LysM-RLK LYRW (not named in Zhang et al, 2009)
The phylogenetic group LYRW contains only members in legume species and in peach (Fig. 4 and S2), suggesting that this group emerged in a common ancestor to this phylogenetically related plant species. Two members of phylogenetic group LYRW were previously found in M. truncatula and were reported in L. japonicus (Lohmann et al., 2010) although only LjLYS20 (Lj1g3v2808030) can be found in the current version of L. japonicus genome. Although the members of this phylogenetic group cluster with the other LYRs rather than with the LYKs used as outgroup, they are very different from the other LYRs. Actually, their kinase domains are closely related to RLKs from the WAK subfamily. They might represent a third independent apparition of LysM-RLKs through fusion of 3 LysM with another kinase domain than in the LYK or LYR group. For this reason, we named this group LYRW. Biological role and biochemical functions are currently unknown for these proteins.
LysM-RLK LYKI (LYK clade VI)
In the phylogenetic group LYKI the number of genes is highly variable between species (Fig. 6 and S3). Legumes display the highest number and diversity of members in this phylogenetic group (9 in M. truncatula, 5 in L. japonicus) whereas we only found 1 protein in Brassicaceae and in B. distachyon.
Fig. 6 PhyML phylogenetic tree of the LYK. Different phylogenetic groups are shown in different colors. Three LYR proteins were used as outgroup sequences.
This phylogenetic group contains AtLYK1/AtCERK1 (At3g21630) that has been widely studied and shown to be required for chitin responses. CO8-induced responses such as ROS production and MAP kinase phosphorylation were completely impaired in Atcerk1 knock-out mutants (Miya et al., 2007, Wan et al., 2008). Moreover, the mutant was more susceptible to both fungal and bacterial pathogens. In addition, pre-treatment with crab shell chitin or purified CO8 decreased pathogen growth in the WT plants whereas this resistance was not induced in Atcerk1 reinforcing the role of AtCERK1 in chitin perception (Miya et al., 2007, Wan et al., 2008).
Contrasted results have been obtained concerning the affinity of AtCERK1 for chitin and COs. High affinity for chitin (Kd of 2 nM) has been reported (Lizasa et al., 2009) for the full length protein fused to GFP, produced in yeast, solubilized and purified to measure binding on chitin beads using a range of protein concentration. However, in the same study competition assays on the chitin beads using CO5, CO6 or CO8 led to half-maximal inhibitory concentration (IC50) of about 100 µM. Similarly, other studies showed a low affinity binding (Kd of 44 µM and 455 µM for CO8) using ITC with purified AtCERK1 ECR produced in insect cells and in E. coli respectively (Liu et al., 2012b, Cao et al., 2014). The huge differences between affinity reported for chitin and COs could be due to the methods used for affinity determination (quantification of AtCERK1:GFP fluorescence bound to chitin beads and ITC) or to differences in affinity for various degree of GlcNAc polymerization (chitin vs CO8).
Although AtCERK1 has been mainly studied for its role in chitin perception, it was originally showed that Atcerk1 knock-out line was more sensitive to the pathogenic bacteria P. syringae (Miya et al., 2007, Wan et al., 2008). In addition, AtCERK1 was shown to be the target of the bacterial effector AvrPtoB, which is able to ubiquitinate and induce its degradation (Gimenez-Ibanez et al., 2009), suggesting that AtCERK1 is also involved in perception of bacterial MAMPs. Indeed, AtCERK1 has been shown to be involved in PGN perception (Willmann et al., 2011) although it does not appear to directly bind PGN (Petutschnig et al., 2010, Willmann et al., 2011).
