3 RESULTS

3.1  Root hair patterns of the Arabidopsis wildtype adapted to the Fe and P availability

Iron- and phosphate-deficient plants increased the number of root hairs (Fig. 2, a-c). Under P starvation, root hairs were also markedly elongated; their length increased from 0.2 ± 0.01 up to 0.6 ± 0.02 mm (Fig. 2c). For a more detailed phenotypic analysis of root hairs, which are developed in adaptation to the Fe and P availability, root hair patterns were analyzed in cross- sections (Fig. 2, d-f; Fig. 3).

	Figure 2 : Root tips (a-c) and hand-cut sections (d-f) of 18-day-old Col-0 plants grown under sufficient nutrient supply (a, d), Fe deficiency (b, e), or P deficiency (c, f). Bar = 250 µm (a-c) and 20 µm (d-e). The arrow head indicates a branched root hair out of focus.

	Figure 3 : Root hair patterns of the primary roots of 18-day-old Arabidopsis plants grown under sufficient nutrient supply (control), without Fe (-Fe) or P (-P) for seven days, or without both nutrients for four days (-Fe & -P). (a) % of root hairs in the H position. 100% correspond to the total number of epidermal cells that had contact to two underlying cortical cells. (b) % of root hairs in the N position. 100% correspond to all epidermal cells with contact to one underlying cortical cell. (c) % of branched root hairs. 100% corresponds to the sum of root hairs in the H and N position. n = 20 plants.

In contrast to seedling roots, in which nearly all epidermal cells in the H position develop a root hair (Galway et al. 1994, Masucci et al. 1996; Table 5), in the aging primary root and in laterals this number was considerably reduced (Fig. 3a; Table 5). Under sufficient nutrient supply, only 34% of the H positions formed a root hair (Fig. 3a). This proportion was slightly increased under Fe deficiency (P value = 0.1 with n = 10 plants, P = 0.005 with n = 20). However, P-deficient plants nearly doubled their root hair number in the H position (Fig. 3a). In the N position of sufficient plants, no root hairs were formed (Fig. 3b). Under Fe and P deficiency, a significant increase of root hairs in this position occurred, which was more pronounced in plants grown without P (P ≤ 0.05; Fig. 3b). In Fe-limited plants, nearly one- half of all root hairs were branched at their base. Branching was never observed in P-starved plants (Fig. 3c; Fig. 4, b-c). Thus, it appears that the increase in the absorptive surface area is realized by different developmental programs in response to Fe and P deficiency, either by the formation of branched root hairs as in the case of Fe plants or by an increase in the number of root hairs as in the case of –P plants. Lateral roots behaved in the same manner (Table 5). If the plants were cultivated in the absence of both Fe and P, the root hair frequency is intermediate between the individual Fe or P deficiencies, indicating a contrasting effect of the two deficiency signals on the modulation of the root hair number (Fig. 3a). The rate of root hair branching was as high as under Fe-limitation alone, suggesting the stimulation of root hair branching under Fe deficiency abolishes the suppression of branching in P-deficient plants (Fig. 3c). The branched root hairs exhibited a regular shape. Always two branches of the same length developed that were arranged in the transversal layer (Fig. 2e; Fig. 4c).

	Figure 4 : Cryo-scanning electron microscope images of Fe-sufficient (a) and Fe-deficient (b, c) Arabidopsis roots. Bar = 20 µm.

To determine whether the root hair phenotype of Fe- or Pdeficient plants is controlled by a local or a systemic signaling pathway, we conducted split-root experiments, in which the root system of adult Arabidopsis plants was divided into two nearly equal parts. One-half of the root system was exposed to control medium and the other half to Fe- or Pdeficient medium. A reaction of the deficient part of the root system is indicative of a local control, while a response on both halves of the root system points to a systemic regulation. In the case of both nutrients, the root hair phenotype of different experimental variations was investigated. Initially, plants with a sufficient shoot were treated for two or seven days; time points corresponding to the state of early or late adaptation to nutrient shortage. Subsequently, plants with an Fe- or Pdeficient shoot and plants without a shoot were analyzed for root hair development. The nutrient-deficient shoot was achieved by utilizing mutants with reportedly disrupted xylem loading of either Fe or P. In the case of Fe, we made use of the frd3 mutant (Green & Rogers 2004). In frd3, all components following the xylem loading (Fig. 1) suffer from a lack of iron, and a systemic deficiency response is triggered. An analougous mutant defective in the xylem loading of P is pho1 (Poirier et al. 1991).

3.2.1  Regulation of root hair development under Fe deficiency

As a control for the induction of the root Fe deficiency response, the ferric-chelate reductase activity in the roots was measured (Fig. 5). The enzyme showed a 24-fold induction in -/Fe plants; the activity in both split-root halves of +/-Fe plants was equal to the +/+Fe controls. This finding indicated the existence of a systemic signal that represses the Fe deficiency responses.

	Figure 5 : Ferric-chelate reductase activity in the roots of plants that were split two days on +/+Fe, -/-Fe, or +/-Fe agar media. The reductase was measured as an end-point determination. Results are an average of three independent experiments, each with 10 plants. Treatment for three days gave the same results.

