Culture-independent molecular approaches reveal a mostly unknown high diversity of active nitrogen-fixing bacteria associated with Pennisetum purpureum—a bioenergy crop

REGULAR ARTICLE

Culture-independent molecular approaches reveal a mostlyunknown high diversity of active nitrogen-fixing bacteriaassociated with Pennisetum purpureum—a bioenergy crop

Sandy Sampaio Videira & Michele de Cássia Pereira e Silva & Péricles de SouzaGalisa & Armando Cavalcante Franco Dias & Riitta Nissinen & Vera Lúcia BaldaniDivan & Jan Dirk van Elsas & José Ivo Baldani & Joana Falcão Salles

Received: 3 April 2013 /Accepted: 24 June 2013 /Published online: 24 July 2013# Springer Science+Business Media Dordrecht 2013

AbstractAims Previous studies have shown that elephant grass iscolonized by nitrogen-fixing bacterial species; however,these results were based on culture-dependent methods,an approach that introduces bias due to an incomplete

assessment of the microbial community. In this study,we used culture-independent methods to survey thediversity of endophytes and plant-associated bacterialcommunities in five elephant grass genotypes used inbioenergy production.Methods The plants of five genotypes of elephantgrass were harvested from the experimental area ofEmbrapa Agrobiologia and divided into stem and roottissues. Total DNA and RNAwere extracted from planttissues and the bacterial communities were analyzed byDGGE and clone library of the 16S rRNA and nifHgenes at both the cDNA and DNA levels.Results Overall, the patterns based on DNA- and RNA-derived DGGE-profiles differed, especially within tissuesamples. DNA-based DGGE indicated that both totalbacterial and diazotrophic communities associated withroots (rhizoplane+endophytes) differed clearly fromthose obtained from stems (endophytes). These resultswere confirmed by the phylogenetic analyses of RNA-derived sequences of 16S rRNA (total bacteria; 586sequences), but not for nifH (186). In fact, rarefactionanalyses showed a higher diversity of diazotrophic or-ganisms associated with stems than roots. Based on 16SrRNA sequences, the clone libraries were dominated bysequences affiliated to members of Leptotrix (12.8 %)followed by Burkholderia (9 %) and Bradyrhizobium(6.5 %), while most of the nifH clones were closelyrelated to the genus Bradyrhizobium (26 %).Conclusions Our results revealed an unexpectedly largediversity of metabolically active bacteria, providing newinsights into the bacterial species predominantly found in

Plant Soil (2013) 373:737–754DOI 10.1007/s11104-013-1828-4

Responsible Editor: Katharina Pawlowski.

In memoriam (Péricles de Souza Galisa)

Electronic supplementary material The online version of thisarticle (doi:10.1007/s11104-013-1828-4) containssupplementary material, which is available to authorized users.

S. S. Videira : P. de Souza GalisaUniversidade Federal Rural do Rio de Janeiro (UFRRJ),Rio de Janeiro, Brazil

S. S. VideiraCentro Universitário de Volta Redonda (UniFOA),Rio de Janeiro, Brazil

M. d. C. Pereira e Silva :R. Nissinen : J. D. van Elsas :J. F. Salles (*)Department of Microbial Ecology, Center for Ecological andEvolutionary Studies (CEES), University of Groningen(RUG), Nijenborgh 7, 9747AG Groningen, The Netherlandse-mail: [email protected]

A. C. F. DiasCenter for Nuclear Energy in Agriculture, Universidade deSão Paulo, CENA/USP, Piracicaba, Brazil

S. S. Videira : P. de Souza Galisa :V. L. B. Divan :J. I. BaldaniEmbrapa Agrobiologia (CNPAB), BR 465, km 07,CEP 23890-000 Seropédica, Rio de Janeiro, Brazil

associationwith elephant grass. Furthermore, these resultscan be very useful for the development of new strategiesfor selection of potential bacteria that effectively contrib-ute to biological nitrogen fixation and enhance the sus-tainable production of elephant grass as bioenergy crop.

Keywords DNA .mRNA . 16S rRNA . nifH .

