A novel ARID DNA-binding protein interacts with SymRK and is expressed during early nodule development in Lotus japonicus.

During the establishment of symbiosis in legume roots, the rhizobial Nod factor signal is perceived by the host cells via receptor-like kinases, including SymRK. The NODULE INCEPTION (NIN) gene in Lotus japonicus is required for rhizobial entry into root cells and for nodule organogenesis. We describe here a novel DNA-binding protein from L. japonicus, referred to as SIP1, because it was identified as a SymRK-interacting protein. SIP1 contains a conserved AT-rich interaction domain (ARID) and represents a unique member of the ARID-containing proteins in plants. The C terminus of SIP1 was found to be responsible for its interaction with the kinase domain of SymRK and for homodimerization in the absence of DNA. SIP1 specifically binds to the promoter of LjNIN but not to that of LjCBP1 (a calcium-binding protein gene), both of which are known to be inducible by Nod factors. SIP1 recognizes two of the three AT-rich domains present in the NIN gene promoter. Deletion of one of the AT-rich domains at the NIN promoter diminishes the binding of SIP1 to the NIN promoter. The protein is localized to the nuclei when expressed as a red fluorescence fusion protein in the onion (Allium cepa) epidermal cells. The SIP1 gene is expressed constitutively in the uninfected roots, and its expression levels are elevated after infection by Mesorhizobium loti. It is proposed that SIP1 may be required for the expression of NIN and involved in the initial communications between the rhizobia and the host root cells.

Legume plants are capable of acquiring nitrogen from rhizobacteria maintained symbiotically in specialized root organ nodules that form through a complex developmental process involving an exchange of signals between the host root cells and the bacteria. At the beginning of nodule organogenesis, specific flavonoid metabolites released by the legume roots notify soil rhizobia that a suitable host is nearby. Inside the bacteria, the flavonoids are recognized by a receptor protein known as NodD. The binding of NodD to these flavonoids activates the protein and promotes the transcription of other nod genes involved in the synthesis and secretion of Nod factors, the rhizobial signaling molecules (Peck et al., 2006). The common feature of all Nod factors is the presence of a chitin backbone and a fatty acyl tail (Denarie et al., 1996). Rhizobial Nod factors, in turn, are capable of inducing a series of specific responses in the host root cells, including root hair deformation, alkalinization of the cytosol, depolarization of the plasma membrane, and calcium influx and spiking (Ehrhardt et al., 1992(Ehrhardt et al., , 1996Kurkdjian, 1995;Felle et al., 1995Felle et al., , 1996Gleason et al., 2006). Root hair deformation and curling are the early morphological changes induced by rhizobia. Purified Nod factors can cause root hair deformation at concentrations as low as 10 212 M (Lerouge et al., 1990;Spaink et al., 1991;Sanjuan et al., 1992;Margaert et al., 1993), although in most cases curling is only observed when the bacteria are present (Relic et al., 1993).
The perception and signal transduction of Nod factors in the host cells have been subject to intense molecular and genetic studies in recent years. Rhizobial Nod factors have been shown to be recognized by lysin motif (LysM)-containing receptor-like kinases (RLKs), such as NFR1/NFR5 from Lotus japonicus and LYK3 and NFP from Medicago truncatula Radutoiu et al., 2003;Arrighi et al., 2006). The LysM RLKs are localized to the plasma membrane with the LysM domain protruding to the extracellular space and the kinase domain facing the cytoplasm. The LysM domain interacts with the lipochitin-oligosaccharide backbone of Nod factors, resulting in Nod factor recognition (Steen et al., 2003(Steen et al., , 2005Radutoiu et al., 2007). Using forward genetics and map-based cloning approaches, a series of host genes involved in the perception of the Nod factor signals, including SymRK (symbiosis RLK), Castor, Pollux, Nup133, Nup85, CCaMK, and Cyclops, have been characterized from the legume L. japonicus (Kistner et al., 2005). Mutations in any of these genes result in defects in nodule initiation. Downstream of this pathway, putative transcription factors encoded by LjNIN, LjNSP1, and LjNSP2 (Schauser et al., 1999;Heckmann et al., 2006) are also required to initiate nodule organogenesis.
SymRK from L. japonicus encodes a protein with three Leu-rich repeat (LRR) domains in the predicted extracellular region. Mutations in the SymRK gene abolish the host interaction with rhizobia (Stracke et al., 2002). Its orthologs, NORK, DMI2, and SYM19, from Medicago sativa, M. truncatula, and Pisum sativum, respectively, have been cloned (Endre et al., 2002;Stracke et al., 2002). The LRR-RKs represent the largest group of receptor kinases in plants, comprising approximately one-half of the predicted receptor kinases in Arabidopsis (Arabidopsis thaliana; Shiu et al., 2004). LRR-RKs have been implicated in diverse plant signaling pathways, including the perception of pathogen signals, brassinosteroid hormones, and the CLAVATA peptide hormone (Oldroyd and Dowine, 2004). It has been proposed that the three LRRs in SymRK may be involved in protein-protein interactions, proteinligand interactions, or autophosphorylation-regulated kinase activation (Yoshida and Parniske, 2005). The observation that the symRK/nork mutant fails to form arbuscular mycorrhiza suggests that SymRK/NORK plays a role in the exchange of signals with both symbiotic bacteria and fungi. Although SymRK is known to be essential in the Nod signaling pathway in L. japonicus, the biochemical mechanism leading to Nod factor-induced transcriptional activation is obscure. Moreover, how the kinase activity of SymRK is regulated remains to be determined. Recently, a NORKinteracting partner in M. truncatula has been identified as 3-hydroxy-3-methylglutaryl-CoA reductase 1, which may link the Nod signaling pathway with the formation of isoprenoid-derived phytohormones (Kevei et al., 2007).
In this study, we demonstrate that a novel protein, designated as SIP1 for SymRK-interacting protein 1, interacts with the kinase domain of SymRK. SIP1 is a transcription factor containing an AT-rich interaction domain (ARID) and may participate in Nod factorinduced transcriptional activation of genes required for nodule initiation.

