Introduction

The legume–rhizobium symbiosis stands as one of the most crucial and intricate mutualistic interactions in nature. Its significance goes beyond the supply of nitrogen fixed by rhizobia, which influences plant growth and ecosystem sustainability, as this symbiosis is a pivotal model for investigating plant-microbe interactions (Yang et al., 2022). The orchestration of this symbiosis involves the perception of rhizobial signal molecules such as lipo–chitooligosaccharidic Nod factors (NFs) and surface polysaccharides (Madsen et al., 2003; Tirichine et al., 2006; Kawaharada et al., 2015), along with the secretion of proteins, including type three protein secretion system (T3SS) effectors translocated into host plants (Sugawara et al., 2018; Zhang et al., 2021).

Plant receptor heterocomplexes consisting of two Lysin-motif receptor kinases (LYKs), play a central role in recognizing various microbial poly- and oligosaccharide molecules. In non–legumes such as Arabidopsis, AtLYK5 and AtCERK1 (Chitin elicitor receptor kinase) are required for perception of chitin, and thereby trigger plant immunity to combat fungal pathogens (Cao et al., 2014). In legumes, structurally similar Nod factor receptors (NFRs) such as NFR1 and NFR5 in Lotus japonicus are vital for recognition of rhizobial NFs and initiation of symbiotic signaling (Broghammer et al., 2012; Bozsoki et al., 2017). Accordingly, nfr1 and nfr5 knockout mutants are almost completely unable to form nodules (Madsen et al., 2003; Radutoiu et al., 2003). Over-expression of either NFR1 or NFR5 can activate NF signaling, resulting in formation of spontaneous nodules in the absence of rhizobia (Ried et al., 2014). The intricate action of NFRs is highlighted by the elictiation of a hypersensitive cell death response when NFR1/NFR5 are simultaneous over-expressed in Nicotiana benthamiana leaves (Madsen et al., 2011). Similarly, defense-like responses in Medicago truncatula nodules were observed upon over–expression of MtNFP (Nod factor perception, the ortholog of NFR5) (Moling et al., 2014), perhaps reflecting a potential role of MtNFP–interacting LYKs involved in plant–pathogen associations. It can therefore be expected that the protein abundance of NFRs must be tightly controlled to avoid resistance responses, thereby ensuring optimal root infection by rhizobia.

T3SS effectors of phytopathogenic Gram–negative bacteria play a critical role in suppressing pattern-triggered immunity (PTI) in host plants (Jones and Dangl, 2006). To avoid diseases, specific resistance (R) genes are employed to directly or indirectly recognize effectors translocated into host cells.

Recognition of such avirulence (Avr) effectors often results in effector-triggered immunity (ETI), a much stronger immune response which is usually associated with programmed cell death (Jones and Dangl, 2006). In contrast to the well–studied effector functions of phytopathogenic bacteria in suppressing immunity through various mechanisms, it is largely unknown how T3SS effectors from symbiotic rhizobia contribute to mutualistic interactions.

T3SS effectors of the YopT–type family, such as YopT of Yersinia pestis, AvrPphB of Pseudomonas syringae pv. phaseolicola and rhizobial NopT proteins, are evolutionary related cysteine proteases (Shao et al., 2002; Dai et al., 2008). YopT cleaves Rho family GTPases leading to their inactivation to disrupt the actin cytoskeleton of human host cells (Shao et al., 2003a). AvrPphB can proteolytically cleave Arabidopsis receptor–like cytoplasmic kinase family proteins including PBS1, PBS1-like proteins and BIK1, to suppress PTI (Shao et al., 2003b; Zhang et al., 2010). Cleavage of PBS1 kinase by AvrPphB can be regarded as a “decoy strategy” as it is associated with RPS5 (RESISTANCE TO PSEUDOMONAS SYRINGAE5)-mediated ETI (Shao et al., 2003b). Depending on the host legume, the NopT effector protease of Sinorhizobium fredii NGR234 can have an opposite function in regulating symbiosis (Dai et al., 2008; Kambara et al., 2009). The heterologous expression of NopT from NGR234 in S. fredii USDA 257 severely blocks nodulation in soybean Nanfeng 15 but not in other soybean cultivars (Khan et al., 2022), implying that NopT can act as an Avr effector. In contrast, NopT positively regulates nodule formation in the interaction between S. fredii HH103 and soybean Suinong 14 (Li et al., 2023). NopT from Bradyrhizobium sp. ORS3257 promotes rhizobial infection in nodules of Aeschynomene indica (Teulet et al., 2019). NopT proteases have been biochemically characterized. Some exhibit autocleavage activity and are subsequently acylated at the newly formed N-terminus (Dai et al., 2008; Dowen et al., 2009; Kambara et al., 2009; Fotiadis et al., 2012; Khan et al., 2022). Recently, soybean PBS1-1 and two proteins in Robinia pseudoacacia were identified as NopT targets (Luo et al., 2020; Khan et al. 2022; Li et al. 2023). However, at the molecular level, it remains largely unknown how NopT modulates rhizobial infection and whether NopTs from different rhizobial strains differ in their effector activities.

A challenge in the investigation of rhizobial effectors lies in the availability of suitable assays and plants with which effector activities can be characterized. In this study, we took advantage of the observation that ectopic expression of NFR1 and NFR5 in N. benthamiana leaves induces programmed cell death (Madsen et al., 2011). We hypothesized that this response could be modulated by co-expressed rhizobial effectors. Among the 15 known or putative T3SS effectors of S. fredii NGR234, only NopT could specifically suppress NFR1/NFR5-induced cell death. This observation suggested that NopT is an effector associated with NFRs. Subsequent experiments demonstrated that NopT interacts with both NFR1 and NFR5 at the plasma membrane and proteolytically cleaves NFR5 at the juxtamembrane (JM) domain to suppress NF-mediated host responses. NFR1 phosphorylates the full-length NopT, thereby inactivating its protease activity. However, the autocleaved NopT form of S. fredii NGR234 retains the ability to cleave NFR5 but loses its ability to interact with NFR1. Intriguingly, various Sinorhizobium strains possess only truncated versions of NopT (related to autocleaved NopT of NGR234) which can evade the inactivation by NFR1. We present in this study a model of mutual regulation, in which the rhizobial effector NopT directly targets NFR5 to dampen symbiotic signaling, whereas NFR1-mediated phosphorylation of NopT counteracts NFR5 cleavage. Our work provides essential insights into the intricate interplay between legumes and rhizobia and shows an example of how a rhizobal effector fine-tunes NF signaling by directly targeting NFRs, and how an NFR inactivates the enzymatic activity of the effector through phosphorylation.

