The temperature sensitive brush mutant of the legume Lotus japonicus reveals a link between root development and nodule infection by rhizobia

The brush mutant of Lotus japonicus exhibits a temperature-dependent impairment in nodule, root, and shoot development. At 26ºC brush formed fewer nodules, most of which were not colonized by rhizobia bacteria. Primary root growth was retarded and the anatomy of the brush root apical meristem revealed a distorted cellular organization and reduced cell expansion. Reciprocal grafting of brush with wild type plants indicated that this genotype only affected the root and that the shoot phenotype was a secondary effect. The root and nodulation phenotype co-segregated as a single Mendelian trait and the BRUSH gene could be mapped to the short arm of chromosome 2. At 18ºC, the brush root anatomy was rescued and similar to the wild type, and primary root length, number of infection threads, and nodule formation were partially rescued. Superficially, the brush root phenotype resembled the ethylene-related ‘thick short root’ syndrome. However, treatment with ethylene inhibitor did not recover the observed phenotypes, although brush primary roots were slightly longer. The defects of brush in root architecture and infection thread development, together with an intact nodule architecture and the complete absence of symptoms from shoots, suggest that BRUSH affects cellular differentiation in a tissue dependent way.


Introduction
Legumes can establish root nodule symbiosis (RNS) with rhizobia bacteria, which fix atmospheric nitrogen to ammonium in exchange for carbon and are housed intracellularly within the legume plant root nodule organ (White et al., 2007). In the legume Lotus japonicus, the symbiotic interaction with Mesorhizobium loti is initiated at root hair cells. Plant root exudates are sensed by rhizobia and stimulate the bacterial production of Nod factor (NF) signaling molecules (Lerouge et al., 1990). NF perception involves NFR1 and NFR5 membrane receptors and leads to rapid physiological and morphological changes, such as membrane depolarization, a rhythmic oscillation of calcium concentration termed 'calcium spiking', and cytoskeleton rearrangements resulting in root hair deformation (Ehrhardt et al., 1992;Heidstra et al., 1994;Ehrhardt et al., 1996;Timmers, 2008). Upon attachment to the root hair tip, rhizobia are entrapped in a process called root hair curling. From this cavern a tubular-like plant plasma membrane derived structure, the infection thread (IT), is formed inside the root hair, into which the rhizobia grow by continuous division (Gage, 2004). In parallel, cortical cell divisions are initiated in the root cortex, leading to the development of a nodule primordium. The IT subsequently grows towards the nodule primordium, penetrates into a cortical cell after a cytoplasmic bridge called the pre-infection thread was formed, which determines the growth direction of the IT through the cortical cell. Ultimately, the bacteria are released into the plant cell by a process resembling endocytosis. Inside the plant cell cytosol the bacteria differentiate into bacteroids and form organelle-like symbiosomes that actively fix atmospheric nitrogen.
Initiation of RNS is strictly controlled by the host plant at the level of the two cell-type specific processes for rhizobial accommodation: plant assisted rhizobial invasion of the epidermis-derived root hair cell followed by IT formation, and development of a nodule primodium in the root cortex. Several different mutants exhibiting a defect in IT and nodule formation have been described phenotypically so far, and a number of genes required for RNS have been isolated (Oldroyd and Downie, 2008). They include receptor kinases, ion channels, nuclear pore components, transcription factors, an ankyrin repeat membrane protein, and the nuclear localized protein kinase CCaMK. Mutants in these genes are impaired in www.plantphysiol.org on August 23, 2017 -Published by Downloaded from Copyright © 2009 American Society of Plant Biologists. All rights reserved.
-6 -nodule formation and IT are either not formed at all or aborted within the root hair.