Functional diversity . Diazotrophic bacteria

Introduction

In the last few decades, much attention has been given toplants capable of efficiently producing biomass in orderto find alternative sources for a sustainable bioenergyprogram (Samson et al. 2005; Morais et al. 2012). Thereare a number of plant species that generate high yields ofbiomasswithminimal inputs, many of these are C4 grasses(Byrt et al. 2011). Grass species with C4 photosynthesis,such asAleman grass (Echinochloa polystachya), elephantgrass (Pennisetum purpureum), fox tail millet (Setariaitalica), Miscanthus (Miscanthus giganteus), sweetsorghum (Sorghum bicolor), sugarcane (Saccharumofficinarum L.), and switchgrass (Panicum virgatum), areideal energy crops because they possess the followingtraits: high conversion efficiency of light into biomassenergy and high water and nitrogen-use efficiency (Tayloret al. 2010).Within these plant species, research conductedin Brazil has shown that some P. purpureum Schum.genotypes are able to accumulate more than 40 Mg ha−1

of dry matter per year when grown in soils with lowavailable nutrient levels, especially nitrogen (Moraiset al. 2009, 2012; Xavier et al. 1998; Reis et al. 2001).Studies carried out by Morais et al. (2012, 2009) used the15N natural abundance technique to verify the hypothesisthat this crop was benefiting from associated biologicalnitrogen fixation (BNF). Their results have indicated thatelephant grass genotypes obtained up to 130 kg of nitrogen(N) ha−1 via BNF in field conditions. Specifically, between18 % and 70 % of N present in elephant grass genotypeswas acquired through plant-associated N2 fixation.

Although the contribution of BNF is well document-ed in elephant grass, few studies have examined thediazotrophic bacteria community associated with theseplants. By using culture-dependent techniques,Kirchhof et al. (1997a, b, 2001), Reis et al. (2001) andVideira et al. (2012) have shown that diazotrophic pop-ulation colonizing roots and stems range from 103 to 107

bacteria g−1 plant tissue fresh weight. Among the

nitrogen-fixing bacteria, the genera Herbaspirillumand Azospirillum were observed to colonize the tissuesof elephant grass endophytically (Reis et al. 2001;2000). More recently, Videira et al. (2012) confirmedthe presence of the N2-fixing bacteria Azospirillum inelephant grass and revealed, for the first time, theco-occurrence of the genera Gluconacetobacter,Burkholderia, Klebsiella, and Enterobacter in differenttissues of two elephant grass genotypes (Videira et al.2012). Moreover, similar numbers of diazotrophic bac-teria were isolated from the roots and stems, suggestingthat both aboveground and belowground tissues wereequally colonized (Videira et al. 2012). For a long time,the microbial diversity of the diazotrophic communitiesassociated with plants, especially grasses, has been stud-ied by traditional microbiological approaches (Brasilet al. 2005; Videira et al. 2009). However, as it ispossible to cultivate only a fraction of the bacterialcommunity on the culture media available, and as manyecologically significant traits are only induced in planta,culture-dependent methods fail to give an accurate pic-ture of the bacterial communities and functions in situ.Thus, DNA- and RNA-based methods are increasinglyapplied. These approaches are providing “snapshots” ofthe molecular make-up of whole microbial communityor of specific microorganisms and genes therein (VanElsas and Boersma 2011). In studies focusing on plant–bacteria interaction, these methods have revealed anenormous and microbial diversity previously unknown,therefore expanding our knowledge on specific process-es or particular functional groups of organisms (Andoet al. 2005; Roesch et al. 2008; Debroas et al. 2009;Burbano et al. 2011; Fischer et al. 2012). Moreover, inorder to understand microbial functioning in the envi-ronment at a molecular level, it is essential to knowwhatgenes are present in a particular system but also theirlevel of expression in a given sample. To achieve thisaim, rRNA- or mRNA-based approaches have beendesigned to discriminate between functionally activeand dormant populations (Sessitsch et al. 2002). TherRNA content represents an approximation of bacterialactivity, whereas the data based on mRNA provideinformation about the active functional groups.According toWagner (1994), the cellular concentrationsof rRNA are correlated with growth rate and activity;hence, the rRNA itself can reveal useful informationabout which community members are active. In thecontext of biological nitrogen fixation, mRNA-basedapproaches targeting the gene coding for a subunit of

738 Plant Soil (2013) 373:737–754

the nitrogenase (nifH gene) constitute a crucial steptowards discovering the active key diazotrophic bacte-rial groups (Knauth et al. 2005; Wartiainen et al. 2008).The great majority of the studies using mRNA-derivednifH were carried out in soils, rhizosphere, and oceansamples (Wartiainen et al. 2008; Hayden et al. 2010;Zehr 2011). Interestingly, the few studies that haveinvestigated the diversity of active diazotrophic commu-nity in planta have identified many bacterial species thathad not been previously detected though standardculture-based methods, thus providing new insights intothe functional diazotrophic bacterial communities insugarcane (Fischer et al. 2012; Thaweenut et al. 2011).

