Mutualistic co‐evolution of T3SSs during the establishment of symbiotic relationships between Vigna radiata and Bradyrhizobia

Abstract This study supports the idea that the evolution of type III secretion system (T3SS) is one of the factors that controls Vigna radiata–bradyrhizobia symbiosis. Based on phylogenetic tree data and gene arrangements, it seems that the T3SSs of the Thai bradyrhizobial strains SUTN9‐2, DOA1, and DOA9 and the Senegalese strain ORS3257 may share the same origin. Therefore, strains SUTN9‐2, DOA1, DOA9, and ORS3257 may have evolved their T3SSs independently from other bradyrhizobia, depending on biological and/or geological events. For functional analyses, the rhcJ genes of ORS3257, SUTN9‐2, DOA9, and USDA110 were disrupted. These mutations had cultivar‐specific effects on nodulation properties. The T3SSs of ORS3257 and DOA9 showed negative effects on V. radiata nodulation, while the T3SS of SUTN9‐2 showed no effect on V. radiata symbiosis. In the roots of V. radiata CN72, the expression levels of the PR1 gene after inoculation with ORS3257 and DOA9 were significantly higher than those after inoculation with ORS3257 ΩT3SS, DOA9 ΩT3SS, and SUTN9‐2. The T3Es from ORS3257 and DOA9 could trigger PR1 expression, which ultimately leads to abort nodulation. In contrast, the T3E from SUTN9‐2 reduced PR1 expression. It seems that the mutualistic relationship between SUTN9‐2 and V. radiata may have led to the selection of the most well‐adapted combination of T3SS and symbiotic bradyrhizobial genotype.

The mung bean (Vigna radiata) is cultivated mostly in South, East, and Southeast Asia by smallholder farmers for its edible seeds and sprouts. The domestication of the mung bean was initiated in the northeast and far south of India approximately 4,000-6,000 years ago (Fuller, 2007). The domesticated mung bean is thought to have spread mainly throughout Southeast Asia from India via different routes (Tomooka, Lairungreang, Nakeeraks, Egawa, & Thavarasook, 1992). Bradyrhizobia are commonly found to establish symbiotic interactions with V. radiata in Thailand (Piromyou et al., 2017;Yokoyama et al., 2006); in particular, Bradyrhizobium sp. SUTN9-2 can form symbiotic nodules with many of the V. radiata cultivars tested.
In contrast, several T3Es from bradyrhizobial strains are major negative effectors for V. radiata symbiosis (Nguyen, Miwa, Kaneko, Sato, & Okazaki, 2017;Songwattana et al., 2017;Wenzel, Friedrich, Göttfert, & Zehner, 2010). Thus, SUTN9-2 is a good model for the symbiotic partnership between V. radiata and Bradyrhizobium. Nevertheless, the genetic basis of how T3SSs are involved in the enhancement and suppression of nodulation in both partners in V. radiata bradyrhizobia symbiotic relationships has not been clearly elucidated. Therefore, the current study is an important step toward understanding the functions of T3SSs in mutualistic relationships.

| RE SULTS AND D ISCUSS I ON S
To examine the evolutionary relationships of bradyrhizobial strains, the nucleotide sequences of the 16s rRNA gene from various reference strains of Bradyrhizobium, Sinorhizobium, Mesorhizobium, and Rhodopseudomonas species were used to construct a phylogenetic tree. Cupriavidus taiwanensis LMG19424 was chosen as the outgroup strain to root the phylogenetic tree ( Figure A1). The DNA sequences were generated, and the most closely related sequences were obtained from the NCBI database. The nucleotide sequences were aligned using the ClustalW program, and the phylogenetic trees based on the 16S rRNA and T3SS gene sequences were constructed using the maximum-likelihood method with PhyML (Guindon & Gascuel, 2003). Based on the 16S rRNA gene sequence similarity, the phylogenetic tree could be divided into two major clusters. Cluster 1 included various groups of Bradyrhizobium species, including non-photosynthetic bradyrhizobia (non-PB), photosynthetic bradyrhizobia (PB), B. elkanii species, and Rhodopseudomonas members. The bradyrhizobial strains SUTN9-2, DOA1, DOA9, ORS3257 and B. diazoefficiens USDA110 belonged to the non-PB cluster. Moreover, strain SUTN9-2 was closely related to strain ORS3257. Both sinorhizobial and mesorhizobial species were located in major cluster 2 of the phylogenetic tree.
