(This chapter has been published online in Systematic and Applied Microbiology, 2018.
DOI information: 10.1016/j.syapm.2018.05.001)
Abstract
The genetic diversity and identification of Bradyrhizobium symbionts of Crotalaria zanzibarica, the most widely-distributed invasive legume in Taiwan, and other sympatric legume species growing along riverbanks of Taiwan were evaluated.
This is the first study investigating the diversity of Bradyrhizobium symbionts in Taiwan. In total, 59 and 54 Bradyrhizobium isolates were obtained from C. zanzibarica and its coexisting legume species, respectively. Based on the multilocus sequence analysis (MLSA) of concatenated four housekeeping genes (dnaK-glnII-recA-rpoB gene sequences, 1901 bp), these isolates displayed 53 unique haplotypes, grouping into 21 clades. Eleven of these clades are congruent to already defined Bradyrhizobium species, while other clades are not congruent to any defined species. The C. zanzibarica isolates belong to 14 MLSA clades, six of which overlapped with the isolates of coexisting legumes. According to the sequences of their symbiotic nodA genes (555 bp) the isolates were classified into three known nodA clades, III.2, III.3 and VII and were further clustered into 10 groups. The C. zanzibarica isolates were clustered into 8 nodA groups, five of which overlapped with the isolates from coexisting legumes. The nodA genes of the isolates from native species were dominated by Asian origin, while those of C. zanzibarica by American origin. In conclusion, C. zanzibarica is a promiscuous host capable of recruiting diverse Bradyrhizobium symbionts, some of which are phylogenetically similar to the symbionts of coexisting legumes in Taiwan. However, there is no evidence that C. zanzibarica established symbiosis with more genetically diverse Bradyrhizobium communities than its coexisting legumes.
Introduction
Bradyrhizobium, a major symbiont of legume taxa distributed in tropical and sub-tropical regions, is the most abundant genus among the known rhizobial genera (Sprent 2007, 2009) and its symbiotic relationship with legume has been widely studied.
However, these studies have been mostly conducted in continental areas and relatively few on island regions. Among the limited reports, the genetic diversity of symbiotic rhizobia in islands was generally found more diverse than that in continental area. For example, Parker and Rousteau (2014) reported that Bradyrhizobium symbionts on the Caribbean island (Guadeloupe) were extremely diverse, which might be generated by multiple colonization events and substantial horizontal gene transfer of symbiotic genes.
Taiwan, a sub-tropical Pacific island, has more than 200 legume species (Huang, 1993), among which the milletoid clade is the most dominant group (127 species), followed by the genistoid clade with 26 species. About a quarter to one third (50-70 species) of the legume species recorded on the island are naturalized (Wu et al., 2003, 2010). The exotic legumes and their nodule symbionts could be co-introduced from native regions into invaded areas (Rodríguez-Echeverría, 2010; Crisóstomo et al., 2013;
Horn et al., 2014). Thus, the introduced legume species might play important roles in affecting the soil rhizobia community of the island. Crotalaria zanzibarica Benth., a perennial leguminous shrub native to Africa, was first recorded in 1934 in Taiwan. After its introduction, the species became the most widely-distributed naturalized legume and has been evaluated as a naturalized legume with the highest invasiveness among the naturalized legumes in Taiwan (Wu et al., 2010). Factors contributing to its widely distribution in Taiwan have not been studied. This plant can be found growing along roadsides, in abandoned fields, but mainly along riverbanks, where there is low soil
fertility. Legume-rhizobia symbiosis has been suggested as one of the critical factors helping exotic legumes colonize novel ranges (Richardson et al., 2000b; Parker et al., 2006; Klock et al., 2015). The capability of forming effective symbioses with diverse rhizobia further plays an important role in increasing the distribution of widespread legume species (Thrall et al., 2000; Klock et al., 2015). Accordingly, C. zanzibarica might have the ability to recruit diverse local rhizobia after its introduction into Taiwan hence establishes symbiosis with more genetically diverse rhizobia communities than other exotic legume plants. Alternatively, exotic rhizobia might have been introduced accompanying the introduction of this plant into Taiwan. Consequently, the symbiotic relationship allows C. zanzibarica to thrive in the nitrogen-poor habitats, such as riverbanks.
