An IRE-like AGC kinase gene, MtIRE, has unique expression in the invasion zone of developing root nodules in Medicago truncatula.

The AGC protein kinase family (cAMP-dependent protein kinases A, cGMP-dependent protein kinases G, and phospholipid-dependent protein kinases C) have important roles regulating growth and development in animals and fungi. They are activated via lipid second messengers by 3-phosphoinositide-dependent protein kinase coupling lipid signals to phosphorylation of the AGC kinases. These phosphorylate downstream signal transduction protein targets. AGC kinases are becoming better studied in plants, especially in Arabidopsis (Arabidopsis thaliana), where specific AGC kinases have been shown to have key roles in regulating growth signal pathways. We report here the isolation and characterization of the first AGC kinase gene identified in Medicago truncatula, MtIRE. It was cloned by homology with the Arabidopsis INCOMPLETE ROOT HAIR ELONGATION (IRE) gene. Semiquantitative reverse transcription-polymerase chain reaction analysis shows that, unlike its Arabidopsis counterpart, MtIRE is not expressed in uninoculated roots, but is expressed in root systems that have been inoculated with Sinorhizobium meliloti and are developing root nodules. MtIRE expression is also found in flowers. Expression analysis of a time course of nodule development and of nodulating root systems of many Medicago nodulation mutants shows MtIRE expression correlates with infected cell maturation during nodule development. During the course of these experiments, nine Medicago nodulation mutants, including sli and dnf1 to 7 mutants, were evaluated for the first time for their microscopic nodule phenotype using S. meliloti constitutively expressing lacZ. Spatial localization of a pMtIRE-gusA transgene in transformed roots of composite plants showed that MtIRE expression is confined to the proximal part of the invasion zone, zone II, found in indeterminate nodules. This suggests MtIRE is useful as an expression marker for this region of the invasion zone.

Nitrogen-fixing root nodules are the result of a complex and unique interaction between leguminous plants and soil rhizobia. During nodule development, rhizobia are brought into roots through infection threads that originate in deformed root hairs that curl to form a so-called ''shepherd's crook.'' Infection thread initiation and growth require living rhizobia that are synthesizing specific Nod factors (Ardourel et al., 1994;Limpens et al., 2003). Infection threads bring rhizobia past several root cell layers and deposit them into newly divided nodule cells within a membranebounded compartment called the symbiosome in a process resembling endocytosis. Within the infected cells, an environment is established where the rhizobia and plant express new proteins that enable biochemical support of nitrogen fixation and assimilation. In indeterminate nodulators such as Medicago truncatula, new infections occur continuously over the lifetime of the nodule, with new infections starting below the meristematic distal end of the nodule at the beginning of the infection zone, zone II. In the proximal end of zone II, both rhizobia and plant cells expand and mature. Exit from zone II is marked by starch accumulation, which the rhizobia apparently use as a carbon source as they continue their maturation to nitrogen fixation capability in zone III . Reviews on legume root nodule development are available (Brewin, 1991;Kijne, 1992;Gage and Margolin, 2000;Brewin, 2004;Gage, 2004).
Nodule-specific (nodulin) and nodule-enhanced genes are expressed exclusively and primarily in nodules, respectively. Most nodulin genes have homologs in nonlegumes, suggesting that nodule-specific genes have been recruited from other developmental pathways. It has been noted that a number of nodulins are also expressed in nonsymbiotic tissue, including in floral tissues (Szczyglowski and Amyot, 2003). In some cases, genes expressed in both nodules and in floral tissue may have a role in tip growth during infection thread and pollen tube growth, respectively (Rodriguez-Llorente et al., 2004). Other tissues with tip growth, a type of cell expansion, include root hair cells, which have been used as a model system to study tip growth, especially in Arabidopsis (Arabidopsis thaliana; Schiefelbein, 2000).
The AGC protein kinases contain the protein kinase A, protein kinase G, and protein kinase C regulatory kinases. In animals and fungi, members of this kinase family act in protein phosphorylation cascades. In animals, a key AGC kinase regulator is the 3-phosphoinositidedependent protein kinase (PDK1), a central growth regulator that integrates signaling events from receptors that stimulate the synthesis of phosphatidylinositol 3,4,5-trisphosphate (Bogre et al., 2003;Mora et al., 2004). Far less is known about AGC kinases in plants.