The aim of this work was to expand our knowledgeon the N2-fixing bacteria associated with elephantgrass by performing a thorough analysis of the totaland nitrogen-fixing bacterial communities by usingculture-independent molecular techniques. For that,we have sampled the roots (rhizoplane+endophyte)and stems (endophyte) of five different elephant grassand targeted the total and active bacterial populationsassociated with them by using DNA- and RNA-basedapproaches, respectively. Moreover, in order to identifypotential nitrogen-fixing species contributing to BNF, wehave analyzed the nifH gene fragments in plant geno-types, Cameroon, Roxo, and CNPGL91F06-3, and esti-mated the contributions of BNF (Morais et al. 2012).

Material and methods

Plant sampling

The present study evaluated the microbial commu-nity in five different elephant grass (P. purpureum)genotypes—Cameroon (G1), Gramafante (G2), BAG02 (G3), Roxo (G4), and CNPGL91F06-3 (G5). Eachgenotypes has a different potential to produce dry matterand accumulate N derived from BNF when cultivatedon low N fertility soil (Table S1). This study wasperformed in the experimental area of EmbrapaAgrobiologia, in Seropédica, RJ, Brazil (22°49′22″S,43°38′42″W, at an altitude of 43 m). The growth condi-tions including soil type, fertilization scheme, and cropmanagement were described by Videira et al. (2012).Three plants of each genotypes were harvested anddivided into stem and root tissues. Young actively grow-ing roots were washed in tap water to remove excess ofsoil and then rinsed with distilled water for three times,

while the middle of the stem (nodes 4–6) was surfacedisinfected with 70 % ethanol and then peeled. Bothtissues were frozen in liquid nitrogen within 30min aftersampling. The samples were stored at −80 °C untilprocessing.

DNA and RNA extraction and cDNA synthesis

Five grams of tissue was ground in liquid nitrogenusing a mortar and pestle until a fine powder wasformed. Extractions were performed separately fromthree individual plants (i.e., replicates) for all fivegenotypes. Total genomic DNA was extracted with0.25 g of powdered tissue and CTAB extraction buffer(2 %w/vCTAB, 1.42MNaCl, 20 mMEDTA, 100mMTris–HCl pH 8.0, 2 % w/v PVP 40, 5 mM ascorbicacid, and 4.0 mM DIECA) as described by Chen andRonald (1999). The total RNA was extracted withTRIzol reagent (Invitrogen, USA) following the man-ufacture’s protocol using 0.25 g of powdered tissue.Total RNAs were treated with PolyATtract® mRNAIsolation Systems (Promega, USA) in order to elimi-nate the majority of mRNA plant derived. The RNAextract was purified by RQ1 RNase-free DNase(Promega, USA). Absence of co-extracted DNA inRNA samples was confirmed by PCR assays targetingthe 16S rRNA genes using the universal primers 799Fand 1492R (Chelius and Triplett 2001).

Five micrograms of total RNAwas used as templatefor the reverse-transcribed PCR (RT-PCR) using Su-perScript III First Strand Synthesis System (Invitrogen,USA) following the manufacture’s guidelines. Thesingle-strand DNA (cDNA) was made by adding0.2 μM of primers specific for the gene nifH–PolR(Poly et al. 2001) or 0.2 μM of random primers forthe 16S RNA gene.

