By continuing to browse this site you agree to us using cookies as described in About Cookies
Wiley Online Library is migrating to a new platform powered by Atypon, the leading provider of scholarly publishing platforms. The new Wiley Online Library will be migrated over the weekend of March 17 & 18. You should not experience any issues or loss of access during this time. For more information, please visit our migration page: http://www.wileyactual.com/WOLMigration/
Lateral root development is a post-embryonic organogenesis event that gives rise to most of the underground parts of higher plants. Auxin promotes lateral root formation, but the molecular mechanisms involved are still unknown. We have isolated a novel Arabidopsis mutant, solitary-root (slr), which has reduced sensitivity to auxin. This dominant slr-1 mutant completely lacks lateral roots, and this phenotype cannot be rescued by the application of exogenous auxin. Analysis with cell-cycle and cell-differentiation markers revealed that the slr-1 mutation blocks cell divisions of pericycle cells in lateral root initiation. The slr-1 mutant is also defective in root hair formation and in the gravitropic responses of its roots and hypocotyls. Map-based positional cloning and isolation of an intragenic suppressor mutant revealed that SLR encodes IAA14, a member of the Aux/IAA protein family. Green fluorescent protein-tagged mutant IAA14 protein was localized in the nucleus, and the gain-of-function slr-1/iaa14 mutation decreased auxin-inducible BA-GUS gene expression in the root, suggesting that SLR/IAA14 acts as a transcriptional repressor. These observations indicate that SLR/IAA14 is a key regulator in auxin-regulated growth and development, particularly in lateral root formation.
Lateral root development is a post-embryonic organogenesis event that gives rise to most of the underground parts of higher plants. Production of lateral roots at the correct positions and in appropriate numbers enables the plant to establish a root system that is capable of efficiently absorbing sufficient water and nutrients under varying soil conditions. As the lateral root derived from the primary root also produces many lateral roots, the plant can establish a root system that has a complex and species-specific architecture (Fitter, 1996). To date, however, the precise mechanisms involved in the establishment of this root system by higher plants have not been elucidated (Malamy and Benfey, 1997a).
Lateral root development is initiated when root pericycle cells undergo cell division (Charlton, 1996). A series of cell division and differentiation processes of these pericycle-derived cells leads to the formation of a lateral root primordium. Once the lateral root primordium is formed, it can develop a mature lateral root through the activity of its lateral root meristem. Thus lateral root initiation is a key event, and understanding how it is controlled will help to elucidate the mechanisms that regulate lateral root development.
Auxins are plant hormones that regulate many aspects of plant growth and development, including lateral root formation, tropic responses, apical dominance, and vascular development (Davies, 1995). Physiological analyses have shown that exogenous auxin promotes cell division in the root pericycle, resulting in the production of many lateral roots (Charlton, 1996; Laskowski et al., 1995). In addition, inhibitors of auxin transport suppress lateral root initiation, indicating that normal polar auxin transport is required for lateral root formation (Casimiro et al., 2001; Reed, R.C. et al., 1998).
Mutational analyses in Arabidopsis have also shown the importance of auxin in lateral root formation. The alf1 (aberrant lateral root formation1) / sur1 (superroot1) / rty (rooty) mutants, which have elevated levels of endogenous auxin, produce many lateral roots, consistent with the observation that auxin promotes lateral root formation (Boerjan et al., 1995; Celenza et al., 1995; King et al., 1995). In contrast, several mutants defective in auxin transport or auxin sensitivity have reduced numbers of lateral roots. For example, the mutation in the auxin influx carrier AUX1 gene results in reduced lateral root formation (Bennett et al; Hobbie and Estelle, 1995). In addition, the auxin-resistant mutants axr4 (auxin resistant4) and tir1 (transport inhibitor response1) also have reduced numbers of lateral roots (Gray et al., 1999; Hobbie and Estelle, 1995; Ruegger et al., 1998). These studies show that the establishment of an appropriate auxin transport and response system is important for lateral root formation.
The Aux/IAA gene family mediates auxin-regulated growth and developmental processes, including lateral root formation. Transcription of the Aux/IAA family members is rapidly induced by auxin (Theologis et al., 1985). Aux/IAA genes were initially discovered in pea, and have now been isolated in several other species, including soybean (Ainley et al., 1988), mung bean (Yamamoto et al., 1992), tobacco (Dargeviciute et al., 1998), cucumber (Fujii et al., 2000), and Arabidopsis. In Arabidopsis, more than 20 Aux/IAA members have been identified (Abel et al., 1995; Conner et al., 1990; Kim et al., 1997; Rogg et al., 2001). Expression analysis of some IAA members showed that their expression pattern profiles (in terms of induction time, transcript level in response to auxin, and organ/tissue specificity) differ, suggesting that each member has a distinct function (Abel et al., 1995). Many Aux/IAA genes are rapidly induced not only by auxin but also by cycloheximide, a translational inhibitor (Abel et al., 1995; Theologis et al., 1985), suggesting that Aux/IAA transcription is repressed by short-lived proteins. As several Aux/IAA proteins are short-lived in vivo, Aux/IAA proteins are thought to regulate auxin-responsive Aux/IAA transcription (Abel et al., 1994).
Mutational analyses in Arabidopsis have identified several gain-of-function Aux/IAA mutants, namely AXR3/IAA17 (Rouse et al., 1998), SHY2/IAA3 (Tian and Reed, 1999), AXR2/IAA7 (Nagpal et al., 2000), MSG2/IAA19 (Tatematsu et al., 1999), and IAA28 (Rogg et al., 2001). These gain-of-function mutants have altered auxin sensitivity and pleiotropic defects in growth and development that include differences in lateral root formation. The shy2-2/iaa3 mutant has reduced numbers of lateral roots (Tian and Reed, 1999), and the iaa28-1 mutant forms fewer lateral roots (Rogg et al., 2001), suggesting a decreased response to auxin. In contrast, the axr2-1/iaa7 mutant has increased numbers of lateral roots (Nagpal et al., 2000), while the axr3-1/iaa17 mutant has increased numbers of adventitious roots (Leyser et al., 1996), suggesting an enhanced response to auxin. Furthermore, some of the gain-of-function mutants also show reduced expression of auxin-inducible genes. For example, the expression of several auxin-inducible genes, including the Aux/IAA genes, is extremely reduced in the axr2-1/iaa7 mutant (Abel et al., 1995; Timpte et al., 1994). These observations support the notion that Aux/IAA proteins regulate auxin-responsive transcription.
