Monophyly of nodA and nifH Genes across Texan and Costa Rican Populations of Cupriavidus Nodule Symbionts

nodA and nifH phylogenies for Cupriavidus nodule bacteria fromnative legumes in Texas and Costa Rica grouped all strains intoa single clade nested among neotropical Burkholderia strains.Thus, Cupriavidus symbiotic genes were not acquired independentlyin different regions and are derived from other Betaproteobacteriarather than from ${alpha}$ -rhizobial donors.

INTRODUCTION

Top

INTRODUCTION

Some lineages of Betaproteobacteria appear to have acquiredthe capacity for symbiotic interaction with legume plants bylateral transfer of symbiosis-related genes from rhizobia inthe Alphaproteobacteria (4, 9). However, it is uncertain howmany lateral gene transfer events have occurred to establishsymbiotic lineages of ß-rhizobia. nodA phylogeniesfor Burkholderia indicate that strains from South Africa andSouth America cluster with two distinct lineages of Alphaproteobacteria(4), implying an independent acquisition of symbiotic genesin the two locations. Less information is available for legumesymbionts belonging to the genus Cupriavidus, which have sofar been found within the native range of their Mimosa hostspecies only in Costa Rica (3), although they also occur onMimosa species introduced into Taiwan (4, 6) and India (12).nod genes of Taiwanese Cupriavidus strains proved to be highlysimilar to those of a Taiwanese Burkholderia strain (4, 6).Because these sequences nested within a clade of several othernodule-forming Burkholderia strains, it appeared that theseCupriavidus nodA genes were acquired from a Burkholderia strainrather than directly from an alphaproteobacterial ancestor (4,6). However, because only three strains have been studied, froma symbiont population that is not known to be native to theregion, it is important to analyze whether the same conclusionshold for indigenous Cupriavidus populations associated withMimosa legumes in their native geographic ranges.

In this study, we analyzed nodule bacteria associated with thelegume Mimosa asperata in south Texas. M. asperata was formerlyconsidered to be a variety of the widely distributed invasiveweed Mimosa pigra (M. pigra var. berlandieri) (11) but is nowrecognized as a distinct species. The range of M. asperata iscentered in Mexico and extends slightly into south Texas, Cuba,and northern Central America (1). Central American populationsof M. pigra associate almost exclusively with strains of Burkholderia,although Cupriavidus is also present at a low frequency in CostaRica (2, 3). Thus, it is of interest to determine whether relatedlegumes in nearby regions harbor Cupriavidus strains. We alsoobtained a small number of nodule bacterial strains from a secondMimosa species native to Texas (Mimosa strigillosa) that isclassified as a close relative of M. asperata and M. pigra (1).

Eighty-one root nodules were collected from multiple plantsof M. asperata at two sites 1 km apart within Santa Ana NationalWildlife Refuge (Hidalgo County) in the lower Rio Grande RiverValley, Texas, and one bacterial isolate per nodule was obtainedas described previously (10). Eleven nodule isolates were alsoobtained from plants of M. strigillosa at two sites in the lowerRio Grande Valley. DNA purification and PCR experiments usedstandard protocols (3). BLAST searches on a 390-bp 5' portionof the 23S rRNA gene sequenced in 14 Texan M. asperata isolates(GenBank accession no. DQ533663 to DQ533681) indicated thatall had high affinities to various strains of Cupriavidus legumenodule symbionts. Eight distinct genotypes were observed amongthe 14 isolates, and all proved to be similar or identical (0to 4 substitutions) to Costa Rican Cupriavidus legume nodulesymbionts analyzed in a prior study (3).

None of the 11 nodule isolates obtained from M. strigillosayielded a 5' 23S rRNA PCR product of the same size as the Cupriavidusstrains. Three of these PCR products were sequenced, and theyproved to be similar either to Sinorhizobium arboris USDA 4878(AY244381; 95 to 96% similarity) or to Sinorhizobium saheliUSDA 4893 (AY244368; 95% similarity).

To screen the remaining 67 M. asperata isolates, PCR assayswere performed with two sets of lineage-specific primers. Primersralf2 and ralr2 have been shown to differentially amplify a187-bp portion of the 23S rRNA gene in nodule-forming Cupriavidus(3). All 67 isolates exhibited an amplification product withthese primers. A second pair of primers (ral16f and ral16r;Table 1) targeted a portion of the 16S rRNA gene characteristicof Cupriavidus and related genera such as Ralstonia. All 67isolates yielded an amplification product with these primersas well. Together, these PCR assays imply that these M. asperatanodule isolates are likely to be strains of Cupriavidus.