Characterization of a SymRK-Associated Protein
The SymRK peptide contains three LRRs, a transmembrane domain, and an intracellular kinase domain. The structure features suggest potential roles of SymRK in the perception of extracellular signals and transduction of the signals through the intracellular kinase domain (Stracke et al., 2002). In an attempt to identify SymRK-associated proteins, we used the kinase domain of SymRK as a bait of the yeast two-hybrid system and screened a Lotus cDNA library constructed in the prey vector pGADT7-Rec2. Approximately five million yeast Saccharomyces cerevisiae colonies expressing the cDNA library were assayed for their ability to grow on selective synthetic dextrose medium (SD/-Leu-Trp-His-Ade). The bait plasmids were isolated from positive colonies and reintroduced back to yeast cells containing the prey plasmid. Colonies that failed to grow in the second round of testing were considered as false positives. After eliminating false positives, several clones were identified as potential interaction partners of SymRKprotein kinase (PK). One positive cDNA was isolated from three independent yeast colonies. It encoded a novel protein designated SIP1 ( Fig. 1A; Supplemental  Fig. S1).
The full-length SIP1 cDNA (GenBank accession no. EU559710) contained an open reading frame of 1,224 nucleotides encoding a peptide of 408 amino acid residues with a predicted molecular mass of 45.7 kD. Analysis of the peptide sequence revealed the presence of a conserved ARID (Fig. 1B) that has been implicated in sequence-specific DNA binding (Gregory et al., 1996). ARID-containing proteins appear to play important roles in diverse biological functions, including cell proliferation and differentiation, and organ development (Herrscher et al., 1995;Gregory et al., 1996;Kortschak et al., 2000). The mouse BRIGHT (B-cell regulator of immunoglobulin H transcription) protein and the fruit fly DRI (Drosophila dead ringer) are two well-characterized ARID-containing transcription factors (Herrscher et al., 1995;Gregory et al., 1996;Valentine et al., 1998;Iwahara et al., 2002;Wilsker et al., 2005). The predicted three-dimensional model of the ARID domain of SIP1 consists of eight a-helices, two b-strands, and four structure-undefined loops (Fig. 1B). The structural features of this 91-residue motif were identical to those found in animal ARIDcontaining proteins such as DRI (Fig. 1B;Iwahara and Clubb, 1999) and indicated that SIP1 might bind to the AT-rich region of plant promoters.
ARID-containing proteins are widely present in plant genomes. Ten such proteins have been found in Arabidopsis (Fig. 1C) and can be grouped into four subfamilies, designated high-mobility group (HMG), EGL-27 and MTA1 homology 2 (ELM2), plant homeodomain (PHD), and heat stress protein 20-like (Hsp20)-related proteins (Fig. 1D). Animal proteins containing HMG and ELM2 motifs have been implicated in DNA binding and transcription regulation (Ding et al., 2003;Stros et al., 2007). PHD-containing proteins are known to be involved in binding to methylated histone H3 (Ramon-Maiques et al., 2007). Hsp20-like proteins, also known as a-crystallin domain-containing proteins, are believed to function in protecting other proteins from denaturation by heat (Scharf et al., 2001). All of these proteins contain at least another motif in addition to ARID. However, there was no other recognizable motif except ARID in the Lotus SIP1. Phylogenetic tree analysis ( Fig. 1D) revealed that SIP1 is closely related to the Hsp20 subfamily of ARID proteins in Arabidopsis. The Hsp20 subfamily is characterized by the presence of an Hsp20-like domain in the C terminus of the protein.
The C terminus of SIP1 does not contain the Hsp20like domain, but has instead evolved to perform a new function, i.e. interaction with SymRK (see below). Taken together, our data suggest that SIP1 may have evolved from the Hsp20 subfamily of ARID-containing proteins during evolution and acquired the ability to act as a transcriptional regulator involved in the Nod factor signaling in L. japonicus.

