The physiological and morphological iron deficiency responses are likely under the control of local and systemic signaling pathways. Bienfait (1987) found a local induction of ferric-chelate reductase, acidification, and root hair and transfer cell formation in potato tubers without sprouts. The first evidence of a shoot-derived signal controlling the Fe uptake system in the root was supplied by the grafting experiments of Grusak and Pezeshgi (1996) with the Fe hyperaccumulating pea mutant dgl. The mutant has a constitutively elevated ferric reductase activity in the root, which depends on the genotype of the shoot. In tobacco overexpressing the iron storage protein ferritin, van Wuytswinkel et al. (1999) found an increase of the root ferric-chelate reductase, which normally occurs under Fe deficiency. Further indication of an as yet unidentified systemic Fe deficiency signal addressing the root is provided by the phenotype of the frd3 mutant, former designated as man1 (manganese accumulator1). As a consequence of the Fe undersupply within the leaf cells, frd3 shows a constitutively induced strategy I response in the root (Delhaize 1996, Rogers & Guerinot 2002). Also chloronerva, which shows intercostal chlorosis as an Fe deficiency symptom due to a lack of NA synthesis, has a constitutively upregulated ferric-chelate reductase in the root (Ling et al. 1999).

Morphological changes of the root are also controlled by the iron status of the shoot. Foliar application of an iron-chelate to plants grown on Fe-deficient medium suppresses the formation of extra lateral roots in addition to the ferric-chelate reductase (Moog et al. 1995). Because down-regulation after replenishment showed a different kinetic response, Moog et al. (1995) concluded the morphological and physiological iron deficiency responses are differentially regulated. This hypothesis was corroborated by Schikora & Schmidt (2001) who showed that the iron-overaccumulating pea mutants brz and dgl, which have constitutively high ferric-chelate reductase, are able to down-regulate their transfer cell frequency under sufficient iron supply. With split-root experiments in tomato, Schmidt and Schikora (2001) found that transfer cells are increased on the Fe-deficient half of the split-root; leading them to propose a local control of transfer cell development by iron.

To determine if the root hair branching of Fe-starved plants is under local or systemic control, we conducted split-root experiments combined with a sufficient or deficient shoot. The presence of a sufficient shoot of +/Fe split wildtype plants was supported by the fact that the activity of the Fe deficiency marker ferric-chelate reductase, which is reportedly induced by a systemic signal, was not elevated in either split-root halves over the +/+Fe control (Fig. 6). This indicates that the plant is able to meet its Fe demand via the Fe-sufficient part of the root system. Both split-root halves of +/-Fe Col-0 wildtype plants displayed a frequency of root hair branching similar to +/+Fe plants (Fig. 5, b, d). Therefore, the root hair branching typical of Fe-deficient plants is not induced by a local Fe limitation if the shoot is Fe-sufficient (Fig. 21a). The induction of root hair branching requires, however, the combination of both a systemic Fe deficiency signal and local Fe depletion, as –Fe-halves of the +/-Fe frd3 mutant showed a significantly increased frequency of branched root hairs (Fig. 5f, Fig. 21b). Local Fe deficiency leads to branched root hairs only if the availability of Fe in the leaf cells is decreased. The ferric-chelate reductase is also systemically activated, but in contrast to root hair branching, the reductase is not downregulated by a local Fe abundance (Fig. 21b). Schmidt and Steinbach (2000) observed rather an upregulation of the enzyme activity by a local presence of Fe when the shoot is Fe-deficient due to lowering the root zone temperature. In split-root experiments, Vert et al. (2003) found an upregulation of FRO2 and IRT1 transcripts in the sufficient part of the root, and Schmidt et al. (1996) observed a slight increase of ferric-chelate reductase activity that declines after three days of treatment. However, an increased enzyme activity on the +Fe half of the split-root system was not found in this study. Perhaps, the plants in the experiments of Schmidt et al. (1996) and Vert et al. (2003) had a deficient shoot resulting from the transfer from Fe-sufficient medium to +/Fe conditions, which has already been overcome in this study.

Together, root hair branching is triggered only by a combination of both a local and a systemic deficiency signal. This is in contrast to the ferric-chelate reductase that is induced also in the presence of a high local Fe abundance. Thus, root hair branching and ferric-chelate reductase may share the systemic signal but they differ in the mechanism responding to the local supply.

