LIN, a novel type of U-box/WD40 protein, controls early infection by rhizobia in legumes

The formation of a nitrogen fixing nodule requires the coordinated development of rhizobial colonization and nodule organogenesis. Based on its mutant phenotype LIN functions at an early stage of the rhizobial symbiotic process, required for both infection thread growth in root hair cells and the further development of nodule primordia. We show that spontaneous nodulation activated by the calcium- and calmodulin-dependent protein kinase (CCaMK) is independent of LIN , thus LIN is not necessary for nodule organogenesis. From this we infer that LIN predominantly functions during rhizobial colonization and the abortion of this process in lin mutants leads to a suppression of nodule development. Here, we identify the LIN gene in Medicago truncatula and Lotus japonicus , showing it codes for a predicted E3 ubiquitin ligase containing a highly conserved U-box and WD40 repeat domains. Ubiquitin-mediated protein degradation is a universal mechanism to regulate many biological processes by eliminating rate-limiting enzymes and key components such as transcription factors. We propose that LIN is a regulator of the component(s) of the Nod factor signal transduction pathway and its function is required for correct temporal and spatial activity of the target protein(s).


INTRODUCTION
The soil bacteria rhizobia are able to establish nitrogen-fixing symbioses with leguminous plants by inducing the formation of a new organ, the nodule, on the roots of the host plant.
Symbiotic infection is initiated and maintained by an exchange of signaling molecules between the host plant and the microsymbionts. In most legumes, flavonoid compounds produced by leguminous plants activate bacterial regulators to induce nodulation (nod) genes required for the synthesis of lipochitooligosaccharide Nod factors. Nod factors initiate multiple early responses on plant hosts including a burst of intracellular calcium levels in root hairs, calcium oscillations (calcium spiking) and the induction of nodulation specific genes whose products are referred to as 'nodulins'. In addition, they induce a rearrangement of the root hair cytoskeleton leading to root hair deformation and curling, which traps surface-attached rhizobia, establishing a site that acts as an infection focus. Bacteria penetrate into the curled root hairs towards the root cortex via host-derived tubular structures, called infection threads. Simultaneously, Nod factors stimulate the reinitiation of mitosis in cortical cells, leading to the formation of nodule primordia, which give rise to the cells that receive the invading bacteria (Oldroyd and Downie 2008). Exopolysaccharides (EPS), capsular polysaccharides (K antigens or KPS), and lipopolysaccharides (LPS) serve as further bacterial signals required for successful infection of nodules in many legumes (Jones et al., 2007). The infecting rhizobia are released into intracellular membrane compartments of plant origin called symbiosomes where they differentiate into bacteroids capable of reducing atmospheric nitrogen to ammonium, which is provided to the plants in exchange for carbon and amino acid compounds (Brewin, 2004).