Here we report the isolation and characterization of a novel Arabidopsis mutant, solitary-root (slr), which has reduced sensitivity to auxin. A dominant slr-1 mutant completely lacks lateral roots and cannot be rescued by the exogenous application of auxin. We provide evidence that the SLR gene encodes IAA14, a member of the Aux/IAA protein family, and that a gain-of-function mutation in IAA14 blocks the early pericycle divisions that initiate lateral root development. This gain-of-function mutation also decreases auxin-induced gene expression. We also show that the slr-1 mutant IAA14 protein tagged with green fluorescent protein is localized in the nucleus. Our studies of the gain-of-function slr-1/iaa14 mutant indicate that SLR/IAA14 is a key regulator of auxin-regulated growth and development, especially in lateral root formation.
The slr-1 mutant is defective in lateral root formation
By studying shoot gravitropic responses of various Arabidopsis mutants (Fukaki et al., 1996a; Fukaki et al., 1997), we isolated a mutant that had no lateral roots, few root hairs, and abnormal gravitropic responses in both roots and hypocotyls (see Experimental procedures). Genetic and mapping analyses of this mutant revealed that these phenotypes are due to a single dominant mutation on chromosome 4 that falls between the mi279 (0. 94 cm south) and g4539 (4.73 cm north) markers (see Experimental procedures). As a mutation on this genomic region that causes such phenotypes has not been previously described, we designated this new mutation slr-1 for solitary-root phenotype.
The most striking phenotype of slr-1 is its lack of lateral roots. In 2-week-old, light-grown, wild-type seedlings on agar plates, the primary roots produced many lateral roots (Figure 1a, left; Table 1). However, the slr-1 primary roots did not produce any lateral roots, resulting in only a primary root (Figure 1a, right; Figure 1b; Table 1). Application of exogenous auxin induces lateral root formation in Arabidopsis (Laskowski et al., 1995). In order to determine if exogenous auxin can affect the lack of lateral roots in slr-1, samples of 4-day-old wild-type and slr-1 seedlings (neither of which has lateral roots at that stage) were transferred to medium containing natural or synthetic auxins: 1 µm IAA, 1 µm NAA, or 0.1 µm 2,4-D. After incubation for an additional 3 days, all these auxin treatments induced the formation of many lateral roots in wild-type seedlings [14.3 ± 1.9 (mean ± SD) lateral roots for 1 µm IAA; 15.9 ± 4.0 for 1 µm NAA; 16.3 ± 3.4 for 0.1 µm 2,4-D, n > 33], but did not induce any lateral roots in slr-1 seedlings (0.0 ± 0.0 lateral roots for 1 µm IAA; 0.0 ± 0.0 for 1 µm NAA; 0.0 ± 0.0 for 0.1 µm 2,4-D, n > 33). A few lateral roots could occasionally be induced, but only if the slr-1 seedlings were grown with high auxin concentrations (1 mm 2,4-D) or when the slr-1 primary roots were cut off (data not shown). Thus, although the slr-1 mutant does not completely lack the ability to develop lateral roots, it is relatively insensitive to the effects of exogenous auxin in promoting lateral root formation.
Table 1. Number of lateral roots in wild-type, slr-1 and slr-1R1 seedlings
Number of lateral roots
Seedlings were grown for 10 days on MS medium. Results shown are the average of more than 25 seedlings ± SD.
Wild type (Col)
13.9 ± 2.9
0.0 ± 0.0
1.2 ± 1.4
The second prominent phenotype of the slr-1 mutant is that root hair formation is extremely reduced compared with wild-type seedlings (Figure 1c); the slr-1 roots have only very few root hairs on the primary root (Figure 1d). However, normal root hairs were observed at the junction between root and hypocotyl in both slr-1 and the wild-type seedlings (Figure 1c,d). In Arabidopsis, root hair formation is promoted by the exogenous application of ethylene (Figure 1e; Masucci and Schiefelbein, 1994; Masucci and Schiefelbein, 1996; Tanimoto et al., 1995). We observed that the application of 0.1 mm 1-aminocyclopropane-1-carboxylate (ACC), an ethylene precursor, could partially induce root hair formation on the slr-1 roots (Figure 1f), indicating that the slr-1 mutant still retains some ability to differentiate root hairs in response to exogenous ACC treatment.
The etiolated slr-1 seedlings have normal apical hooks (Figure 1g,h). However, the growth orientation of the slr-1 roots and hypocotyls tends to be randomized compared to the wild-type (Figure 1i). When the etiolated slr-1 seedlings were horizontally gravistimulated (Fukaki et al., 1996a; Fukaki et al., 1997), the slr-1 roots and hypocotyls showed reduced gravitropic curvatures compared to the wild type (data not shown). These results indicate that normal gravitropism in both root and hypocotyl is reduced by the slr-1 mutation. As the slr-1 mutant had sedimented amyloplasts in the gravity-sensing tissues, namely, the root cap columella and the hypocotyl endodermis (Fukaki et al., 1996a; Fukaki et al., 1998; Sack, 1997), this suggests that the slr-1 mutation does not directly affect the gravity-sensing mechanisms of the plant (data not shown). In addition to the defect in gravitropism, the etiolated slr-1 hypocotyls showed an altered cell elongation profile (data not shown; K. Maruhashi and M.T., unpublished results), indicating that the slr-1 mutation affects the control of hypocotyl cell elongation.
The aerial parts of the slr-1 mutant also have several phenotypes. Compared to the wild type, the slr-1 mutant has small leaves and short and thin inflorescence stems (Table 2). In addition, the slr-1 mutant has reduced numbers of inflorescence stems, indicating that the slr-1 mutation enhances apical dominance (Table 2).
Table 2. Shoot phenotypes of wild-type and slr-1 plants
Wild type (Col)
Seven-week-old plants grown on soil were used to measure lengths of inflorescence stems; for other measurements, 6-week-old plants grown on soil were used. Results shown are average ± SE of 10 plants.