PCR-DGGE

For amplification of the bacterial 16S rRNA gene, theprimers 799F and 1492R were used in the first PCR ofa nested approach (Chelius and Triplett 2001). PCRproducts were analyzed by 1.5 % agarose gel electro-phoresis and fragments of the expected sizes (∼740 bp)were excised and purified using the Wizard® SV Geland PCR Clean-Up System (Promega, USA) accordingto the manufacturer’s instructions. The second reactionwas carried out with the primers 968F-GC and 1401R-1b (Brons and Van Elsas 2008), and 1 μl of the purified

Plant Soil (2013) 373:737–754 739

PCR product from the first reaction was used as atemplate. The mixture and cycling conditions were usedexactly as described by Chelius and Triplett (2001).

For the amplification of the nifH gene, the first PCRwas carried out with the primer FGPH19 and PolR(Simonet et al. 1991; Poly et al. 2001). The secondreaction was carried out with the primers PolF-GC andAQER (Poly et al. 2001), and 1 μl of the first PCRproduct was used as a template. The mixture and cyclingconditions were used as described by Pereira e Silvaet al. (2011). PCR products were analyzed by 1.5 %agarose gel electrophoresis followed by staining withethidium bromide to confirm the size.

DGGE was performed using the Ingeny phorU 2×2apparatus (Ingeny, The Netherlands). Samples (20 μl)were loaded onto 6 % (w/v) polyacrylamide gels withdenaturing gradients from 30 % to 55 % for 16S rRNAand 45–65 % for nifH [where 100 % is 7 M urea and40 % (v/v) deionized formamide] in 0.5× TAE buffer.Electrophoresis was performed at 100 Vat a temperatureof 60 °C for 16 h. Gels were then stained with SBYRGold (Molecular Probes, Netherlands) in 0.5× TAE for1 h at room temperature and visualized under UV illu-mination using ImageMaster VDS system (AmershamBiosciences).

Clone libraries

Clone libraries were constructed using the cDNAsynthetized from the 16S rRNA and nifH gene tran-scripts. For each target gene, six libraries wereconstructed, representing two plant tissues (root andstem) from three representative genotypes (Cameroon,CNPGL91F06-3 and Roxo). The PCR conditions forbacterial 16S rRNA and nifH genes were the same asdescribed above for the DGGE protocol, except thatprimers used for the second PCR reaction did not haveGC clamp. Prior to cloning, the PCR fragments obtainedfrom the triplicates from each treatment were pooledtogether to minimize the effect of PCR drift and laterpurified with Wizard® SV Gel and PCR Clean-UpSystem (Promega, USA). Purified and pooled PCRfragments were then ligated into the pGEM-T-easy vec-tor and transformed into chemically competentEscherichia coli JM109 (Promega, USA) in agreementwith manufacturer’s protocols. Clones containing theinsert (evaluated by blue/white colony) were submittedto a colony PCR with primers M13F and M13R todetermine which clones contained a correct-sized insert

(16S rRNA, 450 pb; nifH, 360 pb). Clones were se-quenced by LGC Genomics (Berlin, Germany).

Phylogenetic analysis

All sequences were submitted to the Ribosomal DataProject (RDP) program (Cole et al. 2009) to detectpossible artifacts and poor quality sequences. The finalsequence data were used in MOTHUR program(Schloss et al. 2009) to define the operational taxonomicunits (OTUs) as well as the richness estimator Chao1and Shannon diversity index. The rarefaction curve wasproduced by plotting the number of OTUs observedagainst the number of clones screened (Schloss et al.2009). Phylogenetic trees based on 16S rRNA and nifHgenes from all samples were obtained in MEGA5 soft-ware (Tamura et al. 2011) using neighbor-joining meth-od (kimura-2 parameter). The bootstrap analysis (1,000replicates) was carried out to represent the evolutionaryhistory of the taxa analyzed. UniFrac analysis, takinginto account the phylogenetic information, wasimplemented to determine whether communities aresignificantly different (Lozupone and Knight 2005).

The RDP Classifier (Wang et al. 2007) was used for16S rRNA sequences to identify the sequences at levelof phylum and class with a confidence threshold of80 %. The phylogenetic relationship of the bacterialcommunity was determined by comparison of individ-ual bacterial 16S rRNA sequences with the publicdatabase from RDP Sequence Match (Schloss et al.2009). The phylogenetic affiliation of diazotrophic bac-terial community—nifH cDNA-derived sequences—wasobtained using the NCBI BLAST server (http://blast.ncbi.nlm.nih.gov/Blast.cgi).