OsCERK1 (Os08g42580) is also involved in chitin and PGN signaling (Shimizu et al., 2010, Ao et al., 2014). Similarly to what has been observed for the Atcerk1 mutant, Oscerk1 mutant displayed altered responses to chitin, soluble CO7-8 or PGN treatment (ROS production, apoplastic alkalinisation, genes regulation and callose deposition). As expected for a chitin-perception defective mutant, Oscerk1 mutant is more susceptible to the fungal pathogen M.
oryzae (Kouzai et al., 2014a). However, unlike AtCERK1, OsCERK1 did not show any binding to insoluble colloidal chitin (Shinya et al., 2012). These functional differences between AtCERK1 and OsCERK1 were supported by the fact that OsCERK1 could not complement Atcerk1 for CO8-induced ROS responses (Shinya et al., 2012). However, chimerical constructs with AtCERK1 ECR and OsCERK1 ICR were able to partially rescue ROS production, demonstrating that AtCERK1 binding properties are necessary for ROS response in Arabidopsis. Moreover, it has been recently published that OsCERK1 is required for CO4 and CO5 perception as these molecules were unable to induce Ca2+ responses in Oscerk1 whereas Oscebip and Osnfr5 were still displaying calcium spiking. Unexpectedly, Osnfr5 is normally colonized by Rhizophagus irregularis (Miyata et al., 2016), even if OsNFR5 was able to complement Ljnfr5 mutant for nodulation, indicating that probably the function is well conserved, at least concerning the ICR.
Interestingly, rice plants with decreased level of OsCERK1 showed almost no fungal penetration at 6 weeks post inoculation (wpi) with AMF (Zhang et al., 2015). Similarly an Oscerk1 knock-out line displayed a mycorrhizal phenotype (Miyata et al., 2014) with no root colonization at 15 days post inoculation (dpi), demonstrating a role of OsCERK1 in early fungal colonization. Some penetration sites and arbuscules were observed at 30 dpi. This suggests that OsCERK1 is involved in perception or signaling of signals produced by AMF such as LCOs or short COs. OsCERK1 is therefore involved in a signaling pathway activated by various molecules (at least COs and PGN, and perhaps LCOs) leading to different biological responses such as defense against pathogens and symbiosis establishment.
In our and in previously published phylogenetic analyses (Fig. 6 and Shimizu 2010), it appears that there is another member of the phylogenetic group LYKI in rice, OsRLK10 (Os09g33630) which is an OsCERK1 paralog. This gene is not present in the monocotyledon B. distachyon. This protein does not contain the YAQ/R motif in the kinase (but rather only the AA sequence AR) which has been shown to be a hallmark of symbiotic function (Miyata et al., 2014), see below. It would be interesting to determine whether OsRLK10 can be functionally redundant with OsCERK1 for one or both of the OsCERK1 functions. This would be particularly important considering that AMS establishment appears not completely abolished in Oscerk1 knock-out mutant (Miyata et al., 2014).
Recently, Bozsoki et al. (2017) showed that the two orthologs in legumes LjLYS6 (Lj6g3v1055580) and MtLYK9 (Medtr3g080050) are involved in defense. Ljlys6 and Mtlyr9 knock-out mutants were impaired in defense. Ljlys6 and Mtlyr9 were more susceptible than WT plants to the fungal pathogen B. cinerea. Moreover, responses to a range of COs from CO4 to CO8 such as ROS production or MAPK3/6 phosphorylation were decreased compared to WT. To determine LjLYS6 affinity for COs, LjLYS6 ECR was produced in insect cells. After purification, it was deglycosylated and labelled with a fluorophore. Affinity for CO8, CO7, CO6 and CO5 was measured by MST using a range of CO concentration. LjLYS6 was found to have a lower affinity for COs with shorter chains, with a Kd of 38 µM for CO8, 227 µM for CO5 and no detectable binding of CO4. This affinity for CO8 is comparable to the affinity found for AtCERK1 ECR produced in insect cells and measured with CO8 by ITC (Liu et al., 2012b). Bozsoki et al. (2017) also made the crystal structure of LjLYS6 ECR and they found similar results that for AtCERK1 ECR (Liu et al., 2012b) although they could not observe LjLYS6 ECR bound to a CO.