	Figure 6 : Split-root experiments related to Fe nutrition. Roots of 25-day-old Arabidopsis were divided into two approximately equal parts and grown on +/+Fe, /Fe, or +/Fe agar medium as indicated in the x-axis of each bar graph. The mean percentage ± standard error of root hairs in the H position and of branched root hairs was determined after two (a, b) and seven days (c, d). The split-root experiment was also conducted with the frd3-1 mutant that has an Fedeficient shoot (e, f). The occurrence of root hairs in the N position was below 1% in all treatments. Numbers above the bars in the charts indicate the P value that results from the respective ttest comparing either +/+Fe with /Fe or the split+ with the split- halves of the root. n = 10.

In the roots of /Fe split plants, the frequency of root hair branching increased compared to +/+Fe split plants (Fig. 6, b, d) as in the case of non-split plants (Fig. 3c). A slight increase in the root hair number was not statistically significant (Fig. 6, a, c). In the +/Fe wildtype, the frequency of branched root hairs was low and did not differ significantly in both split-root halves after two and seven days (Fig. 6, b, d). Because the branching frequency in -/+ split-root plants was similar to that in +/+Fe controls, a systemic repression of root hair branching caused by a sufficient shoot can be assumed. The formation of branched root hairs in +/+Fe split-root plants (Fig. 6, b, d) was higher than in Fe-sufficient non-split control plants (Fig. 3c) and might have been caused by wounding stress resulting from the splitting procedure.

Split-root plants with an Fedeficient shoot like in the case of the frd3-1 mutant revealed a marked increase in the frequency of branched root hairs under /Fe conditions compared to +/+Fe split plants (Fig. 6f). If the systemic Fe deficiency signal of frd3-1 was combined with a local +Fe signal in +Fe halves of +/Fe plants, a local repression on root hair branching was observed (Fig. 6f). In addition, root hairs in the H position were increased compared to the wildtype under all conditions (Fig. 6e). Together, the root hair branching typical of Fe-deficient plants only occurred if both a local and a systemic Fe deficiency signal were present. In other words, any Fe sufficiency signal was able to repress the response, regardless of its local or systemic origin and acted, thus, dominantly.

The most significant effect of P deficiency (/P) on the enlargement of the root surface was an increase in the number of root hairs in the H position (Fig. 7, a, d) and N position (Fig. 7e). Additionally, a significant decrease of the root hair branching frequency was observed (Fig. 7, c, f). In the +/P split wildtype, root hairs in the H position were significantly increased in the –P halves compared to the +P halves after a treatment for two and seven days (Fig. 7, a, d). The suppression of root hair branching also followed a local –P signal (Fig. 7, c, f).

Plants with a P-deficient shoot resulting from the pho1-2 mutation showed a P deficiency response with regard to the frequency of root hairs in the H position under all treatments (Fig. 7g). The systemic P deficiency signal was not suppressed by a sufficient local P concentration, indicating a dominant role for a systemic P deficiency signal over a signaling pathway perceiving a local abundance of the nutrient. Furthermore, the number of root hairs in the N position was significantly increased in –P root-halves of +/P pho1-2 (Fig. 7h), pointing to a local effect in combination with a systemic P deficiency signal. The frequency of branched root hairs was similar to the wildtype, indicating that a local suppression by low P was independent of the systemic P status (Fig. 7, c, f, i, l, o).

	Figure 7 : Split-root experiments related to P nutrition. Plants were grown on +/+P, /P, or +/P agar media. +/+P is the same as +/+Fe in Fig. 2, but was determined separately. The mean percentage ± standard error of root hairs in the H position root hairs in the N position, and branched root hairs was determined after two (a, b, c) and seven days (d, e, f). (g, h, i) Split-root experiments with the pho1 mutant that had a P-deficient shoot and (j, k, l) with wildtype plants and (m, n, o) pho1 whose shoot was removed. Numbers above the bars in the graph indicate the P values that results from the respective t-test comparing either +/+P with -/-P or the split+ with the split- halves of the root.

A local –P signal in wildtype roots without a shoot caused a significant increase of root hairs (Fig. 7j). The pho1-2 mutant without a shoot reacted similarly to the wildtype, providing evidence that the behavior of the mutant was the result of a shoot-born signal, communicating the P status of the shoot to the root (Fig. 7m). Together, the increase in the root hair number of P-depleted plants was triggered by the presence of a P deficiency signal, regardless of its local or systemic origin. In other words, the P deficiency signal acted in a dominant way.

Mutants with defects in the root hair developmental pathway have been investigated mainly in 3-5-day-old roots (Table 2). At this seedling stage, nearly all epidermal cells in the H position form a root hair in the wildtype (Dolan et al. 1994, Galway et al. 1994, Lee & Schiefelbein 2001). This was also observed in the present study when the plants were germinated on Psufficient or Pdeficient media (Table 5). However, if the plants were germinated on Fe-depleted medium no root hairs developed (Table 5). This result was similar to 2-week-old plants that also formed no root hair during the first 3 days after transfer to Fe medium. Thereafter, root hairs developed in both cases. To determine whether defects in seedling root hair development also affect root hair formation of aging plants during the response to nutrient starvation, 19 mutants harboring defects in different stages of seedling root hair development were grown in the presence and absence of Fe and P. Mutants affected in different processes of root hair specification, initiation, bulge formation and tip growth were included in this study (Table 5).