A phylogenetic tree based on sequences of the T3SS gene rhcJ was also constructed ( Figure A2). The rhcJ genes of strains ORS3257, DOA1, DOA9, and SUTN9-2 were obviously distinct from those in the non-PB and PB clusters and the B. elkanii group. To more clearly understand the evolution of T3SSs among bradyrhizobial strains, the genomic arrangements of several T3SS clusters were determined using the program GenomeMatcher (Ohtsubo, Ikeda-Ohtsubo, Nagata, & Tsuda, 2008) at the amino acid level (Figure 1).
The annotated genome sequences of USDA110 (accession number were available in the DDBJ/GenBank/EMBL database. The genome sequence of ORS3257 was received from the MicroScope platform (Vallenet et al., 2016). The T3SS gene clusters were separated into three clusters based on their T3SS structural components ( Figure 1a). The T3SS of bradyrhizobia displayed more notable differences compared with those of S. fredii NGR234 and Mesorhizobium loti MAFF303099. Region I of the bradyrhizobial T3SS cluster was similar to those found in all of the selected bradyrhizobial strains.
However, the T3SS gene organization in region II from the Thai strains and ORS3257 was partially different from that from USDA61 and USDA110. A distinct feature of the USDA110 T3SS cluster was the presence of several open reading frames (ORFs) that were absent in the other bradyrhizobial species. Furthermore, region II lacked nopX and no homologue was present in the USDA110 genome. The T3SS region III clusters of the bradyrhizobial strains were diverse.
The T3Es nopE1 and nopE2 were detected only in USDA110, whereas nopM and nopX were absent in its genome (Figure 1b). Several putative T3Es could not be found in DOA9 and SUTN9-2. It seems that the variation in the T3Es was higher than that in the T3SS structural core components. Based on the phylogenetic tree data and the gene arrangements, the T3SSs of the Thai strains (SUTN9-2, DOA1, and DOA9) and ORS3257 may share the same origin. Thus, the bradyrhizobial strains SUTN9-2, DOA1, DOA9, and ORS3257 evolved their T3SSs independently of the other bradyrhizobia because of biological and/or geological events. If this scenario is true, then it could be hypothesized that horizontal gene transfer (HGT) plays an important role in bradyrhizobial evolution. The T3E genes nopP and nopL were present in all of the bradyrhizobial strains, whereas homologues of these genes are not present in phytopathogenic bacteria (Ausmees et al., 2004;Bartsev, Boukli, Deakin, Staehelin, & Broughton, 2003;Bartsev et al., 2004). These data may also indicate that bradyrhizobia developed their T3Es independently of phytopathogenic bacteria. Therefore, these results may provide evidence confirming bradyrhizobial evolution.
Previous reports showed that Vigna are always found to establish symbiotic interactions with Bradyrhizobium spp. (Appunu, N'Zoue, Moulin, Depret, & Laguerre, 2009;Yokoyama et al., 2006;Zhang et al., 2008) and that bradyrhizobial strains have been isolated from various V. radiata ssp. in Thailand (Yokoyama et al., 2006). Furthermore, V. radiata has been continuously cultivated in Thailand; thus, it is a good plant model for understanding mutualistic properties of Thai bradyrhizobial strains. To obtain more information about the co-evolution between V. radiata and bradyrhizobial T3SSs, the bradyrhizobial strains SUTN9-2 (a native V. radiata symbiont in Thailand), DOA9 (a legume broad host range strain), ORS3257 (a native V. unguiculata symbiont in Senegal; Krasova-Wade et al., 2003), and USDA110 (a Glycine max symbiont) were used in these experiments (Table A11). In our preliminary study of T3SS functions, the expression of rhcN was measured after induction with genistein (20 μM genistein dissolved in DMSO) and mung bean root exudate inductions (1/3 (v/v) of the root exudates). Primers for amplification are listed in Table A22. The effects of T3SS on Bradyrhizobium-V. radiata ssp. symbiosis could be separated into two groups ( Figure A3). The V. radiata cv. KPSII (representing the incompatible group), in which the T3Es from all of the tested bradyrhizobial strains (except SUTN9-2) strongly inhibited nodulation, and V. radiata cv. CN72 (representing the compatible group), in which it seems that the T3Es did not have negative effects on symbiosis (except the T3Es from ORS3257), were selected for root exudate preparation. The results of this experiment revealed that rhcN expression was activated by genistein and V. radiata root exudates in all of the tested bradyrhizobia 12 hr after induction ( Figure   A3). These data also raise the possibility that the T3SS machinery is formed during the early steps of the interaction between bradyrhizobial strains and these two V. radiata cultivars. Therefore, the T3SS is Boxes are color-coded as indicated in the key; white boxes: no detectable homology; yellow boxes: putative T3Es; green boxes: hypothetical proteins; blue boxes: conserved structural components, and gray boxes: transcriptional regulator one of the factors that controls the V. radiata bradyrhizobia interaction at every step (Krause, Doerfel, & Göttfert, 2002).