Analysis of nodulation-defective Medicago truncatula, Lotus japonicus as well as
Pisum sativum mutants has led to insights into the mechanisms by which Nod factors are perceived and trigger subsequent signal transduction cascades (Geurts et al., 2005;Stacey et al., 2006;Oldroyd and Downie, 2008). In M. truncatula and L. japonicus, the Nod factor signal is probably sensed by LysM-type receptor kinases such as NFP/NFR5 (Amor, et al., such as NSP1 (Smit et al., 2005;Heckmann et al., 2006), NSP2 (Kalo et al., 2005;Heckmann et al., 2006) and ERN1 (Middleton et al., 2007) responsible for the transcriptional changes that are required for the initiation of nodule morphogenesis. In addition to the signaling pathway outlined above, legumes have many genes required to enable rhizobia to infect the roots. Only a few of these genes have been identified; the RPG gene required for infection was identified but the function of its product has yet to be defined (Arrighi et al., 2008). Some genes such as NIN and ERN encode predicted regulators that may affect both the Nod-factor signaling and infection pathways (Schauser et al., 1999, Marsh et al., 2007, Middleton et al., 2007, and the hcl mutation (in MtLYK3 gene) affecting a predicted Nod-factor receptor also causes an infection defect (Smit et al., 2007). The NAP and PIR genes are required for normal infection and their loss affects polar growth of some cell types and probably are required for the polar growth of the infection thread (Yokota et al., 2009). The NAP and PIR proteins have been identified as components of the SCAR/WAVE complex, which is involved in polymerization of actin-related proteins (Li et al., 2004).
The lin mutant (C88) was identified as a M. truncatula mutant, in which infection was arrested and reduced to a quarter of the frequency seen in wild-type plants (Penmetsa and Cook, 2000;Kuppusamy et al., 2004). In those cases when infections were initiated, bacteria were arrested after very limited progression within root hairs. Although nodule primordia were formed in the root cortex, differentiation was arrested at an early stage.
Interestingly, infection and initiation of cortical cell division recurred continually in the presence of rhizobia, suggesting that lin is required for appropriate regulation of nodule initiation, and for control of nodule number (Kuppusamy et al., 2004). The early nodulins (RIP1, ENOD20 and ENOD40) were expressed in lin at comparable levels to wild-type, while late nodulins such as MtN6, ENOD2 and ENOD8 failed to be induced in lin. This suggests that LIN acts downstream of the early Nod factor signal transduction pathway, but is required for infection initiation and persistence and nodule differentiation (Kuppusamy et al., 2004). Based on the phenotype of the mutant the LIN function is probably needed either in the process of nodule development/maturation -with indirect effect on the block of infection thread growth or in the invasion process resulting in the interruption of nodule development. It is also possible that LIN acts at both levels as a synchronizing regulator between the parallel processes in the root epidermis and the cortex.
In this work we show the identification of the LIN gene revealing that LIN encodes a large protein containing multiple domains including a U-box (modified RING) domain, indicating a function as an E3 ubiquitin ligase. Spatiotemporal analysis of the promoter activity of LIN demonstrated that the expression of the gene correlated with the early nodule primordia formation and bacterial invasion during the symbiosis. We used a gain-offunction mutation in CCaMK that induces spontaneous nodulation in the absence of rhizobia to show that LIN is not required for nodule organogenesis. This indicates that LIN functions exclusively during rhizobial colonization and the defect in nodule development in lin is a response to aborted infection.

LIN is not required for nodule organogenesis
EMS and fast neutron mutagenised M. truncatula populations were screened to identify new loci required for nodulation; two of the mutants identified had phenotypes similar to lin, namely impaired nodulation and infection with the infection threads arrested in the root hair cells. Though nodule primordia emerged three to four days after inoculating the roots with wild type rhizobia, nodule development was always blocked before the stage of nodule differentiation. No non-symbiotic phenotype was detected in these mutants. Genetic crosses revealed that the two mutants (EMS6:T7 and 14P) carried mutations allelic to lin-1 (Table I), but not allelic to the other nodulation mutations tested (data not shown). EMS6:T7 was derived from EMS mutagenesis and will be defined as lin-2, while 14P was derived from fast neutron mutagenesis and will be defined as lin-3.
To determine if LIN is required for nodule organogenesis we tested whether an autoactive form of CCaMK could induce spontaneous nodulation in lin mutants. In M. truncatula, transgenic expression of an autoactive CCaMK (DMI3 1-311 ), comprised of the kinase domain alone, is capable of inducing spontaneous nodules and has been useful in determining the order of gene product function within the early NF signaling pathway relative to CCaMK (Gleason et al., 2006). When we transformed lin-1, lin-2 and lin-3 mutants with this autoactive CCaMK construct, we could clearly detect nodules on the lin-1 (3/21 plants), lin-2 (9/19 plants) and lin-3 (40/103 plants) mutants. The observation that 52 nodules were induced on 143 lin mutant plants transformed with autoactive CCaMK suggests that LIN is not essential for nodule organogenesis and therefore the defects in nodule development observed in the lin mutants during rhizobial invasion is likely to be a result of the abortion of bacterial infection.