3.3.1  Analysis of mutants with defects in root hair specification

The ERH1 and ERH3 genes are required for correct differentiation of hair and non-hair cells in positions corresponding to the underlying cortex (Schneider et al., 1997). ERH3 encodes a kataninp60 protein involved in the regulation of microtubules and cell wall biosynthesis (Burk et al. 2001, Webb et al. 2002). Under all conditions, both mutants formed more root hairs than did wildtype plants in the H position (Table 5). This might in part be due to a significantly higher number of cortex cells in both mutants, increasing the epidermal cell number in the H position. However, if this were taken into account, the root hair number would also be higher. In the N position, more root hairs were formed as well. This indicates that the ERH genes might be involved in rhizodermal cell differentiation also of elder plants. An increase in root hair frequency in response to Fe and P deficiency was, however, apparent when compared to the sufficient control, but the factor was slightly lower than in the wildtype suggesting the response to the nutrient supply was only slightly influenced in erh1 and erh3. The number of branched root hairs formed in response to Fe deficiency was significantly lower in erh plants.

Table 5: The effect of Fe and P deficiency on root hair formation in the primary roots of 3-week-old Arabidopsis wildtype and mutant plants. Values represent the number (mean ± S_E) of the indicated cell type per cell layer. Ten roots were scored for each genotype and treatment.

^a A comparison with the primary roots of the wildtype under the respective growth condition resulted in a P value ≤ 0.01. In addition, these data are marked in blue. ^b Only small bulges appeared that were not visible in the cross-sections. The value represents root hairs that were elongated. ^c Under P deficiency, the rhd2 mutant also developed small bulges, but in cross-sections, distinct rhizodermal cells showed a strong toluidine blue staining. These cells were counted as root hair cells. ^d mutant background is not Col-0. n.d. = not determined.

WER is a negative regulator of root hair development in the N position. WER encodes a MYB-related protein that is preferentially expressed in non-hair cells and is required for the appropriate level and pattern of CPC and GL2 transcription. In wer seedling roots, nearly all rhizodermal cells differentiate into root hairs cells (Lee and Schiefelbein 1999, 2002). In this study, the root hair number of wer was not as high as described for the seedling roots. The typical phenotype was only observed a few days after germination. Thereafter, under control conditions the aging primary root and newly formed laterals formed only bulges under control conditions that did not elongate. When grown in media lacking either Fe or P, a high number of hairs was formed in wer plants. The number of hairs in H position was similar to that of the wildtype under Fe-deficient conditions and was slightly reduced in P-deficient roots, but the frequency of hairs in N position was considerably enhanced under both Fe- and P deficiency (Table 5). Thus, the WER gene function seemed to be involved in the root hair development of aging plants: although, the effect of the wer mutation was not as strong as in seedling roots.

CPC is a positive regulator of root hair development in the H position. CPC encodes a MYB-like protein that is expressed in the N position due to positive regulation by WER. CPC moves from the N cells into the neighboring H cells, where it represses its own expression and the expression of WER and GL2 (Wada et al. 1997, 2002, Lee & Schiefelbein 2002). Under control conditions, roots of the cpc mutants formed only a few root hairs that were randomly distributed along the root. The appearance of the phenotypes of cpc roots under –Fe and P conditions were similar to the wildtype: although, the number of hairs was lower (P < 0.01; Fig. 8, a-c; Table 5). Ectopic root hairs were not produced in cpc roots.

TTG encodes a small protein with WD40 repeats (Walker et al. 1999). The root hair pattern of ttg seedlings is abolished in H and N positions (Galway et al. 1994). Under all three growth conditions, the frequency of root hairs in the H position was not significantly different from the wildtype (Table 5). The number of hairs in the N position was drastically enhanced. In P-deficient roots, 30% of the epidermal cells in N position were developed into root hairs. This indicated that TTG might be more important for non-hair cell development in the N position than for the specification of non-hair cells in the H position that occurs in older plants in adaptation to the nutrient supply. In addition, roots of P-deficient ttg plants produced root hairs which were clearly longer than those of the wildtype under similar conditions (1.0 ± 0.02 mm, Fig. 8f).

GL2 inhibits root hair development in the N position (Masucci et al. 1996). GL2 encodes a homeobox-containing transcription factor and is preferentially expressed in atrichoblasts (Wada et al. 1997, Masucci et al. 1996). The root hair density of gl2 mutant plants was increased in the H and N position and differed between the growth types (Table 5).

	Figure   8 : Root tips of the cpc and ttg mutants grown under control conditions or in the absence of Fe or P. (a) cpc control, (b) cpc –Fe, (c) cpc –P, (d) ttg control, (e) ttg –Fe, (f) ttg –P. Bar = 250 µm.