Interestingly, a gain-of-function mutation in CCaMK, encoding a calcium and calmodulin-dependent protein kinase, can trigger development of spontaneous nodules in the absence of rhizobia (Tirichine et al., 2006). Spontaneous nodule development can also be induced by exogenous application of the phytohormone cytokinin and by the gain-of-function mutant snf2 of the L. japonicus cytokinin receptor Lhk1 (Tirichine et al., 2007). Conversely, L. japonicus hit1 mutants, which carry a loss-of-function allele of the cytokinin receptor Lhk1, are characterized by excessive IT formation upon rhizobia infection, but fail to initiate nodule primordia (Murray et al., 2007). In addition to the symbiosis phenotype, snf2 mutants display altered root architecture with extra cell layers (Tirichine et al., 2007). Also, in Arabidopsis cytokinin has an effect on root development by limiting root meristem size and activity (Werner et al., 2001).
These findings and many others provide evidence for obvious links between phytohormone signaling in nodule and root development. Meristematic cells proliferate and differentiate into cells of root and shoots. Although the shoot apical and root apical meristems (SAM and RAM, respectively) are under distinct genetic regulation, they share the same developmental mechanism for maintaining the stem cells in response to environmental cues (Veit, 2004;2006). This ability to alter tissue development provides the plasticity necessary for adaption to various environmental conditions. Initiation of nodule primordia is a typical example for post-embryonic development. The legume root nodule is not derived from the RAM but from fully differentiated cortical cells at the infection site that are rejuvenated and enter mitotic cell divisions. The nodule number per plant is modulated by various environmental and nutrition factors, like availability of nitrogen (Barbulova et al., 2007). L. japonicus ASTRAY displaying super-nodulation encodes a gene required for light sensitivity, implying that nodule formation is also regulated by light (Nishimura et al., 2002b). In Pisum sativum, nodule formation on lateral roots is slightly repressed at lower temperature (Fearn and Larue, 1991a). Recent studies suggested that an Arabidopsis gene required for shoot meristematic formation showed high homology to a gene involved in nodule primordium formation, implying that pathways controlling nodulation may be recruited from those controlling SAM (Nishimura et al., 2002a). regulation of nodule positioning and the de novo organogenesis in the root. E.g., auxin is involved in cell division control, differentiation of lateral roots, as well as in nodule formation (de Billy et al., 2001). In Arabidopsis, overproduction of endogenous ethylene results in the suppression of cell division in the quiescent center, consequently leading to growth reduction of the primary root (Ortega-Martinez et al., 2007). The Medicago truncatula sickle mutant forms supernumerical nodules, presumably due to a loss of negative regulation by the plant hormone ethylene.
Sickle encodes a gene orthologous to Arabidopsis EIN2, which is required for ethylene signaling (Penmetsa et al., 2008). Ethylene contributes to the positioning of the nodule on the root (Heidstra et al., 1997). However, in soybean the nodulation phenotype of ethylene insensitive mutants was indistinguishable from wild type (Schmidt et al., 1999). In Sesbania rostrata, ethylene positively acts on infection pocket and nodule primordium formation (D'Haeze et al., 2003). Therefore, the specific role of ethylene in nodulation may vary between different species.
The genetic analysis of intersections between general plant development and nodulation requires the analysis of mutants that are not specifically defective in symbiosis, but show pleiotropic developmental defects. Here, we describe the novel mutant brush, which we identified as a nodulation deficient L. japonicus mutant. A detailed phenotypic analysis indicated that root growth and bacterial infection were seriously hampered. BRUSH is thus not a symbiosis specific gene but is also required for proper root development. Pleiotropic mutants like brush are invaluable tools to identify the involvement of general plant developmental pathways in nodulation including hormone regulation. Surprisingly, although root nodule organogenesis was delayed and most nodules were empty, nodule architecture was not affected as such. brush reveals a novel link between cell differentiation specifically in the root apical meristem and rhizobial infection.