Interaction between SymRK and SIP1
To determine which domain of SIP1 is responsible for its interaction with SymRK-PK, we constructed a series of SIP1 deletions in pGADT7 ( Fig. 2A). The N-terminal half of SIP1 (SIP1N) containing the N terminus and the ARID domain was not found to interact with SymRK-PK. The ARID domain alone, SIP1A, also did not interact with SymRK-PK. However, yeast colonies expressing the C terminus of SIP1, SIP1C, were able to grow on the selection medium lacking His (SD/-Leu-Trp-His) and exhibited significant b-galactosidase activities (Fig. 2C), suggesting that the C-terminal 184 amino acid residues of SIP1 are critical for its interaction with SymRK. This interaction was further confirmed using an in vitro protein-protein interaction assay (Fig. 2D). For this assay, SIP1 and its deletion fragments were expressed as His-or chitinbinding domain (CBD)-tagged recombinant proteins and immobilized to nickel beads or chitin beads. After incubation of the beads with purified SymRK-PK, followed by washing with buffer, proteins retained to the beads were eluted in SDS sample buffer and resolved on SDS-PAGE. The presence of SymRK on the beads was detected by immunoblotting with the anti-SymRK antibody. As shown in Figure 2D, only the full-length SIP1 and SIP1C could pull down SymRK-PK, confirming that the C terminus of SIP1 was responsible for its interaction with SymRK.

SIP1 Binds AT-Rich Double-Stranded DNA
The ARID motif of SIP1 contains a noncanonical helix-turn-helix motif (helices H5 and H6) for potential DNA binding. To test this DNA-binding activity, Figure 1. SIP1 is a novel member of the ARID family of DNA-binding proteins in plants. A, The deduced amino acid sequence of SIP1 contains 408 amino acid residues with a calculated molecular mass of 45.7 kD. The ARID region is shaded in light blue. The conserved a-helices (H1-H8) and b-sheet (b1 and b2 strands) are underlined. The colors of the residues in these regions match those shown in B. B, Comparison of the ribbon models of ARIDs present in plant SIP1 (residues 126-217) and animal DRI (residues 265-388; Iwahara and Clubb, 1999) proteins shows that both are able to form eight a-helices (H1-H8) and two b-strands (b1 and b2). The models were produced using PyMOL (v0.99) software (http://www.delsci.com/rel/099/). C, Schematic representation of functional motifs present in SIP1 and its homologs in Arabidopsis. Contrary to the Lotus SIP1, all Arabidopsis homologs contain other motifs in addition to ARID, suggesting that SIP1 may have a legume-specific function. References: Hsp20, Scharf et al. (2001)  we purified His-tagged SIP1 and performed electrophoretic mobility shift assays (EMSA) with a 32 P-endlabeled double-stranded DNA trimer of NP 3 or TTA 9 , the consensus binding sites of the Drosophila homeodomain protein Engrailed (Gregory et al., 1996). The results showed that SIP1 bound both NP 3 and TTA 9 ( We further examined whether SIP1 contains a transcription activation domain (AD). We constructed plasmids that would express the GAL4 DNA-binding domain fused with SIP1 or its deletion fragments. The plasmids were transferred into the yeast strain AH109 that expressed ADE2, HIS3, lacZ, and MEL1 reporter constructs under GAL4-responsive promoters (CLONTECH). If SIP1 contained an AD, the SIP1 fusion protein should bind to the GAL4-responsive promoters and drive the expression of the ADE2, HIS3, lacZ, and MEL1 reporters. A known transcription activator, NSP1 (Smit et al., 2005;Heckmann et al., 2006), was used as a positive control and proved able to allow yeast colonies to grow on SD/-Trp-His and SD/ -Trp-Ade media (Fig. 3B). However, as shown in Figure 3B, colonies expressing SIP1 and its deletion fragments could not grow on these media, indicating that SIP1 did not contain a transcription AD (Fig. 3B). It would be interesting to test if SIP1 forms hetero-oligomers with other proteins that contain an activator or repressor domain.