Despite marked changes in the root hair frequency in the H and N position among the wer, ttg, gl2, and cpc mutants compared to the wildtype under all growth conditions, the number of root hairs was increased in response to Fe or P starvation. Thus, the nutritional signal can be perceived and translated in this group of mutants. Interestingly, in wer mutants branched root hairs with a frequency typical of –Fe wildtype plants were induced by Fe deficiency. In contrast, cpc, ttg, and gl2 had a significant lower number of branched hairs.

RHL1, RHL2, and RHL3 are subunits of the topoisomerase VI complex involved in endoreduplication, a process in which one to several rounds of DNA replication take place without mitosis. Endoreduplication is important during cell differentiation and expansion (Sugimoto-Shirasu et al. 2005, Joubès & Chevalier 2000). Under the present conditions, rhl1, rhl2, and rhl3 were dwarf and their roots remained hairless when grown on standard medium or under either Fe or P deficiency, suggesting an essential function of endoreduplication in root hair development during adaption to the Fe and P supply (Table 5).

The RHD6 gene promotes root hair initiation and is associated with the establishment of epidermal cell polarity (Masucci & Schiefelbein 1994). In roots of the rhd6 mutant, no hairs were formed under all three growth conditions indicating that the gene was necessary for root hair development independent of the nutrient supply.

TRH1 is required for root hair initiation and tip growth. TRH1 is a potassium transporter that accelerates auxin efflux in the root cap (Rigas et al. 2004, Vincente-Agullo et al. 2004). Plants lacking the TRH1 gene function only formed bulges that did not elongate under control and Fe conditions (Fig. 9, a, b). The trh1 mutant produced normal root hairs under P conditions in a frequency and positional pattern that did not differ from the wildtype (Fig. 4c; Table 5). Thus, TRH1 is possibly not essential for the formation of root hairs in response to P deficiency.

	Figure   9 : Root tips of the trh1, tip1, and rhd3 mutants grown under control conditions or in the absence of Fe or P. (a) trh1 control, (b) trh1 Fe, (c) trh1 –P, (d) tip1 control, (e) tip1 –Fe, (f) tip1 –P, (g) rhd3 control, (h) rhd3 Fe, (i) rhd3 –P. Bar = 250 µm.

The S-acyl transferase, TIP1, is important for vesicle traffic during root hair initiation and tip growth. In tip1 roots, only short root hairs are formed that are often branched (Ryan et al. 1998, Parker et al. 1998, Hemsley et al. 2005). The frequency of these short root hairs did not differ markedly from the wildtype under all growth conditions. Root hairs of P plants were longer than those of the control plants (Fig. 9, df; Table 5). Under control and P conditions but not in medium lacking Fe, tip1 plants produced significantly more cortical and epidermal cells (Table 5). Independent of the growth conditions, the majority of the hairs was branched. Under control conditions, nearly all hairs had two tips (87.5%); the percentage of branched hairs was slightly lower in Fe and P plants (Table 5). The phenotype of tip1 suggests that TIP1 function is important for the proper initiation of only one root hair tip and for tip growth independent of the Fe or P availability. Since the number of cortical and epidermal cells was increased in the tip1 mutant, the gene might also be involved in early developmental stages of these tissues.

RHD2 is an NADPH oxidase important for root hair tip growth (Foreman et al. 2003, Schiefelbein & Somerville 1990). The rhd2 mutant produced short root hairs under control and Fe-deficient conditions that had a normal degree of branching. When grown in P-deficient medium, only bulges were produced that did not elongate. The number of these bulges was similar to the root hair frequency of the wildtype (Table 5). RHD3 encodes a small G protein which is required for ER and golgi vesicle trafficking (). The rhd3 mutation causes short and wavy root hairs (Schiefelbein & Somerville 1990). In the present study, rhd3 mutants also produced short root hairs under all conditions (Fig. 9, g-i). The number of root hairs was similar to the wildtype. As no other characteristics were altered in the rhd2 and rhd3 mutants, both genes may be required for root hair elongation after the appropriate position and number of root hairs in response to the Fe and P supply has been established and after a proper bulge has been formed in the correct position of the epidermal cell.

RHD4 is a further gene active in the maintenance of root hair elongation (Schiefelbein & Somerville 1990). Root hair density under –Fe and –P conditions in rhd4 mutants did not deviate significantly from the wildtype except for the number of root hairs in H position under Fe deficiency, which was significantly higher than in the wildtype. Root hairs of the rhd4 mutant were generally shorter than in the wildtype.

A mutation in the cellulose synthase-like protein KJK/CSLD3 causes a burst of the root hairs after swelling formation (Favery et al. 2001, Wang et al. 2001). When grown under either Fe or P deficiency, kjk mutants displayed a phenotype almost similar to the wildtype with respect to the root hair number, but most of the hairs were ruptured at their tip and had irregular lengths. A considerably lower frequency of root hairs in the H positions was found under control conditions. In contrast to the wildtype, only few branched root hairs developed in response to Fe starvation (Table 5). In contrast to the kjk mutation, bulges of csld3 1 mutants did not elongate. The frequency of the bulges did not significantly deviate from the wildtype under all growth conditions (Table 6). Thus, KJK/CSLD seems to be involved in root hair tip growth independent of the nutrient supply.