Lotus japonicus brush mutants exhibit nodulation and growth defects
Line SL0979-2 was isolated as a nodulation deficient EMS mutant of Lotus japonicus accession Gifu (referred to as wild type) (Perry et al., 2003). When inoculated with Mesorhizobium loti and grown at 26°C, the number of nodules was drastically reduced in the mutant (Fig 1A, B, C). The root hair and calcium spiking responses to NF or M. loti were similar to wild type, indicating no defect in early recognition of the bacteria (Supplemental Fig. S1, Supplemental Table S2). We observed normal colonization patterns of Glomus intraradices (Supplemental Fig. S3), indicating that arbuscular mycorrhiza symbiosis was not disturbed in the brush mutants.
In addition to the nodulation deficiency, brush mutants had shorter shoots, shorter primary roots, and red hypocotyls due to anthocyanin accumulation (Fig. 1A).
The diameter of primary roots was not uniform throughout the length and overall thicker than wild type (Fig 1B, C). The number of lateral roots formed on a primary root varied between plants but like the primary roots they were also relatively short ( Fig. 1A). Root hairs as well as trichome structure appeared normal (Supplemental Fig. S1B, data not shown). Closer observation of mutant root tips showed that the root apical region was thicker and root hair density was higher (Fig 1H, I). Because of the increased root hair density around the root tip we named the mutant brush. All of the non-symbiotic aspects of the root phenotype were independent of inoculation with M. loti (see below).

BRUSH is a novel and single locus gene
The three different phenotypes (short root, short shoot and impaired nodulation) showed normal plant growth and nodule formation, indicating that the brush mutation is recessive. We analyzed 902 self progeny of these F1 individuals for cosegregation of the root and nodulation phenotypes. In all clearly scorable cases, the root and nodulation phenotype occurred together, indicating cosegregation of both -9 -both, the nodule and the root phenotype. This suggests that a single locus mutation is responsible for the observed defects. Further analysis of 1896 F 2 individuals derived from the MG-20/brush crosses linked the mutant locus to marker TM0312 on the short arm of chromosome 2 (Fig. 2). We identified a total of 5 mutant recombinants within an interval of less than 0.1 cM around brush all of which showed both, low nodulation and short roots. Therefore we concluded that all mutant phenotypes were caused by a single mutation or very tightly linked mutations.
Unfortunately, the BRUSH region suffers from an accumulation of repetitive DNA, which makes the construction of physical contig across the gene very difficult.

The nodulation deficiency of brush is temperature-dependent
Since initial experiments indicated a temperature-dependence of the brush phenotype, plants were subsequently grown at 18°C and 26°C as permissive and restrictive temperatures, respectively. Inoculation with Mesorhizobium loti MAFF 303099 constitutively expressing the DsRed fluorescent protein (MAFF DsRed) allowed us to follow rhizobial infection by fluorescence microscopy.
While wild type plants had on average 18 nodules and nodule primordia per root brush mutants had only 2 (Fig. 3G). Of these, only 15% were infected by bacteria compared to almost 100% in the wild type (Fig. 3I). Although root hair curling appeared normal in brush roots, the number of infection threads in brush was reduced compared to wild type (Fig. 3H). Most of the ITs showed abortion within the root hair cells and only very few reached the next cell layer. (Supplemental Fig. S4).
In wild type roots nodule primordia were round and infected as seen by the bacteriaderived red fluorescence (Fig. 1D, E), whereas brush mutant roots mostly developed irregularly shaped, bump-like structures without visible signs of infection (Fig. 1F, G, and 3A, B, C).
Microscopical analysis of sections of nodules and nodule primordia revealed additional brush phenotypes. In wild type, most nodules displayed the brownish leghemoglobin color, suggesting bacterial presence and functional nodules (Fig. 1J).
In brush mutants, only few nodules showed brownish bacteria containing cells ( On the other hand, when brush mutants were incubated at 18°C, most nodules resembled those of the wild type. The brownish color was more pronounced in those nodules, and also the vascular bundle surrounded the nodule central tissue as found in wild type nodules (Fig. 3D, E, arrow). After growth at 18°C, a strong bacteria-derived fluorescent signal was observed within nodule primordia and white immature nodules (Fig. 3F), indicating that also these not fully mature structures were infected by rhizobia. The number of infection threads on brush roots also increased at 18°C compared to 26°C, although this increase was not statistically significant and occurred to a similar extent in wild type plants (Fig. 3H). Wild type plants grown at 18°C developed fewer nodules and primordia compared to plants grown at 26°C (Fig. 3G). In contrast, brush plants showed a significantly increased number of nodules and primordia at 18°C compared to 26°C. Nevertheless, no full rescue of nodule number up to wild type level was observed (Fig. 3G). However, at 18°C the proportion of infected nodules and primordia in brush rose almost six fold to 86% (Fig. 3I), indicating that the rhizobial infection process in brush benefits strongly from growth at the lower temperature.