Statistical analyses

Genetic fingerprints obtained with DGGE were ana-lyzed using GelCompar software (Applied Maths,Sint-Martens Latem, Belgium; Rademaker and DeBruijn 1997). The total band intensity for each lanewas normalized among lanes and data was square roottransformed. Bacterial community matrices as well asnitrogen fixers were analyzed using PRIMER softwarepackage (version 6, PRIMER-E Ltd, Plymouth, UK;Clarke and Gorley 2006). To compare the communitystructure between different genotypes and plant tissues,ANOSIM (analysis of similarities) was performed inPrimer-E. Furthermore, principal component analysis

740 Plant Soil (2013) 373:737–754

(PCA) was performed in Canoco (version 4.0 for Win-dows, PRI Wageningen, The Netherlands).

Nucleotide sequence accession numbers

Sequences of the 16S rRNA genes obtained in this workhave been deposited in the GenBank database under thefollowing accession numbers: KC914098–KC914240.Sequences of the nifH genes obtained in this work havebeen deposited in the GenBank database under thefollowing accession numbers: KC914241–KC914278.

Results

Molecular fingerprinting of bacterial communities

Reproducible DGGE profiles were generated from 16SrRNA gene (DNA) and gene transcripts (RNA)extracted from root and stem parts of the elephant grassgenotypes using specific bacterial primers (Figure S1).The DNA-based DGGE profiles from communitiesassociated with roots produced between 25 and 40detectable bands whereas fewer bands (about 20) werevisible for the stems. On the other hand, RNA-basedDGGE profiles (obtained after RT-PCR) from bothroots and stems contained typically 15–25 bands. Thisdifference in the number of ribotypes detected wasmore expressive for root than stem samples. In addi-tion, the RNA-derived bands were essentially a sub-group of the bands detected in the DNA-based DGGEanalysis, suggesting that the bacterial community met-abolically active in the plant tissues represents a frac-tion of the total bacterial communities colonizing ele-phant grass plants. A subset of exclusive bands wasdetected in both DNA and RNA samples. Bands re-stricted to DNA samples could indicate that some ofthe bacterial groups are not active; whereas, the pres-ence of RNA-exclusive bands could indicate that therewere active bacteria underrepresented in the DNA-derived profiles and below the amplification threshold.

Multivariate analyses of these DGGE profiles indi-cated that for the 16S rRNA gene, DNA-based analysisseparated roots and stems communities, as samplesfrom these two different compartments formed sepa-rate clusters on the first axis (Fig. 1a). The second axisclearly distinguished the DNA- and RNA-derivedDGGE patterns. One-way ANOSIM analysis revealedthat plant compartment explained 66.6 and 36.7 % of

the variation in 16S DNA and RNA-based communi-ties, respectively (Table 1). The effect of plant geno-type on structure of the bacterial communities couldonly be detected for RNA-based communities.

Molecular fingerprinting of nitrogen-fixing bacterialcommunities

The nifH DGGE profiles were reproducible only forDNA replicates obtained from root samples, and theycontained approximately 12 different bands (Fig. S2).The fingerprints based on DNA obtained from stems,as well as those based on RNA (both tissues) showedplant-to-plant variation as illustrated in the Figure S2.These patterns consisted of a few intense and somefaint bands per sample (Fig. S2), suggesting a lesscomplex community.

Multivariate analyses indicated that the first twoprincipal components could explain only 37 % of thevariance of the nitrogen-fixing bacterial communities(Fig. 1b). In general, the PCA analysis showed thatmajor differences in community structure were associ-ated with the template origin, either DNA or mRNA. Asignificant genotype and tissue effect was observedwithin each of these pools. More specifically, for theDNA-based nifH community, a larger percent of vari-ation could be explained by plant tissue (root or stem,35.8 %) than by genotype. Conversely, the nifHmRNA-based community was more influenced by ge-notype than by plant tissue (36.8 % and 15.7 % of thevariation, respectively; Table 1).