In previous studies, I investigated nodule symbionts of C. zanzibarica and demonstrated that this plant was nodulated by Bradyrhizobium in the greenhouse condition and in riverbank of Northern Taiwan (Huang et al., 2016, 2018). This study expands previous studies investigating the genetic diversity and identification of Bradyrhizobium symbionts of C. zanzibarica growing along banks of three major rivers of Taiwan. Though Bradyrhizobium is known to be the most common rhizobium nodulating milletioid and genistoid legumes (Sprent, 2007, 2009), the two most dominant legume groups in Taiwan, the diversity and origins of Bradyrhizobium symbionts have not been investigated in field growing legumes. To infer the possible sources of the Bradyrhizobium symbionts of C. zanzibarica and to increase our understanding of the diversity of Bradyrhizobium in Taiwan, I also sampled native and other exotic legumes, growing adjacent to C. zanzibarica, known to form symbiosis with Bradyrhizobium. To the best of our knowledge, this is the first study investigating
the diversity of Bradyrhizobium symbionts of legumes in Taiwan.
According to the aforementioned background, the objectives of the study were to investigate the diversity of Bradyrhizobium symbionts in Taiwan and to trace possible sources of the Bradyrhizobium symbionts of C. zanzibarica. Specifically, phylogenetic relationships and taxonomic identities of the isolates were analyzed utilizing the multilocus sequence analysis (MLSA) technique with four housekeeping genes, dnaK, glnII, recA and rpoB genes. To infer the possible geographic sources and potential hosts of the isolates, the phylogeny of the symbiotic nodA gene from these isolates was also studied.
Materials and methods
Sampling sites, nodule collection and rhizobial isolation
Xiandan (XD, 24°98′N, 121°52′E), Dajia (DJ, 24°25′N, 120°82′E) and Gaoping (GP, 22°77′N, 120°45′E) Rivers are located in Northern, Central and Southern Taiwan, respectively. Along the bank of each of the three rivers, I selected one sampling area (about 10 × 150 m) containing a population of C. zanzibarica. Soils of the three sites were mostly sandy with slight acidity (pH 5.8 to 6.3) and low electronic conductivity (0.04 to 0.08 S/cm). Legume plants were prevalent but the composition of the legume communities differed among the three sampling sites. In addition to analyzing the nodule symbionts of C. zanzibarica, I also investigated those of the sympatric legume species, occurring at a distance less than 1 m from C. zanzibarica plants, known to associate with Bradyrhizobium. In total, 11 legume species, including plants of distinct tribes and originating from disparate sources, were sampled (Table 3-1).
At each of the three sampling sites, ten to fourteen individuals of C. zanzibarica
were excavated, and 2-5 nodules per individual were collected. For the coexisting legumes, nodules (1-3 nodules per individual) were collected from six individuals of each species. A total of 199 root nodules (92 from C. zanzibarica and 107 from other legumes) were collected for rhizobial isolation.
Fresh nodules were surface sterilized by immersion in 0.5 % SDS for 1 min, 70 % ethanol for 5 min, and finally washed three times by sterile deionized-distilled water (DDW). Nodule suspension was prepared by crushing the nodule in DDW and spread onto yeast extract mannitol (YEM) plate (Vincent, 1970). A single rhizobial isolate was obtained from each nodule and checked for the unity by repeated streaking on YEM plate.
Amplification, sequencing and DNA polymorphism of five genetic markers
Total genomic DNA of each pure rhizobial culture, grown in YEM broth at late exponential phase, was extracted with Geneaid DNA Mini kit (Geneaid Biotech, New Taipei, Taiwan). The primer pairs and PCR conditions for amplifications of the four housekeeping genes (dnaK, glnII, recA and rpoB) and the symbiotic gene, nodA are listed in Supplementary Table S1. PCR mixtures were made up with Taq DNA Polymerase 2x Master Mix RED (Ampliqon, Copenhagen, Denmark) following by the standard protocol. PCR products were first checked on 1.5% agarose gel and purified with Gel/PCR DNA fragments extraction kit (Geneaid Biotech, New Taipei, Taiwan).
All sequencing reactions were performed by using the ABI 3730 DNA sequencer (Applied Biosystems, Foster City, CA, USA). For checking the quality, chromatogram of each sequence was examined by using BioEdit 7.2.5 (Hall, 1999). GenBank accession numbers of the five genes sequences of these isolates are provided in
Supplementary Table S2.
Sequences for each of the five genes were aligned by using MUSCLE (Edgar, 2004) in MEGA version 6 (Tamura et al., 2013). To analyze the sequence polymorphisms, haplotype diversity (Hd) and nucleotide diversity (π, the average number of nucleotide differences per site) for each of the five genes was estimated by using software DnaSP5.10 (Librado and Rozas, 2009). Additionally, nucleotide diversity for synonymous and nonsynonymous substitutions of each of the five genes was also calculated.