View this table:
[in this window]
[in a new window]

TABLE 1. Oligonucleotide primers used in this study

A 1,453-bp portion of the 16S rRNA gene was sequenced in sevenM. asperata strains. (Two cloned PCR products were sequencedseparately for one of the strains [AMP18] due to apparent sequenceheterogeneity among rRNA operons.) The eight sequences wereall >99% similar to one another, differing by 1 to 13 nucleotides(GenBank accession no. DQ530643 to DQ530650). The two clonedPCR products from strain AMP18 differed by one nucleotide substitutionand one 2-bp insertion/deletion polymorphism. Maximum likelihoodanalysis of aligned sequences showed that all of the Texan Cupriavidusstrains formed a well-supported group (61 to 96% bootstrap values)together with other Cupriavidus nodule symbionts from CostaRica and Taiwan (data not shown). One Texan strain proved tohave a 16S rRNA sequence identical to that of a Cupriavidusstrain sampled from Mimosa pudica in Costa Rica (3). The closerelationship between Cupriavidus strains in sites 2,300 km apartin Texas and Costa Rica suggests that certain bacterial lineagesmay have spread across this region recently relative to thetime scale of nucleotide substitution in this gene. Standardestimates of substitution rates in 16S rRNA genes (7) implythat the upper 95% confidence limit on divergence time for twostrains that share an identical 1,453-bp portion of 16S rRNAranges from 5.7 to 8.6 million years ago. This places an approximateupper bound on the time to common ancestry for Texan and CostaRican strains.

A 488-bp portion of the nodA gene was amplified and sequencedin seven Texan and five Costa Rican Cupriavidus strains (chosento represent all major lineages detected in a prior study) (3),using primers nodAcf and nodAcr (Table 1) designed from Cupriavidustaiwanensis sequences (4, 6). Four Texan strains (AMP6, AMP18,AMP45, and CMP29) shared one sequence, which differed by 15nucleotides from a sequence shared by the other Texan strains(CMP2, CMP52, and CMP60). The five Costa Rican strains werehighly similar, with nodA sequences that varied at 0 to 2 nucleotidesites. C. taiwanensis strains LMG19424 and LMG19425 differedby 1 or 2 nucleotides from the Costa Rican strains. Phylogeneticanalysis of the nodA sequences indicated that all Cupriavidusnodule symbionts formed a highly supported group (97 to 100%bootstrap value) together with Burkholderia caribensis strainTJ182 (Fig. 1). Within this group, there was another stronglysupported lineage comprised of C. taiwanensis strains LMG19424and LMG19425, three Texan strains, and the five Costa Ricanstrains. The Cupriavidus nodA sequences nested within a largerclade encompassing all South American Burkholderia sequences(five sequences plus Burkholderia strain PAS44 from a Mimosahost introduced to Taiwan). These results are consistent withthe interpretation that Cupriavidus nodA genes had a singleorigin and may be descended from a South American Burkholderiaancestor.

View larger version (27K):
[in this window]
[in a new window]

FIG. 1. Maximum likelihood phylogeny for nodA genes from Texan Cupriavidus strains (shown in bold) and other proteobacteria. The maximum likelihood model used a transition/transversion ratio estimated from the data, empirical base frequencies, and substitution rates estimated separately for each codon position. Numbers near branches are bootstrap percentages for parsimony (top), neighbor-joining (middle), and likelihood (bottom) analyses. For Cupriavidus and Burkholderia nodule bacteria, the place of origin is indicated (CR, Costa Rica; Fr. Guiana, French Guiana; Ven., Venezuela).

A 473-bp segment of the nifH gene was amplified using primersnhcf3 and nhcr4 (Table 1) designed from C. taiwanensis sequences(4, 6). Patterns of sequence variation in the nifH gene werehighly similar to the nodA results (GenBank accession no. EF374065to EF374088). Four Texan strains (AMP6, AMP18, AMP45, and CMP29)shared an identical sequence, which differed by 12 or 13 nucleotidesfrom the rest of the Texan and Costa Rican strains. All Cupriavidusstrains from Costa Rica were found to be identical to one anotherand also to C. taiwanensis strains LMG19424 and LMG19425. Phylogeneticanalysis of the nifH sequences (Fig. 2) indicated that all Cupriavidusnodule symbionts formed a highly supported lineage (100% bootstrapvalue). Within this lineage, there was the same subgroup evidentin the nodA tree composed of C. taiwanensis strains LMG19424and LMG19425, three Texan strains, and the five Costa Ricanstrains. In contrast to the nodA tree, the nifH sequence ofBurkholderia caribensis strain TJ182 was grouped with SouthAmerican Burkholderia strains (5) rather than within the Cupriavidusclade. Apart from the altered placement of B. caribensis TJ182,the Cupriavidus portions of the tree topology were highly similarfor nodA and nifH phylogenies (Fig. 1 and 2). A partition homogeneitytest (8) on all nodule-forming Burkholderia and Cupriavidusstrains other than TJ182 indicated that there was no evidencefor a difference in nodA and nifH tree topologies (P = 0.99).Thus, both genes appear to share a common genealogical history,and the placement of the Cupriavidus nifH clade suggests thattheir nifH gene may be derived from a Burkholderia ancestorthat lived in tropical America (Fig. 2).

