Auxin and ethylene promote root hair elongation in Arabidopsis


*For correspondence (fax + 812 855 6705; e-mail


Genetic and physiological studies implicate the phytohormones auxin and ethylene in root hair development. To learn more about the role of these compounds, we have examined the root hair phenotype of a number of auxin- and ethylene-related mutants. In a previous study, Masucci & Schiefelbein (1996 ) showed that neither the auxin response mutations aux1 and axr1 nor the ethylene response mutations etr1 and ein2 have a significant effect on root hair initiation. In this study, we found that mutants deficient in either auxin or ethylene response have a pronounced effect on root hair length. Treatment of wild-type, axr1 and etr1 seedlings with the synthetic auxin, 2,4-D, or the ethylene precursor ACC, led to the development of longer root hairs than untreated seedlings. Furthermore, axr1 seedlings grown in the presence of ACC produce ectopic root hairs and an unusual pattern of long root hairs followed by regions that completely lack root hairs. These studies indicate that both auxin and ethylene are required for normal root hair elongation.


Root hair development has been the topic of much study in recent years. In Arabidopsis, root anatomy has been well characterized and the pattern and course of root hair development has been documented ( Dolan et al. 1993 ;Galway et al. 1994 , Galway et al. (1994 ). Root hairs are tubular projections which grow out of a specialized subset of epidermal cells called trichoblasts. ( Peterson & Farquhar 1996). Positional information is important for controlling epidermal cell fate. As epidermal cells proceed through the elongation zone of the growing root, cells that lie over the anticlinal junction between two underlying cortical cells assume the characteristics of trichoblasts while cells that lie above the periclinal surface of a single cortical cell become atrichoblasts ( van den Ber et al. 1995 ;Masucci et al. 1996 ;Tanimoto et al. 1995 ). This developmental regularity, combined with the defined radial pattern of root tissues and the stacked columnar nature of cells along the length of the root, give rise to an ordered pattern of root hair cell files spaced between non-root hair cell files on the surface of the Arabidopsis root.

Many factors, both genetic and environmental, contribute to the wild-type pattern of root hair development ( Bates & Lynch 1996; for review see Dolan & Roberts 1995;Ridge 1995). Several genes have been defined in Arabidopsis by mutational analysis that function in root hair development. The best characterized of these genes are TTG and GL2. Both are shoot and root epidermal developmental regulators ( Koornneef 1981;Rerie et al. 1994 ). They are likely to function as early negative regulators of trichoblast development or, alternatively, as positive regulators of atrichoblast development since mutations in these genes allow almost all root epidermal cells to develop hairs ( Galway et al. 1994 ;Masucci et al. 1996 ;Masucci et al. 1996 ). The ttg mutation abolishes the earliest cytological distinctions between trichoblasts and atrichoblasts while gl2 mutants retain these distinctions. ( Galway et al. 1994 ;Masucci et al. 1996 ). Furthermore, GL2 expression is reduced in a ttg background indicating that TTG acts earlier in the development of the epidermis and may directly regulate GL2 expression. Both GL2, a protein that contains homeodomain and leucine zipper motifs ( DiCristina et al. 1996 ), and TTG are likely to act as transcription factors.

Several other Arabidopsis mutants with abnormal root hair phenotypes have been isolated. One group includes cow1, cpc1, rhd1, rhd2, rhd3 and rhd4, and is characterized by defects in root hair shape and number ( Galway et al. 1997 ;Grierson et al. 1997 ;Masucci & Schiefelbein 1994;Schiefelbein & Somerville 1990;Wada et al. 1997 ). Three root hairless (rhl) mutants have also been found as well as three mutants producing ectopic root hairs (erh) ( Schneider et al. 1997 ). Notably, the rhl mutants also lack root hairs in the collet region, a phenotype not observed in other root hair mutants ( Schneider et al. 1997 ). The RHD3 and CPC1 genes have been cloned and encode putative GTP-binding and Myb-like DNA binding proteins, respectively. ( Wada et al. 1997 ;Wang et al. 1997 ). In addition, the RHL1 gene was recently cloned and shown to encode a novel nuclear protein ( Schneider et al. 1998 ). Isolating the remaining genes and characterizing them on the molecular level will certainly provide further insight into the root hair development pathway from initiation to elongation.