Phylogenetic diversity of bacterial communities

Based on DGGE profiles and the potential to producedry matter and accumulate N derived from BNF(Table S1), three representatives genotypes of elephantgrass—Cameroon (G1), CNPGL91F06-3 (G5), andRoxo (G4)—were included in the 16S rRNA andnifH-cDNA clone libraries analysis. A total of 585 highquality sequences were obtained from RNA-based li-braries (16S rRNA) representing 3 genotypes and 2plant compartments. From this total, 286 were derivedfrom roots and 299 from stems. Detailed statisticalinformation of both root and stem libraries is summa-rized in Table 2. MOTHUR analysis revealed a total of174 OTUs (cutoff of 0.03) from which 107 wereunique sequences. The numbers of OTUs in individuallibraries ranged from 33 to 42 (Table 2). The Chao1

Plant Soil (2013) 373:737–754 741

estimator index indicated higher levels of richness inCameroon (G1) and CNPGL91F06-3 (G5) genotypesthan in Roxo (G4) samples (Table 2). Similar trendswere observed for the Shannon diversity index, except

for the root samples, where the genotype Roxo (G4)was the most diverse. The active bacterial communitiesassociated with roots were more diverse than thatobtained from stems (Fig. 2a; Table 2) in all five

a

b

-0.1

-0.8

1.040,5%

18.7

%1.

0

-0.8

-0.6

0.826,6%

10.4

%0.

8

Fig. 1 Principal componentanalysis (PCA) based on a16S rRNA gene and b nifHgene associated with rootand stem tissues of five ele-phant grass genotypes. De-naturing gradient gel elec-trophoresis (DGGE)banding pattern data wereused to construct a similaritymatrix for PCA analysis.Each sample is codedaccording to colors andshapes of symbols: gray,fragments DNA-derived;black, RNA-derived; sym-bols full, root samples; sym-bols empty, stems samples;circle, genotype Cameroon;square, genotypeGramafante; diamond, ge-notype BAG 02; rectangle,genotype Roxo; and trian-gle, genotypeCNPGL91F06-3

742 Plant Soil (2013) 373:737–754

genotypes. The rarefaction curves based on 16S rRNAtranscripts did not reach an asymptote indicating thatsampling was insufficient to capture the total diversityof the bacterial community (Fig. 2a). Qualitative dif-ferences in the bacterial communities from differentsamples were confirmed by PCA analysis in UniFrac(Bonferroni-corrected P values <0.01), which take intoaccount phylogenetic information. The result showed ahigher similarity of bacterial communities associatedwith root and stem of genotypes Cameroon (G1) andCNPGL91F06-3 (G5) as compared with Roxo (G4;Fig. 3a).

The sequence analysis identified 9 different bacte-rial phyla, 43 families and 85 genera. Of these,Proteobacteria (73.7 %; 431 sequences) and Actino-bacteria (16.4 %; 96 sequences) were the clearly themost prominent; whereas, Acidobacteria (1.9 %),Firmicutes (1.5 %), Deinococcus–Thermus (1.4 %),Bacteroidetes (1.4 %), TM7 (0.5 %), Cyanobacteria(0.3 %), and Gemmatimonadetes (0.2 %) were lessprominent. About 2.7 % was unclassified bacteria(Fig. 4a). Among the clones analyzed, 32 sequencesharbored as-yet-unclassified bacteria. Overall, the mostabundant class was the Betaproteobacteria (27.9 %)and Alphaproteobacteria (23.1 %), which together rep-resented 50 % of the clone library sequences. The thirdmost abundant class was Gammaproteobacteria(12.5 %), followed by Actinobacteridae (10.4 %),Deltaproteobacteria (6 %), and Rubrobacteridae(5.1 %). Bacterial composition differed between rootand stems samples. The majority of the clone sequencesfrom root samples belonged to Gammaproteobacteria(18.5 %), followed by Alphaproteobacteria (16.4 %),Actinobacteridae, and Betaproteobacteria (both 16.1 %;Fig. 4b); whereas, clone libraries of stem samples weredominated by bacteria belonging to Betaproteobacteria(39.1 %) and Alphaproteobacteria (29.4 %; Fig. 4b). Atthe genera level, a detailed overview of the sequencesobtained and their frequency in each library are given inFigs. 4a and 5a. In the combined data set, the genusLeptotrix (12.8 %) was clearly most abundant, followed

Table 1 Results (P values) of one-way ANOSIM for compari-sons of the community structure based on 16S rRNA and nifHstructure assessed by DGGE profiles between the different ge-notypes and tissues of elephant grass