Results

NopT expression in N. benthamiana suppresses the NFR1/NFR5-induced cell death response

To examine the role of effector proteins in regulating rhizobial infection, 15 effector genes (Kimbrel et al., 2013) were cloned from S. fredii NGR234, a strain with an exceptionally broad host-range. Initial experiments were performed to screen the effector proteins for their ability to suppress or enhance the NFR1/NFR5–triggered cell death response of N. benthamiana leaves. In this experiment, each effector was co-expressed with NFR1/NFR5 (Fig. S1A–S1C). Only NopT, a cysteine protease belonging to the YopT family (Shao et al., 2002; Dai et al., 2008), could suppress NFR1 and NFR5-induced cell death (Fig. 1A and S1D–S1F). Interestingly, NopT reduced protein abundance of the intact NFR5 when co–expressed with NFR1 (Fig. S1E), suggesting that NopT promotes proteolytic degradation of NFR5. Cys-93 of NopT is an amino acid residue known to be essential for the protease activity of this effector (Dai et al., 2008; Kambara et al., 2009). In contrast to the wild-type NopT protein, expression of the protease-dead NopT^C93S form in N. benthamiana did not result in suppression of cell death induced by NFR1/NFR5 (Fig. 1B), indicating that the protease activity of NopT is essential for the observed cell death-suppressing activity of NopT. In contrast to NopT, NopM, an effector protein from S. fredii NGR234 reported to possess E3 ligase activity (Xin et al., 2012), induced cell death when expressed in N. benthamiana leaves under our test conditions (Fig. 1C).

Figure 1.

We also tested whether the cell death-suppressing effect of NopT was specific for NFR1/NFR5 expression or whether NopT could act generally to suppress cell death. In addition to NopM, we examined various proteins known to induce cell death when expressed in N. benthamiana leaves, namely Avr3a and R3a (Bos et al., 2006), apoptosis regulator BAX (Bcl-2-associated X protein) (Gan et al., 2009), INF1, an elicitor from Phytophthora infestans (Vleeshouwers et al., 2006) and Arabidopsis AtCERK1 (Cao et al., 2014) (Fig. 1C). In contrast to the NFR1/NFR5–induced cell death, expression of NopT was unable to suppress cell death when each of these proteins was expressed in N. benthamiana leaves (Fig. 1D). Hence, the effect of NopT appears to be specific for the NFRs. These results suggested that the effector protease NopT could act directly on NFR1 and/or NFR5.

NopT interacts with NFR1 and NFR5

Autocleaved NopT of S. fredii NGR234 contains acylation sites required for lipidation and subsequent plasma membrane localization (Dowen et al., 2009; Khan et al., 2022). To investigate whether NopT interacts with NFR1 and NFR5, we first used in vivo bimolecular fluorescence complementation (BiFC) analysis. NopT, NFR1 and NFR5 were C-terminally tagged with nYFP and cYFP. Fluorescence signals representing direct NopT-NFR1, NopT-NFR5 and NopT-NopT interactions were detected at the plasma membrane of N. benthamiana (Fig. 2A and 2B). Interactions between NopT and NFR1 and NFR5 were further verified using a Split-luciferase (Split–LUC) complementation assay (Fig. 2C). Co-expression of NopT-cLUC and NFR1-nLUC or NFR5-nLUC produced strong luminescence signals in transformed leaf discs compared with the negative control expressing AtFLS2-nLUC (Arabidopsis FLAGELLIN SENSING2) and NopT-cLUC (Fig. 2C).

Figure 2.

It is known that acylation of NopT can alter its subcellular localization (Dowen et al., 2009; Khan et al., 2022). Here, we performed Split-LUC assays to examine the interaction between NFR1/NFR5 and NopT^{G50A/C51A/C52A}, a modified NopT form lacking acylation sites. Compared to the wild-type form (NopT^WT), the interactions between NopT^{G50A/C51A/C52A} and NFR1/NFR5 were significantly reduced in Split-LUC assays (Fig. S2A and S2B). However, the protease-dead NopT form (NopT^C93S) showed interactions with NFR1 and NFR5 that were not different from NopT^WT (Fig. S2C). These results were further confirmed by a co-immunoprecipitation (co-IP) experiment with N. benthamiana leaves co-expressing FLAG-tagged NopT with C-terminally HA-tagged NFR1 or NFR5. Both NFR1 and LjNFR5 could be co-precipitated by an anti-FLAG antibody when the sample contained NopT-FLAG (Fig. 2D). Overall, the identified physical interactions between NopT and NFR1/NFR5 indicated that NopT targets NFRs and that NFR5 and/or NFR1 may be proteolytically cleaved by NopT.

NopT proteolytically cleaves NFR5 at the JM domain

A previous study has shown that NopT is autocleaved at its N-terminus to form a processed protein that lacks the first 49 amino acid residues (Dai et al., 2008). To test whether NopT could proteolyze NFR1 and/or NFR5, NopT was co-expressed with NFR1-GFP and NFR-GFP fusion proteins in N. benthamiana leaves. The expressed proteins were separated by SDS-PAGE and subjected to immunoblotting. Proteolytic cleavage of NFR5-GFP but not NFR1-GFP was observed in the presence of co-expressed NopT, whereas protease-dead NopT^C93S showed no effect (Fig. 3A and S3A). A similar in vivo cleavage assay with transgenic L. japonicus roots also showed that expression of NopT but not NopT^C93S caused proteolytic cleavage of NFR5 (Fig. 3B). Based on the molecular weight changes, the cleavage site in NFR5 was predicted to occur within the cytoplasmic domain (CD) of NFR5. To further explore NFR5 cleavage by NopT, the CD of NFR5 (Strep-NFR5^CD-HA) and NopT-FLAG or protease-dead NopT^C93S-FLAG were co-expressed in Escherichia coli cells. Immunoblotting results with an anti-HA antibody showed that the active NopT protease cleaved NFR5^CD resulting in a ∼5 kDa smaller protein. However, when the CD of NFR1 was co-expressed with NopT in the same experiment, no NFR1 cleavage was detected (Fig. 3C and 3D). Migration of the bands representing full-length NopT or NopT^C93S was significantly retarded on the gel when the CD of NFR1 was co-expressed (Fig. 3C), suggesting that NopT was phosphorylated by NFR1 in E. coli. The NFR5^CD cleavage product with a ∼5 kDa lower molecular weight in the cleavage assay suggested that the cleavage site is located within the juxtamembrane (JM) domain of NFR5. To test this hypothesis, we replaced the JM with the SUMO tag to creat SUMO-NFR5^KC-HA (KC, kinase domain and C-terminal tail region, a modified NFR5^CD without the JM) for proteolytic assay. Indeed, immunoblot analysis showed that co-expressed NopT was unable to cleave SUMO-NFR5^KC-HA (Fig. S3B). The JM domain of NFR5 was then fused to SUMO and GFP to create a recombinant protein, SUMO-NFR5^JM-GFP. In an in vitro cleavage assay with recombinant proteins from E. coli, GFP was immunodetected in the flow though sample when SUMO-NFR5^JM-GFP was incubated with NopT but not with NopT^C93S (Fig. 3E). The interaction between NopT^C93S and SUMO-NFR5^JM-GFP was confirmed by an in vitro pull-down assay (Fig. S4). Overall, these data strongly indicate that the cleavage site of NFR5 for NopT is located in its JM domain.