SIP1 Binds Specifically to the NIN Promoter
The expression of the NIN (nodule inception) and CBP1 (calcium-binding protein 1) genes is induced by Nod factors (Schauser et al., 1999;Webb et al., 2000;Radutoiu et al., 2003). To test if SIP1 plays a role in the induction of NIN and CBP1 by Nod factors, we cloned the promoter regions of both genes and assayed for their affinity with SIP1 in the yeast one-hybrid system. The 500-bp NIN promoter (Borisov et al., 2003) and the 482-bp CBP1 promoter (Webb et al., 2000) were inserted in front of the HIS3 reporter, generating NIN pro THIS3 and CBP1 pro THIS3 constructs. SIP1 was expressed as a fusion protein with the GAL4 AD. This plasmid (pGAD-SIP1), along with one of the reporter constructs, was cotransformed into yeast Y187 cells. If SIP1 could bind to the promoter, the GAL4 AD would drive the expression of the HIS3 reporter, and the yeast cells would grow on SD/-Trp-Leu-His plates supplemented with 3-amino-1,2,4-triazole (3-AT). The results showed that the cotransformants of pGAD-SIP1 and NIN pro THIS3 (Fig. 4C, colony 3) were indeed able to grow on the selective plates, whereas the cotransformants of pGAD-SIP1 and CBP1 pro THIS3 (Fig. 4C, colony 5) did not grow, indicating that SIP1 SymRK-PK was expressed as a fusion protein with the GAL4 DNA-binding domain in the bait vector (pGBKT7). SIP1 and its deletions were expressed from the prey vector (pGADT7). Yeast AH109 cells harboring the two plasmids were streaked on the SD/-Leu-Trp plate containing X-Gal (80 mg/mL; X-Gal plate) and were cultured in SD/-Leu-Trp broth for b-galactosidase activity assays. The combination of pGBKT7-53/pGADT7-SV40 was used as a positive control and pGBKT7-Lam/pGADT7-SV40 as a negative control (CLONTECH). D, In vitro pull-down assay for the interaction between SymRK-PK and SIP1. His-tagged SIP1 bound to nickel beads was mixed with purified SymRK-PK protein. After washing, the proteins pulled down by the nickel beads were separated on SDS-PAGE (top). A similar gel was used for western blot using anti-SymRK antibodies (bottom). [See online article for color version of this figure.] specifically binds to the NIN promoter but not the CBP1 promoter.
A search of the promoter region (about 4 kb upstream of the first ATG) of the NIN gene identified three potential AT-rich motifs, located in the regions from 22,299 to 22,287, from 2393 to 2367, and from 269 to 259 bp, respectively (Borisov et al., 2003;Fig. 4A). To determine whether SIP1 binds to these sequences in vitro, we synthesized three oligonucleotides corresponding to Oligo1 to Oligo3 (Fig. 4A) and performed EMSA assays using purified SIP1. As shown in Figure 4B, SIP1 specifically recognized and bound to Oligo1 and Oligo3 but not Oligo2.
Because the 500-bp NIN promoter fragment used in the yeast one-hybrid assay (Fig. 4A) did not contain Oligo1 (Fig. 4C), we hypothesized that Oligo3 should be the AT-rich site for SIP1 binding. We further reasoned that deletion of the Oligo3 site from the NIN promoter should diminish the ability of the yeast cells to grow on SD-His13-AT medium. We removed a 70-bp fragment (21 to 269 bp) from the NIN promoter and showed that the DNIN promoter lost the binding site for SIP1 and yeast cells were no longer able to grow on the selection medium (Fig. 4C, colony 4). Taken together, our data demonstrate that SIP1 specifically recognizes the Oligo3 AT-rich site of the NIN promoter and may be required for the Nod factorinduced NIN gene expression.

SIP1 Dimerization
The mouse BRIGHT protein forms a tetramer that binds to DNA (Herrscher et al., 1995) but exists as a dimer in the absence of DNA (Nixon et al., 2004). To determine whether SIP1 dimerizes in the absence of DNA, we coexpressed two SIP1 fusion proteins, one with the GAL4 DNA-binding domain (SIP1-BD) and the other with the GAL4 AD (SIP1-AD) in the yeast two-hybrid system. The N and C termini and the ARID domain of SIP1 were also fused with the GAL4 AD, generating SIP1N-AD, SIP1C-AD, and SIP1A-AD. As shown in Figure 5A, full-length SIP1-AD interacted with SIP1-BD, suggesting that SIP1 dimerizes. The interactions were also detected when the C terminus of SIP1 was used instead (SIP1C with SIP1; SIP1C with SIP1C; Fig. 5A), suggesting that the C-terminal 184 residues are responsible for SIP1 dimerization in the absence of DNA. In contrast, the N terminus (223 residues) of SIP1, which contains the ARID domain, was not required for SIP1 dimerization.
To further confirm this interaction, we immobilized glutathione S-transferase (GST)-SIP1, CBD-SIP1N, and CBD-SIP1C fusion proteins on GST beads or chitin beads and incubated the beads with His-tagged SIP1, SIP1N, and SIP1C, respectively. After washing, the proteins retained on the beads were subjected to immunoblotting with the anti-His-tag antibody. As shown in Figure 5B, GST-SIP1 and CBD-SIP1C could pull down SIP1 (lanes 1 and 3), whereas CBD-SIP1N could not (lane 2). Moreover, CBD-SIP1C was able to pull down SIP1C (lane 6), suggesting the C termini are sufficient to form dimers.