Defects in the RHD1 locus lead to larger bulges at the base of root hairs (Schiefelbein & Somerville 1990). RDH1 is a UDP-D-glucose-4epimerase that galactosylates xyloglucan and arabinogalactan, which are pectin and cell wall structural protein constituents, respectively (Seifert et al. 2002). Under the present conditions, the primary root of rhd1 plants was completely devoid of hairs (Fig. 10a), with the exception of –P plants, in which some hairs were occasionally observed (Table 5). In laterals, root hair density of rhd1 roots was not significantly different from the wildtype under all growth conditions, when only those trichoblasts were considered that succeeded in initiating root hairs (Table 6). The non-hair cells were characterized by wide bulges that comprised the entire epidermal cell wall (Fig. 10b). This phenotype suggests that RHD1 is important for proper epidermal cell development but that root hair initiation and tip growth of laterals might not strictly depend on RHD1 function.

Table 6 : The number of root hairs or root hair bulges per mm. The plants were cultivated under sufficient nutrient supply (control) or in media without Fe (Fe) or P (P). Root hairs were counted in the root hair zone of root tip within the 2^nd mm behind the apex. n = 20.

	Figure 10 : Roots of the rhd1 mutant grown in P deficient medium. (a) The primary root did not form root hairs. (b) Epidermal cells of lateral roots exhibited large bulges and normally shaped root hairs. Bar = 100 µm.

The chimeric leucine-rich repeat/extensin protein, LRX1, is localized in the cell wall of root hairs. LRX is a possible regulator for proper cell wall assembly (Baumberger et al. 2001). The lxr mutants show an irregular root hair length with hairs often being shorter, and with swellings either at the basis or along the stalk. Root hairs of lrx roots were often ruptured at their tips. The mutant responded to Fe and P depletion with an increase in the number of root hairs. However, fewer hairs than in the wildtype were formed in response to P deficiency. Under –Fe conditions, most of the hairs (80%) in the lrx mutant were branched. A significant higher number of branched hairs was also formed under control conditions (Table 5). LRX may, thus, be involved in root hair initiation and elongation after the proper number of root hairs in the correct position has been determined.

In 5-day-old seedlings, nearly all epidermal cells in the H positions form a root hair (Dolan et al. 1994, Galway et al. 1994, Lee & Schiefelbein 2001; Table 5; Fig. 11a). In the late meristematic region of the root, nearly all H positions show a more intense cytoplasmic staining and lower degree of vacuolization (Dolan et al. 1994, Galway et al. 1994, Masucci et al. 1996). In the aging primary root and in laterals, however, only three of eight rhizodermal cells in the H position develop a root hair under sufficient nutrient supply or Fe starvation. The majority of cells in the hair position had, therefore, no root hair (Table 5, Fig. 11b). Under P deficiency, five of eight H positions possess a root hair (Table 5, Fig. 11c). This raised the question, if the root hair patterning in mature, sufficient and Fe- or Pdeficient plants was also reflected in the spatial distribution of the differential cytoplasmic staining and degree of vacuolization in the meristematic region. However, similar to seedlings, nearly all epidermal cells in the H position exhibited a stronger staining with toluidine blue than epidermal cells in the N position (Fig. 11, df). The same observation held true for the lower degree of vacuolization (Fig. 11, df). Therefore, the Fe or P nutritional status did not alter the differential cytoplasmic staining and lower degree of vacuolization.

The GL2 gene, which is a negative regulator of root hair development, and the CPC gene, a positive regulator, are expressed in nearly all N positions of 5-day-old seedling roots (Masucci et al. 1996, Wada et al. 2002). The lack of GL2 and CPC expression in the H positions of seedlings is correlated with the occurrence of root hairs (Masucci et al. 1996). Thus, it remains to be clarified, if the non-hair cells in the H position occurring in plants older than 5 days are correlated with an expression of CPC or GL2 in the H position. To determine the expression pattern of CPC and GL2, plants carrying GUS transgenes were cultivated in the presence and absence of either Fe or P. Both GL2-GUS and CPC-GUS plants showed vertical stripes of stained cells as previously described by Masucci et al. (1996) and Wada et al. (2002) in primary and lateral root tips. No difference in the intensity of staining was visible between nutrient sufficiency and Fe or P deficiency in GL2-GUS and CPC-GUS plants. Cross-sections revealed that GL2-GUS and CPC-GUS were expressed under all conditions in nearly all N positions, whereas nearly all H positions showed no GUS staining (Fig. 11, g-l; Table 7). Thus, the expression pattern was the same as that described for 5-day-old seedlings. The root hair patterning of GL2-GUS and CPC-GUS plants did not deviate from the wildtype (not shown). Therefore, root hairs in the H position (Fig. 11, b-c) were not correlated with the expression patterns of GL2-GUS and CPC-GUS (Fig. 11, d-i; Table 7). The spatial expression of CPC and GL2 was independent of age and nutrient status of the plants.