Length of inflorescence stems (cm)
42.7 ± 1.4
29.2 ± 2.1
Internode length (cm)
2.1 ± 0.2
1.5 ± 0.1
Number of primary inflorescence stems
3.8 ± 0.3
1.3 ± 0.2
Diameter of inflorescence stems (mm)
0.97 ± 0.04
0.57 ± 0.04
The slr-1 mutant has reduced sensitivity to auxins
The slr-1 phenotypes in both roots and shoots strongly suggest that the slr-1 mutation directly alters auxin sensitivity, as auxin promotes both lateral roots (Charlton, 1996; Laskowski et al., 1995) and root hair formation (Masucci and Schiefelbein, 1994), and mediates gravitropism and apical dominance (Davies, 1995). To investigate this possibility, we examined the effects of exogenous auxin and various other hormones/chemicals on slr-1 root growth. As already described, the slr-1 roots showed reduced sensitivity to all examined natural and synthetic auxins (IAA, 2,4-D, NAA) compared to the wild-type roots (Figure 2a, data for 2,4-D and NAA are not shown). In contrast, the slr-1 roots had almost normal sensitivity to a synthetic cytokinin benzyl adenine (BA) and the ethylene precursor ACC (Figure 2b,c), but were slightly hypersensitive to abscisic acid (ABA) (Figure 2d). We also found that the slr-1 roots were resistant to an anti-auxin, p-chlorophenoxy-isobutyric acid (PCIB), and an auxin-transport inhibitor, 2,3,5-triiodobenzoic acid (TIBA), but that they responded almost normally to the other auxin transport inhibitors N-1-naphthylphthalamic acid (NPA) and 9-hydroxyfluorene-9-carboxylic acid (HFCA) (data not shown). These observations suggest that the inhibitory effects of TIBA on auxin transport in the slr-1 differ from those of NPA and HFCA. This difference might be related to the possibility that TIBA itself has weak auxin activity and is transported polarly (Thompson et al., 1973).
Thus a dominant mutation in the SLR gene specifically confers auxin resistance that may be directly responsible for the pleiotropic phenotypes characterizing the slr-1 mutant.
The slr-1 mutation blocks early cell divisions in lateral root initiation
Arabidopsis is a good model plant for the study of lateral root development because all developmental stages have been described in detail by using several lateral root initiation markers (Casimiro et al., 2001; Malamy and Benfey, 1997b). According to Malamy and Benfey (1997b), the formation of the lateral root primordium is initiated by anticlinal cell divisions (in a transverse orientation) of pericycle cells (stage I). Subsequently, some of the divided pericycle cells at stage I undergo periclinal cell divisions (in a longitudinal orientation) that produce another cell layer between the pericycle and endodermis layers (stage II). After stage II, the lateral root primordium is precisely formed through sequential cell division and differentiation processes (stages III and IV), after which a mature lateral root can be developed through the activity of its lateral root meristem (Malamy and Benfey, 1997b).
To determine which stages in lateral root formation are affected in the slr-1 mutant, we analysed the expression patterns of cell cycle and cell differentiation markers in slr-1 primary roots. First we used the cyclin-GUS (β-glucuronidase) chimeric reporter protein, CycB1;1::GUS (Colón-Carmona et al., 1999). As the Arabidopsis CycB1;1 is expressed only around the G2/M transition (Doerner et al., 1996), this marker enables us to find the dividing cells in both the root pericycle during lateral root initiation and in the root apical meristem (Figure 3a) (Casimiro et al., 2001; Dubrovsky et al., 2000; Ferreira et al., 1994; Gray et al., 1999). In the 5-day-old light-grown wild-type seedlings, CycB1;1::GUS activity was detected in the dividing cells where lateral root initiation occurs, indicating that these cells are undergoing stage I (Figure 3b). Older wild-type seedlings with a longer primary root, or seedlings treated with exogenous auxin, had an increased number of lateral root initiation sites with CycB1;1::GUS activity (Figure 3c; Figure 4a; Table 3). In contrast, the 5-day-old light-grown slr-1 seedlings did not have CycB1;1::GUS activity in/around the root pericycle cells, except at the primary root apical meristem (Figure 3d,e). However, in the older slr-1 seedlings, a small number of initiation sites with CycB1;1::GUS activity, associated with small pericycle cells produced by anticlinal cell division, could be detected (Figure 4a). However, the number of auxin-induced initiation sites with CycB1;1::GUS activity was still reduced in the slr-1 compared to the wild type (Figure 3f,g; Table 3). These observations indicate that the slr-1 mutation partially blocks anticlinal cell divisions of the pericycle in lateral root initiation.
Table 3. Effects of auxin treatment on early cell divisions during lateral root initiation in wild-type and slr-1 seedlings
Number of initiation sites with GUS activity
5-day-old (no treatment)
5-day-old (1 day IAA)
Results shown are the average of initiation sites with CycB1;1::GUS or End199 activity ± SD of 4-day-old, 5-day-old (no treatment), and 5-day-old (1 day treatment with 1 µm IAA) seedlings. Six to 11 seedlings were used for each treatment.
4.1 ± 1.5
7.1 ± 2.1
14.8 ± 2.6
0.0 ± 0.0
0.1 ± 0.3
1.5 ± 1.4
3.2 ± 1.1
7.0 ± 1.7
12.7 ± 3.4
0.0 ± 0.0
0.0 ± 0.0
0.0 ± 0.0
We then assessed whether the slr-1 mutant undergoes stage II in lateral root initiation, where some of the divided pericycle cells divide periclinally to produce another cell layer between the pericycle and endodermis layers. We used End199, an enhancer trap line that has the GUS insertion upstream of the SCARECROW gene that is required for radial patterning (Di Laurenzio et al., 1996). This line expresses GUS in the endodermal layer, the cortex/endodermal initials, the quiescent centre in the root tip (Figure 3h) and the daughter cells of the periclinal cell divisions at stage II, and in several types of cells after subsequent stages of lateral root formation (Malamy and Benfey, 1997b). In 5-day-old, light-grown, wild-type seedlings, End199 expression was detected in the dividing cells where lateral root formation occurs, indicating that these cells have already undergone stage II (Figure 3i). In contrast, in 5-day-old, light-grown, slr-1 seedlings, End199 expression was not detected in or around the root pericycle cells, except in the primary root apical meristem (Figure 3k,l). Even in older seedlings or in auxin-treated seedlings, End199 activity was never detected around the slr-1 root pericycle (Figure 3m; Figures 4b; Table 3), whereas the initiation sites with End199 activity were increased in the wild type (Figure 3j; Figure 4b; Table 3). Thus sites of periclinal cell division are consistently lacking in the slr-1 roots, despite the fact that a few anticlinal cell divisions were observed in the slr-1 pericycle. These results indicate that the slr-1 mutation completely blocks periclinal cell divisions.