Induction of SIP1 Gene Expression by Rhizobial Infection
SymRK is expressed constitutively in the roots of Lotus, and its mRNA level does not change upon the treatment with Nod factors for 24 and 48 h or after the inoculation with Mesorhizobium loti (Stracke et al., 2002). The NIN gene, on the other hand, is not expressed in the uninfected roots and is induced by treatments with Figure 3. DNA binding of SIP1. A, EMSA was performed using purified SIP1 and 32 P-end-labeled oligonucleotide probes (TTA 9 and NP 3 ). NP 3 : (TCAATTAAATGA) 3 ; (TTA) 9 : (TTATTATTA) 3 . B, Transcription activation assay of SIP1 was performed in the yeast strain AH109, which contains the HIS3 and ADE2 reporter genes under distinct GAL4-responsive promoter elements. Plasmids each expressing a fusion protein of the GAL4 DNA-binding domain (BD) with SIP1 or its deletion fragments were transformed into AH109. The yeast cells were streaked on SD/-Trp (control medium), SD/-Trp-His, and SD/-Trp-Ade (selection media). Yeast cells harboring pGBKT7-lam or pGBKT7-p53 (CLONTECH) served as negative controls, whereas NSP1 (Smit et al., 2005)  Nod factors or rhizobial infection (Schauser et al., 1999;Radutoiu et al., 2003). Using quantitative PCR, we examined the SIP1 and NIN mRNA levels in different tissues of L. japonicus. SIP1 mRNA was expressed constitutively in leaves and control (uninoculated) roots (Fig. 6A). In the young roots (2 d old), SIP1 expression levels were low but increased to a steady state after 6 to 8 d. In the stem, its expression levels were relatively low but detectable. In contrast, NIN expression levels were relatively low in stems, leaves, and control roots (Fig. 6B).
We then focused on the expression levels of SIP1 and NIN in roots after infection with M. loti. When inoculated with M. loti, an induction of SIP1 mRNA was observed as early as 5 h post inoculation (hpi). In 24 hpi, the expression levels dropped down to a steady level, which was slightly higher than that observed in the control roots. This expression pattern was distinct from that of NIN (Fig. 6B), which exhibited significant induction 5 h after rhizobial inoculation and maintained a high expression level in inoculated roots. It is important to note that the timing of the SIP1 induction (5 hpi) correlates well with that of NIN. After the initial induction, NIN continued to be expressed at relatively high levels. In conclusion, the SIP1 gene is expressed constitutively in roots and leaves, and its expression levels are elevated in roots transiently (5 hpi-1 d postinoculation [dpi]) after rhizobial infection. It may play a role in the induction of the NIN gene during the process of rhizobial entry and nodule organogenesis.

Subcellular Localization of SIP1
To determine the subcellular localization of SIP1, we expressed SIP1 as a fusion protein with the Discosoma red fluorescent protein (DsRed) under the control of the cauliflower mosaic virus 35S promoter. The fusion protein was transiently expressed in the onion (Allium cepa) epidermal cells via particle bombardment, and its expression was monitored using a confocal laserscanning microscope. As expected in the control cells expressing DsRed alone, the red fluorescence was detected only in the cytoplasm (Fig. 7, D-F). In the onion epidermal cells expressing the SIP1-DsRed fusion protein, the red fluorescence was concentrated to the nuclei (Fig. 7, A-C). This result is consistent with its potential function in DNA binding and transcription regulation. However, it remains to be determined how SIP1 interacts with SymRK in Lotus roots after rhizobial infection.  (Oligo3). B, EMSA was performed using purified SIP1 and 32 P-end-labeled Oligos. The components of the binding reactions are indicated at the top of the lanes. C, In vivo assay of SIP1 binding to the NIN promoter. SIP1 was expressed as a fusion protein with the GAL4 AD (pGADT7-SIP1) in yeast Y187 cells harboring the reporter construct NIN pro THIS3, DNIN pro THIS3, or CBP1 pro THIS3, which expresses the HIS3 reporter gene under the promoter of the NIN or CBP1 gene. The NIN promoter is a 500-bp fragment (21 to 2500 bp), whereas the DNIN promoter contains a 430-bp fragment (270 to 2500 bp) lacking the Oligo3 AT-rich site. The CBP1 promoter (462 bp) was amplified by PCR (Webb et al., 2000). Yeast cells were examined for growth on the SD/-Trp-Leu (control) and SD/-Trp-Leu-His plates in the presence of 40 mM 3-AT (selection). Yeast cells harboring p53-HIS2 and pGADT7-53 were used as a positive control, whereas those containing p53-HIS2 and pGADT7-SIP1 served as negative control. [See online article for color version of this figure.] perceive the Nod factor signals released by the infecting rhizobia (Yoshida and Parniske, 2005). NIN is a key transcription factor required for rhizobial entry into the root cells and for nodule organogenesis (Schauser et al., 1999;Radutoiu et al., 2003). In this report, we describe a novel DNA-binding protein, SIP1, which may provide a link between SymRK and NIN. We demonstrate that SIP1 interacts with the intracellular kinase domain of SymRK (Fig. 2) and specifically binds to the Oligo3 AT-rich site of the NIN promoter (Fig. 4). Because SIP1 does not contain a transcription AD (Fig. 3), it may form hetero-oligomers with other unidentified proteins that contain an activator or repressor domain. Plu-1, a human ARID-containing protein, also lacks an AD but contains transcriptional repressor properties and forms hetero-oligomers with other transcription factors (Tan et al., 2003). The initial signal communications between the rhizobia and roots involve a number of different molecular exchanges culminating in calcium spiking in the host cells. SIP1, by linking the SymRK receptor kinase to the NIN transcription factor, may play a pivotal role for the successful establishment of rhizobia-legume symbiosis.