Phylogenetic identification of Bradyrhizobium isolates by using four housekeeping genes
For species identification, the dnaK, glnII, recA and rpoB gene sequences of type strains of all 39 validly published Bradyrhizobium species were download from the NCBI GenBank database (https://www.ncbi.nlm.nih.gov/). However, due to lack of rpoB gene sequences, four type strains of B. erythrophlei, B. ferriligni, B. ganzhouense and B. lupini were excluded from the phylogeny reconstruction. In the preliminary analysis, B. oligotrophicum, B. denitrificans, B. jicamae, B. lablabi, B. retamae, B.
icense, B. valentinum and B. paxllaeri are not closely related to our isolated strains (data not shown). For brevity, these 8 species were also excluded from further analysis.
Therefore, a total of 27 type strains of Bradyrhizobium species were used as the reference strains. The GenBank accession numbers of the four housekeeping genes of the reference strains are listed in Supplementary Table S3. The homologous sequences from Rhodopseudomonas palustris BisB5 (NC_007958) were used as an outgroup for each of the gene phylogenies.
Maximum likelihood (ML) phylogenetic trees and best fit substitution models of these four housekeeping genes were reconstructed and evaluated using the software MEGA6. The best substitution models of four gene sequences were listed as the following. The Tamura-Nei model plus Gamma rate distribution (TN93 + G) was for dnaK sequences; The Tamura-Nei model plus Gamma rate distribution and invariant site (TN93 + G + I) was for glnII sequences; The Tamura 3-parameter model plus Gamma rate distribution and invariant site (T92 + G + I) was for both recA and rpoB sequences.
To visualize the conflicting phylogenetic signals among the four housekeeping genes, I concatenated dnaK, glnII, recA and rpoB gene sequences of 113 isolates and performed NeighborNet analysis by using SplitsTree v4.13.1 (Huson and Bryant, 2006). Based on a network for the concatenated dataset (Supplementary Figure S1), multiple reticulations indicated that horizontal gene transfer of housekeeping genes might occur among these isolates, which was consistent with previous findings (Andam and Parker, 2008; Koppell and Parker, 2012). Due to single gene phylogenies usually failing provide sufficient information for taxonomic conclusions of Bradyrhizobium (Menna et al., 2009; Rivas et al., 2009; Azevedo et al., 2015), the concatenated sequence for the dnaK, glnII, recA and rpoB genes was also used to reconstruct an ML tree, as has been suggested from previous studies.
An ML tree based on the concatenated dnaK-glnII-recA-rpoB gene dataset was reconstructed under the model General Time Reversible plus Gamma rate distribution and invariant site (GTR + G + I). To test the strength of the phylogeny of all ML trees, the bootstrap method based on 1000 replicates was used. Approximate likelihood ratio test (aLRT) was used to assess robustness of ML tree in PhyML3.0 (Guindon et al., 2010).
In addition, the concatenated dnaK-glnII-recA-rpoB gene tree was assessed by Bayesian Inference (BI) with MrBayes version 3.2.2 (Ronquist et al., 2012) using the nucleotide substitution model GTR + G + I. For analysis, I used 3,000,000 Markov Chain Monte Carlo (MCMC) generations and trees were sampled every 250 generations.
Posterior probabilities were calculated by sampling 250 post-burnin trees.
nodA phylogeny and possible origin of isolates
In order to infer the possible origin of the isolates, the nodA gene sequences of reference strains similar to those of the isolates were download from NCBI GenBank based upon BLAST results. In addition, the sampling information (host and geographic origin) of each reference strain was also retrieved from NCBI GenBank and published literatures. The nodA gene sequence from Methylobacterium nodulans ORS2060 (AF266748) was used as the outgroup. The nodA phylogeny among the isolates and reference strains was assessed using both ML and BI analyses. The ML tree was reconstructed under T92 + G model with 1,000 bootstrap replicates. Bayesian analysis was assessed by using the Hasegawa–Kishino–Yano model plus Gamma rate (HKY + G) and run for 1,000,000 MCMC generations, sampling every 250 generations with a relative burnin of 25%.
Results
DNA polymorphism patterns of Bradyrhizobium isolates
A total of 113 putative rhizobial isolates were obtained from the root nodules of 11 legume species growing along the three riverbanks (Table 3-1). Among these isolates, 59 were collected from C. zanzibarica. All isolates formed visible colonies on YEM
plates after 5-7 days of incubation at 30℃, consistent with the slow-growing phenotype of Bradyrhizobium.