Figure 3.

The cleavage of NFR5 by NopT is dependent on multiple residues of the JM domain

The “DKLLSGV” motif in the JM domain of NFR5 (residues Asp-288 to Val-294; Fig. S5) shows highest similarity with the autocleavage region of NopT (DKMGCCA). However, a protein variant of NFR5^CD in which the “DKLLSGV” motif was replaced by 7 alanine residues could still be cleaved by NopT (Fig. S6A). Similar to NopT, the AvrPphB effector of P. syringa has an autocleavage site (Shao et al., 2003b). We therefore also examined a version of NFR5^CD in which the “DKLLSGV” motif was replaced by that of AvrPphB. In the cleavage assay with E. coli cells, AvrPphB did not cleave NFR5^CD. However, the modified NFR5^CD form with the 7 residues from AvrPphB was cleaved by AvrPphB and the cleavage product had a lower molecular weight than that formed by NopT (Fig. S6A). These findings indicated that the cleavage site for NopT is located upstream of the “DKLLSGV” motif in NFR5.

As mentioned above, NopT undergoes autocleavage but it is not clear whether this is due to intermolecular or intramolecular proteolysis. In addition to Cys-93, His-205 and Aps-220 of NopT are conserved catalytic residues (Dai et al., 2008; Kambara et al. 2009). Therefore, NopT forms lacking these catalytic residues were co-expressed with NopT in E. coli. The immunoblot analysis showed that NopT could not proteolyze NopT^C93S, NopT^H205A or NopT^D220A (Fig. S6B), suggesting that NopT autocleavage is due to intramolecular proteolysis. Hence, the residues at the autocleavage site of NopT may not help to predict the cleavage site of NopT substrates.

In an attempt to define the NopT cleavage site in NFR5, we created a series of modified NFR5 forms covering the JM domain of NFR5 (Val-269 to Cys-320) and the first seven amino acid residues of the kinase domain (Lys-321 to Tyr-327). Initially, 17 variants of NFR5^CD were created in which three adjacent amino acids in the JM domain were replaced with three alanine residues. When these proteins were co-expressed with NopT in E. coli, all 17 NFR5^CD variants could be cleaved by NopT but not NopT^C93S (Fig.S6C-S6F). In a similar experiment, we created seven NFR5^CD forms which contained a deletion in the JM domain of 10 residues each. Surprisingly, cleavage products were still observed for all seven examined proteins (Fig. S6G). These results suggest that the JM domain may have multiple sites that can be cleaved by NopT, which is different from the specific cleavage site identified for AvrPphB. In another experiment, recombinant His-SUMO-NFR5^CD-GST protein was co-expressed with NopT in E. coli and its cleavage product was subjected to N-terminal protein sequencing using Liquid Chromatography Mass Sprectrometry (LC-MS) analysis. Based on alignment of identified peptides, the NopT cleavage site in NFR5 was mapped to four basic amino acids in the JM domain (RRKK; amino acid residues 271-275) (Fig. S7). However, this result was not consistent with the observation that NopT expressed in E. coli could still proteolyze a co-expressed NFR5^CD form in which the residues from Tyr-268 to Leu-277 were deleted (Fig. S6G). Overall, the experiments performed suggest that NopT preferentially proteolyzes NFR5 at the RRKK motif, but other sites in the JM domain can also be cleaved by NopT. Since NFR5 belongs to a subfamily of LYK family proteins that lacks kinase activity (Yang et al., 2022), we wondered whether NopT could proteolyze other LYK proteins of this subfamily, which might be helpful to characterize the biochemical function of NopT. We therefore investigated whether the CDs from Arabidopsis AtLYK5, L. japonicus LjLYS11 and M. truncatula MtNFP expressed in E. coli can be proteolyzed by NopT. As shown in Fig. S8, NopT, but not NopT^C93S, was able to cleave the CDs of AtLYK5 and LjLYS11. Replacement of the JM domain in NFR5^CD with the JM domain from either AtLYK5 or LjLYS11 resulted in cleavage by NopT (Fig. S8). These data show that NopT cleaves AtLYK5 and LjLYS11 at their JM domains. Based on sequence analysis of the JM region of NFR5 and its homologous proteins, five conserved residues (Ser-283, Gly-294, Tyr-304, Ala-310, Thr-311) were identified and mutated as NFR5^CD-5m for the proteolytic assay (Fig. S5B). As shown in Fig. 3F, NFR5^CD-5m, a NFR5^CD variant with point mutations of S283Y, G294Q, Y304S, A310I, and T311Y, could not be cleaved by NopT. Interestingly, NopT could not proteolyze MtNFP^CD (Fig. 3G). We then swapped the JM domains in NFR5^CD and MtNFP^CD, generating NFP^JM-NFR5^KC and NFR5^JM-NFP^KC (KC stands for the kinase domain and the C-terminal tail region). As shown in Fig. 3G, NopT could proteolyze NFP^JM-NFR5^KC but not NFR5^JM-NFP^KC in E. coli. In an in vitro cleavage assay, the cleavage product of SUMO-NFP^JM-GFP proteolyzed by NopT was detected in the flow though sample (Fig. S9). These data suggest that the cleavage of NFR5 by NopT depends on the conformational structure of NFR5^KC and/or other specific amino acids in NFR5^KC. We then investigated whether two additional recombinant proteins with large swapping regions in NFR5^CD and NFP^CD (NFR5^268-445-NFP^458-595 and NFP^270-457-NFR5^456-595) could be cleaved in the E. coli proteolysis test (Fig. S10). Notably, both proteins were cleaved by NopT, suggesting that a specific conformation of the NopT substrate is required for proteolysis. Taken together, we concluded from our experiments that NopT can proteolyze NFR5 at the JM domain and that mutiple nonadjacent residues in NFR5^JM in combination with a specific KC conformation are required for proteolysis.

NopT phosphorylated by NFR1 is proteolytically inactive

The perception of rhizobial NFs by the heterocomplex of NFR1 and NFR5 involves trans-phosphorylation events from the NFR1 kinase to the pseudo-kinase NFR5 (Madsen et al., 2011). Since NopT is associated with both NFR1 and NFR5, we hypothesized that NopT is a phosphorylation target of NFR1, which may interfere with transphosphorylation events between the two receptors. NopT^C93S was therefore expressed in L. japonicus wild-type and NFR1 knockout mutant plants (nfr1-1) and analyzed on Zn²⁺-Phos-tag gels. On such gels, a Zn²⁺-Phos-tag bound phosphorylated protein migrates slower than its unbound nonphosphorylated form. Slower migrating bands corresponding to phosphorylated NopT^C93S were observed, particularly when wild-type plants were inoculated with rhizobia (Fig. 4A and 4B). These findings indicated that NopT phosphorylation in planta is largely dependent on NFR1 and that the protein was probably phosphorylated at multiple sites.

Figure 4.