The Medicago genome project has identified many potential genes with roles in nodule development by virtue of their being found as expressed tag sequences (ESTs) only in cDNA libraries prepared from tissue that includes nodules (Fedorova et al., 2002;El Yahyaoui et al., 2004;Lohar et al., 2006). One of the genes identified in this manner is MtIRE, a Medicago homolog of the AtIRE gene. In this work, we present results showing the cloning of the complete MtIRE cDNA and its relationship to AtIRE and IRE-like genes. We investigate MtIRE expression in plant tissues during nodule development and in symbiotically defective Medicago mutants. Our findings show that MtIRE does not have a role in root hair growth in Medicago and suggest the role of MtIRE is likely to be in maturation of infected nodule cells in zone II, before effective symbiotic nitrogen fixation occurs.

Isolation and Sequence of the Complete MtIRE Gene
In 2002, Fedorova et al. reported 340 tentative consensus (TC) sequences comprising ESTs found only in cDNA libraries that were made from M. truncatula nodule-containing tissue (Fedorova et al., 2002). One of these was TC33166, currently annotated as TC103185 in The Institute for Genomic Research MtGI 8.0 release (www.tigr.org). TC103185, 1,383 nucleotides long, consists of four ESTs; its strongest BLASTX hit is the Arabidopsis AtIRE protein kinase-like gene (GenBank accession no. AB037133; Oyama et al., 2002). Using the working draft of overlapping bacterial artificial chromosome (BAC) clones mth2-13b8 and mth1-8d23 (GenBank accession nos. AC122727 and AC133139, respectively) containing the genomic copy of TC103185, sequence from TC103185, and the AtIRE gene, primers were chosen to reverse transcribe and amplify M. truncatula A17 nodule mRNA corresponding to MtIRE of progressively increasing sizes. These efforts resulted in a cDNA containing the complete MtIRE coding region of 3,504 nucleotides. Subsequently, 5#-and 3#-RACE were carried out to identify the 5# and 3# ends of the MtIRE transcript. 5# RACE identified three different 5# ends corresponding to untranslated 5# regions of 174, 120, and 115 bases upstream of the predicted translational start codon. 3#-RACE yielded a single 3# end that contained 296 nucleotides 3# to the predicted translational stop. The complete MtIRE cDNA of 3,974 nucleotides corresponding to the longest transcript detected is available in the GenBank database (accession no. AY770392).

Characterization of MtIRE Gene Structure and Its Encoded Protein
The Medicago genome project has mapped BACs mth2-13b8 and mth1-8d23 containing MtIRE to chromosome 5 (www.medicago.org/genome). Comparison of the MtIRE genomic and cDNAs revealed that the MtIRE gene spans a genomic region of 9.1 kb and consists of 17 exons and 16 introns (Fig. 1A). This organization is almost identical to the AtIRE gene, which is shown for comparison in Figure 1B. Both genes have 17 exons of similar sizes, but in the Medicago MtIRE gene the introns are significantly larger than in the AtIRE gene. The Arabidopsis genome has four other IRE-like genes. Their genome organization is similar to AtIRE and to MtIRE although slight differences are noted, especially extra exons in AtIRE_2 and the lack of the first three exons in AtIRE_3 (Fig. 1C).
The third start codon (ATG) from the 5# end of the transcript is the presumed start site for the longest open reading frame of the MtIRE gene. MtIRE encodes a deduced protein of 1,168 amino acid residues. ExPASy (www.expasy.org/tools/scanprosite; Gattiker et al., 2002) and BLAST (www.ncbi.nlm.nih.gov/BLAST/; Altschul et al., 1990) tools were used to analyze the deduced MtIRE protein (Fig. 1D). It contains a Glu-rich region at its N-terminal end, basic-type (Kalderon et al., 1984) and bipartite-type (Robbins et al., 1991) nuclear localization signals, a zinc finger-like sequence (C-X 2 -C-X 11 -H-X 3 -C; Fig. 1D; Bohm et al., 1997), and numerous potential phosphorylation sites, including those for casein kinase II, protein kinase C, cAMP-, cGMP-, and phosphoinositide-dependent protein kinases. MtIRE contains a putative Ser/Thr protein kinase domain toward its C-terminal end (Fig. 1D). Within the Ser/ Thr protein kinase domain is an activation loop motif that is a putative target of PDK1. C terminal to the kinase domain is the PDK1-interacting fragment (PIF) found in some PDK1 substrates (Fig. 1D). Both the activation loop motif and the PIF are signature sequences of the AGC family (Bogre et al., 2003;Mora et al., 2004).