	Figure 11 : (a-c) Schematic representation of one transverse cell layer from the Arabidopsis root hair zone with only cortical and epidermal cells. Root hairs in the H position are marked in blue. (a) In five-day-old seedlings nearly all eight epidermal cells in the H position form a root hair. (b) In roots older than five days that are cultivated under sufficient nutrient supply or Fe deficiency, only three of eight H positions develop a root hair in a randomly distributed manner. (c) Pdeficient roots older than five days have on average five root hairs in the H position per cell layer. (d-f) Differential cytoplasmic staining with toluidine blue is shown in cross-sections from the late meristematic region of three-week-old Arabidopsis. (d) control, (e) Fedeficient, and (f) Pdeficient roots. The differential cytoplasmic staining is visible in the epidermis, which is surrounded by cells of the lateral root cap. The lower intensity of staining in (e) is due to a lower thickness of the cross-section (2µm). (g-l) Shown are cross-sections from the late meristematic region of three-week-old GL2-GUS and CPC-GUS plant roots adapted to Fe and P nutrition. (g) control GL2-GUS, (h) Fe-deficient GL2-GUS, (i) P-deficient GL2-GUS, (j) control CPC-GUS, (k) Fe-deficient CPC-GUS, and (l) P-deficient CPC-GUS plants. Bar = 20 µm.

Table 7 : The percentage of H and N cells showing GUS staining. Twelve-day-old GUS plants were cultivated in the presence and absence of either Fe or P. The spatial GUS expression was analyzed in cross-sections from the late meristematic region of the root. n = 6.

3.4.1  Mutant screening and genetic analysis

To identify potentially new genes that are involved in the root hair development under P limitation, a mutant screening was performed. Mutants were searched for that were not able to develop root hairs when the plants were germinated on –P medium but formed normal root hairs after the plants were transfered to +P medium. An EMS-mutagenized and a TDNA mutant population were screened. Putative mutants were propagated. After retesting the progeny of putative mutants, 4 EMS and 2 TDNA mutants were considered to be clearly impaired in the root hair development under P starvation, which was abolished in the presence of P (Table 8). One of the EMS mutants was characterized in more detail. The mutant was called per2 ( p hosphat e deficiency r oot hair defective2).

Table 8 : The statistical data from the primary and secondary mutant screening.

^a For estimation of the mutant rate, chlorotic and albino plants were counted.

The F1 offspring of a cross between the per2 mutant and the Col-0 wildtype developed normal root hairs under P deficiency. The F2 generation segregated in a ratio of 23% plants exhibiting the per2 phenotype and of 77% plants appearing normal; a 1:3 ration that is consistant with a single recessive nuclear mutation.

The F1 progeny of crosses between the per2 mutant and the 5 remaining mutants from the screening also displayed the wildtype phenotype indicating none of the mutants was allelic to per2 and all mutations were recessive.

	Figure   12 : The root hair phenotype of the per2 mutant. (a, b) Stereo microscope images (a) per2, -P, (b) per2 +P. (c, d) Micrographs of hand-cut sections from the root hair zone. (c) per2, P, (d) Col-0, P. (e-m) Cryo-SEM images. (e) Col0, control, (f) Col0, Fe, (g) Col0, P, (h) per2, control, (i) per2, Fe, (j) per2, P, (k) per2, control, (l, m) per2, P. Bars = 250 µm (a, b), 100 µm (e, g), 50 µm (c, d, h, i, j, l, m), and 30 µm (f, k).

Under the stereo microscope, only small bulges or short root hairs that did not elongate were visible in P-deficient per2 plants. In the presence of P, the root hairs appeared to be nearly normal, although a bit elongated (Fig. 12, a, b). In cross-sections, the bulges were located in the H positions and did not differ in their frequency from the wildtype (Fig. 12, c, d; Fig. 13). However, the epidermal cells of per2 seemed to be slightly enlarged (Fig. 12, c, d). A more detailed analysis of the root surface with the scanning electron microscope revealed a regular shape of the root hairs formed in per2 under sufficient nutrient supply or Fe starvation (Fig. 12, h, i, k). In P-deficient per2 plants, the root hairs were shorter, thicker, and deformed. Nearly all root hairs showed material accumulations at their tips. These accumulations could results from improperly secreted or assembled cell wall constituents. Occasionally, two growth points were visible in one root hair indicating the direction of vesicle transport could be altered in per2 (Fig. 12, j, l, m).

	Figure 13 : The percentage of wildtype root hairs and per2 bulges in the H position. 14-day-old plants were germinated in the presence of P (grey bars) or without P (black bars).