Next, we performed assays to investigate whether expressed NFR1^CD directly phosphorylates co-expressed NopT. NFR5^CD and a kinase-inactive NFR1^CD form (NFR1^CD-K351E; Lys-351 is a conserved residue required for ATP binding) were included into these experiments. When NFR1^CD was expressed in E. coli, extracted NopT (Fig. 3C and 4C) and NFR5^CD (Fig. S11) exhibited retarded migration on SDS-PAGE gels, indicating that NFR1 was able to phosphorylate both proteins. A band shift was observed when the phosphorylated full-length NopT was in vitro dephosphorylated by calf intestinal alkaline phosphatase (CIAP). However, the CIAP treatment caused no obvious band shift of the autocleaved NopT form (Fig. 4C). The phosphorylation sites of NopT were then identified using an in vitro phosphorylation assay followed by LC-MS analysis. Several serine and threonine residues in NopT were identified to be phosphorylated by NFR1^CD (Table S1). Taken together, these results showed that full-length NopT is a phosphorylation target of the NFR1 kinase. Finally, we wondered whether NopT phosphorylation by NFR1 influences its proteolytic activity. Plasmids encoding NopT variants were constructed in which the phosphorylated residues were substituted to either alanine (to block phosphorylation) or aspartate (to mimic the phosphorylation status). The different NopT forms were then expressed in E. coli to analyze NopT autocleavage and proteolysis of co-expressed NFR5^CD. Aspartate substitutions at three phosphorylation sites of NopT (Thr-97, Ser-119 and Thr-257) resulted in the loss of the autoproteolytic activity, as well as the ability to cleave+ NFR5^CD (Fig. 4D). In contrast, corresponding NopT proteins with alanine substitutions retained autocleavage activity and the ability to proteolyze NFR5^CD (Fig. 4D). These data indicate that phosphorylation of full-length NopT by NFR1 inhibits its proteolytic activity, thereby keeping NopT unprocessed and NFR5 uncleaved.

NopT dampens rhizobial infection

As NopT of S. fredii NGR234 could directly target and cleave NFR5, we expected that NopT may be an important player regulating rhizobial infection in legumes. NF signaling is known to function during early stages of rhizobial infection (Geurts et al., 2005; Wang et al., 2012; Cai et al., 2018). We focused our experiments on early stages of rhizobial infection. Inoculation of L. japonicus plants with a GFP- or GUS-labelled nopT mutant of NGR234 (NGR234 ΔnopT) resulted in a massive infection of root hairs, whereas infection by the NGR234 wild-type strain was less frequent (Fig. 5A and 5C). We also performed experiments with L. japonicus plants containing pNIN:GUS which exhibit β-glucuronidase (GUS) activity when the NIN (NODULE INCEPTION) promoter activity is induced by NF signaling (Schauser et al., 1999). GUS staining of roots showed that the NIN promoter activity was stronger in response to NGR234ΔnopT inoculation relative to inoculatioin by wild-type NGR234 (Fig. 5B and 5D). Likewise, inoculation with NGR234ΔnopT resulted in an increased number of nodule primordia (Fig. 5G). In contrast, over-expression of Not under the control of the T7 promoter in the NGR234 parent strain reduced the number of nodule primordia when compared to inoculation with the wild-type strain (Fig. 5G). These findings indicate that NopT negatively influences symbiosis at early symbiotic stages. However, we could not find any uninfected (empty) nodules when roots were inoculated with these strains.

Figure 5.

To confirm that the massive infection of roots by NGR234ΔnopT is due to the loss of NopT, we complemented the NGR234ΔnopT by expressing wild-type NopT. The L. japonicus plants inoculated with this complemented strain had fewer infection foci, indicating restoration of the wild-type phenotype (Fig. 5E). We also tested whether the proteolytic activity of NopT affected rhizobial infection. The protease-dead version of NopT^C93S and three NopT forms with phospho-mimimetic residues (i.e., NopT^T97D, NopT^S119D, and NopT^T257D lacking proteolytic activity) were individually expressed in NGR234ΔnopT. Compared with the high number of infection foci induced by NGR234ΔnopT in L. japonicus plants, inoculation with these strains resulted in significantly fewer infection foci, but more than with the wild-type strain (Fig. 5E).

We also expressed nopT under the control of a ubiquitin promoter in hairy roots of L. japonicus and inoculated the transgenic roots with DsRed-labeled M. loti MAFF303099, which does not possess a nopT gene in its genome. As shown in Fig. 5F, reduced numbers of infection foci and infection threads were observed in the NopT expressing roots, indicating that NopT negatively affects the symbiosis between L. japonicus and M. loti strain.

To investigate whether the cleavage of NFR5 by NopT reduces rhizobial infection of L. japonicus roots, NFR5 and NFR5^5m (uncleavable variant with 5 amino acid substitutions in the JM domain; Fig. 3F) under the control of the native NFR5 promoter, were expressed in hairy roots of NFR5 knockout mutant plants (nfr5-3). NFR5 expression resulted in numerous infections and a significant increase of infection foci was observed upon inoculation with the NGR234ΔnopT mutant (Fig. 5H). However, NFR5^5m expression showed no effects on rhizobial infection (Fig. S12), suggesting that the amino acid substitutions in NFR5^5m caused a conformational change that rendered the protein inactive in symbiotic signaling. As the NFR5^JM-NFP^KC fusion protein was not cleaved by NopT (Fig. 3G), we also constructed hairy roots of nfr5-3 mutant plants in which NFR5-NFP^KC (NFR5 variant with its KC domain replaced by the KC of MtNFP) was expressed under the control of the NFR5 promoter. Remarkably, inoculation of these plants showed no difference between NGR234ΔnopT and the NGR234 wild-type strain in terms of rhizobial infection foci (Fig. 5H), suggesting that the cleavage of NFR5 by NopT reduced the degree of rhizobial infection. Taken together, these data indicate that the protease activity of NopT, its phosphorylation status and the proteolysis of the NFR5 target are important for the regulation of rhizobial infection.

NopT from other S. fredii strains also cleave NFR5

Since NopT of S. fredii NGR234 (NopT_NGR234) cleaves NFR5, we wondered whether homologs from other rhizobial species also possess such a proteolytic activity. Phylogenetic analysis showed that NopT proteins of Sinorhizobium and Bradyrhizobium are located in different clades (Fig. S13). B. diazoefficiens USDA110, a typical Bradyrhizobium strain used to study nodulation of soybeans, produces two NopT proteins (Fotiadis et al., 2012). In contrast to NopT_NGR234, however, neither NopT1_USDA110 nor NopT2_USDA110 were able to cleave co-expressed NFR5 in E. coli (Fig. 6A). This finding suggests that NopT homologs from Bradyrhizbium species have lost the ability to cleave NFR5 and probably act on other, unknown host target proteins. The genome of various Sinorhizobium species (e.g., S. fredii USDA257 and S. fredii HH103) possess genes that encode truncated NopT forms, which are almost identical to the autocleaved version of NopT_NGR234 (Khan et al., 2022). Remarkably, NopT_USDA257 and NopT_HH103 were both able to cleave NFR5 in E. coli (Fig. 6A). However, both NopT proteins were unable to suppress cell death triggered by NFR1 and NFR5 expression in N. benthamiana leaves (Fig. 6B). Moreover, expression of NopT_USDA257 in NGR234ΔnopT had no effect on the infection ability of NGR234ΔnopT, as observed in inoculation tests with the L. japonicus line containing pNIN:GUS (Fig. 6C and 6D). These results suggest that NopT_USDA257 and NopT_HH103 exhibit similar proteolytic activity to NopT_NGR234, but target different host proteins to regulate rhizobial infection.