The brush root phenotype is temperature-dependent
Due to the temperature sensitivity of the symbiotic phenotype, we asked whether the brush primary root growth defect was also rescued at lower temperatures. brush primary roots were longer at 18°C than at 26°C but not fully restored to wild type level ( Fig. 4A). Thin sections of roots were analyzed to assess their cellular organization.
At 26°C wild type roots displayed distinctive and organized cell layers ( Fig. 4B). At the meristematic region called root initial, the putative quiescent center was surrounded by small actively dividing cells, which later differentiate in the different cell types ( of brush was apparent in the epidermal cell layer, which was not well defined. brush epidermal cells appeared much larger than wild type and were radially expanded.
Collectively these developmental defects gave rise to a severely distorted overall root architecture.
Wild type roots grown at 18°C showed no structural differences to roots grown at 26°C, although the overall root length was shorter (Fig. 4A, B, D). Surprisingly, at 18°C brush mutant roots showed complete restoration of cellular structure, shape and size to wild type levels ( Also, in the root hair region, cells expanded longitudinally similar to wild type (Supplemental Fig. S6E). Transverse sectioning revealed that cell shapes and positions in all cell layers were wild type like (Supplemental Fig. S6F). This analysis revealed that the root phenotype of brush is strongly temperature-dependent. The root phenotypes were independent of inoculation with M. loti (data not shown). -12 -

The brush mutant phenotype is determined by the root genotype
To examine whether the brush mutant phenotypes were determined by shoot-or root-genotype, we performed grafting experiments between brush and wild type plants. Two weeks after grafting, the roots were inoculated with M. loti MAFF 303099 and nodule formation was analyzed after 4 weeks at 26°C. When wild type (wt) as a rootstock was grafted with brush as a scion (wt/brush), brush shoots grew normal, as did the wt/wt grafted plants (Fig 5A, B, C). However, when stock brush was grafted with wild type scion (brush/wt), the shoots looked pale and stunted even 4 weeks after inoculation, probably due to the lack of nodulation of the brush roots. This was similar to the situation in brush/brush-grafted plants, which were stunted and nonnodulating ( Fig. 5A, B, C, D). On wt/brush plants, almost the same nodule number was observed as on wt/wt roots (Fig. 5D), indicating that the brush phenotype is root autonomous and determined by the root genotype.