In the phylogenetic group LYKI, there are two orthologous genes LjNFR1 (Lj2g3v2904690) and MtLYK3 (Medtr5g086130) that originates from a duplication event specific to legumes (De Mita et al., 2014). Both are involved in LCO (Nod-factor) perception in the RNS. Ljnfr1 mutants were impaired in nodulation and in the earliest responses to LCOs (Radutoiu et al., 2003). Apoplast alkalinisation, occurring immediately after LCO application and later responses such as root hair deformation (few hours after LCO application), were not observed in Ljnfr1 mutant lines. Whether LjNFR1 also plays a role in AMS is a matter of debate. It has been shown that Ljnfr1 mutant line displayed a lower colonization ratio compared to the WT 5 wpi (Zhang et al., 2015). Moreover, AMS marker genes and Myc-LCO-induced calcium spiking were reduced in an Ljnfr1 mutant line compared to the WT (Zhang et al., 2015). In contrast, no difference in colonization ratio or fungal structure morphology was observed between a triple Ljnfr1-Ljnfr5-Lys11 mutant line and the WT (Rasmussen et al., 2016). LjNFR1 LCO binding was analyzed with the same strategy as for LjNFR5 (phylogenetic group LYRIA). High affinity for LCO structures derivated from the M. loti main LCOs was found with a Kd of 4.9 nM using SPR and with a Kd of 0.61 nM using MST (Broghammer et al., 2012).
A functional difference between LysM-RLKs of the phylogenetic group LYKI from Arabidopsis or from other plants was demonstrated. In all species except in Brassicaceae, there is at least one member of the phylogenetic group LYKI that contains a YAQ/R motif in the kinase domain while AtCERK1 and Brara.E02055 have respectively TV and IV as AA instead of YAQ or YAR at this position (Fig. 7). The YAQ/R motif has been demonstrated to be important for nodulation. Expression in an Lfnfr1 mutant of a chimerical protein containing the LjNFR1 ECR and the AtCERK1 ICR was unable to restore nodulation (Nakagawa et al., 2011) while an LjNFR1-OsCERK1 chimera was (Miyata et al., 2014). Replacement in AtCERK1 of the TV AA by YAQ allowed the chimera LjNFR1-AtCERK1 (YAQ) to restore nodulation in Ljnfr1. This suggests that the YAQ/R motif is associated with a symbiotic either in the RNS as in MtLYK3 and LjNFR1 or in the AMS as in OsCERK1. Although they bear the YAQ motif, LjLYS6 and MtLYK9 were shown to be no essential in the RNS (Bozsoki et al., 2017) suggesting they rather play a role in the AMS.
Fig. 7 Amino acid sequence alignment of the members of the phylogenetic group LYKI. The YAQ/R motifs is boxed in black.
The LjNFR1 ortholog in M. truncatula, MtLYK3, has been also demonstrated to be involved in nodulation. Mtlyk3 knock-down (Limpens et al., 2003) or missense (Smit et al., 2007) lines were impaired in nodule formation and in rhizobial colonization of M. truncatula roots but not in LCO responses such as apoplast alkalinisation or calcium spiking. This led to the hypothesis that the genetic control of the rhizobial colonization is different between M. truncatula and L. japonicus and that MtLYK3 is involved LCO perception during rhizobial colonization but not in the LCO perception preceding colonization. However, a tandem duplication of MtLYK3 has occurred in M. truncatula. Two genes MtLYK3 and MtLYK2 (Medtr5g086310/Medtr5g086330) are LjNFR1 orthologs (Fig. 6; De Mita et al., 2014). MtLYK2 might be redundant with MtLYK3. Actually, MtLYK2 contains as MtLYK3, the YAQ motif in its kinase domain. Although MtLYK2 is less expressed than MtLYK3 in roots (Limpens et al., 2003), it is possible that the MtLYK2 expression level in Mtlyk3 mutant lines is enough to ensure the LCO responses preceding colonization but not the rhizobial colonization. This would explain the phenotypic difference between Mtlyk3 and Ljnfr1. Complementation experiments have shown that a chimeric protein containing LjNFR1 ECR and MtLYK3 ICR can restore the absence of nodulation in a Ljnfr1 mutant (Nakagawa et al., 2011) suggesting a conservation of the function of these legume LysM-RLKs.