Figure 6.

Discussion

Plant-microbe interactions are complex, employing distinct “chemical weapons” or strategies to communicate mutually, thereby securing more resources for their survival (Jones and Dangl, 2006; Schubert et al., 2020). T3SS effectors of phytopathogenic bacteria are injected into plant cells to facilitate bacterial infection by targeting host proteins of the plant’s immune system, thereby suppressing PTI (Jones and Dangl, 2006; Tang et al., 2017). However, the role of rhizobial T3SS effectors in regulating the mutualistic symbiosis between legumes and rhizobia remains largely unknown. The aim of this study was to gain essential insights into the regulation mechanisms by which rhizobial effectors act directly on symbiotic signaling pathways to promote or dampen infection of host cells.

Understanding how plants recognize different microbes as “friends” or “foes” is a central question in biology. The perception of microbial signals by plant receptor proteins is essential for triggering specific signaling pathways to resist invading pathogens or to establish symbiosis with beneficial microbes (Zipfel and Oldroyd, 2017). Direct targeting of these receptors by T3SS effectors represents an efficient strategy of phytopathogenic bacteria to suppress PTI and promote bacterial infection. In the well-studied Arabidopsis-Pseudomonas syringae interaction, T3SS effectors target flagellin receptor FLS2 and its coreceptor BAK1 (BRI1-[Brassinosteroid insensitive1]-associated receptor kinase 1) to suppress flagellin-induced PTI (Zipfel et al., 2004; Sun et al., 2013). For example, AvrPto targets FLS2 and HopB1 cleaves BAK1 (Xiang et al., 2008; Li et al., 2016). Likewise, AvrPtoB, an E3 ligase in P. syringae, targets FLS2 to promote its degradation and suppress PTI (Goehre et al., 2008). Despite the evolutionary importance of T3SS effectors in targeting host receptors, we were surprised to find that NopT, a rhizobial member of the C58 protease effector family, can directly target NFRs and manipulate the symbiotic signaling pathway in legumes. The interaction between NopT of S. fredii NGR234 and NFRs of L. japonicus was identified by using a unique screening system, in which co-expression of NFR1/NFR5 in N. benthamiana leaves leads to cell death. NopT, but not other effectors of NGR234, could suppress this cell death response. Consistent with these initial observations, inoculation of L. japonicus with the S. fredii NGR234ΔnopT mutant resulted in increased infection of L. japonicus roots. Moreover, over-expression of NopT in the roots was found to reduce rhizobial infection by M. loti. The interaction between NopT and NFR1/NFR5 was verified by BiFC, Split-LUC and co-IP experiments.

Interestingly, using different assays, we found that NopT cleaves NFR5 at the JM domain. However, the efficiencies of NopT cleavage of NFR5 in L. japonicus and N. benthamiana were different (Fig. S14), which could be related to the uncontrollable activation of NFRs in N. benthamiana or phosphorylation of NopT by another kinase leading to decreased proteolytic activity. The modification of five conserved residues in the JM domain and the replacement of NFR5^KC by NFP^KC resulted in NFR5^CD forms that were resistant to proteolysis by NopT in E. coli cells. These results suggest that cleavage of NFR5 depends on both the JM domain and the KC region. Although belonging to the same protease family, the NopT cleavage site in NFR5 was found to be polybasic as in YopT substrate (Shao et al., 2003a; Schmidt, 2011), whereas the cleavage site of AvrPphB substrate is canonical with seven adjacent amino acids involved (Kim et al., 2016). These differences reflect a broad variety of YopT-type effector proteases and their substrates in host cells. Our study shows, that NopT targets the NFR1/NFR5 complex thus reduces NF signaling and formation of nodule primordia in L. japonicus. NopT-NFR interactions may reflect a regulation mechanism to prevent rhizobial hyperinfection in host species other than L. japonicus.

Similar to the Pseudomonas effector AvrPphB, NopT of S. fredii NGR234 is able to cleave soybean PBS1-1 and expression of this effector in S. fredii USDA257 negatively influences symbiosis with certain soybean cultivars (Khan et al., 2022). Indeed, NopT of NGR234 acts as an Avr effector in some nonhost plants. Transient expression of NopT of strain NGR234 in Arabidopsis and tobacco triggers strong immune responses and cell death (Dai et al., 2008; Kimbrel et al., 2013; Khan et al., 2022; Fig. S15). In Arabidopsis, NopT-induced cell death was found to be dependent on the PBS1-1 kinase and the resistance protein RPS5 (Khan et al., 2022). Likewise, when constitutively expressing NopT and NopT^C93S in L. japonicus, we only obtained stable transgenic lines expressing NopT^C93S, indicating that the protease activity of NopT had a negative effect on plant development. In contrast, generation of hairy roots expressing NopT was possible. Such differences may be explained by different NopT substrates in roots and aerial parts of the plant. Indeed, our study shows that NopT not only cleaves NFR5 but is also able to proteolyze Arabidopsis AtLYK5 and L. japonicus LjLYS11. These two receptors possess chitin-binding affinities and trigger PTI responses (Cao et al., 2014; Gysel et al., 2021). NFR5, AtLYK5, and LjLYS11 are kinase-dead LYKs belonging to the same LYK subfamily. Thus, NopT appears not only to suppress NF signaling, but may also interfere with signal transduction pathways related to plant immunity. Depending on the host legume species, NopT could suppress PTI or induce ETI, thereby modulating rhizobial infection and nodule formation. Interactions between NopT and proteins related to the plant immune system may represent an important evolutionary driving force for host-specific nodulation and explain why the presence of NopT in NGR234 has a negative effect on symbiosis with L. japonicus but a positive one with other legumes.

The regulation of kinase-dead pseudokinases in plants may be more complex compared to the well-studied control of protein kinase activities by phosphorylation. The cleavage of kinase-dead receptors may be an overlooked strategy to regulate their function in plant biology. Regulation of protein activities by cleavage is well studied in animals, e.g., for Notch signaling, where proteolysis of Notch molecules leads to downstream signaling (Kopan and Ilagan, 2009). Arabidopsis BAK1 undergoes proteolytic cleavage, which is important for both brassinosteroid signaling and induction of PTI responses (Zhou et al., 2019). The cytoplasmic domain of NFR5 is a pseudokinase that contributes to activation of NF signaling. Our work indicates that NopT proteolytically cleaves NFR5 and suppresses NF signaling. Such a fine-tuning process could be advantageous for the symbiosis in host plants other than L. japonicus, for example to prevent hyperinfection. However, the precise function of NFR5 cleavage by NopT and the fate of the released cytoplasmic domain remain to be clarified.