Relationship with Other IRE and AGC Protein Kinase Genes
BLAST (Altschul et al., 1990) was used to search for MtIRE homologs in the Arabidopsis, rice (Oryza sativa), and tomato (Lycopersicon esculentum) genomes, as well as the unfinished Medicago genome (Young et al., 2005). In the Medicago genome, 20 other ESTs or TCs (www. tigr.org) were found that had homology to AGC protein kinases (Table I). Unfortunately, none of these represents a complete cDNA; many contain a recognizable full or partial Ser/Thr protein kinase domain. Of these, three were found by BLAST or ClustalW (Thompson et al., 1994) to be IRE-like ESTs or TCs that belong to the AGC Other subfamily of AGC genes (Bogre et al., 2003). The other 17 were more similar to different AGC protein kinases than to the IRE/AGC Other family (data not shown).
MtIRE, IRE homologs, and the Arabidopsis AGC kinases (Bogre et al., 2003) were subjected to phylogenetic analysis using ClustalW (Thompson et al., 1994) and the neighbor-joining method. The Medicago IRE genes were found to group as a single clade with the AtIRE, rice OsIRE, and tomato LeIRE genes (Fig. 2). This suggests that MtIRE, AtIRE, and other IRE-like genes diverged from the other AGC kinases before the monocots and dicots diverged, estimated to be about 170 million years ago (Sanderson et al., 2004). MtIRE was found to form a subclade with AtIRE within the IRE clade (Fig. 2), suggesting that these two genes are orthologous.
The N termini of MtIRE, AtIRE, and OsIRE are highly diverged (not shown), while the putative nuclear localization sequences, zinc finger, PIF motif, and Ser/ Thr protein kinase domain are conserved (Fig. 3). Within the Ser/Thr protein kinase domain, the activation loop signature motif is conserved among the IRE genes, including MtIRE. However, the putative PDK1 kinase target Ser residue of the activation loop was found to be missing in both AtIRE and OsIRE; it is MtIRE amino acid sequence. Dotted underline, Glu-rich region; boxed shaded type, basic nuclear localization signal; single underline, zinc finger; double underline, bipartite nuclear localization signal; bold type, Ser/ Thr protein kinase domain; boxed bold type, activation loop signature; asterisk, Ser residue in activation loop that is putative phosphorylation target; double dotted underline, PIF motif. present in MtIRE (Fig. 3D) and also present in the other IRE genes for which full sequence is available (not shown).

MtIRE Is a Single Copy Gene in Medicago
To experimentally determine the MtIRE copy number in M. truncatula, Southern-blot analysis was carried out. M. truncatula genomic DNA was restricted with several enzymes, blotted, and probed with a cDNA probes derived from exon 4 of MtIRE (Fig. 1). The results obtained at low stringency (not shown) are similar to those obtained at high stringency ( Fig. 4), showing one or two bands per lane. These results are consistent with MtIRE being a single copy gene in M. truncatula, and, with the exception of the pattern obtained with HincII enzyme, consistent with predicted in silico restriction pattern. In some cases, HincII cleavage is affected by DNA methylation, which may account for the inconsistency with the in silico prediction. Probing the blot with a different cDNA derived from the putative Ser/Thr kinase domain ( Fig. 1) yielded similar results of one or two bands per lane, also consistent with MtIRE being single copy (not shown).

MtIRE Expression Is in Nodules and Flowers
To examine MtIRE expression, semiquantitative reverse transcription (RT)-PCR was carried out on total RNA extracted from different plant tissues of M. truncatula A17 wild-type plants. RT-PCR primers were chosen to flank introns (Fig. 1). The transcript of the MtIRE gene was detected only in nodules and flowers and not in other organs of the plant, including uninoculated roots (Fig. 5A).
To determine when during nodule development MtIRE is first expressed, a time course of MtIRE expression in roots with developing nodules was carried out. The results show that MtIRE is first expressed at 4 d postinoculation (dpi) of nitrogen-starved roots with Sinorhizobium meliloti (Fig. 5B). This is well before the onset of nitrogen fixation in growth conditions used for these experiments, at about 8 dpi.