SymRK is a member of a large family of LRR RLKs in plants and is required for the legume root cells to
SIP1 represents a new member of the conserved ARID family of proteins in plants. The ARID domain was initially identified in the mouse BRIGHT protein Figure 5. Dimerization of SIP1. A, Yeast AH109 cells were cotransformed with different combinations of the bait and prey constructs containing either the full-length SIP1 or its deletion fragments. The yeast cells were transferred from SD/-Leu-Trp-His-Ade to SD/-Leu-Trp plate containing X-Gal (80 mg/mL). Yeast cells harboring pGBKT7-53/ pGADT7-SV40 were used as a positive control, whereas those containing pGBKT7-Lam/pGADT7-SV40 served as a negative control. B, In vitro protein-protein interaction assay. SIP1 and its deletion fragments were expressed as GST-or CBD-tagged proteins and immobilized to glutathione-or chitin beads. SIP1 and its deletion fragments were also expressed and purified as His-tagged proteins, followed by elution from the nickel-column with imidazole. The eluted soluble SIP1 or its deletion fragments were then mixed with glutathione beads or CBD beads containing an immobilized peptide. After washing with buffer, proteins retained to the beads were solubilized in SDS sample buffer and separated on SDS-PAGE (top), and the interacting proteins were detected using anti-His antibodies (bottom). [See online article for color version of this figure.] Figure 6. Expression of SIP1 and NIN genes in L. japonicus. Roots were harvested 2, 5, and 12 hpi, and 1, 2, 4, 6, and 8 dpi with M. loti. Roots treated with water served as the mock control. Total RNA was isolated stems (S), leaves (L), control roots (R), and Rhizobium-inoculated roots (IR). Steady-state transcript levels of SIP1 (A) and NIN (B) were measured by quantitative PCR on a real-time PCR system . The ATPase gene (AW719841) was used as an internal control. Relative values of transcripts normalized to the control roots (4 dpi) are shown.   Herrscher et al., 1995) and Drosophila DRI (Gregory et al., 1996). ARID-containing proteins have now been shown to be present in all sequenced eukaryotic organisms, including mammals, insects, plants, nematodes, and yeast (Wilsker et al., 2002). Although ARID is an a-helix-based DNA-binding domain, the structures and functions of ARID-containing proteins are highly diverse (Wilsker et al., 2002). These proteins appear to play vital roles in the regulation of development and tissue-specific gene expression.
The ARID domains, especially their 3-D structures, are conserved between animals and plants (Fig. 1B). However, other parts of the molecules can be very different among ARID-containing proteins in an organism. In Arabidopsis, the 10 ARID-containing proteins vary in length from 319 residues in At3g13350 to 786 residues in At2g17410. They can be grouped into four subfamilies on the basis of their phylogenetic relationship and the presence of functional motifs (Fig. 1D). L. japonicus SIP1 is closely related to the Hsp20 subfamily in Arabidopsis, although it does not contain an Hsp20like domain. The Hsp20 motif is found in the C termini of the Arabidopsis ARID proteins (Fig. 1C). In L. japonicus SIP1, the C terminus has evolved into a different domain that is responsible for interacting with SymRK (Fig. 2) and for the homodimerization (Fig. 5). SIP1 represents a new ARID-containing transcription factor in plants and is potentially involved in linking the Nod signal perception to the transcriptional regulation during nodule development in L. japonicus.
The ARID domain has not been found in Archea and eubacteria, but is widespread in protozoa, metazoans, green algae, fungi, plants, and animals. There are 15 ARID proteins in the human genome and two in S. cerevisiae (Wilsker et al., 2002(Wilsker et al., , 2005. The four Arabidopsis ARID subfamilies (Fig. 1D) do not correlate closely with the seven subfamilies of the human ARID proteins (Wilsker et al., 2005). At3g43240 is a single member of the Arabidopsis subfamily that possesses an ARID and a PHD domain. In humans, the most closely related proteins are the four ARID proteins in the JARID1 subfamily, which contain two or more PHD domains each, in addition to several other functional motifs (Wilsker et al., 2005), and are also twice as large as At3g43240. The remaining three subfamilies of Arabidopsis ARID proteins are plant specific, because no known human ARID protein contains an Hsp20, HMG, or ELM2 domain (Wilsker et al., 2002(Wilsker et al., , 2005. This diversity in peptide length and motif arrangements suggests that the specific ARID subfamilies have apparently evolved independently after the divergence of plant and animal lineages. ARID proteins in higher eukaryotes are involved in differentiation and transcriptional regulation of gene expression (Wang et al., 2007). At least two human ARID proteins are known to be induced during tumorigenesis. Plu-1 (JARID1B) was identified as an upregulated gene product in human breast cancer and is not expressed in normal adult tissues except testis. RBP1L1 (ARID4B) was identified as a tumor antigen and is highly expressed in all cancer cell lines (Wilsker et al., 2002;Tan et al., 2003). Nodule initiation involves the de-differentiation of the root cortical cells into the meristematic cells and the activation of a series of nodule-specific genes. Our study shows that SIP1 is constitutively expressed in roots, and its expression is highly induced 5 h after rhizobial infection (Fig. 6). Although L. japonicus SIP1 is not homologous to the human Plu-1 and RBP1L1 except within the ARID domain, they share some similarity in their function in reprogramming the differentiated cells back to the dividing status, i.e. tumorigenesis in human and nodule organogenesis in the legume.
There are three AT-rich motifs in the promoter of the NIN gene, two of which showed affinity with SIP1 in in vitro assays (Fig. 4). In the yeast one-hybrid system, the 500-bp promoter region of the NIN gene was found Figure 7. Subcellular localization of SIP1. SIP1 was expressed as a fusion protein with DsRed under the control of the 35S promoter. The plasmid was delivered to the onion epidermal cells via particle bombardment. The plasmid (35STDsRed) expressing DsRed alone served as a control. The onion cells were observed with a confocal laser-scanning microscope, and pictures were taken in the light transmission channel (A and D) or using the 583/558-nm filter (B and E). The photos were superimposed using Photoshop software (C and F). Note that DsRed is distributed in the cytosol, whereas SIP1:DsRed goes to the nuclei.
to contain cis-DNA elements for SIP1 binding. The binding of SIP1 to the NIN promoter was apparently brought about via the Oligo3 AT-rich motif, which is localized to the promoter region between 259 to 269 bp. Because SIP1 does not have a transcription AD (Fig. 3B), it is likely that SIP1 may function as a DNA sequence-specific binding protein and may form an oligomer with another transcription activator or repressor. We are currently in the process of searching for these potential SIP1-interacting partners and examining the biological functions of SIP1 and its associated proteins in nodule organogenesis in L. japonicus. Further analyses of the signaling pathway involving SIP1 may eventually lead to the identification of new molecular events required for the later steps in the symbiotic establishment in legume roots.