Another class of mutants, those involved in auxin and ethylene responses, also affect root hair development and have been implicated both by genetic and physiological studies. The auxin response mutations aux1, axr1, axr2 and axr3 all affect root hair growth ( Leyser et al. 1996 ;Lincoln et al. 1990 ;Okada & Shimura 1994;Wilson et al. 1990 ). The ethylene response mutant, ctr1, has ectopic root hairs ( Tanimoto et al. 1995 ). Moreover, the ethylene precursor, aminocyclopropane carboxylic acid (ACC), induces ectopic root hair formation in wild-type seedlings while the ethylene biosynthesis inhibitor, aminovinylglycine (AVG), abolishes root hairs ( Masucci & Schiefelbein 1996;Tanimoto et al. 1995 ). These observations have led to the hypothesis that ACC or ethylene act as a diffusible signal directing root hair production in those epidermal cells supertending the boundary between two underlying cortical cells ( Tanimoto et al. 1995 ). Interestingly, the application of 2,4-D or IAA does not cause ectopic root hair development despite the fact that both hormones induce ethylene biosynthesis in roots ( Masucci & Schiefelbein 1996). In addition, neither the etr1 nor ein2 mutants have reduced numbers of root hairs, suggesting that ethylene is not required for initiation of hair growth ( Masucci & Schiefelbein 1996).

We are interested in understanding the mechanisms by which auxin and ethylene act to control root hair development. Toward that goal we have conducted morphological analyses of root hairs in wild-type and phytohormone mutant seedlings of Arabidopsis thaliana. The mutants axr1, aux1, etr1, ein2 and eto1 were all found to have abnormalities in root hair elongation. We have also examined the effects of auxin and ethylene on root hair length by growing seedlings on 2,4-D or ACC supplemented media. Furthermore, we describe the effect of the sar1 mutation on root hair development in the auxin mutants. The results of our studies indicate that auxin and ethylene are both positive regulators of root hair elongation in Arabidopsis.


Effects of auxin response mutations on root hair development

Genetic studies in Arabidopsis have implicated auxin in the process of root hair development ( Cernac et al. 1997 ;Masucci & Schiefelbein 1996;Wilson et al. 1990 ). Cernac et al. (1997) showed that the axr1 mutants have reduced numbers of root hairs. However, it was not clear from their data if the mutant was deficient in root hair initiation or elongation. In an earlier study, Masucci & Schiefelbein (1996) examined the weak axr1–3 allele and found a normal number of root hairs, suggesting that the apparent reduction observed by Cernac et al. (1997) was due to a defect in elongation. To distinguish between these possibilities, we measured root hair numbers and lengths in a large sample of axr1 seedlings.

When viewed with a dissecting microscope, the strong allele axr1–12 has less than half the number of root hairs per millimeter as wild-type roots ( Cernac et al. 1997 ). However, when viewed at a 400 × magnification, 46.5% ± 2.8 of the axr1–12 root epidermal cells had initiated a root hair compared to 50.0% ± 2.3 in wild type. These results confirm that the axr1 mutations have little effect on root hair initiation. In contrast, axr1–12 had a dramatic effect on root hair length ( Figs 1 and 2, Table 1). The root hairs on the mutant were significantly shorter than wild-type root hairs and much more variable in length. When plotted as a histogram, the lengths of wild-type root hairs resemble a typical Gaussian distribution with values centered around a predominant mean value, whilst axr1–12 root hairs display a much flatter distribution ( Fig. 3). The difference between the mean root hair length values for wild-type and axr1–12 is significant. We conclude, therefore, that axr1–12 seedlings are severely deficient in root hair elongation. The axr1 phenotype is not allele specific as the same defects are seen in axr1–3, a weak allele (data not shown, Lincoln et al. 1990 ).

Figure 1.

Roots of 7-day-old Arabidopsis seedlings grown on unsupplemented nutrient media.

The photographs show wild-type root hairs compared with various hormone mutants. (a) Columbia wild-type; (b) axr1–12;(c) aux1–7;(d) etr1;(e) sar1–1;(f) axr1–12 sar1–1;(g) aux1–7 sar1–1;(h) eto1–1. Scale bars = 500 μm.