Samples Global R

Genotype Tissue

16S gene

DNA derived 0.070NS 0.666***

RNA derived 0.314*** 0.367***

nifH gene

DNA derived 0.216*** 0.358***

RNA derived 0.368*** 0.157**

ANOSIM analysis of similarity, NS not significant

*P<0.05; **P<0.01; ***P<0.001, significance level

Table 2 Statistical analysis of clone libraries based on the transcripts form 16S rRNA and nifH genes from root and stem samples fromthree genotypes of elephant grass

Samples 16S rRNA gene nifH gene

Number ofsequences

OTUsa Chao(richness)

Shannon(diversity)

Number ofsequences

OTUsb Chao(richness)

Shannon(diversity)

RG1 106 42 64.09 3.15 30 11 12 2.25

RG4 92 36 52.00 3.22 29 8 8.3 1.81

RG5 88 38 71.00 3.18 29 8 18 1.46

SG1 101 41 68.63 3.12 26 7 7.5 1.67

SG4 107 33 63.66 2.83 27 10 13 1.8

SG5 91 35 85.60 2.99 22 12 21.3 1.8

Root 286 93 192.00 3.88 88 25 31 2.7

Stem 299 79 146.56 3.38 75 28 48 2.87

OTUs operational taxonomic units, R root, S stem, G1 cameroon, G4 roxo, G5 CNPGLaOTUs were defined at 97 % sequence identityb OTUs were defined at 99 % sequence identity

Plant Soil (2013) 373:737–754 743

by Burkholderia (9 %) and Bradyrhizobium (6.5 %).These genera were found in all tissues sampled, withexception of root from genotype Roxo (G4; Fig. 5a). Alarge number of unclassified sequences within differentfamilies and classes were observed. Figure 5a displaysmore details about the distribution of all 16S rRNAsequences analyzed in this study.

Venn diagrams were constructed (Fig. S3) to verifyhow the metabolically active members of a bacterialcommunity varied between the root and stem, using allsequences that occur more than once in each library. Thedata from root library revealed that 22.5 % of the taxo-nomical groups were shared among the 3 elephant grassgenotypes evaluated (Fig S3a). Within these genotypes,

the highest frequency of sequences was an unclassifiedGammaproteobacteria (46 sequences), followed bygenera Actinoplanes and Steroidobacter (21 sequences),Conexibacter (12 sequences), and Amycolatopsis (11sequences). For stem samples, 20 % of the bacterialgroups was detected in all three libraries, and the generaLeptothrix (55 sequences), Burkholderia (40 se-quences), unclassified Alphaproteobacteria (40 se-quences), and Bradyrhizobium (27 sequences) showedthe highest frequency (Fig. S3b). Analysis of each ge-notype stem and roots sequences combined showed 14genera and 3 families within the three genotypes(Fig. S3c). The genera Leptothrix (77 sequences),Burkholderia (54 sequences), and Bradyrhizobium (39

0

10

20

30

40

50

0 20 40 60 80 100 120

Num

ber

of O

TU

s (9

7%)

Number of 16S clones sequenced

RG1

RG4

RG5

SG1

SG4

SG5

0

5

10

15

0 5 10 15 20 25 30 35 40

Num

ber

of O

TU

s (9

9%)

Number of nif H clones sequenced

RG1RG4RG5SG1SG4SG5

a

b

Fig. 2 Rarefaction curvesfrom 16S rRNA (a) andnifH (b) mRNA-derivedclone libraries from root andstems. The number of OTUsdetected vs the number ofsequences sampled in eachroot and stem library. Theabbreviations in the figurerepresent: root (R) and stem(S) of the genotypes Camer-oon (G1), Roxo (G4), andCNPGL (G5)

744 Plant Soil (2013) 373:737–754

sequences), as well as the unclassifiedAlphaproteobacteria (45 sequences), were identified inall roots and stem samples, with the exception of rootsample from the genotype Roxo (RG4; Fig. S3c).