Positional cloning of LIN in M. truncatula
Preliminary mapping data previously positioned lin-1 (C88) between molecular markers DSI and SCP on the lower arm of linkage group 1 (LG1) within a relatively large distance of several cM (Kuppusamy et al., 2004). Using newly generated markers and a new segregating population of 290 F2 individuals and 508 F3 individuals of selected F2 plants, we mapped LIN within the region on LG1 between markers 4E6R and e53J20F ( Fig. 1. A). This region is spanned by the sequenced BACs: mth2-69D21, mte1-13O17 and mte1-53J20 (Fig 1. A, B). Within this region there were numerous good candidates for LIN among the predicted genes including genes encoding AP2 transcription factors, a bZIP transcription factor, a protein kinase, a putative RING zinc finger protein and genes with unknown function, but represented with nodule expressed sequence tags (ESTs) in the Medicago database. No mutation was detected in lin-1 for any of these candidate genes.
Further sequencing of this region in lin-1 and comparisons with the sequence from wild-type revealed a single nucleotide difference in lin-1 in the predicted coding region MTCON310-47 (indicated by the star in Fig. 1B). Since no EST had been reported for this gene, we validated that this region was actively transcribed in M. truncatula. PCR amplification products could easily be obtained from plant cDNA isolated from roots four days after S. meliloti inoculation. The cDNA of MTCON 310-47 could be assembled from the sequences of overlapping fragments. The intron/exon boundaries were similar to what had previously been predicted for the genomic sequence, with the exception that exon 12 was found to be 24 bp longer (indicated by an open triangle in Fig. 1C) resulting in a predicted total protein of 1488 amino acids. The mutation we identified in lin-1 was in the last nucleotide position of intron 4 (indicated by an arrow in Fig. 1C). RT-PCR amplification of the affected region using primers Lin-3F (in exon 3) and Lin-3R (in exon 5) on lin-1 RNA samples of inoculated roots revealed that indeed there was an error in the RNA splicing resulting in a longer transcript ( Fig. 2A). Sequencing of this longer fragment showed that due to the mutation intron 4 was not spliced out during mRNA processing (while correct splicing of intron 3 was detected in the same fragment), causing a premature stop codon to be introduced. Amplification of all other parts of the lin-1 cDNA resulted in the expected fragment sizes (e.g. the fragment amplified by Lin-8R and Lin-8R primers in Sequencing of the genomic DNA amplified from the lin-2 mutant revealed a point mutation in the first exon that appears at nucleotide position 662 in the cDNA sequence, introducing a premature stop codon into MTCON 310-47 very early in this gene (indicated by an arrow in Fig. 1C). In addition, Southern analysis of lin-3 revealed an apparent large deletion or rearrangement in this same gene (Fig. 2B). RT-PCR experiments on lin-3 mutant demonstrated the presence of a short mRNA (1568 nucleotide) transcribed from this allele (indicated by an arrow in Fig. 1C). Another mutant line, C105 originating from the same population (Penmetsa and Cook, 2000) and having a similar phenotype to C88 (lin-1) is allelic with C88 (R. Dickstein, personal communication); sequencing of MTCON 310-47 in C105 revealed the same mutation, suggesting that C88 and C105 are siblings. The identification of mutation events in three different alleles of lin provides strong evidence that MTCON 310-47 is indeed LIN. Normal looking pink nodules could be observed in the complemented plants indicating that they were functional; nodulation, scored only using those transgenic roots showing GFP fluorescence, was significantly different from the controls lacking MtLIN (Table II)  There was no detectable signal of GUS activity throughout the most parts of the uninoculated roots, except a very faint signal just above the detection limit at the apical region (Fig. 6A). Three days after inoculation with rhizobia, GUS activity seemd to be associated with dividing cortical cells leading to the formation of nodule primordia (Fig. 6B).