Figure 2.

Root hair phenotypes of hormone response mutants.

(a) Photomicrograph of a 10-day-old axr1–12 root grown on unsupplemented media. Arrow indicates a trichoblast located above the elongation zone with an initiated, non-elongated root hair. Scale bar = 100 μm.

(b) Photomicrograph of a bifurcated (twin) root hair of a 7-day-old aux1-7 seedling grown on unsupplemented media. Scale bar = 250 μm.

(c) Photomicrograph of a 7-day-old aux1–7 sar1–1 seedling grown on unsupplemented media. Arrows indicate two root hairs initiating from the same trichoblast. Scale bar = 50 μm.

Table 1. Root hair length in wild-type and mutant seedlings
GenotypeRoot hair length (μm) ± SE
  1. aNumber of root hairs measured.

wild-type 499 ± 5.5 (427)a
axr1–12 310 ± 8.3 (575)
aux1–7 201 ± 4.4 (764)
sar1–1 500 ± 5.9 (595)
aux1–7 sar1–1 246 ± 4.9 (796)
axr1–12 sar1–1 592 ± 6.2 (592)
etr1 203 ± 4.1 (659)
eto1–1 752 ± 9.1 (295)
Figure 3.

Distributions of root hair lengths in 7-day-old Arabidopsis seedlings grown on unsupplemented media.

(a) Columbia; (b) axr1–12;(c) aux1–7;(d) etr1;(e) ein2–1.

The sar1–1 mutant of Arabidopsis suppresses nearly every aspect of the axr1 phenotype ( Cernac et al. 1997 ). Based on double mutant analysis, SAR1 functions after AXR1 in auxin response ( Cernac et al. 1997 ). One of the phenotypes that sar1 suppresses is the failure of axr1–12 seedlings to elongate root hairs ( Cernac et al. 1997 ). We performed a detailed analysis of the effects of the sar1–1 mutation on root hair development in the axr1–12 and aux1–7 mutants.

The lengths ( Table 1) and distribution (data not shown) of sar1–1 root hairs closely resembles that of wild-type seedlings. Occasionally, we observed an ectopic root hair in sar1–1 seedlings (arrow, Fig. 1e). Similar to the study by Cernac et al. (1997) , the double mutant axr1–12 sar1–1 has a root hair phenotype resembling that of the sar1–1 single mutant. The distribution of root hair lengths is like that of either sar1–1 or wild type (data not shown). Surprisingly, we found that the length of hairs was significantly greater in axr1–12 sar1–1 seedlings than in either wild type or sar1–1.

The auxin resistant mutant, aux1–7, has also been reported to have shorter root hairs than wild-type seedlings ( Okada & Shimura 1994). Unlike axr1–12, aux1–7 root hairs assume a more normal length distribution ( Fig. 3). aux1–7 root hairs did not elongate to the maximum of about 0.75 mm that wild-type or axr1–12 achieve on unsupplemented media ( Fig. 3). Both axr1–12 and aux1–7 produce some bifurcated or ‘twin hairs’ which originate from the same cell and have split into two distinct shafts very near the base of the hair ( Fig. 2d). The AUX1 gene product is related to transmembrane amino acid transporters and has been proposed to function as an auxin influx carrier in Arabidopsis roots ( Bennett et al. 1996 ). Thus the aux1–7 phenotype might be explained as the result of loss of auxin redistribution into trichoblasts where it is required for normal root hair elongation.

We found that the sar1–1 mutation has some interesting effects on root hair development in the aux1–7 background. Root hairs in the double mutant are significantly longer than aux1–7 and there is an increase in the number of root hairs per millimeter compared with aux1–7 (data not shown) At least part of the increase in hair number is due to ectopic root hair formation in the double mutant (i.e. root hairs initiating in several adjacent files, Fig. 1g) as well as the occasional initiation of more than one root hair within the same trichoblast ( Fig. 2c). The data suggest that sar1–1 is able to partially suppress the aux1–7 reduction in root hair numbers and root hair length.