To examine whether the tested bradyrhizobial strains were V. radiata symbionts, they were inoculated into various V. radiata hosts (Table A33). The injectisome mutant strains ORS3257 (ORS3257 ΩT3SS: rhcN disruption; Okazaki et al., 2016), SUTN9-2 (SUTN9-2 ∆T3SS: rhcJ deletion; Piromyou et al., 2015), DOA9 (DOA9 ΩT3SS: rhcN disruption) (Songwattana et al., 2017), and USDA110 (USDA110 ∆T3SS: nolB, rhcJ, nolU, and nolV deletion; Krause et al., 2002) were used in this experiment (Table A11). It seems that the T3SSs of strains ORS3257, DOA9, and USDA110 showed negative effects on V. radiata (incompatible) symbiosis, but not on the V. radiata (compatible) group ( Figure 2). Interestingly, the ORS3257 wild-type strain could not form nodules in V. radiata ssp., whereas the ORS3257 ΩT3SS F I G U R E 2 Nodulation and plant growth promotion by Vigna radiata cv. KPSII and V. radiata cv. CN72 inoculated with wild-type (WT) and the three mutant TTSS strains. Total dry weights (a, and d), nodule number (b and e), and nitrogen fixation (c and f) are shown for the two different cultivars (abc, KPSII; def, CN72). Significance at p < 0.05 is indicated by the means and standard deviation bars (n = 3) could not be detected after inoculated with ORS3257, while DOA9 could form a small number of nodules (approximately 5 nodules per plant). The nitrogen fixation of DOA9 was also lower than that resulting from SUTN9-2 and USDA110 inoculations ( Figure 2f). Based on the results with the T3SS mutants, it seems that bradyrhizobial T3SSs were less important for V. radiata bradyrhizobia symbiosis (except for V. radiata CN72). However, all of the tested bradyrhizobial strains still maintained the T3SS injectisome in their genomes, while they evolved their T3Es independently of other bradyrhizobia.
Perhaps the bradyrhizobial T3SS was important for symbiosis with other legumes Viprey et al., 1998). Therefore, some T3Es from ORS3257 and DOA9 showed negative effects on nodulation efficiency. Furthermore, the nodulation results implied that the mung bean cultivar is one of the factors that controls the compatibility of V. radiata bradyrhizobia symbiosis. This phenomenon reflects the bradyrhizobial host specificity of Thai bradyrhizobial strains for Thai V. radiata ssp. cultivars ( Figure 2 & Table A33). In addition, the current symbiotic state of SUTN9-2 has perfectly adapted to every tested V. radiata ssp. cultivar, whereas DOA9 cannot form effective nodules in any of the tested cultivars.
The T3SS mutation experiments showed that the T3SSs had cultivar-specific effects on nodulation properties. Wild-type ORS3257 cannot form nodules with V. radiata cv. KPSII or cv. CN72, but both V. radiata cultivars readily formed nodules after ORS3257 ΩT3SS inoculation (Figure 2b,e). Similarly, DOA9 ΩT3SS had improved nodulation in V. radiata cv. KPSII and cv. CN72. Thus, the T3SSs of ORS3257 and DOA9 displayed negative effects on V. radiata cv.