Taken together, these observations indicate that the slr-1 mutation impedes the early stages in lateral root initiation by partially and completely blocking the anticlinal (stage I) and periclinal (stage II) cell divisions of pericycle cells, respectively.
SLR encodes IAA14, a member of the Aux/IAA protein family
To confirm that SLR encodes IAA14, we generated transgenic plants that express the slr-1-derived mutant IAA14 cDNA (mIAA14) under the control of the 2.0 kb IAA14 promoter. This promoter was found to be sufficient for the expression of the IAA14 gene in the root, but not in the hypocotyl and shoots (see below). All T1 and T2 transformants (pIAA14::mIAA14) had almost the same phenotypes as the slr-1 mutant: no or reduced lateral root formation; little root hair formation; and a reduced gravitropic response by the roots (Figure 5c,d). The shoot and hypocotyl of these plants were essentially normal (data not shown). In contrast, the transgenic plants that express the wild-type IAA14 cDNA under the same promoter (pIAA14::IAA14) were indistinguishable from wild-type plants (Figure 5e). Thus the expression in the wild-type plant of the mutant IAA14 cDNA under its own promoter is sufficient to confer the slr-1 phenotype, and we conclude that SLR encodes IAA14.
The SLR/IAA14 protein consists of 228 amino acids and bears the conserved domains I–IV (Figure 5a). Of the over 20 IAA proteins that have been identified in Arabidopsis, SLR/IAA14 is the most similar to AXR2/IAA7, AXR3/IAA17 and IAA16 (78, 66 and 62% identical, respectively, in amino acid sequence), and these four members fall into the same subclade in the phylogenetic tree of the Aux/IAA proteins (Rogg et al., 2001). Notably, domain II of SLR/IAA14 is completely identical to that of AXR2/IAA7 (Figure 5b).
Additional evidence that SLR encodes IAA14 was obtained from the isolation of an intragenic suppressor mutant of slr-1. Mutagenized slr-1 M2 seedlings were screened for suppressor mutants of slr-1 that can produce lateral roots (see Experimental procedures). This led to the identification of several suppressor mutants, including slr-1R1. This mutant produces a few lateral roots (Figure 5g; Table 1) and has normal root hairs (Figure 5h). Sequence analysis revealed that the IAA14 gene of slr-1R1 contained an additional nucleotide substitution that was predicted to convert the 38th aspartic acid in the N-terminal region between domains I and II to asparagine (Figure 5a). This aspartic acid residue between domains I and II is conserved among several IAA proteins, including the three proteins closest in sequence to SLR/IAA14: AXR2/IAA7, AXR3/IAA17 and IAA16 (data not shown). Thus the slr-1R1 mutant is an intragenic suppressor of slr-1. The existence of independent mutant alleles also confirms that SLR encodes IAA14.
We also produced transgenic plants that ectopically express the slr-1-derived mutant IAA14 gene under the control of the cauliflower mosaic virus 35S promoter (35S::mIAA14), which is constitutively active in most plant tissues (Odell et al., 1985). Most of the 35S::mIAA14 T1 seedlings had slr-1-like phenotypes: fewer root hairs and reduced gravitropic responses by the roots (Figure 5f). However, their root and shoot growth was arrested, resulting in seedling death. Notably, some of the transgenic T1 seedlings expressing the wild-type IAA14 gene under control of the same promoter (35S::IAA14) also had slr-1-like phenotypes similar to 35S::mIAA14 (data not shown). The likely reason for this is discussed below. Thus the ectopic expression of the mutated IAA14 gene causes the slr-1-like phenotype, supporting the notion that a gain-of-function mutation in IAA14 causes the slr-1 phenotype.
SLR/IAA14 expression pattern
Abel et al. (1995) have examined the expression of IAA14 by RNA blot analysis, and have shown that (i) IAA14 is expressed in almost all organs, including the roots, leaves, stems, and flowers; (ii) IAA14 expression in etiolated seedlings is slightly up-regulated by exogenous auxin; and (iii) IAA14 expression is induced by cycloheximide, a translational inhibitor. After we confirmed these observations by RNA blot analysis (data not shown), we produced transgenic plants expressing the 2.0 kb IAA14 promoter linked to the GUS reporter gene (pIAA14::GUS) to examine the tissue specificity of SLR/IAA14 expression. In 7-day-old, light-grown, pIAA14::GUS seedlings GUS activity was detected throughout the root elongation zone, except in the root cap and meristematic regions (Figure 6a). In addition, pIAA14::GUS was expressed in the vascular tissues of the primary roots (Figure 6b). When lateral roots were initiated at the pericycle layer, strong GUS staining was observed in the dividing cells (Figure 6c), consistent with our observation that the slr-1 mutation inhibits lateral root formation. When the pIAA14::GUS seedlings were treated with exogenous auxin (1 µm IAA), GUS activity increased, although there was still no expression in the root cap and meristematic region (Figure 6d), indicating that the 2.0 kb IAA14 promoter is responsible for auxin-induced GUS activity. In the presence or absence of auxin, GUS activity in the hypocotyl and shoot organs was never detected, which is consistent with our observation that the pIAA14::mIAA14 transgenic plants had an essentially normal phenotype in their hypocotyl and shoots (data not shown). Thus the 2.0 kb IAA14 promoter regulates the root expression of IAA14, but other regulatory regions are necessary for IAA14 transcription in the hypocotyl and shoot organs.
SLR/IAA14 is localized in the nucleus
Aux/IAA proteins have conserved nuclear localization signals (Abel and Theologis, 1996; Abel et al., 1994), and some of the Aux/IAA proteins (pea PS-IAA4/6, Arabidopsis IAA1 and AXR3/IAA17) were shown to be localized in the nucleus by transient expression assays using tobacco protoplasts or onion epidermal cells (Abel et al., 1994; Ouellet et al., 2001). To determine if the IAA14 protein is also localized in the nucleus, we generated transgenic plants expressing the mutant slr-1-derived IAA14 protein tagged with the green fluorescent protein (GFP) under the control of the 2.0 kb IAA14 promoter (pIAA14::mIAA14-GFP). These plants also had the slr-1 root phenotypes: no or reduced lateral root formation; reduced root hair formation; and reduced gravitropic responses by the root, indicating that the mutant IAA14-GFP fusion protein is functional in planta (data not shown). As shown in Figure 6(e), in the root epidermal cells of seedlings, strong GFP fluorescence was detected at the position where the nucleus was stained by 4′, 6′-diamidino-2-phenylindole (DAPI; not shown). Thus SLR/IAA14 is a nuclear protein.