View larger version (26K):
[in this window]
[in a new window]

FIG. 2. Maximum likelihood phylogeny for nifH genes from Texan Cupriavidus strains (shown in bold) and other proteobacteria. Numbers near branches are bootstrap percentages for parsimony (top), neighbor-joining (middle), and likelihood (bottom) analyses. For Cupriavidus and Burkholderia nodule bacteria, the place of origin is indicated (CR, Costa Rica; Fr. Guiana, French Guiana; Ven., Venezuela).

Nodulation and nitrogenase assays.

Top

Nodulation and nitrogenase...

Seeds from five Mimosa species (M. pigra, M. pudica, M. invisa,M. strigillosa, and M. quadrivalvis) were inoculated with sevenTexan Cupriavidus isolates using standard protocols (13). Thefive Mimosa species developed nodules with most or all of theseven Cupriavidus strains tested (Tables 2 and 3). M. invisaformed the most nodules on average (18 to 144 nodules per plantacross bacterial strains). M. quadrivalvis and M. strigillosashowed limited nodulation for all seven isolates.

View this table:
[in this window]
[in a new window]

TABLE 2. Nodule numbers of five Mimosa species inoculated with Cupriavidus strains

View this table:
[in this window]
[in a new window]

TABLE 3. Acetylene reduction activity of five Mimosa species inoculated with Cupriavidus strains

Nodules of six out of the seven Cupriavidus strains showed substantialnitrogenase activity, as measured by acetylene reduction assayson plants of both M. pudica and M. invisa (Table 3). Plantsof M. pigra exhibited little or no nodule nitrogenase activitywith any bacterial strain. A consistent absence of nodule nitrogenaseactivity was also observed for plants of M. quadrivalvis andM. strigillosa.

The identity of nodule occupants was verified by reisolatingone bacterial culture from a surface-sterilized nodule for eachof the seven Cupriavidus strains and then sequencing a 390-bpportion of the 23S rRNA gene. Two of the reisolated cultureswere obtained from nodules formed on M. strigillosa, and theothers came from M. pudica or M. invisa hosts. For all sevenstrains, the 23S rRNA sequence from the nodule isolate was identicalto that of the inoculant strain. This confirmed that these Cupriavidusisolates were responsible for inducing legume root nodule developmentin this experiment.

Three strains from M. strigillosa that had 5' 23S rRNA sequencessimilar to those of Sinorhizobium arboris or S. saheli werealso tested for nodule formation ability on M. strigillosa.The mean number of nodules per plant ranged from 6 to 16 forthe three strains, and nodules of all three strains exhibitedpositive acetylene reduction activity (0.04 to 0.16 µmolper plant per minute).

Conclusions.

Top

Conclusions.

This is the first report of Cupriavidus strains being the predominantroot nodule symbionts for a legume population in its nativegeographic range and is also the first report of Cupriavidusnodule symbionts for a legume indigenous to the continentalUnited States. The results help to clarify the origins of nodulesymbiosis in Cupriavidus. All Cupriavidus nodule symbionts,including strains indigenous to two distant regions in the NewWorld and strains isolated from Mimosa plants introduced toTaiwan (4, 6), formed a strongly supported monophyletic groupin both the nodA and nifH phylogenies (Fig. 1 and 2). This suggeststhat different regional populations of Cupriavidus nodule bacteriadid not independently acquire symbiotic genes from differentsources but rather may be derived from a single ancestor. TheCupriavidus clade was found to nest within a set of Burkholderiastrains indigenous to South America for both the nodA and nifHtrees (Fig. 1 and 2). Thus, it appears that Cupriavidus acquiredthese genes by horizontal transfer from Burkholderia ratherthan from alphaproteobacterial nodule bacteria and that thisevent most likely occurred in tropical America.

ACKNOWLEDGMENTS

We are grateful to J. Pfeil for assistance with sequencing andto David Sutton for providing Mimosa pigra seeds. We also thankEuan James for providing bacterial strains and Chris Best atSanta Ana National Wildlife Refuge for assistance with collectingnodule samples.

Financial support was provided by NSF grant DEB-0235766.

FOOTNOTES

* Corresponding author. Mailing address: Department of Biological Sciences, State University of New York, Binghamton, NY 13902-6000. Phone: (607) 777-6283. Fax: (607) 777-6521. E-mail: mparker{at}binghamton.edu

Published ahead of print on 25 May 2007.

REFERENCES

Top