Effects of ethylene mutants on root hair development

Although the plant hormone ethylene has been strongly implicated in Arabidopsis root hair development, a root hair defect in the ethylene response mutants etr1 and ein2 has not been described. Both mutants are completely resistant to ethylene treatment with respect to all phenotypes observed including hypocotyl and root growth inhibition, leaf chlorophyll loss, ethylene biosynthesis and induction of seed germination ( Bleecker et al. 1988 ;Roman et al. 1995 ). We analyzed the root hair morphology of these two mutants, as well as the ethylene overproducing mutant eto1–1 ( Guzmán & Ecker 1990). etr1 has significantly reduced root hair lengths compared to wild-type seedlings while still displaying a normal root hair length distribution pattern ( Table 1, Fig. 3). etr1 trichoblasts occasionally fail to initiate a root hair (see Fig. 1c) and the hairs do not elongate past a maximum of about 0.5 mm ( Fig. 3). ein2–1 seedlings have an average root hair length and distribution pattern similar to that of etr1 ( Fig. 3). These results indicate that ethylene contributes to hair elongation.

In contrast to the etr1 phenotype, eto1–1 trichoblasts are significantly longer than wild-type, achieving an average length of just over 0.75 mm compared to 0.5 mm for wild-type (data not shown). The distribution of root hair lengths is also normal in eto1–1 seedlings (data not shown).

Effects of hormone treatments on wild-type and mutant phenotypes

Because both auxin and ethylene response mutants have root hair elongation defects, we reasoned that the application of either 2,4-D or ACC to seedlings might also affect root hair elongation. Okada & Shimura (1994) have observed increased root hair lengths when seedlings were transferred to media containing auxin or auxin transport inhibitors. Our results again quantitatively agree with these observations. Figure 4(a) shows the result of treating wild-type, a×r1–12 and etr1 seedlings with increasing amounts of 2,4-D. All three genotypes show a modest increase in root hair length in response to low concentrations of 2,4-D. We wished to look at the effects of higher concentrations of 2,4-D on root hair elongation but wild-type and etr1 seedlings roots do not elongate when germinated on high 2,4-D concentrations. We therefore transferred 7-day-old wild-type and etr1 seedlings to media containing 0.05 μm or 0.10 μm 2,4-D. Figure 4(b) shows that each genotype responds dramatically to higher levels of 2,4-D. Wild-type and etr1 root hairs nearly double in average length while axr1–12 root hairs are over four times as long ( Fig. 4b). Furthermore, the root hair length distribution of axr1–12 is restored to normal on higher concentrations of 2,4-D, while the distributions of wild-type and etr1 remain normal ( Fig. 4c, data not shown).

Figure 4.

Figure 4.

    Effect of 2,4-dichloropheno×yacetic acid (2,4-D) on root hair lengths.

    (a) Wild-type (solid line), axr1–12 (dotted line) and etr1 (dashed line) seeds were germinated on unsupplemented media or supplemented with increasing concentrations of 2,4-D. Data points represent the average length of 75–100 root hairs plus and minus standard error.

    (b) Graph shows the effect of higher concentrations of 2,4-D on root hair elongation. Each bar represents the average length of 75–100 root hairs plus standard error. Wild-type (solid bars), axr1–12 (shaded bars), and etr1 (dashed bars). Note that y-axis scales are different in (a) and (b).

    (c) Distribution of root hair lengths of 7-day-old axr1–12 seedlings grown on 0.1 μm 2,4-D.

    Treatment of seedlings with ACC also had a dramatic effect on root hair length. As shown in Fig. 5(a), increasing concentrations of ACC lead to increased root hair lengths in all three genotypes. 5 μm ACC caused an increase in root hair length of > 80% in wild-type seedlings and 200% in axr1–12 seedlings ( Fig. 4a). In this trial, etr1 root hair length increased slightly on 5 μm ACC ( Fig. 4a), but in other trials root hair lengths were increased by as much as 100% to a maximum of approximately 0.4 mm (data not shown).

    Figure 5.

    Figure 5.

      Effect of 1-aminocyclopropane-1-carboxylic acid (ACC) on root hair length.

      (a) Wild-type (solid line), axr1–12 (dotted line) and etr1 (dashed line) seeds were germinated on unsupplemented media or supplemented with increasing concentrations of ACC. Data points represent the average length of 75–100 root hairs plus and minus standard error.