Mutual regulation between T3SS effectors and host target proteins has been investigated for various plant-pathogen interactions. For example, the conserved Pseudomonas effector AvrPtoB can directly ubiquitinate the key components of plant immunity to promote bacterial virulence (Goehre et al., 2008). However, the activity of AvrPtoB in promoting virulence could either be enhanced or dampened by host proteins. AvrPtoB can be phosphorylated by SnRK2.8 to promote its virulence activity (Lei et al., 2020). On the other hand, the plant lectin receptor-like kinase LexRK-IX.2 and the Pto kinase phosphorylate AvrPtoB to inactivate its ubiquitin E3 ligase activity and undermine the effector’s ability to suppress PTI (Ntoukakis et al., 2009; Xu et al., 2020). A similar strategy is used by NFR1 to inhibit the protease activity of NopT via phosphorylation. NFR1 expressed in E. coli phosphorylates full-length NopT of S. fredii NGR234 at several residues. Three NopT forms with a phosphomimetic substitution (i.e., NopT^T97D, NopT^S119D, and NopT^T257D) lacked autoproteolytic activity and the ability to cleave NFR5. The expression of these NopT variants in the NGR234ΔnopT mutant showed little effects in comparison with expression of wild-type NopT. These data suggest that L. japonicus utilizes NFR1 to phosphorylate NopT in order to dampen its catalytic activity and protect NFR5 from cleavage, which favors rhizobial infections. However, NopT^ΔN50, which is similar to autocleaved NopT, retained the ability to interact with NFR5 but not with NFR1. Moreover, full-length NopT, but not autocleaved NopT, migrated slower on gels when NFR1^CD was co-expressed in E. coli, suggesting that autocleaved NopT may escape being phosphorylated and inactivated by NFR1. Future studies are required to explore the interactions between acylated NopT and NFR1. When expressed in N. benthamiana, GFP-tagged NopT forms lacking their autocleavage site (and thus non-acylated) were found to be localized within the cytoplasm, whereas NopT was targeted to the plasma membrane (Khan et al., 2022). Accordingly, NopT without acylation sites showed weaker interactions with NFRs in our study, suggesting that acylation of NopT, leading to plasma membrane location, promotes NopT-NFR interactions.

NopT homologs vary in different S. fredii strains. USDA257 and HH103 produce truncated NopT versions different from NGR234 due to a 19-bp deletion (Khan et al., 2022). Accordingly, NopT of these strains lacks autocleavage and acylation sites and possess a different N-terminal secretion signal sequence (Akeda and Galán, 2005). Nevertheless, NopT_HH103 is a functional T3SS effector as mutant analysis indicated a symbiosis-promoting role in certain soybean cultivars (Li et al., 2023). Similar to NopT_NGR234, NopT_USDA257 and NopT_HH103 were able to cleave NFR5 in E. coli. Since NopT_USDA257 and NopT_HH103 cannot be acylated, they likely accumulate in the cytoplasm of host legumes and are therefore less efficient in cleaving plasma membrane-bound NFR5 in comparison to acylated NopT_NGR234. This could also explain why NopT_USDA257 was unable to suppress cell death triggered by NFR1/NFR5 in N. benthamiana leaves. It can be expected that NFR1 does not interact with and phosphorylate NopT_USDA257 or NopT_HH103, as these effectors are similar to NopT^ΔN50 of NGR234, which cannot interact with NFR1. Consistent with these findings, expression of NopT_USDA257 in NGR234ΔnopT did not alter the infection phenotype of NGR234ΔnopT. It can be hypothesized that the host targets of these truncated NopT effectors might be different from those of NopT_NGR234. Bradyrhizobium NopT effectors are also different from NopT_NGR234, as NopT of B. diazoefficiens USDA110 was unable to cleave NFR5). NopT effectors may play a crucial role in Bradyrhizobium-legume interactions. A Bradyrhizobium sp. ORS3257 mutant deficient in NopT production induced only ineffective nodules on roots of Aeschynomene indica and most of the formed nodules were not infected (Teulet et al., 2019). Overall, these findings indicate versatile functions of NopT homologs in different strains and suggest that NopT controls different regulation processes in specific host cells.

We present here a model of mutual regulation, in which NopT proteolyzes NFR5 to suppress NF signaling, whereas protease activity of full-length NopT is suppressed by NFR1 via phosphorylation (Fig. 6E). NopT impairs the function of the NFR1/NFR5 receptor complex. Cleavage of NFR5 by NopT reduces its protein levels. Possible inhibitory effects of NFR5 cleavage products on NF signaling are unknown but cannot be excluded. Inactivation of NopT protease activity via phosphorylation by kinases such as NFR1 appears to be a countermeasure of the host to enable symbiotic signaling and rhizobial infection. Such feedback regulation could be the result of evolutionary pressure to produce ‘improved’ NFRs, which are resistant to cleavage by Not and have an increased capacity to phosphorylate NopT. On the other hand, suppression of NopT protease activity by NFRs may drive broad-host-range strains such as NGR234 to produce higher levels and structurally modified NFs.

Taken together, this study provides insights into a legume-rhizobium interaction, in which the bacterium deploys an effector protease to dampen symbiotic signaling, while the host plant counteracts by phosphorylating the effector, leading to its inactivation. Our findings highlight the function of a bacterial effector protease in regulating a symbiotic signaling pathway in legumes. This opens up the perspective of developing specific kinase-interacting proteases to reprogram and fine-tune cellular signaling processes in general.

Materials and Methods

Germination and growth of Lotus japonicus

Lotus japonicus (ecotype Gifu B-129) was provided by the Center for Carbohydrate Recognition and Signaling (https://lotus.au.dk/), and used as wild-type for nodulation assays. The Gifu pNIN:GUS transgenic line and nfr mutants (nfr1-1 and nfr5-3) were kindly provided by Dr. Jens Stougaard from the Aarhus University, Denmark (Madsen et al., 2003; Radutoiu et al. 2003; Heckmann et al., 2011). All L. japonicus seeds were treated with concentrated sulfuric acid for 10 min, followed by surface sterilization in 1% (w/v) NaClO for 8 min. After incubation at 4°C in the dark for 2 d, the seeds were placed on 0.8% (w/v) agar containing half-strength Murashige & Skoog (MS) medium for germination at 22°C for 2 d in the dark. Seedlings were then transferred to growth pots containing vermiculite supplied with half-strength B&D (Broughton & Dilworth) medium without nitrogen under long-day conditions (16-h light/8-h dark) at 22°C. Ten-day-old seedlings were inoculated with rhizobial cultures (OD₆₀₀=0.02) for infection experiments.