Taken together, these data suggest a divergent evolution of the members of the phylogenetic group LYKI for which the ancestral protein might have a dual role in defense and the AMS. During the evolution, proteins have been specialized in pathogen recognition in Brassicaceae. In legumes, the genes experienced several duplication events, and members might have also been specialized in symbiosis establishment or in defense responses.
LysM-RLK LYKII (not named in Zhang et al, 2009)
In the Phylogenetic group LYKII, we found one ortholog in each species analyzed, except in Brassicaceae (Fig. 6 and S3). In the peach genome, gene duplication occurred and two copies are present. The only characterized member of the phylogenetic group LYKII is the L. japonicus member LjEPR3/LjLYS3 (Lj2g3v1415410) which has been shown to be implicated in the recognition of bacterial exopolysaccharides (EPS) and colonization by rhizobia (Kawaharada et al., 2015, Kawaharada et al., 2017). Knock-out or missense Ljepr3 mutants developed more nodules in presence of a rhizobial M. loti strain which produces an abnormal EPS structure (exoU) and which is almost unable to colonize L. japonicus (Kawaharada et al., 2015). However, the number of nodules found in this study was extremely low. In contrast to the M. loti exoU, a M. loti strain unable to produce EPS (exoB) was able to colonize L. japonicus, despite that WT plants inoculated with a M. loti exoB strain showed abnormal ITs (i.e.: branched or forming balloon-like structure at the epidermal-cortical cell boundary) and intercellular rhizobial colonization in nodules. Although in lower extent, this phenotype was also observed in the Ljepr3 mutants inoculated with WT M. loti. The quantitative phenotypic differences between Ljepr3 and WT plants suggest that EPR3 is however not the only actor in EPS perception (Kawaharada et al., 2017). LjEPR3 expression is induced by Nod-factors and by rhizobia. In the presence of rhizobia, its expression pattern in roots corresponds to the zone susceptible to rhizobial colonization, around the ITs and in the nodule primordia. This result suggests that LjEPR3 is required all along the infection process (Kawaharada et al., 2017).
The ortholog in M. truncatula MtLYK10 (Medtr5g033490) is also induced by Nod-factors and rhizobia and in lower extent by Myc-factors and during the AMS (Mt gene atlas, Mtr.25148.1.S1_at). Similarly, the orthologs in monocots (Os01g36550 and Bradi2g40627) are induced during the AMS (Güimil et al., 2005), gene annotated OsAM191, and unpublished data). Absence of member of the phylogenetic group LYKII in Brassicaceae, together with the role of LjEPR3 in rhizobial colonization and the induction of the expression of various orthologs in presence of root symbionts, suggest that the members of this phylogenetic group play a role in root endosymbioses. LjEPR3 ECR was expressed in insect cells by using a baculovirus system. Binding to EPS was measured by biolayer interferometry using purified LjEPR3 ECR (Kawaharada et al., 2015) and a Kd of 2.7 µM was found. Currently, LjEPR3 is a unique example of a LysM-RLK/P binding a molecule not containing GlcNAc. Whether the recognition of EPS by LysM-RLK is specific to L. japonicus or legumes remains unknown. It is therefore of interest to study a putative role in AMS of non-legume members of the phylogenetic group LYKII and to determine their biochemical properties, especially their ability to bind EPS.
Table of contents :
Chapter 1: SlLYK10 regulates the AMS in tomato
Chapter 2: SlLYK10 binds LCOs with high affinity
Chapter 3: AMS LysM-RLKs might have been directly recruited for the RLS