Figure 21 : Influence of local and systemic signals related to iron (a, b) and phosphate homeostasis (c, d) on root hair development and deficiency markers. (a) If the shoot is Fe sufficient, a local Fe limitation can not activate root hair branching (RHB). Also the Fe-deficiency marker ferric-chelate reductase (FCR) is inactive. (b) Only if a systemic Fe deficiency signal is combined with a local deficiency signal, RHB is induced. FCR is active also in the presence of local Fe. (c) Also in the presence of a sufficient shoot local P depletion leads to an increase in the root hair number (RHN). The P starvation markers Mt4/At4 are not induced under these conditions. (d) If the shoot is P deficient, RHN is increased and Mt4/At4 induced.

Noteworthy is the constitutively high number of root hairs in the H position observed in the frd3 mutant (Fig. 5e), which was not found in Fe-deficient wildtype roots (Fig. 5, a, c) and, therefore, could not be attributed solely to Fe deficiency. The influenced root hair number could have originated from additional effects caused by the frd3 mutation. Accordingly, Delhaize (1996) showed that the man1 mutant mainly accumulates manganese, but also copper, zinc, magnesium, and sulphur, and He et al. (2005) found a stimulating effect of Mn and Zn on the root hair density. Thus, the elevated root hair number of frd3 might rather be due to a defective Mn or Zn homeostasis.

The increased root hair number of Pdeficient plants is induced by a local signal even if the shoot is sufficient, as the number of H cells in the H position was significantly higher in the P-halves of +/P split wildtype plants than in the +P-halves, which were equal to the +/+P controls (Fig. 7, c, h, Fig. 21c). This finding is consistent with Bates & Lynch (1996) who found a similar local control of the root hair elongation under P depletion. Using a geometric model, Ma et al. (2001b) calculated a synergistic effect of root hair length and density on P acquisition efficiency and argued for a coordinated regulation. A sufficient shoot of +/-P wildtype plants is presumed basing on the fact that the expression of the phosphate starvation marker Mt4, and the P starvation-inducible phosphate transporters LePT1 and LePT2 in the roots are systemically suppressed (Burleigh & Harrison 1999, Liu et al. 1998, Fig. 21c). In the present study, it was, moreover, shown that a long-distance P signal cannot be overruled by a sufficient local P supply (Fig. 21d). Support for this statement comes from the pho1 mutant that exhibited a constitutively elevated root hair number under all treatments similar to /-P Col0 (Fig. 7g). This increased root hair number was not apparent when the shoot was removed (Fig. 7m). The shoot-derived P deficiency signal is, therefore, dominant to a local sufficiency signal, which might allow the plant to meet an increased P demand, e.g. during seed maturation from nutrient-rich patches. Also At4 can not be downregulated in the pho1 mutant (Burleigh & Harrison 1999, Fig. 21d). The increase of the root hair number and of At4 induction may be regulated by a common systemic signal. In addition, a local P depletion was able to trigger root hair formation. Because the same phenotype was induced by a systemic or a local deficiency signal both pathways may discharge into the same target.

Together, this study established further differences between Fe and P regulated root hairs. Root hair branching of Fe starved plants is repressed by either a local or a systemic sufficiency signal, while the increase in the root hair number of P depleted plants is triggered by a systemic or a local deficiency signal. In both cases, root hair development may share common systemic signaling components with physiological starvation markers, but are differentially regulated from them with respect to a local signaling mechanism.

The erh1 and erh3 mutants formed more root hairs in H and N position than the wildtype (Table 5) indicating that the ERH1 and ERH3 genes are important for proper epidermal cell development also in older plants. The increase in the root hair number under Fe and P deficiency was similar to the wildtype; although, the size of the increase was slightly lower (Table 5). This suggests the genes may not directly be involved in the nutrient-sensitive changes of the root hair number. ERH3 is a katanin-p-60 protein involved in microtubule organization and cell wall biosynthesis. In the erh3 mutant, the identities of H and N cells and the cell identities of other root tissues are changed (Burk et al. 2001, Webb et al. 2002). Thus, the ERH3 gene may be required at an early developmental stage, which is mainly independent of the nutrient status. The nutritional signal may enter the root hair developmental pathway downstream of ERH3 action (Fig. 22, arrowhead 1).