Complementation of the Medicago lin mutants on transgenic hairy roots
Six days after inoculation strong overall GUS staining was detected in the young, emerging nodules, where infection of plant cells by rhizobia takes place (Fig. 6C). In elongated mature nodules (21 days after inoculation), strong GUS activity was detected but was mainly restricted to a relatively broad area of nodule apices including the infection zone. Much lower expression was detected in the nitrogen-fixing zone (Fig. 6D). and occupation, and further regulatory elements in a larger promoter segment in the 5' upstream region of LIN would be responsible for its expression and function at these later stages.  (Pickart, 2001;Frugis and Chua, 2002;Vierstra, 2003;Smalle and Vierstra, 2004). In higher plants, a large gene family composed of diverse isoforms encodes the E3 ubiquitin ligases. In the Arabidopsis genome there are twelve hundred genes encoding the E3 components, whilst forty-one genes encode E2 components and only two genes encode the E1 components (Vierstra, 2003;Kraft et al., 2005). Therefore it is thought that the E3 ubiquitin ligases must play a central role in selecting the appropriate candidate proteins during the ubiquitination process (Zeng et al., 2006 Recently, nsRING, a novel RING finger protein required for rhizobial invasion and nodule formation was identified in L. japonicus (Shimomura et al., 2006). The possible involvement of nsRING in phytohormone-related signaling was suggested. In M. truncatula, it has been shown that the SINA family of E3 ligases is important for infection thread growth and symbiosome differentiation (Den Herder et al., 2008). Overexpressing SINAT5DN from

DISCUSSION
Arabidopsis caused a significant reduction in nodule number, indicating a negative regulatory role for this protein. Based on the mutant phenotype we can assume that LIN negatively regulates certain transcription factors and/or signal transduction proteins required for bacterial invasion, and which need fine tuned regulation for their proper action. Good candidates were identified in a suppression subtractive hybridization approach that revealed candidate transcription factors: a bHLH, a WRKY and a C2H2 zinc-finger protein, that were upregulated, following S. meliloti inoculation in lin (Godiard et al., 2007). Understanding of LIN targets should provide insights into the maintenance of infection thread growth and may also provide insights into the mechanisms by which infection thread growth and nodule organogenesis are coordinated.

Plant growth and bacterial strains
Medicago truncatula cv Jemalong genotype A17 was used as the wild-type control for phenotypic and genotypic analysis. The plants were grown as previously described by Cook the transgenic roots. Plants harboring 4-5 cm long hairy roots (approx. two weeks after A. rhizogenes treatment) were transferred into pots and let grow and strengthen for another 3-4 weeks in Turface. One week before the date of inoculation by rhizobia nitrogen was omitted from the liquid media. Plants were inoculated afterwards with S. meliloti 1021 strain carrying a hemA promoter::lacZ fusion (Leong et al., 1985) allowing the detection of bacteria within plant tissues by histochemical staining for lacZ activity. Roots were screened and scored for the presence of nodules or nodule primordia 21 days after inoculation.
Transgenic hairy roots expressing GFP constitutively from the inserted T-DNA were selected and monitored for nodulation and for nodule occupation by S. meliloti.

Histochemical localization of the LIN promoter activity
To generate pLIN-GUS fusion construct, a 1.2 kb segment upstream the coding region of

Supplemental Data
The following materials are available in the online version of this article.          DNA fragments with primers Lin-3F and Lin-3R include intron 4 that is spliced out from the wild type allele but not spliced from the mutant. Amplification with primers Lin-8F and Lin-8R resulted in identical fragments representing exons 11-14 on the same cDNA templates. (B) Southern blot using EcoRV-digested genomic DNA of the wild type (wt) and lin-3 mutant plants (as well as nsp2-4 mutant included as another control) and a probe of 500 bp from exon 5 amplified with primers Lin-3Fb and Lin-12Rc revealed an apparent deletion or rearrangement affecting the large part of the gene.