defects were first observed (Table 2). Mutations affecting the shape of root hairs, like some members of the rhd-class (Schiefelbein and Somerville, 1990), do not appear to influence hair positioning and are therefore not referred to. The classification shown in Table 2 leads to the proposal of the following model for the pathway leading to root epidermal cell differentiation (Fig. 9). Very early processes in epidermal cell fate specification are most easily explained if the RHL genes promote trichoblast development in all epidermal cell files. However, the partial response of the rhl mutants to ACC indicate that the pathway is not strictly linear. It shows that the rhl-mutants can perceive the ethylene signal to a very limited degree and that there is some ethylene dependent competence to form root hairs in these mutant backgrounds. Although none of the rhl alleles has yet been confirmed to be a null mutation the few root hairs on ACC treated rhl roots are likely to be due to a root hair initiation pathway largely independent of RHL gene functions. The epistasis results are also formally consistent with an interpretation that places the RHL genes downstream of ethylene action in both cell files, although the severe pleiotropic phenotypes of the rhl mutants and the fact that the cell files in the RHL mutants are not visually distinguishable in the meristem both argue for the RHL genes acting earlier than this.

Table 2.

Classification of root epidermal mutants with altered hair density

Classification of root epidermal mutants with altered hair density
Classification of root epidermal mutants with altered hair density
Fig. 9.

Preliminary model for hair initiation in Arabidopsis primary roots. Open arrows show an alternative point of action of RHL1, RHL2 and RHL3 that is consistent with genetic results. Genes are all shown in boxes. Arrows indicate positive regulation (+), blunted lines indicate inhibitory action (–).

Fig. 9.

Preliminary model for hair initiation in Arabidopsis primary roots. Open arrows show an alternative point of action of RHL1, RHL2 and RHL3 that is consistent with genetic results. Genes are all shown in boxes. Arrows indicate positive regulation (+), blunted lines indicate inhibitory action (–).

A number of genes, conferring ectopic root hairs when mutated, are assumed to negatively regulate hair formation or positively regulate non-hair formation in distinct cells. The earliest among these are likely to be identified by the ttg, erh1and erh3 mutations. Their effects are already visible at the level of cell fate determination in the meristematic zone where tri-choblasts and atrichoblasts are differentiating. After this initial cell fate specification the GL2 gene product is active only in atrichoblasts to secure the hairless cell fate. The ERH2/POM1 gene product possibly also functions at this level to control the initiated developmental program in atrichoblasts by regulating cell expansion in the adjacent cortical cells. Finally root hair formation mediated by high concentrations of growth factors, in particular ethylene, appears to affect all cells irrespective of their predetermined fate. However, trichoblasts and atri-choblasts might be distinguishable by quantitative differences in susceptibility to growth factors. By growing seedlings in the dark to lower the effective ethylene concentration it has been shown for gl2 that root hair formation in cells in non-hair positions is dependent on a higher ethylene level than root hair formation in trichoblast positions (Di Cristina et al., 1996).

By studying 8 novel mutants that are impaired in root hair initiation and by examining the possible interactions of the corresponding genes, progress has been made towards understanding the different steps in the pathway leading to cell differentiation in the root epidermis and its links with other developmental processes.

We would like to thank Sue Bunnewell for preparing media for plates and for excellent photographic work and Nigel Orme for the graphic design of Fig. 9. We acknowledge Claire Grierson for collaboration in the initial phase of the project and Karin Metzlaff for sharing work in the first part of the screen. We are grateful to Ken Feldmann who made the rhl1 mutant available. Our thanks also go to John Schiefelbein for providing seeds and sharing useful conversations. K. S. is grateful to members of Phil Benfey’s lab and especially to Marie-Theres Hauser for sharing protocols, donating seeds and discussing important aspects throughout the work. We acknowledge both the Ohio and the Nottingham Arabidopsis stock centres for providing seeds. K. S. was supported by fellowships from EMBO (ALTF 5681992), HFSP (LT-135/93) and by a grant from the Gatsby Charitable Foundation. L. D. was funded by a PMB II grant and B. W. and K. R. acknowledge funding from the BBSRC.