The H and N cell identities are determined by a transcription factor cascade in the late meristematic region involving a MYB protein, which is either WER or CPC, the WD40 repeat protein TTG, and the bHLH proteins GL3 and EGL3. Activation of this cascade leads to GL2 expression in the N position whereby inhibiting root hair formation (Schiefelbein & Lee 2006). Under the tested growth conditions, the root hair patterning of the wer, cpc, ttg, and gl2 mutants differed from the wildtype in H and N positions. With the exception of ttg, the Fe and P deficiency-specific increase in the root hair number in the H position was also different in this group of mutants. These observations could be interpreted in two ways. One possibility is that the formation of root hair cells adapted to the Fe and P supply is directly linked with the action of the transcription factor cascade leading to GL2 expression (Fig. 22, arrowhead 2). However, all these mutants were able to increase their root hair frequency in response to Fe or P starvation; although, not to the same degree as the wildtype. This observation indicates that the WER/CPC/GL2 pathway may not be a direct target of the nutritional signals.

With the exception of wer, the erh1, erh3, cpc, ttg, and gl2 mutants were not able to develop branched root hairs. These genes could be a prerequisite for the branching procedure or be directly involved in branching. Ohashi et al. (2003), in contrast, observed the formation of branched root hairs in the gl2-5 mutant, but these root hairs were formed rather irregular and not in a transversal layer. The difference between the gl2-1 phenotype in this study and the gl2-5 phenotype observed by Ohashi et al. (2003) could have resulted from using different mutant alleles or growth conditions.

The RHL genes are part of the topoisomerase VI complex and are important for ploidy-dependent cell growth. All the rhl mutants all did not develop root hairs under the conditions described (Table 5), indicating endoreduplication is required for root hair formation independent of age or nutrient supply (Fig. 22).

RHD6 is involved in positioning the root hair initiation site near the apical end of the epidermal cell (Masucci & Schiefelbein 1994). In the present study, RHD6 gene function was essential for root hair initiation independent of age or nutrient status. Under all conditions, the rhd6 mutant was devoid of root hairs (Table 5). The nutritional signal could be a prerequisite for RHD6 action or act together with or downstream of RHD6.

TRH1 is a potassium transporter that accelerates auxin transport in the root cap (Rigas et al. 2001, Vincente-Agullo et al. 2004). Under control and –Fe growth conditions, root hair initiation and tip growth were dependent on TRH1 function (Fig. 9, a-b; Table 5). In contrast, when grown under P deficiency, the trh1 mutant developed normal root hairs (Fig. 9c; Table 5). Thus, TRH1 is important for root hair initiation and tip growth in sufficient and Fe-deficient but not of P-deficient plants (Fig. 22). The behavior of trh1 is similar to the response of the axr1, axr2, and ein2 mutants under the same growth conditions (Schmidt & Schikora 2001). This supports the hypothesis of Schmidt and Schikora (2001) that root hair development of Fe- and P-deficient plants is regulated by different pathways and that root hairs formed under P starvation are at least partially regulated independent of components belonging to the auxin signaling pathway.

The tip1, rhd2, rhd3, and rhd4 mutants developed short root hairs under all growth conditions at a frequency that did not deviate from the wildtype (Fig. 9; Table 5). The S-acyl-transferase TIP1, the NADPH oxidase RHD2, and the small G protein RHD3 are assumed to be involved in different processes of vesicle traffic during the tip growth of root hairs (Hemsley et al. 2005, Foreman et al. 2003, Hu et al. 2003). KJK/CSLD, RHD1, and LRX are involved in proper cell wall assembly during root hair initiation and tip growth (Wang et al. 2001, Favery et al. 2001, Schiefelbein & Somerville 1990, Seifert et al. 2002, Baumberger et al. 2001). Also in this group of mutants, the number of root hairs was unaltered (Table 5, Table 6), but the root hair shape was impaired as previously described. Hence, it appears that after the proper position and frequency of root hairs in response to the nutrient supply has been determined, the same root hair-specific vesicle transport and cell wall synthesis machinery performs the elongation of the root hairs. This suggests that the nutritional signal enters the root hair developmental pathway upstream of root hair tip growth (Fig. 22, arrowhead 3).

Together, the data indicate that Fe and P influence root hair development at the stage of root hair specification or root hair initiation upstream of components involved in tip growth.