Partial sequences of the four housekeeping genes, dnaK (224 bp), glnII (539 bp), recA (527 bp) and rpoB (611 bp), were obtained from all 113 isolates. Among these isolates, eight isolates (CzHD6, ZAHD2, CzDJD1, CzDJG1, CmDJA1, AaGPB1, CeGPD and CeGPC) did not form specific PCR product of nodB-D box, consequently, the partial nodA gene (555 bp) was only sequenced in 105 isolates. Table 2 shows the haplotype (Hd) and nucleotide diversity (π) of the C. zanzibarica isolates and those of isolates from other coexisting legumes at each of the sampling sites. In general, haplotype diversity calculated with the five genetic markers are relatively similar within the analyzed groups, while among the five markers, symbiotic nodA gene displayed much higher nucleotide diversity than the four housekeeping genes. This result was not only due to high synonymous substitution of nodA gene, but also caused by considerable variation in non-synonymous sites of this gene (Table 3-2). The level of DNA polymorphism of isolates from C. zanzibarica was either similar to or lower than that of isolates from other coexisting legumes at the same sampling site. Hence, there is no evidence that C. zanzibarica established symbiosis with more genetically diverse Bradyrhizobium communities than its coexisting legumes.
Phylogenetic analysis and taxonomic identification by using four housekeeping genes Thirty-two, 46, 41 and 43 haplotypes of dnaK, glnII, recA and rpoB genes, respectively, were identified among the 113 isolates. The ML trees reconstructed individually with each of the four housekeeping genes (Supplementary Figures S2 to
S5) rarely provided well-supported resolution among the haplotypes of isolates and the
27 named Bradyrhizobium taxa. On the other hand, concatenated dnaK-glnII-recA-rpoB gene sequences (1901 bp) of the 113 isolates displayed 53 unique multilocus haplotypes, designed as MLSA-hap 1-53. ML analysis of the combined sequences dataset produced more robust and well-supported phylogeny for these Bradyrhizobium isolates (Figure
3-1). The resulting phylogenetic tree contained two major clades (A and B), including
the type strains of B. japonicam and B. elkanii, commonly referred to as B. japonicam and B. elkanii lineages, respectively (Vinuesa et al., 2008; Rivas et al., 2009). Most of the isolates in this study were situated in B. japonicam lineages (Clade A), and further split into 18 terminal clades (defined as A.1-A.18). Only 8 isolates, all collected from the legume hosts of tribe Desmodieae, belonged to the B. elkanii lineages (Clade B.1-B.3). Several clades formed well-supported groups (ML aLRT confidence test values > 0.7, ML bootstrap support > 70% and Bayesian inference posterior probability> 0.8) with named Bradyrhizobium strains. Accordingly, Clade A.1-A.5 were classified as B. yuanmingense, Clade A.7 as B. liaoningense, Clade A.12 as B. daqingense, Clade A.15 as B. arachidis, Clade A.18 as B. manausense, Clade B.1 as B. pachyrhizi and Clade B.2 as B. elkanii. Other clades were situated outside the defined strains. These clades might represent novel genospecies.
The number of isolates belonging to each of the 21 MLSA clades varied from 1 to 35 (Table 3-3). Among the 21 clades, 14 were limited to a single sampling site, 6 were found in two sampling sites, and only one clade, clade A.3, extended across all three sites. Clade A.3 was the most abundant clade containing 31% of the isolates. At the XD site, most of the isolates from C. zanzibarica and its coexisting legumes were separated into non-overlapping clades, with one exception of clade A.8. At the DJ site, 3 clades, A.2, A.3 and A.10 were shared by C. zanzibarica and its coexisting legumes. In contrast,
all the four clades (A.1, A.2, A.3 and A.12) found in GP site were shared by C.
zanzibarica and its coexisting legumes (Table 3-3).