MtIRE Expression in Nodulating Roots of M. truncatula Nodulation Mutants
To further pin down the stage of nodule development with which MtIRE gene expression is associated, M. truncatula nodulation mutants were examined for MtIRE expression in nodulated root systems or root areas susceptible to nodulation 10 dpi after inoculation with S. meliloti, also by semiquantitative RT-PCR. As shown in Figure 6A, MtIRE was found to be expressed in both supernodulators tested, sunn (Penmetsa et al., 2003) and skl (Penmetsa and Cook, 1997). Mutants defective in Nod factor signal transduction, like dmi2 (Catoira et al., 2000), fail to express MtIRE. MtIRE gene expression was not detected in nodulated root systems of mutants that have defects in nodule invasion: lin (Kuppusamy et al., 2004), sli , nip Veereshlingam et al., 2004), and Mtsym1 (TE7; Benaben et al., 1995; Fig. 6A). In lin, sli, and nip nodules, almost all nodules contain rhizobia trapped within infection threads, whereas in Mtsym1 nodules, rhizobia are endocytosed into host cells and undergo limited replication but fail to elongate into bacteroids. M. truncatula dnf mutants, dnf1 to 7 Starker et al., 2006), deficient in nitrogen fixation, were examined in a similar fashion. In comparison to the mutants blocked very early in nodule development or during rhizobial invasion, all of the dnf mutants were found to express MtIRE (Fig.  6B). However, the level of MtIRE expression was found to be lower than in wild type in some of the dnf mutants. The dnf mutants were grouped into three classes, based on the extent of MtIRE expression (Fig.  6B). In the first class are the dnf1-1, dnf1-2, and dnf5 mutants that had significantly lower expression of MtIRE in their nodulated root systems. The second class contains dnf7, which has lower expression of MtIRE. The third class includes dnf2, dnf3, dnf4, and dnf6 mutants that have wild-type levels of MtIRE expression in their nodulated root systems.
To correlate MtIRE expression in the mutant nodules with phenotypes of the mutants, microscopic analyses of mutant nodules were carried out by examining X-Gal-stained 50-mm sections of nodules or whole roots of mutant plants inoculated with S. meliloti/hemATlacZ (Boivin et al., 1990), also at 10 dpi (Fig. 7). While similar analyses have been reported for some of the mutants used in this study, it was important to verify the phenotypes under the growth conditions employed in these experiments.
As shown in Figure 7A, for the skl mutant, more nodules than wild type, all staining so darkly blue they almost look black, were detected, as reported previously (Penmetsa and Cook, 1997). For dmi2, no invasion was seen; as reported previously, root hairs developed bulbous tips (Catoira et al., 2000;Fig. 7B). For the four mutants with defects in rhizobial invasion, lin, sli, nip, and Mtsym1, a gradient of rhizobial infection was observed, with lin infections confined to the root hair cell containing the initial infection thread, similar to published observations (Kuppusamy et al., 2004;Fig. 7C). In the sli mutant, infections progressed further, with infection threads penetrating the outer cortical cells and occasionally reaching inner cortical cells (Fig. 7D). The rare invaded nodules seen previously  at much later times postinoculation were not observed in our growth conditions at the 10-dpi time point. In the nip mutant, infection threads filled nodule primordia that showed signs of defense responses by accumulating brown pigments, as described previously Fig. 7E). For Mtsym1 (Benaben et al., 1995; Fig. 7F), nodule primordia filled with infection threads were readily observed, as were nodules with rhizobia released into plant cells. Brown pigments were observed, although the accumulation of these pigments was less than in the nip mutant and were distributed differently from those in nip, mostly around infection threads in Mtsym1. These results are similar to those previously described by some researchers (Benaben et al., 1995). The small, uninvaded Mtsym1 bumps seen by others (Benaben et al., 1995; were only rarely observed in our conditions. The dnf mutants' nodules host cells all showed evidence of released rhizobia around a central vacuole (Fig. 7, G-R). Nodules from the dnf1 and dnf5 mutants appeared to have the most serious defects. dnf1-1 and dnf1-2 nodules were characterized by large accumulations of brown-orange pigments in the most proximal zone developed in these nodules. To a lesser extent, these pigments also accumulated in the nodule parenchyma (Fig. 7, G and H). In the dnf5 mutant, the rhizobia occupying infected cells appeared to be sparser than in wild-type nodules, and some accumulation of pigments was observed (Fig. 7N). The dnf2 mutant nodules appeared to have lower rhizobial occupancy but were clearly invaded by rhizobia. Only a small pale-yellow band that may be polyphenolic accumulation near the invasion zone was noted (Fig. 7I). For the dnf3, dnf4, dnf6, and dnf7 mutants, at least two different phenotype nodules were observed in our growth conditions. One type appeared defective with lower rhizobial occupancy in the infected cells (Fig. 7, J, L, O, and Q), while the second type was hard to distinguish from wild type (Fig. 7, K, M, P, and R). For the case of dnf7 mutants, nodulated root systems displayed only a few of the wild-type phenotype nodules (as in Fig. 7R). The defective dnf7 nodules look quite abnormal, with rhizobia accumulating in a band toward the apex of the nodule (Fig. 7Q).