SLR/IAA14 represses auxin-induced gene expression
To determine if the SLR/IAA14 protein is involved in auxin-induced gene expression, we examined whether the slr-1 mutation affects expression of the auxin-inducible BA-GUS marker (Oono et al., 1998). The BA-GUS construct contains synthetic auxin-responsive elements that are derived from the A and B domains of the promoter in the PS-IAA4/5 gene, and thus can be used to examine auxin sensitivity (Oono et al., 1998; Rogg et al., 2001). In the absence of exogenous auxin, BA-GUS gives no GUS activity in both the slr-1 and wild-type roots (data not shown). When wild-type seedlings containing the BA-GUS gene were treated with exogenous 10 µm IAA, strong GUS activity was induced at the root elongation zone, as previously described by Oono et al. (1998) (Figure 7a). In the slr-1 root, however, auxin-induced GUS activity was very reduced (Figure 7b). Thus the slr-1/iaa14 mutation decreases auxin-induced BA-GUS expression in the root, suggesting that the mutant IAA14 protein could repress auxin-induced gene expression.
A gain-of-function mutation in SLR/IAA14, an Aux/IAA gene, causes altered auxin responses and defects in growth and development
We have identified an Arabidopsis mutant, slr-1, which lacks lateral roots and has few root hairs, reduced gravitropic responses in the root and hypocotyls, and increased apical dominance. It also responds poorly to exogenously applied auxin. Mapping and sequencing studies revealed that the SLR gene encodes the IAA14 protein, and is thus a member of the Aux/IAA gene family. Gain-of-function mutations in other Arabidopsis IAA genes, namely AXR3/IAA17 (Rouse et al., 1998); SHY2/IAA3 (Tian and Reed, 1999); AXR2/IAA7 (Nagpal et al., 2000); MSG/IAA19 (Tatematsu et al., 1999); and IAA28 (Rogg et al., 2001), also cause altered auxin responses and pleiotropic defects in growth and development. All of these iaa mutations, including slr-1/iaa14, map to the second domain (II) of the four domains that are conserved between the IAA proteins (Nagpal et al., 2000; Rogg et al., 2001; Rouse et al., 1998; Tatematsu et al., 1999; Tian and Reed, 1999; this study). Some Aux/IAA proteins are known to be unstable in vivo (Abel et al., 1994), and thus it is possible that these mutations in domain II serve to stabilize the encoded proteins, which results in altered auxin responses (Nagpal et al., 2000; Rogg et al., 2001; Rouse et al., 1998; Tatematsu et al., 1999; Tian and Reed, 1999). This hypothesis has been supported by recent biochemical experiments (Colón-Carmona et al., 2000; Ouellet et al., 2001; Worley et al., 2000). For example, mutant SHY2-2/IAA3 proteins accumulate abundantly in the shy2-2/iaa3 mutant plant, whereas wild-type SHY2/IAA3 proteins do not accumulate in the wild-type plants (Colón-Carmona et al., 2000). Furthermore, the in vivo half-life of the mutant AXR3-1/IAA17 protein is seven times that of the wild-type AXR3/IAA17 protein (Ouellet et al., 2001). In support of these observations, the slr-1 mutant phenotype segregates as a dominant trait, and expression of the mutant slr-1 IAA14 cDNA under its own promoter is sufficient to confer the slr-1 phenotype in the wild-type plant. Taken together, we conclude that the altered auxin responses and pleiotropic developmental defects in the slr-1/iaa14 mutant are caused by a gain-of-function mutation in IAA14 that probably stabilizes the encoded protein.
Possibly linked to this are our observations of transgenic plants expressing the mutant and wild-type IAA14 cDNA under the CaMV 35S promoter. This promoter is constitutively active in most plant tissues. As when the mutant IAA14 cDNA was under control of its promoter, wild-type plants transformed with 35S::mIAA14 had slr-1 root phenotypes: fewer root hairs and a reduced gravitropic response by the roots. However, the 35S::mIAA14 plants had a more severe phenotype than slr-1 as their root and shoot growth was arrested, resulting in seedling death. This lethal phenotype might reflect the effect of the ectopic expression of the mutant IAA14 protein by 35S promoter. We also noted that several lines of transgenic plants expressing the wild-type IAA14 under the 35S promoter also showed slr-1 features. We speculate that, despite their natural instability, wild-type IAA14 proteins may accumulate in these 35S::IAA14 plants due to overexpression by the 35S promoter. If this is the case, it is consistent with the hypothesis that accumulation of IAA proteins confers altered auxin responses. Biochemical studies on protein stability are needed to further characterize IAA14 protein function.
In this study we also isolated the slr-1R1 mutant, which is an intragenic suppressor of slr-1 whose phenotype is intermediate between that of the slr-1 and wild-type plants. The slr-1R1 mutation was located at an amino acid residue between domains I and II in the IAA14 gene. This residue is conserved among several IAA proteins, including the three that fall into the same subclade as SLR/IAA14: AXR2/IAA7, AXR3/IAA17 and IAA16. Although several intragenic revertant alleles have been identified in the other iaa mutants, mutations in the region between domains I and II have never been found (Nagpal et al., 2000; Rouse et al., 1998; Tian and Reed, 1999). Ouellet et al. (2001) have showed that a mutation in domain I alters the ability of the IAA17 protein to homodimerize, but does not affect its nuclear localization. Although it is not known whether the slr-1R1 mutation affects the stability, cellular localization or homo/heterodimerization ability of IAA14, our finding of the slr-1R1 allele suggests that the conserved region between domains I and II may be important for the function of several Aux/IAA proteins. In the slr-1R1 mutant, defects in root hair formation are almost suppressed (Figure 5h), but lateral root formation is still poor (Table 1). This differential suppression of defects indicates that the slr-1R1 intragenic mutation is more critical for restoring root hair formation than lateral root formation. It is possible that lateral root and root hair formation require different molecular actions of SLR/IAA14 and/or mechanisms involving SLR/IAA14.
Aux/IAA proteins have distinct and overlapping functions in growth and development
Physiological and mutational analyses have shown that there is cross-talk between several plant hormones. In Arabidopsis, most of the auxin-resistant mutants, including the iaa mutants, also resist several other hormones. For example, the axr3-1/iaa17 mutant is resistant to both auxin and ethylene, and has unusual responses to cytokinin (Leyser et al., 1996), while axr2-1/iaa7 is resistant to auxin, ethylene and ABA (Wilson et al., 1990). In contrast, slr-1/iaa14 is specifically resistant to auxin, but is slightly hypersensitive to ABA. These differences in hormone sensitivity also indicate that each Aux/IAA protein participates in distinct cross-talk pathways between auxin and the other hormones.