      (b) Distribution of root hair lengths of 7-day-old axr1–12 seedlings grown on 5 μm ACC.

      (c) Photograph of a 7-day-old axr1–12 seedling root grown on 5 μm ACC. Patches of long root hairs followed by areas devoid of hairs are observed. Scale bar = 500 μm.

      (d) Photograph of the root tip of a 7-day-old axr1–12 seedling grown on 5 μm ACC. Root hairs can be seen initiating in all cell files. Scale bar = 500 μm.

      The axr1–12 seedlings displayed a novel response to 5 μm ACC. Some regions of the root have very long root hairs while other regions completely lack hairs ( Fig. 5c). This phenotype was not observed in 2,4-D treated axr1–12 seedlings. Thus, axr1–12 root hair initiation and elongation were effectively rescued by 2,4-D but not by ACC treatments. axr1–12 roots also showed ectopic root hair formation on 5 μm ACC whereas wild-type and etr1 seedlings did not ( Fig. 5d).


      During the last 5 years significant progress has been made in our understanding of the genetic regulation of root hair development in Arabidopsis. Still, little is known about either the signaling processes which direct epidermal cell differentiation or the specific effects of hormones on its later stages. Auxin and ethylene are both positive regulators of root hair development that exert their influence on root hair development after the TTG/GL2 pathway has established the early differentiation events ( Masucci & Shiefelbein 1996). Unlike ethylene, the importance of auxin in the process has been questioned because applying auxin to seedlings does not cause ectopic root hair formation ( Masucci & Shiefelbein 1996;Tanimoto et al. 1995 ; this study). However, the observations that both axr1 and aux1 have shorter hairs, as well as the effects of exogenous auxin on hair length, suggest that auxin is necessary for normal root hair elongation.

      Interactions among hormone response pathways complicate the interpretation of the data. The rhd6 root hair deficiency is rescued by treatment with either auxin or ACC ( Masucci & Schiefelbein 1994). Furthermore, aux1 is ethylene resistant and some aspects of the aux1 phenotype may be a consequence of reduced ethylene response ( Pickett et al. 1990 ). It is also possible that seedling responses to 2,4-D treatment are indirect since 2,4-D stimulates ethylene production.

      In previous reports, a model has been presented in which ethylene is required for initiation of root hairs in trichoblast files. This model is based on the production of ectopic root hairs in the ctr1 mutant and pharmacological studies using inhibitors of ethylene synthesis or action. Our results present the first clear genetic evidence that ethylene response is required for normal root hair elongation. However, it is striking that both the ein2 and etr1 mutants do produce many normal root hairs, despite the fact that both mutants are insensitive to ethylene. There are several possible explanations for this result. One possibility is that the mutants retain a residual ethylene response sufficient for some root hair development. The fact that in some trials etr1 seedlings responded to ACC supports this view. In addition, treatment of wild-type seedlings with AVG, an inhibitor of ethylene biosynthesis, results in the loss of root hairs ( Masucci & Schiefelbein 1996;Tanimoto et al. 1995 ).

      The axr1 phenotype is a challenge to explain. Some root hairs were as long as wild-type whilst most were extremely short. One explanation for this result is that with respect to root hair elongation, the axr1 mutation is leaky and that some function is retained in the cells where root hairs have elongated. Alternatively, the loss or partial loss of the auxin response pathway in the axr1 mutant may cause the cells to be much more sensitive to subtle differences in auxin levels or other environmental factors. If individual cells are subjected to a changing environment or are exposed to different amounts of auxin as the root grows, then root hair elongation might be different in trichoblasts within the same cell file.