The brush phenotype is independent of ethylene, ABA, and GA
The brush mutants phenotypes might be due to a difference in production or perception of the plant hormone ethylene, which was found to be a negative regulator of root growth (Ortega-Martinez et al., 2007) and of rhizobial infection (Oldroyd et al., 2001). Therefore, we tested the ethylene-responsiveness of the mutants compared to the wild type. brush seedlings were subjected to the ethylene precursor 1aminocyclopropane 1-carboxylic acid (ACC) and grown in the dark. These growth conditions normally induce the so-called 'triple response' observed upon etiolating the seedlings (radial swelling of the stem, absence of normal geotropic response, exaggerated apical hook curvature, and inhibition of root and stem elongation) (Guzman and Ecker, 1990) in a dose-dependent manner (Fig. 6A). brush seedlings showed the same or a slightly higher ethylene sensitivity as the wild type (Fig. 6A), indicating no general defect in ethylene response.
Since the inhibition of primary root growth in brush could be caused by excessive endogenous ethylene production, we applied aminoethoxyvinylglycine (AVG), an ethylene synthesis inhibitor, at different concentrations to reduce the endogenous ethylene level. Application of more than 1 μM AVG caused retardation of root growth in both wild type and brush (data not shown). Below that concentration, in wild type as well as brush plants root growth was enhanced by the -13 -treatment in a dose-dependent manner, although mutant root lengths never reached wild type levels (Fig. 6B). Thin sectioning of the root apical meristem revealed that AVG treatment did not restore the mutant root architecture back to wild type (Supplemental Fig. S7). We also tested whether the brush nodulation defect could be restored by AVG. The different doses of AVG enhanced nodulation in both, wild type and brush, in a similar way, but no apparent recovery of the mutant phenotype to wild type levels of fully colonized mature nodules was observed (Fig. 6C). We obtained similar results upon silver ion (Ag + ) treatment (Supplemental Fig. S8), which interferes with ethylene perception (Rodriguez et al., 1999).
Other pathways that might be potentially disturbed in brush and responsible for the root growth phenotype are abscisic acid (ABA) signaling, exemplified by the Medicago truncatula latd short root mutant (Liang et al., 2007), or the gibberellin pathway. For the latter, a lack of bio-active gibberellic acid (GA 3 ) in plants results normally in dwarfism (Koornneef and van der Veen, 1980). However, exogenous application of ABA or GA to brush roots failed to rescue the brush root mutant phenotype (Supplemental Fig. S9 and S10).
The short root phenotype could also be a consequence of a generally impaired nutritional status. To test this, plants were grown on agar plates containing 1%, 2%, or 4% sucrose. Root growth increased in both, wild type and brush plants, upon addition of sucrose in a dose-dependent manner, but mutant root length was never restored to wild type levels (Fig. 6D) and the root architecture remained disturbed (data not shown). Treatment of roots with 1% mannitol, which influences osmotic regulation, did not enhance growth of either wild type or brush roots (data not shown).

The brush phenotype is independent of auxin
Auxins stimulate cell differentiation, number of lateral roots, and nodule formation (Aloni et al., 2006;Oldroyd and Downie, 2008). Since retardation of root growth and nodulation deficiency in brush could be caused by disturbed auxin homeostasis, plants were subjected to different concentrations of the synthetic auxin analog 1naphthaleneacetic acid (NAA). When 0.01 µM of NAA was added to wild type plants, primary root length slightly increased in comparison to non-treated plants, although not significantly (Fig. 7A). At higher NAA concentrations root growth was inhibited (Fig. 7A). brush plants responded to NAA in the same way as wild type (Fig. 7A) -14 -However, at 0.01 µM of NAA no root growth recovery was observed. Addition of NAA also did not lead to a recovery of nodule formation in brush (Fig. 7B). Since addition of high concentrations of NAA inhibited root growth, the short root phenotype in brush might be caused by overproduction of endogenous auxins. To test this possibiltity, brush plants were treated with the auxin inhibitor 2,3,5-triiodobenzoic acid (TIBA). No significant increase in root length was observed in both, wild type and brush roots (Fig. 7C). However, lateral root formation was decreased in both wild type and brush in a dose dependent manner (Fig. 7D) -15 -

BRUSH defines a genetic link between cell differentiation at the root apical meristem and rhizobial infection
In this study, we performed a detailed phenotypic analysis of the L. japonicus brush mutant, which exhibited three distinctive phenotypes: short root, short shoot, and low number of nodules. Importantly, the majority of nodules was lacking rhizobial infection. We found that the short root phenotype was caused by a severe distortion of the root architecture. brush root cells failed to elongate properly and to organize into single files that make up the typical root structure. Interestingly, this differentiation defect was specific for the root. Although rhizobial infection and the frequency of nodule formation were suppressed, the overall architecture of the nodule primordia in brush was wild-type like, indicating that cortical cell division and differentiation were not affected. This points to a specific defect that only affects the root architecture but not the initiation of nodule primordia. During microscopical analysis of the brush root apical region the typical root cell layers could still be distinguished but longitudinal cellular expansion was disturbed . Moreover, grafting experiments revealed that shoot growth could be restored by replacement of mutant with wild type roots, suggesting that the shoot phenotype is a secondary effect of the root phenotype. We conclude that brush exerts a specific effect on cell expansion in the root. Moreover, we could pinpoint the symbiotic defect of brush to the infection of the nodule tissue by rhizobia. These specific defects in brush reveal a novel link between cellular development at the root and bacterial infection.