Plasmid construction

All plasmids used in this study were generated using MultiF Seamless Assembly Mix (Abclonal, # RK21020). The plasmid pGWB514 was used as an original plasmid for construction of all binary vectors (Nakagawa et al., 2007). Briefly, different tags including HA tag, Myc tag, FLAG tag, Strep tag, nLUC, cLUC and GFP, were amplified using a forward primer containing a KpnI site and ligated into pGWB514 digested with XbaI and SacI using the seamless cloning method described above. The generated plasmids were named pG5XX-HA, pG5XX-Myc, pG5XX-FLAG, pG5XX-nLUC, pG5XX-cLUC, pG5XX-GFP, respectively. The coding sequences of L. japonicus NFR1 and NFR5 were cloned into pGWB514 and pG517 between XbaI/KpnI to generate NFR1-HA and NFR5-Myc fusion constructs under the control of the cauliflower mosaic virus 35S promoter. The DNA fragment containing pro35S:NFR5-Myc-NosT was then amplified and cloned into pG514-NFR1 at SbfI. The final plasmids contained pro35S:NFR1-HA-NosT and pro35S:NFR5-Myc-NosT. Rhizobial genes encoding effectors from S. fredii NGR234 were cloned into pG5XX-FLAG between XbaI/KpnI. To detect the interaction between NFR1/NFR5 and NopT in N. benthamiana, the full-length NFR1/NFR5, NopT and AtFLS2 coding sequences were cloned into pG5XX-Nluc, pG5XX-Nluc, pSPYCE(MR) and pSPYNE(R)173 (Waadt et al., 2008). For cleavage assays in N. benthamiana leaves, the NFR5 and NopT coding sequences were cloned into pG5XX-FLAG GFP and pG5XX-Strep, respectively. For cleavage assays in E. coli, the sequence encoding the cytoplasmic domain of NFR5 was cloned into pACYC duet Strep-HA (Han et al., 2017), and the sequences encoding the kinase domain of NFR5 fused to the HA tag and the JM domain of NFR5 fused to the GFP tag were cloned into pSUMO. The point mutations in NFR5^CD and NopT were created by site-directed mutagenesis PCR using the templates pACYC duet Strep-NFR5^CD-HA and pET28a NopT-FLAG. For NFR/NopT cleavage experiments with L. japonicus plants, NopT-FLAG or NopT^C93S-FLAG was cloned into pUB-GFP between XbaI and KpnI, and 35S-NFR5-Myc-NosT was cloned into pUB-NopT-GFP at PstI. For Not phosphorylation experiments with E. coli, the sequence encoding the cytoplasmic domain of NFR1 fused to a Myc tag was cloned into pCDF duet. For complementation experiments with the NGR234ΔnopT mutant (Dai et al., 2008), a 1317-bp DNA fragment upstream of NopT and the coding sequence of a given NopT sequence was fused using overlap PCR and cloned into pHC60 between XhoI and KpnI. For complementation experiments with the nfr5-3 mutant, a 1316-bp NFR5 promoter DNA fragment was fused to NFR5-NFP^KC or NFR5^5musing overlap PCR and cloned into pUB-Cherry between PstI and KpnI.

Transient gene expression in Nicotiana leaves

The plasmids of INF1, Avr3, BAX, and R3a for transient expression in Nicotiana benthamiana were kindly gifted by Dr. Juan Du from Huazhong Agricultural University. Agrobacterium tumefaciens strain EHA105 carrying various constructs were mixed with an Agrobacterium culture harboring the P19 suppressor in an infiltration buffer (10 mM MgCl₂, 10 mM MES-KOH pH 5.8 and 200 μM acetosyringone). After incubation for 2 h at room temperature, Agrobacterium cultures were infiltrated into the leaves of 4-week-old N. benthamiana and N. tabacum plants. Leaves were harvested and analyzed two-days post inoculation. The cell death suppression experiments with N. benthamiana plants were performed as described previously (Wang et al., 2011).

Hairy root transformation

Hairy root transformation of L. japonicus plants was performed as described previously (Li et al., 2018). In brief, surface-sterilized seeds were germinated and grown on MS plates without sucrose (23°C in the dark for the first 3 days and 23°C at a 16-h light/8-h dark period for the following 2 days). The seeds were then cut at the middle of the hypocotyl and co-cultivated with A. rhizogenes LBA1334 carrying pUB-GFP, pUB-NopT-FLAG-GFP, pNFR5:NFR5-mCherry, pNFR5:NFR5^5m-mCherry and pNFR5:NFR5-NFP^KC-mCherry, respectively (23°C in the dark for the first 3 days and 23°C at a 16-h light/8-h dark period for the following 2 days). The plants were then transferred onto solid B5 medium. After about 15 days, non-fluorescent hairy roots (lacking expression of GFP or mCherry) were removed. Where indicated, the plants were inoculated with rhizobia: S. fredii NGR234, NGR234ΔnopT or Mesorhizobium loti MAFF303099, and analyzed at indicated time points.

Analysis of NFR5 cleavage by NopT and variants

In experiments with E. coli, Strep-NFR5^CD-HA or Strep-NFR5^CD-HA variants were co-expressed with NopT-FLAG or FLAG tagged NopT variant or AvrPphB-FLAG by using Duet vectors (Novagen; Han et al., 2017). Bacterial cultures were treated with 0.5 mM IPTG at 22°C for 15 h. After washing the bacterial cells with PBS buffer, the bacterial pellet was resuspended in extraction buffer consisting of Tris-HCl (pH 7.3), 150 mM NaCl, 5 mM EDTA, 0.5% (w/v) SDS and 0.5% (v/v) Triton X-100. After adding SDS loading buffer and boiling at 100°C for 5 min, the extracted proteins were subjected to immunoblot analysis using anti-HA-peroxidase (Sigma, clone 3F10) or anti-FLAG (Sigma, F1804) antibodies. An anti-groEL antibody (ABclonal, A0969) was used to confirm equal loading of proteins on gels.

For in vitro cleavage assays, His-SUMO-NFR5^JM-GFP and NopT-FLAG-His proteins were expressed in E. coli and purified. The reaction mixture contained 3 μg NopT-FLAG (or NopT^C93S-FLAG) and 1 μg SUMO-NFR5^JM-GFP in extraction buffer without SDS and EDTA. After incubation at 22°C for 7 h, 10 μL of a Ni-charged resin (GenScript, Cat. No. L00223) were added, and the cleavage products in the flow through were detected by immunoblot analysis using an anti-GFP antibody (ABclonal, AB_2770402).

To analyze cleavage of NFR5 in L. japonicus roots, hairy roots co-expressing NopT-FLAG or NopT^C93S-FLAG with NFR5-Myc were harvested at about 15 days after induction of hairy roots without rhizobia inoculation and subjected to immunoblot analysis. Expressed proteins were extract by extraction buffer and immunodetected with anti-FLAG antibodies and anti-Myc (Bio-Legend, # 626808) antibodies.