Bell
,
C. J.
and
Ecker
,
J. R.
(
1994
).
Assignment of 30 microsatellite loci to the linkage map of Arabidopsis
.
Genomics
19
,
137
144
.
Bünning
,
E.
(
1965
). Die Entstehung von Mustern in der Entwicklung von Pflanzen. In
Encyclopedia of Plant Physiology
, Vol.15 (ed.
W.
Ruhland
) pp.
383
408
,
Springer-Verlag
,
Berlin
.
Dellaporta
,
S. L.
,
Wood
,
J.
and
Hicks
,
J. B.
(
1983
).
A plant DNA minipreparation: version II
.
Plant Mol. Biol. Rep.
1
,
19
21
Di Cristina
,
M.
,
Sessa
,
G.
,
Dolan
,
L.
,
Linstead
,
P.
,
Baima
,
S.
,
Ruberti
,
I.
and
Morelli
,
G.
(
1996
).
The Arabidopsis Athb-10 (GLABRA) is an HD-Zip protein required for regulation of root hair development
.
The Plant Journal
10
,
393
402
.
Dolan
,
L.
,
Janmaat
,
K.
,
Willemsen
,
V.
,
Linstead
,
P.
,
Poethig
,
S.
,
Roberts
,
K.
and
Scheres
,
B.
(
1993
).
Cellular organization of the Arabidopsis thaliana root
.
Development
119
,
71
84
.
Dolan
,
L.
,
Duckett
,
C. M.
,
Grierson
,
C.
,
Linstead
,
P.
,
Schneider
,
K.
,
Lawson
,
E.
,
Dean
,
C.
,
Poethig
,
S.
and
Roberts
,
K.
(
1994
).
Clonal relationships and cell patterning in the root epidermis of Arabidopsis
.
Development
120
,
2465
2474
.
Feldmann
,
K. A.
(
1991
).
T-DNA insertion mutagenesis in Arabidopsis: mutational spectrum
.
Plant J.
1
,
71
82
.
Galway
,
M. E.
,
Masucci
,
J. D.
,
Lloyd
,
A. M.
,
Walbot
,
V.
,
Davis
,
R. W.
and
Schiefelbein
,
J. W.
(
1994
).
The TTG gene is required to specify epidermal cell fate and cell patterning in the Arabidopsis root
.
Dev. Biol.
166
,
740
754
.
Hauser
,
M.-T.
and
Benfey
,
P. N.
(
1993
). Genetic regulation of root expansion in Arabidopsis thaliana. In
NATO-ASI series, Vol. H81, Plant Molecular Biology
(ed.
G.
Coruzzi
and
P.
Puigdomenech
), pp.
31
40
.
Springer-Verlag, Berlin Heidelberg
.
Hauser
,
M.-T.
,
Morikami
,
A.
and
Benfey
,
P. N.
(
1995
).
Conditional root expansion mutants of Arabidopsis
.
Development
121
,
1237
1252
.
Hülskamp
,
M.
,
Miséra
,
S.
and
Jürgens
,
G.
(
1994
).
Genetic dissection of trichome development inArabidopsis.Cell
76
,
555
566
.
Kieber
,
J. J.
,
Rothenberg
,
M.
,
Roman
,
G.
,
Feldmann
,
K. A.
and
Ecker
,
J. R.
(
1993
).
CTR1, a negative regulator of the ethylene response pathway in Arabidopsis, encodes a member of the Raf family of protein kinases
.
Cell
72
,
427
441
.
Konieczny
,
A.
and
Ausubel
,
F. M.
(
1993
).
A procedure for mapping Arabidopsis mutations using co-dominant ecotype-specific PCR-based markers
.
Plant J.
4
,
403
410
.
Koorneef
,
M.
(
1981
).
The complex syndrome of ttg mutants
.
Arabid. Inform. Serv
18
,
45
51
.
Leyser
,
H. M. O.
,
Pickett
,
F. B.
,
Dharmasiri
,
S.
and
Estelle
,
M.
(
1996
).
Mutations in the AXR3 gene of Arabidopsis result in altered auxin response including ectopic expression from the SAUR-AC1 promoter
.
Plant J.
10
,
403413
.
Lloyd
,
A. M.
,
Walbot
,
V.
and
Davis
,
R. W.
(
1992
).
Arabidopsis and Nicotiana anthocyanin production activated by maize regulators R and C1
.
Science
258
,
1773
1775
.
Masucci
,
J. D.
and
Schiefelbein
,
J. W.
(
1994
).
The rhd6 mutation of Arabidopsis thaliana alters root-hair initiation through an auxin- and ethylene associated process
.
Plant Physiol.
106
,
1335
1346
.
Masucci
,
J. D.
and
Schiefelbein
,
J. W.
(
1996
).
Hormones act downstream of TTG and GL2 to promote root hair outgrowth during epidermis development in the Arabidopsis root
.
Plant Cell
8
,
1505
1517
.
Masucci
,
J. D.
,
Rerie
,
W. G.
,
Foreman
,
D. R.
,
Zhang
Meng
,
Galway
,
M. E.
,
Marks
M. D.
and
Schiefelbein
,
J. W.
(
1996
).
The homeobox gene GLABRA2 is required for position-dependent cell differentiation in the root epidermis of Arabidopsis thaliana
.
Development
122
,
1253
1260
.
Rerie
,
W. G.
,
Feldmann
,
K. A.
and
Marks
,
M. D.
(
1994
).
The GLABRA2 gene encodes a homeo domain protein required for normal trichome development in Arabidopsis
.
Genes and Development
8
,
1388
1399
.
Schiefelbein
,
J. W.
and
Somerville
,
C.
(
1990
).
Genetic control of root hair development in Arabidopsis thaliana
.
The Plant Cell
2
,
235
243
.
Tanimoto
,
M.
,
Roberts
,
K.
and
Dolan
,
L.
(
1995
).
Ethylene is a positive regulator of root hair development in Arabidopsis thaliana
.
Plant J.
8
,
943948
.
Wilson
,
A.
,
Pickett
,
F. B.
,
Turner
,
J. C.
and
Estelle
,
M.
(
1990
).
A dominant mutation in Arabidopsis confers resistance to auxin, ethylene and abscisic acid
.
Mol. Gen. Genet.
222
,
377
383
.