brush is a novel temperature sensitive mutant
The brush mutant phenotypes were temperature sensitive: at lower temperatures, root architecture, nodule number and rhizobial infection were improved. In legumes, several temperature sensitive mutants have been isolated such as sym5 in pea (Fearn and LaRue, 1991a) and nup133 and nup85 in L. japonicus (Kanamori et al., 2006;Saito et al., 2007). In our experiments, the lower growth temperature of 18°C did not significantly affect nodule formation in L. japonicus Gifu. In contrast, we saw a strong positive effect on brush (Fig. 3G), similar to the temperature effect in other temperature sensitive mutants. It appears unlikely that brush is allelic to the P. -16 -sativum sym5 mutant, since in this case nodulation can be recovered by addition of Ag + or AVG (Fearn and Larue, 1991b). brush is not an allele of nup133 or nup85, since these genes map to chromosome 1 in L. japonicus, whereas brush is on chromosome 2.
The brush phenotype differs from that of previously described developmental

mutants of Arabidopsis thaliana
Thin sectioning of primary roots of brush revealed an aberrant root architecture ( -17 -phenotype as described for MOR1. Interestingly, Robledo and collegues demonstrated that the cell-bound bacterial cellulase CelC2 is essential for the primary infection process in Rhizobium leguminosarum (Robledo et al., 2008). These data suggest that cellulytic activity is required for both proper root development and rhizobial infection, It is therefore possible that BRUSH encodes a cell wall degrading enzyme or cytoskeletal component that is specifically required for normal development of the root architecture.
In legumes, rearrangement of the MT is essential for root hair curling, IT formation, and the re-entrance into mitosis of cortical cells at the incipient nodule primordium (Timmers, 2008). In contrast to the MT and cellulose mutants mentioned above which display root hair or trichome defects, the root hair and trichome development of brush was normal (Supplemental Fig. S1, Supplemental Table S2 and data not shown). Moreover, root hair response to NFs and rhizobia appeared wild type like. Also the grafting experiments clearly revealed a root-determined phenotype for brush (Fig. 5). Mycorrhiza development of the brush mutant at 26ºC did not show any differences to wild type (Supplemental Fig. S3). Since penetration of hyphae into the root cortical cell requires proper re-orientation and organization of the MT (Genre et al., 2005), these data further support the idea that the cortical MT are not be affected in brush.
The brush phenotype appears independent from the phytohormones ethylene,