One of the phenotypes of the slr-1 mutant is its increased apical dominance. This appears to be anomalous given that the other phenotypes of this mutant are more reminiscent of a reduced auxin response. This suggests that the role the mutant IAA14 protein plays in auxin sensitivity and/or signalling that results in apical dominance differs from the role it plays in other auxin-regulated responses. Similar contradictory phenotypes were observed in the axr3/iaa17 mutants, which had increased apical dominance and more adventitious roots, but whose roots grew less markedly in response to exogenous auxin (Leyser et al., 1996). These observations suggest that each IAA protein might take part in distinct pathways in each tissue/organ.
Control of lateral root formation by SLR/IAA14
Our observation with the CycB1;1::GUS and End199 markers revealed that a gain-of-function mutation in the SLR/IAA14 blocks lateral root initiation by partially blocking anticlinal cell division of the pericycle (stage I) and completely blocking periclinal cell division of the divided pericycle cells (stage II). Although both cell divisions are affected by the slr-1 mutation, the difference in the extent of the inhibition suggests that the regulation of the anticlinal and periclinal cell divisions involved in lateral root initiation differs. This has also been suggested by recent observations made using the CycB1;1::GUS marker. Dubrovsky et al. (2000) showed that Arabidopsis root pericycle cells maintain their proliferative activity in the differentiation zone, but only some of the dividing pericycle cells (about 12.5%) undergo lateral root primordium formation, at least in the conditions of growth that those authors used. Although End199 was not used in their analysis, their results suggest that not all dividing pericycle cells in stage I undergo stage II. In contrast, in the conditions of growth that we employed, we found that most of the dividing pericycle cells undergo lateral root primordium formation in wild-type seedlings (Figure 4). This discrepancy from the observations of Dubrovsky et al. (2000) might reflect the different growth conditions between their studies and ours. We grew seedlings on medium containing 1% sucrose under constant light, whereas they grew seedlings on medium containing 3% sucrose under long-day light (16 h light, 8 h dark). It is known that environmental conditions dramatically affect lateral root formation in Arabidopsis (Zhang and Forde, 2000). Further studies on the environmental regulation of lateral root formation may shed more light on this issue. The identification of new factors that are either specifically involved in one cell division or commonly involved in both cell divisions will be also necessary to understand the mechanisms regulating lateral root initiation.
There are several ways the mutant SLR/IAA14 protein could block lateral root formation in the slr-1 mutant. First, the mutant IAA14 protein might change the auxin sensitivity of the pericycle cells by altering auxin-responsive transcriptional networks that involve Aux/IAA and ARF proteins. This, in turn, might affect the anticlinal and periclinal cell divisions in lateral root initiation. Second, the slr-1 mutation might block lateral root formation by affecting polar auxin transport. It is known that auxin transport is important for lateral root formation because (i) auxin transport inhibitors block lateral root initiation (Casimiro et al., 2001; Reed, R.C. et al., 1998); and (ii) several mutants defective in auxin transport (aux1 and tir3) have reduced numbers of lateral roots (Hobbie and Estelle, 1995; Ruegger et al., 1997). However, the slr-1 roots responded normally to the auxin transport inhibitors NPA and HFCA, suggesting that normal polar auxin transport is not affected by the slr-1 mutation. Thus it is unlikely that the slr-1 mutation might block lateral root formation by directly affecting polar auxin transport. Third, the mutant IAA14 protein might directly repress the gene expression necessary for the divisions of the pericycle cells. In Arabidopsis, the ALF4 (ABBERANT LATERAL ROOT FORMATION 4) gene is essential for lateral root initiation (Celenza et al., 1995). A loss-of-function alf4 mutant has no lateral roots, but its root growth is inhibited by auxin to a normal degree, indicating that ALF4 is a positive regulator for lateral root initiation (Celenza et al., 1995). It is possible that the mutant IAA14 protein might repress ALF4 activity in slr-1.
Control of root hair formation by SLR/IAA14
The slr-1 mutant has few root hairs, apart from normal hairs at the junction between the root and hypocotyl. This phenotype could be partially rescued by the application of exogenous ACC, an ethylene precursor, indicating that the slr-1 mutant retains some ability to differentiate root hairs in response to ethylene. The GLABRA2 (GL2) gene, which is required for cell specification of non-hair cell identity in the epidermis, is expressed in the non-hair cell file (Masucci and Schiefelbein, 1996). It has been proposed that auxin and ethylene act downstream of GL2 in promoting root hair formation, because the mutations affecting hormone sensitivity to auxin and/or ethylene do not alter the expression of the GL2 gene (Masucci and Schiefelbein, 1996). For example, the axr2-1/iaa7 mutant resistant to both auxin and ethylene has reduced root hair formation, but has a normal expression pattern of the GL2 promoter::GUS reporter in the non-hair cell file, indicating that the axr2-1/iaa7 mutation acts downstream of GL2 (Masucci and Schiefelbein, 1996). Our observation that the slr-1 mutant can develop at least some root hairs after ACC treatment, despite its resistance to auxin, strongly suggests that the slr-1/iaa14 mutation also acts downstream of GL2. Supporting this is that GL2 promoter::GUS reporter is normally expressed in the non-hair cell file of the slr-1 mutant (unpublished results).
The slr-1 mutant appears to have normal sensitivity to ethylene because (i) the slr-1 mutant developed some root hairs in response to ACC; (ii) like the wild-type plant, slr-1 mutant root growth was inhibited by the exogenous application of ACC (Figure 2c); and (iii) apical hook formation of etiolated seedlings, which is an ethylene-related response, was not affected by the slr-1 mutation (Figure 1h), and this formation could be exaggerated by application of exogenous ACC, as was also observed for the wild-type plant (unpublished results). As the slr-1 mutation results in altered responses only to auxin and not to ethylene, this strongly suggests that auxin acts upstream of ethylene in root hair formation.