KPSII and cv. CN72 nodulation. In the case of USDA110, the T3SS showed negative effects on V. radiata cv. KPSII but not on cv. CN72 (positive effect). The T3SS of SUTN9-2 had no effect on nodulation in both mung bean cultivars. Our results revealed that features of the T3SS seem to be important determinants of root nodule formation in V. radiata. To explore the T3SS-dependent regulation of V. radiata defense mechanisms, we compared the expression of the Pathogenesis-Related 1 (PR1) gene in V. radiata CN72 roots inoculated with the wild-type and T3SS mutant strains at 2 dai ( Figure A4). The PR1 gene was expressed at a very low level in the uninoculated control; however, the expression level was significantly enhanced by inoculation with most of the bradyrhizobial strains except SUTN9-2. In V. radiata CN72 roots, the PR1 gene expression levels following inoculation with wild-type ORS3257 and DOA9 strains were significantly higher than those after inoculation with the T3SS mutant strains  in the secretion of Nod factor, which is then tuned based on recognition by the V. radiata host (Geurts & Bisseling, 2002;Göttfert, Grob, & Hennecke, 1990;Long, 1996). Based on these phenomena, we hypothesized that Nod factor is the main factor required for V. radiata ssp. symbiosis; however, the T3SS is also important for host specificity. To better understand the relationship between the T3SS and the nodulation (nod) genes in V. radiata ssp. symbiosis, the Nod clusters were preliminary identified at the amino acid level using the GenomeMatcher program (Ohtsubo et al., 2008; Figure   A5). The Nod clusters of strains SUTN9-2 and ORS3257 were similar to that found in USDA110, whereas the Nod cluster of DOA9 was diverse. In addition, the nodulation-mutant strains SUTN9-2 (SUTN9-2 ΩnodABC) and DOA9 (DOA9 ΩnodB) lost their symbiotic properties with V. radiata ssp. (data not show). It seems that V. radiata ssp. were promiscuous plants with diverse Nod factors. These results were consistent with previous reports that V. radiata is one of the most promiscuous plants (Yokoyama et al., 2006;Zhang et al., 2008).
In addition, the SUTN9-2 T3Es are more likely to avoid recognition and/or suppression by the V. radiata defense mechanism. The T3Es of bradyrhizobial strains display mutualistic co-evolution with V. radiata ssp. Our data support the idea that mutualism can result in host specificity and that bradyrhizobial mutualists may be under pressure from the host that limits diversification. This model could explain why the T3Es and tts boxes are diverse among bradyrhizobial species. Therefore, this work will provide a very logical transition into further study of how variations in the specific T3E content contribute to immune recognition.
Since SUTN9-2 could nodulate all tested V. radiata cultivars, its T3Es seem to have effects on PR1 expression. This property might be linked to its nodulation competition. However, the function of T3SSs in enhancing competition is still a mystery. To assess the competition of nodulation among bradyrhizobial strains, the nodule number and nodule occupancy of each pair, with cross-inoculation with the same amount of living cells (10 6 CFU/ml), were carried out using Leonard's jar experiments (Figure 3). Since V. radiata cv. CN72 was more promiscuous than cv. KPSII (Figure 2), it was selected for this experiment.
Single inoculation with ORS3257 did not lead to the formation of symbiotic nodules in V. radiata cv. CN72, whereas nodulation without necrotic symptoms was detected after single inoculation with SUTN9-2. Ineffective nodules in V. radiata CN72 were mostly found after DOA9 inoculation (Figure 3a To determine the nodule occupancy, bacterial genomic DNA was directly extracted from the surfaces of sterilized nodules. Next, the bradyrhizobia inside the nodules were identified using BOXAIR1-PCR ( Figure A6 and Figure 4c; Versalovic, Schneider, Bruijn, & Lupski, 1994). The nodule occupancy of SUTN9-2 was not significantly different than resulting from single inoculation with SUTN9-2 or from any of the co-inoculations (Figure 3b). In contrast, the nodulation efficiency of DOA9 was entirely lost when it was co-inoculated with ORS3257. The nodule number derived from USDA110 was also significantly reduced in the ORS3257:USDA110 co-inoculation experiment. One possibility is that ORS3257 secreted some effector proteins that triggered the plant immunity and, consequently, the DOA9 and USDA110 strains also lost some of their capacity to form symbiotic nodules with V. radiata cv. CN72. On the other hand, the nodule number derived from SUTN9-2 was not reduced after co-inoculation with ORS3257. These results imply that SUTN9-2 could ignore the plant immune response stimulated by ORS3557 and/or that are still unclear; therefore, ORS3257 infection processes will further be explored.