auxin, abscisic acid, and gibberellin
It is known that plant phytohormones play an important role in plant growth and nodulation. It has been suggested that ethylene directly suppresses cell division in the quiescent center of the RAM, resulting in retarded primary root growth (Ortega-Martinez et al., 2007). Initially, we speculated that the brush phenotypes might be caused by the production of, or sensitivity to ethylene, since they resemble typical ethylene-related phenotypes (Fig. 1A). For example, the tsr syndrome in Vicia sativa is characterized by thick and short roots with abundant root hairs very similar to brush. However, TSR only develops in the presence of rhizobia or Nod factor (Zaat et al., 1989), while in brush the root phenotype is independent of Nod factor. The TSR syndrome was found to be ethylene dependent (Zaat et al., 1989 -18 -architectural defects in the mutant. In contrast, the previously isolated mutants of P. sativum, sym5, which is an ethylene hypersensitive mutant (Fearn and Larue, 1991b), or sym17 or sym16, which are ethylene overproduction mutants (Lee and Larue, 1992;Guinel and Sloetjes, 2000), displayed a nodulation phenotype rescued by AVG addition. Taken together, these results implied that BRUSH exerts an ethylene-independent function.
It has also been suggested that other hormones such as GA and ABA contribute the root elongation and plant growth (Liang et al., 2007) However, addition of GA and ABA on brush mutant plants did not recover root development. Auxin hormone landscapes are also crucial for lateral root and nodule primordium initiation and thus auxin signaling might be directly or indirectly affected in brush. However, the addition of the synthetic auxin analog NAA did not recover nodule formation or primary root growth in brush mutants. Since the number of lateral root was also not altered in brush (Fig. 1A) a defect in auxin production or perception seems unlikely.
Addition of TIBA, an inhibitor of polar auxin transport, led to a decrease in lateral root formation indicating normal auxin signaling in brush, as well. Therefore, we conclude that hormonal disturbances are not the primary cause for the brush phenotype.
It has been suggested that gene functions required for nodule formation are recruited from plant developmental pathways (Szczyglowski and Amyot, 2003).
Nodulation mutants that show pleiotropic phenotypes, such as har1, astray, crinkle, and lot1 in Lotus japonicus, and sickle and latd in M. truncatula (Krusell et al., 2002;Nishimura et al., 2002a;Tansengco et al., 2003;Oka-Kira et al., 2005;Ooki et al., 2005;Prayitno et al., 2006;Liang et al., 2007), provide evidence for this statement.     -20 -upon excitation at 545 nm. In contrast, round pink nodules with vascular bundle formation were apparent (arrows in D) when plants were grown and inoculated at the permissive temperature of 18°C, whereas at 26°C mostly white nodule bumps were observed (A-C). At the root surface infection events were visible (arrowhead in C).     Seeds were germinated upon scarification using sand paper, surface sterilization with 2% (w/v) NaClO for 6 minutes, and subsequent washing with sterile water, to be finally incubated in water at room temperature for 6 hours to overnight, depending on the water absorption of the seeds. Germinated seeds were transferred to 1% (w/v) Bacto agar (GIBCO) prepared in rectangular (18 cm  -23 -Leica MZ 16FA (Leica), for fluorescent images we used a DSR filter for excitation at 545 nm.

Temperature sensitivity analysis
All plants were germinated and planted in sterilized pots as described above.
Temperature effects were analyzed by incubation of plants in a temperaturecontrolled growth chamber with a constant temperature of 26°C or 18°C, with 16/8 hr photoperiods.

Grafting of brush and wild type L. japonicus plants
Grafting experiments were carried out as described by Nishimura et al., 2002.

Ethylene sensitivity and sucrose effect analyses
For triple response analysis, plants were germinated as described above. 2 days old seedlings were transferred to a ½ B5 Bacto agar plate containing 1aminocyclopropane 1-carboxylic acid (ACC) and incubated at 26°C in the dark for an additional 4 days. For analysis of AVG effects on nodulation and root length, a 10,000 times concentrated stock of (S)-trans-2-amino-4-(2-aminoethoxy)-3-butenoic acid hydrochloride (Sigma-Aldrich) was prepared in water and then diluted to the final concentration in the medium supplied to each pot. The root length was measured from the junction of hypocotyl and root to the tip of the primary root with Digital caliper (Mitutoyo, Japan). Plants were incubated in a growth chamber at 26°C or 18°C with 16/8 hr photoperiods. For sucrose analyses, 2 days old plants were transferred to concentrations of 1%, 2%, and 4% (w/v) sucrose in ½ B5 Bacto agar.
The plates were covered up to the plant hypocotyls with a sheet of black paper to avoid direct illumination of the roots. The initial positions of the root tips were marked on the plate. Plates were placed in an upright position and root length was measured from the initial mark to the tip of the primary root.