Root epidermal cells originate from the epidermal initials in the meristem. A periclinal division initially forms two daughter cells, of which one remains an initial and the other undergoes a second periclinal division producing two cells, one outside the other. Each of these daughter cells then divides anticlinally to produce cell files, the outer one generates the lateral root cap and the inner one the epidermis (Dolan et al. 1993). Dependent on the underlying cortical cells, trichoblast and atrichoblast files are formed. The trichoblasts are shorter than the atrichoblasts and divide more often. They have a denser cytoplasm and a lower degree of vacuolization. Occasionally, the trichoblasts divide longitudinally; thereby the daughter cell in the N position switches its identity into an atrichoblast (Dolan et al. 1994, Berger et al. 1998a). These studies have been performed with 5dayold seedlings that develop a root hair in nearly all H positions (Dolan et al. 1994, Galway et al. 1994, Lee & Schiefelbein 2001). In plants older than 5 days, only about one-half of epidermal cells in the H position developed a root hair and this frequency was sensitive to the Fe and P supply (Fig. 3a; Table 5). It was, therefore, determined if the root hair cells in the H position can already be distinguished from the non-hair cells in the H position by a stronger cytoplasmic staining and lower degree of vacuolization in the late meristematic region. However, both features did not reflect the root hair patterning in adaptation to the Fe or P supply (Fig. 11, bf). The non-hair cells in the H position showed the same cytological features as the root hair cells and they have not completely been converted from H cells into N cells. The differential cytoplasmic staining and the degree of vacuolization reflect, therefore, more a developmental state unaffected by nutrient signals rather than an early indication of future root hair outgrowth. Thus, the nutritional signal enters the root hair developmental pathway after cell characteristics like a denser cytoplasm and a lower degree of vacuolization have been established.

ERH3 is required for the correct positioning of cells with a denser cytoplasm in the H position. Because this feature is not altered under the investigated conditions, the nutritional signal may enter the developmental pathway downstream of ERH3 action (Fig. 22, arrowhead 1).

In seedlings, the gl2 mutant develops root hairs in the N position, as do mutations in its upstream regulators WER and TTG, but displays, in contrast to them, the differential cytoplasmic staining of the wildtype. Thus, the gl2 mutation alters only a subset of N cell differentiation processes and does not cause the complete conversion of N into H cells (Masucci et al., 1996). In this study, the gl2 mutant showed an increased number of root hairs in H and N position when compared to the wildtype under all growth conditions (Table 5). Under Fe and P starvation, the root hairs in the H position were not increased to the same degree in the gl2 mutant as in the wildtype suggesting an involvement of GL2 in the nutrient-sensible root hair development. Using GUS plants, we determined if the non-hair cells in the H position are linked with GL2 expression in the H position. Additionally, we investigated the GUS expression pattern of CPC, a negative regulator of GL2 (Wada et al. 2002). Similar to the situation in the postembryonic root, nearly all H positions showed no CPC-GUS or GL2-GUS staining (Fig. 11, g-l; Table 7). Fe or P had, thus, no impact on the spatial expression pattern of CPC and GL2. The data indicate that the CPC- and GL2-related pathway primarily controls the nonhair specification in the N position as discussed by Masucci et al. (1996) and not the development of non-hair cells in the H position observed in this study. It is possible that the transcription factor cascade leading to GL2 expression is part of a common mechanism that specifies the positional information irrespective of age or nutrient status. The nutritional signal rather modulates the abundance of hair or non-hair cells in the H position. However, it is also possible that the root hairs in Arabidopsis might be regulated by fine tuning the level of GL2 or CPC in the N position that was not visible with the GUS staining.

In the rhl mutants, GL2 is correctly expressed leading Sugimoto-Shirasu et al. (2005) to conclude that RHL function is independent of the GL2 activity. Since the rhl mutants are not able to induce a denser cytoplasm and lower degree of vacuolization (Schneider et al. 1997), as did Fe- and P-deficient plants in the present study, RHL function may be a prerequisite for Fe- and P-dependent root hair development, which would than act downstream of RHL (Fig. 22, arrowhead 4).

In summary, the Fe and P nutritional status did not affect the spatial distribution of GL2, CPC, and the mechanism causing the differential cytoplasmic staining in the late meristematic region of the root. Thus, Fe and P may influence root hair development after these characteristics have been established. Together with the mutant analysis (Table 5), the results indicate that the nutritional signal may modulate root hair development at the stage of root hair initation (Fig. 22, between arrowhead 3 and arrowhead 4).