nodA gene phylogeny and possible origins
ML analyses of the nodA gene sequences of the present isolates and reference strains produced a well-resolved tree with three major clades (Figure 3-2). Each clade contained previous published Bradyrhizobium nodA gene sequences, and thus, the three clades could be identified as nodA sub-clades III.3, III.2 and clade Ⅶ (Moulin et al., 2004; Stępkowski et al., 2005, 2007). More than half of the present isolates (64 out of 105 isolates) belonged to a highly diverse clade, nodA sub-clade III.3, and were further divided into 8 groups (assigned as III.3a-h). The nodA Group III.3a included 27 isolates from diverse sources, including isolates from three sampling sites and from both native and exotic legumes. This nodA group clustered with the reference strains originating from Thailand, India, Japan and China (Figure 3-2), indicating these nodA gene sequences were prevalent in Asia. In both nodA Group III.3c and III.3d, all isolates were collected from C. zanzibarica and were exclusively grouped together with multiple peanut symbionts in China (Figure 3-2). On the other hand, isolates in the nodA Group III.3b, III.3e and III.3g were mainly C. zanzibarica symbionts and clustered with the reference strains from distant geographic regions, hence their specific origins could not be resolved (Figure 3-2). One C. zanzibarica isolate, strain CzDJC1 was not grouped together with any reference strain, forming nodA Group III.3f (Figure 3-2). In nodA Group III.3h, all isolates belonged to B. elkanii lineage (Clade B, Figure 3-1) and clustered with B. elkanii USDA76 and reference strains from diverse sources (Figure
3-2).
In addition, 38 isolates were classified as nodA sub-clade III.2 strains which were dominant by C. zanzibarica isolates (26 isolates) and were found in all three sampling sites (Figure 3-2 and Table 3-4). Consistent with previous reports, all reference strains in nodA sub-clade III.2 were from the Americas (Moulin et al., 2004; Stępkowski et al., 2005, 2007). Particularly, the nodA gene sequences of these isolates were nearly identical to that of strain NC92, which was a peanut symbiont originated in Bolivia (Urtz and Elkan, 1996). Furthermore, it had been reported that the strains belonged to nodA clade VII exclusively originated from Central and South America (Stępkowski et al., 2007), however, our data showed that 3 isolates from Desmodium spp. and reference strains from Australia and China were also situated in this clade (Figure 3-2).
The nodA groups of isolates from C. zanzibarica and other legumes were rarely, partially and totally overlapping in XD, DJ and GP sites, respectively (Table 3-4).
Discussion
C. zanzibarica was nodulated by diverse Bradyrhizobium genospecies
Aserse et al. (2012) reported that several Crotalaria spp. growing in their native regions (Ethiopia) were associated with diverse Bradyrhizobium symbionts. A non-classical rhizobium, Methylobacterium nodulans, was also reported to nodulate native Crotalaria plants in Senegal (Jourand et al., 2004). In this study, I found that the 59 isolates collected from exotic C. zanzibarica in Taiwan all belonged to the genus Bradyrhizobium, in contrast, I did not find any Methylobacterium isolate. Thirty-four C.
zanzibarica isolates in MLSA clades A1-A5 formed a well-supported group with B.
yuanmingense 10071T (ML bootstrap support = 77% and BI posterior probabilities = 1,
Figure 3-1). Accordingly, B. yuanmingense appears to be a dominant species of C.
zanzibarica symbionts in Taiwan. Besides, several other species were also identified among the C. zanzibarica isolates, including B. liaoningense (2 isolates), B. daqingense (4 isolates), B. arachidis (1 isolate) and B. manausense (4 isolates). Among the defined species, B. yuanmingense (Yao et al. 2002) and B. liaoningense (Xu et al., 1995) were first described in China and had been reported as legume symbionts widely across distant geographic regions, such as Africa (Aserse et al., 2012; Grönemeyer et al. 2014), America (Koppell and Parker, 2012), Asia (Vinuesa et al., 2008; Noisangiam et al., 2012) and Australia (Stępkowski et al., 2012). B. arachidis (Wang et al., 2013), B.
daqingense (Wang et al., 2013) and B. manausense (Silva et al. 2014) were identified more recently, and thus, information about these species is relatively limited. Although most of the C. zanzibarica isolates (45 out of 59 isolates) analyzed in this study were closely related to defined species, the remaining 14 isolates in MLSA clades A.8, A.9, A.10, A.11, A.14 and A.16 might represent novel Bradyrhizobium genospecies (Figure
3-1). Particularly, clade A.10, comprising 8 isolates collected from two distant C.
zanzibarica populations and two isolates from Crotalaria micans and Indigofera spicata, is a strongly supported clade and distinct from any known species. Thus, the genospecies, clade A.10, might be well adapted to diverse soil conditions and legume hosts in Taiwan, and is therefore worth further analyzing for proposal of a novel species.
Multiple Bradyrhizobium lineages were observed in all three sampling sites (Table
Multiple Bradyrhizobium lineages were observed in all three sampling sites (Table