Auxin sensitivity effect analyses
All plants were germinated and planted in sterilized pots as described above. 1,000 times concentrated stocks of 1-naphthaleneacetic acid (NAA, Sigma-Aldrich) and 2,3,5-triiodobenzoic acid (TIBA, Sigma-Aldrich) were freshly diluted in ethanol and www.plantphysiol.org on August 23, 2017 -Published by Downloaded from Copyright © 2009 American Society of Plant Biologists. All rights reserved.
-24 -added directly to the liquid medium. Since TIBA inhibits lateral root formation, 2% sucrose was added in these experiments to maximize lateral root formation.

Microscopic analysis of nodules and roots
Nodule sections were examined by bright field microscopy. For this, nodules were freshly harvested and vacuum infiltrated for 20 minutes with fixation solution [PBS buffer containing 2.5% (v/v) glutaraldehyde (Sigma-Aldrich)] and left at room temperature for about 1 hour. Samples were subsequently embedded in 5% (w/v) agarose in water. For sectioning, samples were briefly fixed with 2.5% (v/v) glutalaldehyde in PBS solution. Vibratome VT1000S (Leica) was used with settings of 40 µm slice thickness, frequency of 2.0 and vibration speed of 2.0. Sections were mounted to glass slides and analyzed using an Olympus BP50 microscope at 5x objective magnification (Olympus).
For examination of root apical meristem architecture we used the method described in Tansengco et al. (2003). Semi-thin sections (2-4 μm) were cut from a Technovit 7100 (Heraeus Kulzer) embedded sample block with a Microtome RM 2125RT(Leica). Sections were mounted to glass slides and stained with 0.05% (w/v) Toluidine blue O (SPI supplies) in Borox buffer at pH 8.4. All these samples were observed with a Leica DMI 6000B at 20-40x objective magnification (Leica).

Linkage analysis
brush was crossed with L. japonicus MG-20 accession Miyakojima, which was described as a suitable partner for map-based cloning purposes (Hayashi et al., 2001;Kawaguchi et al., 2001). Plants used for linkage analysis were grown under the following conditions. Upon sterilization, seeds were directly planted in Seramis containing M. loti (final OD 600 = 0.001) and one week later the plants were reinoculated with rhizobia. Since mutant plants showed temperature-dependent growth behavior, they were grown at 26°C constantly. Nodulation of the mapping population was checked after one month following the first inoculation. The F 1 generation was self-fertilized, and a population of 1896 F 2 individuals was used for linkage determination. The segregation of phenotypes was also analyzed within the self progeny of four back cross individuals (BC1) resulting from two independent backcrosses of brush with Gifu wild type. The observed segregation ratio was 74:24 www.plantphysiol.org on August 23, 2017 -Published by Downloaded from Copyright © 2009 American Society of Plant Biologists. All rights reserved.
-25 -(Wild type:brush), with clear co-segretation of root and nodule phenotype. The mapping analysis was subsequently carried out using Simple Short Repeat (SSR) markers followed by separation on an ABI3730 DNA Analyzer allowing further analysis using the GeneMapper software (Applied Biosystems). SSR marker information was obtained from the Kazusa DNA Institute homepage (Sato et al., 2008).

Statistical analysis
All ANOVA and Tukey HSD tests were performed with the Vassar statistical homepage provided by Vassar College. (http://faculty.vassar.edu/lowry/VassarStats.html).

Supplemental material
Supplemental Figure S1: Root hair morphology of brush and wild type roots.
Supplemental Table S2: Calcium spiking response in brush root hairs.
Supplemental Figure S4: Infection thread formation and colonization of root hairs on brush plant roots.
Supplemental Figure S5: Temperature-dependent brush phenotype at the root apical meristem.
Supplemental Figure S6: Temperature-dependent brush phenotype at the root hair zone.
Supplemental Figure S7: AVG treatment of brush and wild type roots.
Supplemental Figure S8: Silver ion treatment of brush and wild type roots.
Supplemental Figure S9: ABA treatment of brush and wild type roots.
Supplemental Figure S10: GA 3 treatment of brush and wild type roots.