      The differences in the aux1 and axr1 phenotypes probably reflect the difference in the function of the respective proteins. AUX1 encodes a putative auxin transport protein that may be required to redistribute auxin during differentiation of the epidermis and root hair elongation. AXR1 encodes a protein with similarity to the N-terminal half of ubiquitin-activating enzyme, E1, and is required, in combination with a second protein called ECR1, for the activation of an Arabidopsis ubiquitin-related protein called RUB1 ( del Pozo et al. 1998 ;Leyser et al. 1993 ). An analogous protein modification functions in yeast to promote progression through the cell cycle via an AXR1 homologue, ENR2 ( Lammer et al. 1998 ). Immunolocalization studies indicate that AXR1 is localized to the nucleus suggesting that the targets of RUB modification are nuclear proteins. Thus, RUB modification of nuclear proteins in the root hair appears to be required for root hair growth ( Gray & Estelle 1998). The aux1 mutants, on the other hand, may simply fail to distribute auxin into trichoblasts where it is required to promote root hair elongation via the AXR1 and possibly other pathways.

      axr1–12 is also hypersensitive to the effects of ACC on root hair development, displaying an exaggeration of the phenotype of seedlings grown on unsupplemented medium ( Fig. 5c). For some trichoblasts, ACC promotes elongation of root hairs to about twice that of wild-type roots while others completely lack hairs ( Fig. 5c). This suggests that auxin and ethylene are interacting to promote root hair elongation. While it can be argued that root hair elongation in the presence of hormones is a stress response of the root system, these data in combination with mutant analyses indicate that auxin and ethylene promote root hair elongation. Future studies, including those that focus on the cellular responses to auxin and ethylene, may help clarify the role of these compounds in root hair development.

      Experimental procedures

      Plant materials and growth conditions

      Wild-type or mutant Arabidopsis thaliana (ecotype Columbia) seeds were surface sterilized in 10% bleach for 15 min and rinsed three times with sterile distilled water. Seeds were dispersed onto Petri plates containing nutrient medium ( Wilson et al. 1990 ) containing 1% agarose and 1% sucrose. Seeds were cold treated at 4°C for 2–4 days and grown vertically in an incubator at 20°C under 16 h illumination.

      Morphometric analysis

      Seedlings were grown vertically as described above for 7 days. Seedlings were stained with dilute toluidine blue (< 0.05%) and placed onto a glass microscope slide under a cover slip. The number of root hairs in a 2 mm section at the approximate midpoint of the root were counted under a dissecting microscope. Trichoblasts (from the stained seedlings) located above the elongation zone of the root were measured using an ocular micrometer with a Zeiss phase microscope at 500× magnification. Root hairs that grew along the surface, but not penetrating, agarose nutrient media were photographed at 70× magnification at the midpoint of the root using an Olympus SZH10 stereo dissecting microscope equipped with an Olympus 35 mm camera (model PM-C35B) and Kodak black and white ASA125 film. Root hairs from the photographs were measured with a ruler and their lengths in μm determined by comparison with an ocular micrometer photographed at the same magnification. Odd root hair phenotypes were photographed using an Olympus dissecting scope as described above or by using a Zeiss Axioplan microscope.

      Significance tests

      Significance tests based on normal distributions comparing the means of two large samples (n > 30) were carried out using the following equation: d = mean1-mean2/√(s12/n1 + s22/n2) where n = number of measurements, mean = sum of individual measurements/n, and s = standard deviation for each sample. When d > 2.576 then the null hypothesis (no difference in sample means) was rejected at the 99% confidence level. When n < 30, Student t-tests were carried out. In each case the t-test result agreed with the original ‘d’ test. All differences were significant to at least a value of P < 1%. This indicates that the probability of the differences in the two sample means was due to chance alone was less than 1%. A detailed description of each test can be found in Bailey (1995).

      Distribution of root hair lengths

      Root hair length measurements from three separate trials on unsupplemented media were compiled for all genotypes measured. Measurements for four trials of axr1–12 on ACC and for three trials on 2,4-D were compiled.

      Hormone treatments

      Seeds of each genotype were germinated and grown vertically in an incubator for 7 days on unsupplemented nutrient media as described above, or under the same conditions on nutrient media supplemented with 2,4-D or ACC. For 0.1 μm 2,4-D study, wild-type and etr1 seedlings were grown for 7 days on unsupplemented nutrient media as described above then under sterile conditions were transferred to the same nutrient media supplemented with 2,4-D and grown for another 7 days. Root hairs were photographed and measured as described above. Data were plotted using Cricket Graph 1.6 software.


      The authors would like to acknowledge the members of the Estelle laboratory for their advice and encouragement throughout this study. This research was supported by the NSF (IBN-9604398) and NIH (GM43644).