Plant Materials
Seeds of Lotus japonicus MG-20 were surface-sterilized in 2% sodium hypochlorite for 8 min and washed seven times with sterile water. The seeds were left to germinate for 48 h at 22°C on sterile water-soaked filter paper in petri dishes in a dark room. Seedlings were subsequently planted in pots on sterile sands supplemented with nitrogen-free Fahraeus medium (Fahraeus, 1957) and were grown in a growth chamber maintained at 22°C with a 16-h-light/8-h-dark cycle. Five-day-old seedlings were inoculated with approximately 10 7 CFU/mL of Mesorhizobium loti NZP2037. Roots were collected from plants 2, 4, 6, 8, and 12 d after rhizobial inoculation and were frozen in liquid nitrogen.

cDNA Library Construction
Total RNA was isolated from the equally mixed roots collected 2, 4, 6, 8, and 12 d after rhizobial inoculation using the TRIzol reagent (Invitrogen). Total RNA (2 mg) was reverse transcribed into single-stranded cDNA using oligo(dT) as the primer. The cDNA was size-fractionated using BD Matchmaker library construction and screening kits (BD Biosciences CLONTECH). The cDNA fragments longer than 500 bp were cotransformed with linearized pGADT7-Rec into yeast (Saccharomyces cerevisiae) AH109 cells. The transformants were selected on SD/-Leu medium according to the manufacturer's instruction. The transformation efficiency was approximately 2 3 10 6 CFU/3 mg pGADT7-Rec.

Library Screening
The SymRK-PK cDNA (GenBank accession no. AF492655) was amplified by reverse transcription-PCR with the following gene-specific primers: 5#-GGGCATATGATGGAGAGGTACAAAACCTTG-3# and 5#-AAAGAATTCT-CTCGGCTGTGGGTGAG-3#. PCR was carried out using ExTaq DNA polymerase (TaKaRa) with an initial denaturation step of 95°C for 5 min, followed by 30 cycles of 95°C for 15 s, 55°C for 30 s, and 72°C for 1 min. After the last cycle, an extension step for 10 min at 72°C was carried out. Purified PCR products were inserted into the Nde1-EcoR1 site of pGBKT7 vector. Screening of interaction clones was carried out according to the manufacturer's instructions (CLONTECH). A total of 1 3 10 7 transformants from the cDNA library were screened for growth on the SD/-Leu-Trp-His-Ade media. Positive clones were verified by retransformation of the rescued plasmid into the AH109 containing the bait plasmid (pGBKT7-SymRK-PK). Plasmid pGBKT7-Lam (CLONTECH) was used as a negative control. Colonies growing on the SD/-Leu-Trp-His-Ade media were transferred to selective media-containing X-Gal (80 mg/mL) or were left on filters as described . The blue colonies were characterized.