To look more closely at mutant nodules not expressing MtIRE versus those with low levels of MtIRE expression, higher magnifications were used to compare the phenotypes of the nip and Mtsym1 nodules to those of dnf1 and dnf5 (Fig. 8). Both nip and Mtsym1 nodules contain rhizobia mostly confined within infection threads (Fig. 8, A and B). At the subcellular level, both mutants have been shown by electron microscopy to have rhizobial release into host cells; in nip, this is a rare event, but it is frequent in Mtsym1 (Benaben et al., 1995;Veereshlingam et al., 2004). In contrast, the dnf1-1, dnf1-2, and dnf5 nodules have host plant cells with released rhizobia around a central vacuole, as in wild type. In the dnf1 mutants, rhizobia also accumulate as dark blue patches in the intercellular spaces (Fig. 8, C and D). In the dnf5 nodules, a lower rhizobial occupancy than wild type was observed with rhizobia accumulating in thinner rings around the central vacuole than they do in wild type (Fig. 8, compare E with F). Brown pigmentation indicative of polyphenolic accumulation is evident in the nip, dnf1, and dnf5 mutants (Fig. 8, B-E).

Spatial Localization of Nodule MtIRE Expression
To examine the spatial localization of MtIRE expression, an MtIRE promoter-GUS reporter construct (pMtIRE-gusA) was made and expressed in transgenic roots in composite M. truncatula wild-type plants. Plants were inoculated with rhizobia carrying a constitutive hemA gene, and nodulated roots were harvested at 15 dpi. Nodules formed asynchronously and more slowly in the composite plants than in untransformed roots, and nodules were found at various stages of development in the composite plants at 15 dpi in our growth conditions. Fixed tissue was stained with 5-bromo-4-chloro-3-indolyl-b-glucuronic acid (X-gluc) to visualize pMtIRE-gusA. The largest wildtype nodules were costained with Red-Gal to visualize the rhizobia. The results (Fig. 9A) show that blue staining indicative of MtIRE expression is confined to a narrow zone toward the apical end of the mature nodule, with red-staining rhizobia both apical and distal to the X-gluc-staining region. Transgenic nodules from composite plants were also stained with iodide to localize the starch, defining the interzone II-III region of the nodule . By comparing the results of iodide staining (Fig. 9B) with that for Red-Gal and X-gluc staining, it can be seen that pMtIRE-gusA expression localizes apically to the iodide staining, showing that MtIRE expression starts at the proximal end of the infection zone, zone II, but ends before the interzone II-III region. Costaining with X-gluc and iodide confirms that MtIRE expression is distal to the iodide staining region (data not shown), demonstrating the MtIRE expression is confined to the zone II invasion region of the nodule. As expected, and in confirmation of the semiquantitative RT-PCR expression results, no expression was detected in root hairs, even those containing infection threads (Fig. 9C).
Expression of the pMtIRE-gusA construct in composite dnf1-2 and dnf7 plants was evaluated to test the validity of using MtIRE as a marker for invaded nodule cell tissue. The results showed pMtIRE-gusA Note that OsIRE contains two bipartite nuclear localization signal sequences. C, Basic-type nuclear targeting sequence. Asterisks show basic Arg and Lys residues. D, PIF motif. Asterisks show the conserved amino acid residues of the motif. E, Ser/Thr kinase domain. Roman numerals denote the 12 distinct subdomains, as described in Hanks and Hunter (1995). White-boxed residues are conserved residues found in Ser/Thr kinase domains and the gray-boxed Ser residue in subdomain I is Ser in all three IRE sequences, but G in the conserved kinase family (Hanks and Hunter, 1995). The position of the activation loop is noted with the amino acids in the activation loop signature motif (Bogre et al., 2003) underlined. expression in both mutants as expected. dnf1-2 has smaller nodules than does dnf7 (and wild type) and pMtIRE-gusA staining was confined to the proximal end of the largest dnf1-2 nodules observed (Fig. 9D). For the largest observed dnf7 nodules, staining was observed in approximately the nodule middle (Fig.  9E), suggesting that dnf7 nodules are capable of further development than are those of dnf1-2.
A pictorial interpretation of the visualized staining pattern for wild-type nodules is presented in Figure  9F. pMtIRE-gusA is confined to the proximal side of zone II, and thus may be a marker for this developmental zone.