To determine, if in addition to the impaired root hair development also other P starvation responses are affected in the per2 mutant, the length of the primary root and the number of lateral roots were measured (Fig. 14). The primary root length of the per2 mutant was not altered (Fig. 14a). In contrast, the lateral root number was increased under sufficient P supply (P ≤ 0.01). Also P-deficient per2 showed an increased number of lateral roots, but this was not significant (P = 0.08; Fig. 14, b-c). Thus, the per2 mutation seemed to affect also other P deficiency responses in addition to root hair development. To test, if the impaired root hair formation and the increased lateral root development of the per2 mutant were caused by the same mutation, F2 plants from a backcross to Col-0 were selected for the per2 phenotype. Plants with an impaired root hair development on –P medium were allowed to self. The progeny were then checked for the frequency of lateral roots. Also these backcross plants exhibited an increased lateral root development (Fig. 14b), which could be a first indication for a genetic linkage of the impaired root hair development and the increased lateral root frequency of the per2 mutant.

	Figure 14 : The root system architecture of the per2 mutant. The primary root length (a) and the number of lateral roots (b) of per2 and the Col-0 wildtype grown in vertical culture for 15 days in the presence (grey bars) or absence (black bars) of P. Any lateral root outgrowth that was visible without magnification was counted as a lateral. Also included in the experiment were F3 plants that were the progeny of F2 lines from a backcross with Co0 that had been selected for the per2 root hair phenotype. (c) The increased lateral root development of the per2 mutant on +P medium is shown in an image of 16dayold plants. Bar = 1 cm. Two repetitions of the experiment gave the same results.

One of the classical features of P deprivation is the accumulation of anthocyanins in the leaves (Poirier & Bucher 2002). To test if this response was altered in per2, the absorbance of anthocyanins extracted from the leaves of per2 and Col-0 was measured (Fig. 15). No differences were obvious in the P-sufficient mutant. When cultivated under P deficiency, an increase in the anthocyanin absorbance in per2 was observed. Allthough this increase was not significant (P > 0.05), it was apparent in two independent experiments. Plants from a backcross that exhibit the per2 root hair phenotype also showed an increased anthocyanin absorbance indicating the reaction could be linked with the same mutation that causes the impaired root hair phenotype.

	Figure 15 : The anthocyanin content of the per2 mutant grown vertical culture for 20 days in the presence (grey bars) or absence (black bars) of P. Also included in the experiment were F3 plants that were the progeny of F2 lines from a backcross with Co0 that had been selected for the per2 root hair phenotype. Repetition of the experiment gave the same result.

To test wether the per2 mutation affects the uptake of phosphate or of other elements from the medium, an element analysis with ICP was conducted (Fig. 16). The content of none of the elements investigated was changed in the root and shoot material of per2. Thus, only the phophorus content is shown (Fig. 16, a-b). Unfortunately, the climate conditions were changed by others during repetition of the experiment. Allthough the experiments could not be repeated under the same growth conditions, no effect of the per2 mutation on the phosphorus content or on the content of other elements was obvious under the different growth conditions. Also the shoot fresh weight of per2 was not altered in the different experiments (P > 0.05). Thus, under growth conditions tested, the per2 mutation had no impact on the phosphorus accumulation or biomass production of the plants.

	Figure 16 : (a, b) The phosphorus content of per2 and Col-0 that had been cultivated on Petri dishes (a) or in liquid culture medium (b). (a) Two independent experiments were conducted. In the first experiment, per2 and Col-0 were treated for 20 days. The second experiment includes also the backcross (bc). During this repetition, the climate conditions had been changed by others to a continuous light period and 23°C. Because direct competion of the replicas was not impossible, the culture period was then extended to 27 days to increase potential effects of P deficiency on the P content of the mutant. (b) The plants were also grown in liquid culture. Also a replenishment study was included. One part of the plants were harvested after 7 days of treatment with P deficiency together with the sufficient controls. The remaining plants were replenished with medium containing sufficient amounts of P and were harvested after one day. Two experimental replicas were conducted. During the second experiment, plant material was limiting and was only sufficient to conduct the replenishment experiment.

 

As the per2 root hair phenotype could be repressed by a sufficient P supply (Fig. 12), it was tested if this suppression is mediated by a systemic or a local signal. A local abundance of phosphate can be mimicked by phosphite (Phi), the reduced but metabolically inert form of phosphate, which is readily taken up by plant cells (Carswell et al. 1996). Phi specifically suppresses anthocyanin accumulation, expression of At4 and nucleolytic enzymes, and high affinity transport while the phosphate content decreases and growth is inhibited (Carswell et al. 1997, Ticconi et al. 2001). When the per2 mutant and Col-0 were grown on medium containing Phi instead of phosphate, both showed a severely inhibited root and shoot growth, as it was described by Ticconi et al. (2001). However, the root hair elongation of per2 was the same as in the wildtype (Fig. 17). Thus, the impaired root hair elongation of per2 can be overcome by mimicking a local P abundance indicating the PER2 gene might belong to a local rather than to a systemic signaling pathway.

	Figure 17 : The effect of the phosphate analogon phosphite on the root hair elongation of per2. The plants were germinated on medium containing phosphite instead of phosphate and were analyzed after 14 days. Bar = 250 µm.