To more clearly understand the function of the T3SS in SUTN9-2 nodulation, the competition for nodule formation (cross-inoculation) between wild-type and T3SS mutant strains was explored in V. radiata cv. CN72 (Figure 4). After single inoculation, SUTN9-2 and SUTN9-2 ∆T3SS could perform symbiotic nodules, whereas senescence nodules could be detected when V. radiata cv. CN72 was inoculated with DOA9 ( Figures 3 and 4). The DOA9 ΩT3SS strain could form significantly more nodules compared to DOA9. It seems that the T3Es of DOA9 could suppress V. radiata cv. CN72 nodulation. The nodule number derived from SUTN9-2 was not significantly different compared to that resulting from co-inoculation (SUTN9-2 with DOA9 and SUTN9-2 with DOA9 ΩT3SS). In contrast, the number of nodules derived from SUTN9-2 ∆T3SS was drastically reduced when it was co-inoculated with DOA9 root ( Figure A7b). H 2 O 2 production was clearly found at the junction of the lateral roots and root hair zone after each bradyrhizobial inoculation. The roots were strongly stained (brown color) after inoculation with DOA9, whereas the brown color was drastically reduced after inoculation with SUTN9-2. After co-inoculation (SUTN9-2:DOA9), the brown color seems paler than that caused by single DOA9 inoculation.
This result indicated that DOA9 strongly induced H 2 O 2 accumulation during the early stage of infection. However, H 2 O 2 accumulation was also detected following SUTN9-2 inoculation; therefore, the defense response was triggered transiently even during the compatible V. radiata bradyrhizobia interactions. In addition, SUTN9-2 could suppress H 2 O 2 production following co-inoculation. Based on these results, we could confirm that SUTN9-2 evolved its T3Es (signaling) to interact with V. radiata ssp. receptors and that these interactions likely weaken the plant immunity; therefore, some DOA9 cells take this opportunity to form V. radiata cv. CN72 root nodules (dual nodules; Figure 4 b,c).
Consequently, it seems that SUTN9-2 is the best-adapted strain for V. radiata symbiosis. However, the symbiotic mechanisms of SUTN9-2 are still partially unclear; therefore, the symbiotic relationships between SUTN9-2 and V. radiata will further be explored.
We assume that successful establishment of V. radiata-bradyrhizobia symbiosis depends on how the bradyrhizobia have adapted to the special conditions on and in the V. radiata ssp. roots. Perhaps the T3Es from ORS3257 and DOA9 are directly bound by plant receptor CC-nucleotide-binding sites (NBS-LRRs) inside the plant cells (Flor, 1971). In this situation, NBS-LRRs strongly induce the V. radiata ssp.
defense response, which ultimately blocks nodule formation. On the other hand, SUTN9-2 could ignore the plant immune response and/ or developed detoxification systems to overcome the effects of the plant defense mechanisms on nodule development. Nevertheless, the T3Es from USDA110 showed negative and positive effects on V. radiata ssp. symbiosis ( Figure 5). Therefore, the host legumes and/ or the environmental conditions are the main selective forces that drive the evolution of genes encoding functions involved in the symbiotic relationships of the microsymbionts. The mutualistic partnerships between V. radiata and their symbionts showed co-evolution between SUTN9-2 and V. radiata; thus, their mutualism may lead to selection of the most adapted combination of T3SS and symbiotic bradyrhizobial genotypes.

ACK N OWLED G EM ENT
This work was financially supported by Suranaree University of Technology (SUT). Appreciation is also extended to Issra Pramoolsook for advice and comments on the language of the manuscript.

CO N FLI C T O F I NTE R E S T
The authors declare that there is no conflict of interest.

AUTH O R S CO NTR I B UTI O N
P. P. and N. T. conceived and designed the experiment. P. P., P. S., and K. T. performed the experiment. P. P., and P. S analyzed the data.E.
G., M. G., and N. T. provided bacteria used in this experiment.P. A. T.
provided mung bean used in this experiment.P. P., P. T., N. B., and N.
T. contributed to the critical discussion about the results. P. P., and N. T. wrote the manuscript. All authors read and approved the final manuscript.

E TH I C S S TATEM ENT
This article does not contain any studies with humans or animals performed by any of the authors.

DATA ACCE SS I B I LIT Y
The