DISCUSSION
In 2002, Fedorova et al. reported nodule-specific TC sequences found in Medicago. Among these was a TC with high homology for the AtIRE gene, with a predicted signal transduction function (Fedorova et al., 2002). Because the AtIRE gene has a known role in root hair elongation and is thought to function in regulating the duration of tip growth (Oyama et al., 2002), we thought that the MtIRE gene might have a similar role in Medicago or one regulating infection thread growth, which can be viewed as inward apical growth, similar to root hair tip or pollen tube growth.
MtIRE is a member of the AGC family of Ser/Thr protein kinases involved in signal transduction. In many animal AGC kinases, activation depends on sequential phosphorylation at two sites, one within the PIF motif, found C terminal to the kinase domain, and one within the activation loop of the kinase domain (Mora et al., 2004). Although well studied in animals, the signaling pathways regulated by AGC kinases in plants are not yet well understood. In Arabidopsis, seven of at least 39 AGC kinases have been partially characterized as having important roles in development (Huala et al., 1997;Christensen et al., 2000;Briggs and Christie, 2002;Oyama et al., 2002;Bogre et al., 2003;Anthony et al., 2004Anthony et al., , 2006Takemiya et al., 2005;Devarenne et al., 2006;Zegzouti et al., 2006), and a number of other AGC kinases are under study (Zegzouti et al., 2006). MtIRE is the first AGC kinase studied in legumes.
The large plant AGC kinase family is subdivided into distinct phylogenetic classes (Bogre et al., 2003), with the IRE genes, including MtIRE, clustering in a single clade (Fig. 2). Of these genes, AtIRE and AtIRE H1 have been previously studied (Oyama et al., 2002). In addition to sequence motifs characteristic of AGC kinases, MtIRE encodes a putative zinc finger-like sequence and nuclear localization sequences (Fig. 3). This suggests that MtIRE protein, like some of the other IRE genes, could localize to the nucleus, similar to the NDR proteins of the AGC family that also contain nuclear localization signals (Tamaskovic et al., 2003). Unique to this IRE-like gene, MtIRE encodes a Glu-rich sequence near its N terminus. Other Glu-rich proteins interact with Ca 21 (Endo et al., 2004;Jo et al., 2004), and the Glu-rich region of MtIRE could have a  (2), no RNA. Below the RT-PCR results is an ethidium bromide-stained agarose gel with 1 mg of total RNA from each sample used for the RT-PCR. B, Total RNA from nodulating roots was analyzed at the indicated times after inoculation of roots with S. meliloti. M, (2), as in A. Below the RT-PCR results is a gel with 1 mg of total RNA from each sample used for the RT-PCR. similar function. Although Ca 21 signaling has a welldocumented role in the Nod factor signaling between rhizobia and legume roots (Levy et al., 2004;Gleason et al., 2006;Tirichine et al., 2006), only a few studies have investigated the role of Ca 21 in intermediate and later stages of nodule development. In Medicago, the Ca 21 /calmodulin-dependent protein kinase DMI3 was recently shown to have a role in infection during rhizobial release from infection threads into host cells (Godfroy et al., 2006). In determinate soybean (Glycine max) nodules, two calmodulin genes were found to be expressed in the infection zone and proposed to be essential for Bradyrhizobium release into symbiosomes (Son et al., 2003). In Medicago, calmodulin transcripts are expressed in root nodules (Fedorova et al., 2002), and novel Ca 21 -binding proteins are found in the symbiosome space (Liu et al., 2006). Ca 21 has been found to modulate symbiosome membrane intake and efflux (Udvardi and Day, 1997).
Expression studies in developing nodules showed MtIRE expression is first detectable at 4 dpi in our growth conditions (Fig. 4). This time coincides with the time when symbiosomes are beginning to develop and is long before the onset of nitrogen fixation, at about 8 dpi in our conditions.
To correlate MtIRE expression with the phenotypes formed by nodulation mutants, it was necessary to define the mutated phenotypes in our growth conditions. Our findings show that at 10 dpi, sli nodules are blocked during invasion of the nodule primordia by rhizobia, different from the rare invaded nodules that are observed at a later time point . The dnf mutant nodules are blocked after the rhizobia have invaded the nodules through infection threads and deposited rhizobia within host cells (Figs. 7 and 8). The extent of dnf mutant nodule development that we observed correlates well with previous plant and bacterial gene expression studies in the dnf mutants Starker et al., 2006).