To characterize the interaction between per2 and genes involved in root hair specification, double mutants with gl2 and with erh3 were generated. In both cases, the F2 offspring was analyzed for the root hair phenotype on –P medium (Fig. 18). The segregation ratio was 9:3:3:1. The per2gl2 and the per2erh3 double mutants both showed the phenotype of either parent. This additive interaction indicated that per2 acts in a pathway that is independent of gl2 or erh3.

	Figure 18 : The characterization of a potential genetic interaction of the per2 mutation with mutations in genes regulating epidermal cell specification. The per2 mutant was crossed with gl2 (a) or erh3 (b). The F2 progeny was germinated on –P medium and the phenotype was analyzed after 14 days. The percentage of plants exhibiting the indicated phenotype is given. (a) The F2 offspring of a cross between per2 and gl2 segregated into 4 different phenotypes. Plants that showed many and elongated root hairs represented the homozygous wildtype and the heterozygous mutant genotypes. The recessive gl2 mutation is phenotypically linked with the lack of trichomes (Koornneef et al. 1982). Thus, plants that showed many root hairs but no trichomes represented the homozygous gl2 mutants. One-quarter of the plants showing the per2 root hair phenotype also had no trichomes indicating that those mutants carrying a homozygous mutation in both loci have an additive phenotype. n = 910 plants. (b) The root system of the recessive erh3 mutant is stunted and the roots are radially expanded (Schneider et al. 1997, Burk et al. 2001). The F2 progeny of a cross between per2 and erh3 also segregated in four phenotypical groups. Those plants that exhibited the per2 phenotype segregated into plants with normal root growth and plants, which showed the erh3 root morphology. Thus, the homozygous per2erh3 genotypes showed both the per2 and the erh3 phenotypes suggesting an additive interaction. n = 684 plants.

Low resolution mapping of the per2 mutation was performed by linkage analysis with the SNP marker set described by Törjek et al. (2003). F2 plants from a mapping population with C24 that displayed the per2 root hair phenotype were analyzed with a subset of of these markers spanning the whole Arabidopsis genome (Fig. 19a). The per2 penotype was linked with the markers MASC02947 and MASC02841 located on the upper arm of chromosome 3 (Fig. 19a). Analysis with further markers located in that region revealed the highest recombination frequencies with MASC04279 and MASC02841. The genetic map distances between per2 and the markers were 9.2 and 9.8 cM, respectively (Fig. 19b).

	Figure 19 : The low resolution mapping of the per2 mutation with the SNP marker set described by Törjek et al. (2003). The SNPs were detected with the SNaPshot method. (a) Analysis of 213 F2 plants from a mapping population derived from a cross between per2 (Col-0 background) and C24 that showed the per2 root hair phenotype revealed a co-segregation with markers on the upper arm of chromosome 3. (b) Genetic map distances in cM between per2 and markers from the upper arm of chromosome 3 calculated with the Kosambi function.

For fine mapping, further F2 plants displaying the per2 phenotype from a mapping population with Cvi were tested for recombination events between the markers MASC04279 and MASC02841. Recombinant lines were then analyzed by closer SNP markers (Fig. 20). Newly identified SNPs are listed in Table 9.

	Figure 20 : The fine mapping of per2. The upper arm of chromosome 3 is shown ranging from nucleotide no. 6,500,000 to 8,000,000. The F2 plants from the low resolution mapping and 987 further plants exhibiting the per2 phenotype from a mapping population with Cvi were analyzed for recombinations between the markers MASC04279 and MASC02841. The location of these recombinations was then analyzed with a higher resolution of SNP markers that were either obtained from the NASC SNP database (http://www2.mpiz-koeln.mpg.de/masc/search_masc_snps.php) or identified by sequence analysis (a). The name of the markers was chosen according to the BAC clone they were located on. The numbers below the SNP marker names indicate the chromosomal location of the polymorphic nucleotide. Three lines were heterozygous at the marker location F3H11-7.22 and homozygous at F3H111r, two lines were heterozygous at MQC12 and homozygous at MQC121, and one line was heterozygous at MQC121 and homozygous at K10D20-7.16. The frame indicates markers where no recombination was found.

Table 9 : The SNPs between Cvi and Col-0 identified by sequence analysis.

^a The indicated polymorphism was between C24 and Col-0 and not between Cvi and Col-0.

No recombinations were found between nucleotide 7162174 and 7209300, inclusively. In conclusion, a location of per2 outside this area is unlikely since only plants with the recessive per2 phenotype had been selected for the marker analysis. The per2 phenotype of the recombinant lines was confirmed in the F3 progeny. Candidate genes that are located between nucleotide 7162174 and 7209300 are listed in Table 10.

Table 10 : The candidate genes for the per2 mutation.

The genes At3g20550 and At20630 could be excluded as candidates for per2 since their mutation causes curly leaves or an embryo-lethal phenotype, respectively, which was both not observed in per2. By comparing the sequences of per2 and Col0, no SNPs were found in the genes At3g20500, At3g20510, At3g20555, At3g20590, and At3g20610. Thus, these genes could also be excluded as possible candidates.