SLR/IAA14 may act as a transcriptional repressor of auxin-induced gene expression
Aux/IAA proteins have the ability to interact not only with Aux/IAA proteins, but also with ARF proteins through their conserved domains III and IV (Abel et al., 1995; Kim et al., 1997). Although Aux/IAA proteins are not shown to bind to DNA directly, ARF proteins bind to auxin-responsive elements (AuxREs) (Ulmasov et al., 1997a; Ulmasov et al., 1997b; Ulmasov et al., 1999b). Several observations suggest that interactions between Aux/IAA proteins, between ARF proteins, or between Aux/IAA and ARF proteins regulate auxin-responsive transcription (Guilfoyle et al., 1998). For example, both Aux/IAA and ARF proteins regulate the activity of AuxRE-containing promoters in transient assays (Ulmasov et al., 1997b; Ulmasov et al., 1999a). In addition, auxin-responsive transcription was reduced by gain-of-function mutations in several IAA genes (Abel et al., 1995; Timpte et al., 1994) and by a loss-of-function mutation in ARF7 (Harper et al., 2000; Stowe-Evans et al., 1998). In this study, we showed that the gain-of-function slr-1/iaa14 mutation also reduced auxin-inducible BA-GUS reporter expression in the root. As the BA-GUS construct contains synthetic AuxRE promoters (Oono et al., 1998), our results suggest that SLR/IAA14 protein regulates auxin-induced gene expression, probably as a repressor, through its interaction with ARF proteins.
Further analysis of the molecular function of the SLR/IAA14 protein will help us to understand the mechanisms behind various auxin-regulated developmental processes, especially those operating in lateral root development.
Plant materials and growth conditions
Arabidopsis ecotypes used in this study were Columbia (Col), Wassilewskija (Ws), and Landsberg erecta (Ler). Seeds were surface-sterilized and plated on MS medium containing 1.0% sucrose solidified with 0.5% agar, as described previously (Fukaki et al., 1996a). If plant hormones or chemicals were added, they were derived from 1 µm to 100 mm stocks in DMSO. Kanamycin and hygromycin were added to a final concentration of 25 µg ml−1 from a 50 mg ml−1 stock in water. Plates and plants transferred to soil were grown at 23°C under constant white light as described previously (Fukaki et al., 1996a).
Isolation of the slr-1 mutant
The slr-1 mutation was initially found in the original cop4-1 mutant line (Hou et al., 1993). The cop4-1 mutant was reported to have several phenotypes, including the opening of the apical hook in darkness (hookless phenotype), and agravitropic responses in both roots and hypocotyls (Hou et al., 1993). When we studied the shoot gravitropic response of the original cop4-1 line (Fukaki et al., 1996b), we found that this line had no lateral roots and few root hairs. All F1 seedlings from the cross between the original cop4-1 line and the wild-type plant had a normal apical hook development in darkness, consistent with the recessiveness of the cop4-1 mutation (Hou et al., 1993). However, all F1 seedlings had no lateral roots, few root hairs, and agravitropic responses, suggesting that these phenotypes are due to a dominant mutation that differs from the cop4-1 mutation. In the F2 progeny, the hookless cop4-1 phenotype was found to separate genetically from the other phenotypes, namely the lack of lateral roots, the few root hairs, and the agravitropic responses. All these other phenotypes co-segregated as a dominant trait, indicating that these phenotypes were due to a single dominant mutation (slr-1).
Phenotypic characterization of the slr-1 mutant
Root growth during treatment with various hormones and chemicals was measured under a dissecting microscope. The number of lateral roots was also counted under a dissecting microscope.
GUS activity of CycB1;1::GUS, End199 and BA-GUS was detected by incubating the seedlings containing the transgenes at 37°C for 2 h in 50 mm sodium phosphate pH 7 containing 1 mm potassium ferricyanide, 1 mm potassium ferrocyanide and 0.3 mg ml−1 5-bromo-4-chloro-3-indolyl-glucuronide (X-Gluc). After incubation, the samples were rinsed with 70% ethanol and observed under a Nomarski microscope (Nikkon, Tokyo, Japan).
Mapping and cloning of the SLR gene
DNA from individual wild-type F2 plants from a cross of slr-1 (Col) to the wild type (Ler) was isolated and analysed for co-segregation with various CAPS (Konieczny and Ausubel, 1993) and SSLP markers (Bell and Ecker, 1994). Polymorphisms between Col and Ler ecotypes were found with the use of the g4539 CAPS marker and a PCR product amplified from the region 6 kb north to the mi279 marker with two primers: 5′-TTGGAGACTCGTGAGAGGAGA-3′ and 5′-TTCCTTTATCTCGCACTCTCT-3′. Of 214 chromosomes, two recombinations were found between the mi279-linked marker and the slr-1 mutation. Of 212 chromosomes, 10 recombinations were found between g4539 and the slr-1 mutation. These markers mapped the slr-1 mutation to the genomic region where two IAA genes (IAA1 and IAA14) exist. The IAA14 gene was amplified from the wild-type plant and slr-1 mutant DNA with two pairs of primers: 5′-GTAGAAACGGCAAATTACGTA-3′ and 5′-TCTCTTCTTGCTGTCTATTA-3; 5′-GGTACTCATGTTTTTATAATA-3′ and 5′-TCTTAAGAAATTGTTATGTGA-3′. These primers were based on the genomic sequence containing the IAA14 gene that was denoted the IAA7-like gene on the ESSA I FCA contig (Accession No. Z97335). PCR products were purified and directly sequenced by standard methods using these PCR primers and additional internal primers.
To isolate wild-type IAA14 and slr-1 mutant IAA14 (mIAA14) cDNAs, total RNA was extracted from 3-day-old wild-type and slr-1 seedlings by RNeasy Plant Mini Kit (Qiagen, Hilden, Germany). cDNAs were synthesized with random primers and reverse transcriptase using SUPERSCRIPT II Reverse Transcriptase (Life Technologies, Rockville, MD, USA). From both the IAA14 and mIAA14 cDNAs, a region containing the IAA14 coding region was amplified using two end primers: 5′-CTCGAGCATATTCTGATTTAAGACATA-3′ and 5′-ACTAGTAATCAATGCATATTGTCCTCT-3′. The resultant 0.7 kb PCR fragments, which contained XhoI and SpeI restriction sites at their 5′ and 3′ ends, were cloned into the pGEM-T Easy vector (Promega, Madison, WI, USA). The sequence of the wild-type IAA14 cDNA was the same as the annotated IAA14 mRNA sequence (Accession No. AF34718), and the sequence of the slr-1 mutant IAA14 cDNA had the same mutation as found in the IAA14 gene of slr-1.