Comparing MtIRE expression in mutant nodulating root systems (Fig. 6) with phenotypes of the mutant nodules ( Figs. 7 and 8) shows that MtIRE expression is associated only with nodules that are able to achieve successful invasion, release of rhizobia into infected cells, and some development of the resulting symbiosomes, i.e. the dnf mutants. MtIRE expression was not observed in the sli or nip mutants that have little to no rhizobial release from infection threads Veereshlingam et al., 2004). Nor was MtIRE expression found in Mtsym1 (TE7) mutants that have rhizobial release from infection threads into symbiosomes but no obvious replication of rhizobia inside symbiosomes or elongation of rhizobia into bacteroids (Benaben et al., 1995). The finding that MtIRE expression is lower in dnf1, dnf5, and dnf7 mutants may point to defects in these mutants that affect the physiology of infected nodule cell maturation in zone II.
In attempts to discern a function for MtIRE, we and others (S. Gantt, personal communication) have attempted MtIRE RNAi-induced gene silencing. The results have been inconsistent, with phenotypes that varied from root hair defects to defective nodulation to no detectable effect (data not shown). We have tried three different silencing constructs, with two chosen from MtIRE regions that are apparently unique to MtIRE and one from the putative Ser/Thr kinase domain. The lack of consistent results from any RNAi construct tested suggests possible functional redundancy in one or more of the other Medicago IRE-like genes. We note that another group has also reported similar issues with the RNAi technique (Liu et al., 2006).
Localization experiments using pMtIRE-gusA in transformed hairy root composite plants showed that MtIRE expression localizes to a band in the proximal part of zone II, the infection zone. To date, there are very few details about the biology or biochemistry of the symbiotic interaction in this nodule region. In this area of zone II, rhizobia are already released from infection threads, symbiosomes are starting to develop and the host cells are expanding, but the rhizobia are not capable of nitrogen fixation . Localization of pMtIRE-gusA expression in dnf1-2 and dnf7 plants showed that while nodules from both mutants express pMtIRE-gusA, the dnf1-2 plants appeared to halt nodule development after the zone where MtIRE is expressed, while the dnf7 plants progressed further in nodulation. Thus, MtIRE may Figure 6. MtIRE gene expression in nodulation mutants. A, Total RNA from 10-dpi nodulated root systems of skl, sunn, dmi2, sli, lin, nip, Mtsym1 (TE7), and wild-type (A17) plants was analyzed by semiquantitative RT-PCR in the presence of primers specific for MtIRE and for rRNA as positive control for RNA in the RT-PCR reaction. M, Markers whose sizes are on the left. Below the RT-PCR results is an ethidiumstained agarose gel with 1 mg of total RNA from each sample used for the RT-PCR. B, Total RNA from 10-dpi nodulated root systems of dnf1-1, dnf1-2, dnf2, dnf3, dnf4, dnf5, dnf6, dnf7, and wild-type (A17) plants was analyzed as in A. Below the RT-PCR results is a gel with 1 mg of total RNA from each sample used for the RT-PCR. be a good marker for the ability of nodules to progress to proximal zone II development.
MtIRE was found to be a single copy gene in M. truncatula and a member of the IRE clade of the AGC kinase family. The phylogenetic analysis suggests that MtIRE may be the ortholog of the AtIRE gene. We performed the phylogenetic analysis with the avail-able, but partial, cDNA sequence for the three other Medicago IRE genes. When the full sequences of these cDNAs are available, it may turn out that one of them is a closer homolog to AtIRE than is MtIRE.
Our results demonstrated MtIRE expression in nodules and flowers (Fig. 3), similar to a number of other nodulins (Allison et al., 1993;Crespi et al., 1994;Figure 7. Phenotypes of nodules from nodulation mutants. Plant roots from the indicated mutants or wild type were grown in the presence of S. meliloti/pXLDG4 containing a constitutive hemATlacZ construct and stained at 10 dpi with X-Gal. A, skl; B, dmi2; C, lin; D, sli; E, nip; F, Mtsym1 (TE7); G, dnf1-1; H, dnf1-2; I, dnf2; J, dnf3; K, dnf3 (similar to wild type); L, dnf4; M, dnf4 (similar to wild type); N, dnf5; O, dnf6; P, dnf6 (similar to wild type); Q, dnf7; R, dnf7 (similar to wild type); S, A17 (wild type). Bars 5 100 mm. A to D, Whole mounts; E to S, 50-mm sections. Zucchero et al., 2001;Rodriguez-Llorente et al., 2004). The MtIRE expression pattern has diverged from the orthologous AtIRE gene that is expressed in every organ of the wild-type plant, with higher expression in roots (Oyama et al., 2002), suggesting that MtIRE has been recruited to nodule development from another role. Several previous studies have highlighted recruitment of genes from other plant programs to symbiotic processes as an evolutionary mechanism of the nitrogen-fixing symbiosis (Szczyglowski and Amyot, 2003;Rodriguez-Llorente et al., 2004;Liu et al., 2006). These include the ENOD genes, genes that transduce signals from both the arbuscular mycorrhizal and rhizobial symbioses, genes that control nodule number, and genes for leghemoglobin, all of which appear to have been derived from other functions in plant ancestors that gave rise to legumes. AtIRE regulates the duration of root hair cell expansion in Arabidopsis (Oyama et al., 2002). Our data show that the MtIRE gene has unique expression during nodule development in zone II, where the expansion and development of host cells and symbiosomes take place. Because of this, we propose that the recruited MtIRE gene has a role in these processes and, because it apparently encodes an AGC kinase, we propose that this role is in signal transduction.