b-Galactosidase Assay
Yeast cells grown in liquid selection media were pelleted and washed twice with Z-buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 2 , 10 mM KCl, 1.0 mM MgSO 2 , pH 7.0). The cells were resuspended in 300 mL of Z-buffer and permeabilized by two freeze-thaw cycles in liquid nitrogen and 37°C. Cell extracts were added 0.7 mL of Z-buffer containing 50 mM b-mercaptoethanol and 160 mL of O-nitrophenyl b-D-galactopyranoside (4 mg/mL) in Z-buffer. After incubation at 30°C for 30 min or until the yellow color appeared, the reaction was terminated by the addition of 0.4 mL of 1.0 M Na 2 CO 3 . The reaction mixture was centrifuged for 10 min at 14,000 rpm to remove cell debris. b-Galactosidase activity in the supernatant was measured at OD 420 and expressed in Miller units (Miller, 1972).
To assay SIP1 dimerization, GST-tagged SIP1 was expressed and purified on glutathione-Sepharose 4B (Sigma) as described elsewhere (Zhang et al., 2000). CBD-tagged SIP1N and SIP1C were immobilized on chitin beads. The beads were incubated with purified His-SIP1, His-SIP1N, and His-SIP1C proteins for 1 h on ice with gentle shaking. Following washing, the retained proteins were resolved on 10% PAGE and detected using the anti-His-tag antibody.

EMSA
EMSA of SIP1 was performed on 5% nondenaturing polyacrylamide gels as described previously (Gregory et al., 1996). DNA-binding activity of SIP1 was examined using g-32 P-end-labeled double-stranded oligonucleotide trimer of the consensus Engrailed binding site NP 3 [(TCAATTAAATGA)X3] or trimer of TTA 9 [(TTATTATTA)X3; Gregory et al., 1996]. To test if SIP1 specifically binds to the promoter of the NIN gene, Oligo1, Oligo2, and Oligo3 (see Fig. 4A for sequences) were synthesized and used for EMSA. His-tagged SIP1 was purified with nickel-agarose beads and eluted using 500 mM imidazole followed by dialysis. Approximately 10 ng of SIP1 protein was added in a mixture containing 0.3 mM of g-32 P-labeled double-stranded DNA in 10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 1.0 mM EDTA, 1.0 mM dithiothreitol, 50 mg bovine serum albumin, and 5% glycerol. After incubation at room temperature for 20 min, the samples were subjected to electrophoresis on a 5% polyacrylamide/bis (29:1) gel in 0.53 Tris-borate/EDTA buffer at a constant voltage of 100 V. The gel was dried, and the radioactive bands were detected using PhosphorImager (FUJIFILM).
The plasmid was transformed into yeast Y187 cells harboring pGADT7-SIP1, which expresses SIP1 fused with the GAL4 transcription AD. The DNA-binding activity of SIP1 was determined by the expression of the HIS3 reporter.

Measurement of Transcript Level by Quantitative PCR
Total RNA was isolated from stems, leaves, control roots, and roots inoculated with M. loti, using a TRIZOL reagent (Invitrogen). RNA samples were treated with DNase I to remove potential contaminating genomic DNA, followed by extraction with phenol:chloroform. The DNA-free RNA samples were confirmed by lack of a PCR product using primers specific to the untranscribed NIN gene promoter (5#-GTTTTCAAGAATGGGAGGGG-3#, 5#-CTCCTCTGGTTTCATTGGTG-3#). First-strand cDNA was prepared using Oligo(dT) primer, and quantitative PCR was performed on a Mini Opticon real-time PCR system (Bio-Rad) using One-Step SYBR PrimeScript RT-PCR kit II (Takara). The same cDNA pool was used for amplification of all tested transcripts. The relative quantification software (Bio-Rad) was used to determine the efficiency-corrected relative transcript concentration, normalized to a calibrator sample . ATP synthase was used as a reference gene (AW719841). Melting curve analysis and sequencing of the amplified products was used to determine the identity of the amplified PCR products. Intron spanning primers were used for transcript amplification of the ATP synthase (5#-CAATGTCGCCAAGGCCCATGGTG-3# and 5#-AACACCACT-CTCGATCATTTCTCTG-3#), SIP1 (5#-CTTGCAGGTAAGTTCACAAC-3# and 5#-GTACGAGTCCAATCAGATCCAG-3#), and NIN (5#-AATGCTCTTGAT-CAGGCTG-3# and 5#-AGGAGCCCAAGTGAGTGCTA-3#) genes. The same experiment was performed three times, and the averages of measurements were presented.
Sequence data from this article can be found in the GenBank/EMBL data libraries under accession number EU559710.

Supplemental Data
The following materials are available in the online version of this article.