For NFR5/NopT cleavage experiments with N. benthamiana plants, leaves expressing proteins (NopT-FLAG or NopT^C93S-FLAG with NFR5-GFP) were analyzed. Leaf discs were collected at 2 days post infiltration with agrobacteria and extracted in the extraction buffer (Tris-HCl (pH 7.3), 150 mM NaCl, 5 mM EDTA, 0.5% (w/v) SDS and 0.5% (v/v) Triton X-100) and 2% protease inhibitors (Sigma, P9599). NFR5 was immunodetected with anti-Myc and anti-GFP (ABclonal, AB_2770402) antibodies. The NopT or NopT^C93S were detected with anti-FLAG antibody. The actin proteins were detected with anti-Actin antibody (Abclonal, # AC009).

Phosphorylation assays

For NopT phosphorylation in E. coli, His-NFR1^CD-Myc or His-NFR1^CD–K351E-Myc was co-expressed with NopT-FLAG-His by using Duet vectors (Novagen). Protein expression was induced by 0.5 mM IPTG and cells were grown at 28°C for 18 h. After simultaneous purification of NFR1^CD and NopT using Ni-charged resin, the proteins were incubated with CIAP (TaKaRa, 2250B) at 37°C for 3 h. The mobility shifts of NFR1^CD and NopT were detected with anti-Myc (Bio-Legend, 626808) and anti-FLAG (Sigma, F1804) antibodies by immunoblot analysis.

In phosphorylation experiments with L. japonicus plants, NopT^C93S-FLAG was expressed in hairy roots of the nfr1-1 mutant and wild-type plants, respectively. After removal of roots without fluorescence signals, plants were kept on half-strength B&D medium for 5 days. Roots were inoculated with Mesorhizobium loti MAFF303099 or treated with water. Total proteins were extracted using the extraction buffer (Tris-HCl (pH 7.3), 150 mM NaCl, 5 mM EDTA, 0.5% (w/v) SDS and 0.5% (v/v) Triton X-100) supplemented with 2% protease inhibitor cocktail (Sigma, P9599) and 2% phosphatase inhibitor cocktail (Yeasen, 20109-A). Proteins were then precipitated with TCA to remove various contaminants (e.g., EDTA and surfactant)s. Phosphorylation of the NopT^C93S protein was analyzed by 50 μM Zn²⁺-Phos-tag SDS-PAGE gel (Kato & Sakamoto, 2019) and detected with the anti-FLAG antibody by immunoblot analysis. As loading control, actin proteins were immunodetected with an anti-Actin antibody.

BiFC and Split–LUC assays

In the BiFC assay with N. benthamiana plants, the eYFP fluorescence of expressed fusion proteins was recorded 2-3 days post infiltration with A. tumefaciens using a confocal microscope (Leica TCS SP8). For the Split–LUC assay, N. benthamiana leaves were collected at 2 days post infiltration and sprayed with a solution of 1 mM D-luciferin (Promega, E1603) and 0.02% (v/v) Triton X-100. After a dark adaptation of 5 min, bioluminescence images were acquired using a Tanon Bio–Imaging System (Tanon, Shanghai, 4600).

LC-MS analysis

For analysis of NopT phosphorylation sites, NopT-FLAG-His was co-expressed with His labelled NFR1^CD or NFR1^CD-K351E in E. coli. The proteins were purified using Ni-charged resin (GenScript, Cat. No. L00223). The bands containing NopT^P and NopT (control) proteins, were cut out, destained by 50 mM triethylammonium bicarbonate (TEAB) solution (50% (v/v) acetonitrile in 50 mM TEAB), and washed with 100% acetonitrile until the gel turned white. The protein bands were then in-gel digested with sequencing-grade trypsin. LC-MS analysis was performed by Novogene Co., Ltd. (Beijing, China).

For identificaiton of the cleavage site in NFR5, NopT and His-SUMO-NFR5^CD-GST, were co-expressed in E. coli and purified using Glutathione Resin (GenScript, # L00206). About 80 μg of the cleaved protein was collected for analysis. The protein bands were digested with trypsin,glu-C, chymotrypsin and pepsin. LC-MS analysis was performed by Oulu Biotechnology Co., Ltd. (Shanghai, China).

Examination of rhizobial infection

The morphology of rhizobial infection foci and infection threads have been described previously (Rae et al., 2021). In brief, an infection focus is formed when the tip of a root hair curls around a single bacterium, forming an infection pocket. The number of infection foci and infection threads was determined using a Leica fluorescence microscope (DM2500). Whole roots were mounted on microscope slides and infection foci and infection thread counts were expressed on a per-root basis.

Histochemical GUS staining of L. japonicus Gifu B-129 or Gifu pNIN:GUS roots inoculated with different rhizobial strains was performed using the chromogenic substrate 5-bromo-4-chloro-3-indolyl-β-D-glucuronide as described (Heckmann et al., 2011). After evacuation of air for 20LJmin, root samples were incubated in GUS staining solution overnight (at 37°C in the dark). The GUS-stained tissues were observed and photographed using a Motic Swift M200D compound microscope (Gifu B-129 roots inoculated with rhizobial strains harboring pT7-GUS) or a Nikon SMZ18 stereo microscope (Gifu pNIN:GUS inoculated with unlabeled rhizobia).

Acknowledgements

We greatly thank Prof. Zhongming Zhang for his valuable suggestions on this project. We thank Drs. Eric Giraud, Ertao Wang, and Yan Liang for critical reading of this manuscript. We thank Dr. Jens Stougaard for kindly providing L. japonicus pNIN:GUS and nfr mutant seeds. Dr. Juan Du is thanked for providing plasmids for expression of Avr3a and R3a, BAX, and INF1. The qPCR and microscopy data were acquired from the Core Facility Center at the National Key Lab of Agricultural Microbiology.

Funding

This work was supported by the National Key R&D Program of China (2019YFA0904700), the National Natural Science Foundation of China (32090063), and the Fundamental Research Funds for the Central Universities (2662022SKYJ002 and 2662021JC010). Work in the Stacey’s lab was supported by the NSF Plant Genome Research Program (2048410)..

Author Contributions

H.B., Y.C., and G.S. conceived the idea and designed the research, H.B., Y.W., Y.L., and Y.Y. performed experiments; H.B., Y.W., H. L., H.Z., and Y.C. analyzed the data; Q.W. and S.X. identified the cleavage site by MS. H.B., Y.C., S.F.W., C.S., and G.S. wrote the paper.

Supplemental Figures

Supplemental Fig. 1.

Supplemental Fig. 2.

Supplemental Fig. 3.

Supplemental Fig. 4.

Supplemental Fig. 5.

Supplemental Fig. 6.

Supplemental Fig. 7.

Supplemental Fig. 8.

Supplemental Fig. 9.

Supplemental Fig. 10.

Supplemental Fig. 11.

Supplemental Fig. 12.

Supplemental Fig. 13.

Supplemental Fig. 14.

Supplemental Fig. 15.

Supplemental Table 1.

Supplemental Table 2.