Histochemical Staining of Wild Type and Nodulation Mutants
Nodules elicited with S. meliloti/pXLDG4 containing the constitutive hemAT lacZ gene were stained with X-Gal, as described previously , by using standard methods (Boivin et al., 1990). The buffer used in our protocols was 80 mM PIPES, pH 7.2, instead of cacodylate. For sectioning, nodules were embedded in 5% low-melting point agarose and 50-mm-thick sections were obtained with a 1000 Plus Vibratome (Vibratome). Sections were observed with an Olympus BX50 microscope using bright field or dark field (for dmi2 roots). Digital micrographs were processed using Adobe Photoshop.
MtIRE Promoter-GUS Reporter Construct and M. truncatula Root Transformation pRD022 was created by subcloning the SphI fragment from pCAMBIA2301 (Hajdukiewicz et al., 1994) containing the cauliflower mosaic virus promoter-GUS coding region into pUC18. A large region upstream (3,458 bp) of the MtIRE Figure 8. Higher magnification of phenotypes of nodulation mutants' nodules. Plant roots from the indicated mutants or wild type were grown in the presence of S. meliloti/pXLDG4 containing a constitutive hemATlacZ construct and stained at 10 dpi with X-Gal. A, Mtsym1 (TE7); B, nip; C, dnf1-1; D, dnf1-2; E, dnf5; F, A17 (wild type). Bars 5 100 mm. All are 50-mm sections.
gene was amplified from the mth2-13b8 BAC clone (obtained from the Clemson Stock Center, www.genome.clemson.edu). The forward primer MtIREp F has an EcoRI site at its 5# end, 5#-TGGAATTCCTGCATGGCGCGAGCAAAATGT, and the reverse primer MtIREp R has an NcoI site at its 5# end, 5#-ACCATGGTGA-GAGATGAAAGGAAGAGAG. After PCR amplification (94°C 2 min, followed by 35 cycles of 94°C 30 s, 60°C 30 s, 72°C 3 min), the resulting product was digested with EcoRI and NcoI and ligated to EcoRI, NcoI-digested pRD022, replacing the cauliflower mosaic virus promoter with the MtIRE promoter. This plasmid was then digested with EcoRI and BstEII and ligated with EcoRI, BstEIIdigested pCAMBIA2301 to create pCIP005. pCIP005 was transformed into Agrobacterium rhizogenes strain ARqua1 (Quandt et al., 1993) using the freezethaw method (Hofgen and Willmitzer, 1988). M. truncatula roots were transformed A. rhizogenes strains  containing pCIP005 or a positive control containing the pENOD11-gusA construct .
Histochemical Staining of Root Nodules on Transformed Roots Figure 9. Localization of pMtIRE-gusA expression in wild-type and dnf1-2 and dnf7 mutants. Composite M. truncatula plants with transgenic roots were grown in the presence of S. meliloti containing a constitutive lacZ gene. A, Double staining for gusA with X-gluc (blue), for the localization of MtIRE promoter activity, and lacZ with Red-Gal (red), for the rhizobia localization. The arrow points to the X-gluc staining. Bar 5 100 mm. B, Staining with iodide reveals the position of interzone II-III. Bar 5 100 mm. C, Root hair with an infection thread stained with X-gluc and Red-Gal shows rhizobia staining but no pMtIRE-gusA staining. Bar 5 20 mm. D, dnf1-2 nodule stained with X-gluc only shows pMtIRE-gusA staining in proximal part of nodule. Bar 5 100 mm. E. dnf7 nodule stained with X-gluc only shows pMtIRE-gusA staining in middle of nodule. Bar 5 100 mm. F, Interpretative diagram of A and B showing the position of MtIRE expression in the proximal region of the invasion zone, zone II.