To generate both the IAA14 promoter::IAA14 cDNA and IAA14 promoter::mIAA14 cDNA constructs, the 2.0 kb IAA14 promoter region was amplified from wild-type Col DNA using two end primers: 5′-GGGTCGACGTAGAAACGGCAAATTACGTA-3′ and 5′-GGGTCGACTCTCTTCTTGCTGTCTATATA-3′. These primers were based on the genomic sequence containing the IAA14 gene, as described above. The resultant 2.0 kb PCR fragment, which contained SalI sites at both ends, was cloned into pBluescript II SK(+) (Stratagene, CA, USA) and the sequence was confirmed. The 2.0 kb SalI fragment containing the IAA14 promoter was inserted in the SalI site in the plant transformation vector pBI101 (Clontech, Palo Alto, CA, USA). The 0.7 kb IAA14 or mIAA14 coding region described above was then inserted between the 2.0 kb IAA14 promoter and the nopaline synthase 3′ region by replacing the GUS coding region. These constructs were introduced into Agrobacterium tumefaciens MP90 by electroporation, and the strain containing the construct was used to transform wild-type Col plants using the floral dip method (Clough and Bent, 1998). T1 seeds were selected on medium containing 25 µg ml−1 kanamycin, and resistant seedlings were transferred to soil and allowed to set seeds. Homozygous lines were selected by examining the kanamycin resistance of T3 seedlings.
To generate both the CaMV 35S promoter::IAA14 and CaMV 35S promoter::mIAA14 cDNA constructs, the 0.7 kb XhoI–SpeI fragment containing either the IAA14 or the mIAA14 coding region described above was inserted into the XhoI and SpeI sites of the plant transformation vector pDH127, which was modified from pBI121 (Clontech) and has XhoI and SpeI sites between the 35S promoter and the nopaline synthase 3′ region. These constructs were introduced into A. tumefaciens MP90. The strain containing the construct was used to transform wild-type Col plants, and T1 seeds were selected on medium containing 25 µg ml−1 hygromycin, and resistant seedlings transferred to soil and allowed to set seed.
Isolation of intragenic suppressors of slr-1
About 5000 seeds of the slr-1 mutant were mutagenized in 0.2% (v/v) ethyl methanesulfonate (EMS) for 16 h at room temperature. After rinsing the seeds with water, they were surface-sterilized and plated on MS plates. M1 plants grown for 3 weeks in plates were transferred to soil and allowed to set M2 seeds. Approximately 301 500 M2 seeds were surface-sterilized and sown on MS plates. Among the 2-week-old M2 seedlings, slr-1 suppressor candidates that had lateral roots were screened. Many candidates were identified and one of these, the slr-1R1 mutant, was characterized in this study (the other suppressor mutants will be published elsewhere). To locate the intragenic suppressor mutation in IAA14, the genomic region containing the IAA14 coding region was amplified from the slr-1R1 genomic DNA, and the PCR product was purified and directly sequenced using the PCR primers and additional internal primers described above.
Expression of SLR/IAA14
To generate the SLR/IAA14 promoter-GUS reporter construct, the 2.0 kb IAA14 promoter fragment linked to SalI sites at both ends was inserted in the SalI site upstream of GUS in the plant transformation vector pBI101, as described above. This construct was introduced into A. tumefaciens MP90. The strain containing the construct was used to transform wild-type Ws plants, and homozygous T3 kanamycin-resistant lines were selected as described above. GUS staining was performed as described above.
Nuclear localization of SLR/IAA14 protein
To generate the mutant IAA14 (mIAA14)-GFP translational fusion, the modified GFP fragment was inserted in-frame between the last codon and the stop codon of the mIAA14 coding region in the pGEM T Easy vector described above. To do this, inverse PCR amplification of the mIAA14 coding region was performed by amplifying the whole pGEM-T Easy vector containing the mIAA14 coding region using two primers, 5′-TGAACAAAAAAAAAAGAGGACAATAT-3′ and 5′-TGATCTGTTCTTGAACTTCTCCATTG-3′. The 0.7 kb GFP (S65T) (Niwa et al., 1999) coding region with a short linker sequence (GlyGlyGly) at both ends was synthesized by PCR and cloned into the SmaI site in pUC19 (Takara, Japan). Both the 0.7 kb SmaI fragment containing the GFP coding region and the whole T-Easy vector fragment containing the mIAA14 coding region at its 3′ end were ligated together. The sequence of the resultant 1.4 kb mIAA14-GFP fusion was confirmed. The 1.4 kb mIAA14-GFP fragment was inserted between the 2.0 kb IAA14 promoter and the nopaline synthase 3′ region by replacing the GUS-coding region in the IAA14 promoter–GUS construct described above. This construct was introduced into A. tumefaciens MP90 and the strain containing the construct was used to transform wild-type Col plants as described above. T1 seeds were selected on medium containing 25 µg ml−1 kanamycin, and resistant seedlings were transferred to soil and allowed to set seed. T2 seedlings resistant to kanamycin were examined for GFP fluorescence. For GFP analysis, roots were counterstained with 10 µg ml−1 propidium iodide (Sigma, St Louis, MO, USA) and placed on slides in a drop of water. GFP fluorescence was imaged by confocal microscopy (Carl Zeiss, LSM510). The FITC channel (green: GFP) was overlaid onto the rhodamine channel (red: propidium iodide) to identify the GFP-expressing cells. For DAPI staining, roots were stained with 1 µg ml−1 DAPI and the DAPI channel (blue) was used to identify the nucleus.
We wish to thank Xing-Wang Deng for kindly providing the original cop4-1 seeds, Peter Doerner for the CycB1;1::GUS seeds, Philip N. Benfey for the End199 seeds, Yutaka Oono for the BA-GUS seeds, John W. Schiefelbein for the GL2-GUS seeds, Yasuo Niwa for the GFP vector, and Takehide Kato and Yasufumi Daimon for other vectors. We also thank Keiko Uno for screening for slr-1 suppressors, and Ryo Matsui for laboratory assistance. We also thank Miyo Terao-Morita and Takehide Kato for helpful comments on the manuscript. This work was supported in part by a grant to M.T. from the ‘Research for the Future’ program of the Japan Society for the Promotion of Science, a grant to M.T. from the Novartis foundation (Japan) for the Promotion of Science, and a Grant-in-Aid to H.F. for Scientific Research for Encouragement of Young Scientists from the Ministry of Education, Science, Culture, Sports and Technology of Japan.