Diversity of active diazotrophic bacterial communities

A total of 163 sequences were obtained from librariesbased on the nifH mRNA pools extracted from 3

Fig. 3 Principal coordinateanalyses for the 16S rRNA(a) and nifH (b) RNA-derived clone libraries fromroot and stems. The ordina-tion was constructed usingUnifrac distances weightedby the relative abundances.The abbreviations in the fig-ure represent: root (R) andstem (S) of the genotypesCameroon (G1), Roxo (G4),and CNPGL (G5)

Plant Soil (2013) 373:737–754 745

genotypes and 2 plant compartments; 88 were derivedfrom roots and 75 were derived from stems (Table 2).In the MOTHUR analysis (cut off value of 0.0), a totalof 25 and 28 OTUs were identified for root and stemssamples, respectively (Table 2). Overall, the nifH com-munities associated with stems showed a higher rich-ness than those associated with roots, but these samplesdid not differ in terms of diversity (Table 2). Rarefac-tion analyses showed that while RG1, RG4, and SG1samples attained a considerable degree of saturation inthe number of OTUs, the curves for RG5, SG4, andSG5 libraries did not reach an asymptote (Fig. 2b).Results from a PCA analysis in UniFrac revealed thatthe nitrogen-fixing bacterial community presented intissues of all 3 genotypes was quite variable (Fig. 3b).

The homology analysis of nifH sequences againstNCBI databank sequence data revealed that the clonelibraries were dominated by the Proteobacteria(87.5 %). At the class level, most of nifH sequences

showed highest similarities to sequences fromAlphaproteobacteria (47.2 %), followed byGammaproteobacteria (23.3 %), Betaproteobacteria(17.2 %), and Cyanobacteria (1.2 %). At the generalevel, the majority of the sequences showed homologyto NCBI database entries. Within the genera found, thenifH transcript widely distributed in all six libraries wasclosely related to the genus Bradyrhizobium (Figs. 5band 6). The genera and species that dominate the sam-ples varied between genotypes. A detailed overview ofthe sequences obtained and their numbers of occurrencein each library are given in Figs. 5b and 6.

Discussion

In order to investigate the total and active bacterialcommunities associated with elephant grass tissuesand their potential to contribute to biological nitrogen

a

b

0%

20%

40%

60%

80%

100%

RG1 RG4 RG5 SG1 SG4 SG5 Root Stem

Gammaproteobacteria

Deltaproteobacteria

Betaproteobacteria

Alphaproteobacteria

Rubrobacteridae

Actinobacteridae

0%

20%

40%

60%

80%

100%

RG1 RG4 RG5 SG1 SG4 SG5 Root Stem

unclassified_Bacteria

TM7

Tenericutes

Proteobacteria

Gemmatimonadetes

Firmicutes

Deinococcus-Thermus

Cyanobacteria

Bacteroides

Actinobacteria

Acidobacteria

Fig. 4 Relative abundanceof bacterial composition ofeach 16S rRNA gene clonelibrary at the phylum leveldetermined by using theRDP Classifier tool with an80 % confidence level (a)and the class level abun-dance of two most dominantphyla, Proteobacteria andActinobacteria. b The x axisrepresents each sample ofroot (R) and stem (S) of thegenotypes Cameroon (G1),Roxo (G4), and CNPGL(G5)

746 Plant Soil (2013) 373:737–754

a bRG1 RG4 RG5 SG1 SG4 SG5 R S

ALPHAPROTEOBACTERIA1 Bradyrhizobium japonicum2 Bradyrhizobium sp BTAi13 Bradyrhizobium genosp. TUXTLAS-234 Bradyrhizobium sp IRBG 2285 Bradyrhizobium sp ORS2786 Bradyrhizobium sp ORS2857 Bradyrhizobium sp MAFF 2103188 Bradyrhizobium sp ORS3319 Mesorhizobium loti

10 Xantobacter autotrophicus11 Azospirillum brasilense

BETAPROTEOBACTERIA12 Azohydromonas australica13 Burkholderia silvatlantica14 Burkholderia vietnamiensis

GAMAPROTEOBACTERIA15 Klebsiella pneumoniae16 Enterobacter sp MTP_05051217 Azotobacter vinelandii18 Pseudomonas stutzeri

CYANOBACTERIA19 Tolypotrix sp LCRNK_2620 Anabaena varialis

Unidentified nitrogen- fixing bacteria

ScaleNumber of sequences 1 2 3 4 5 6-9 10-19 >20

RG1 RG4 RG5 SG1 SG4 SG5 R SACIDOBACTERIA

1 Gp12 Gp3

FIRMICUTES3 Staphylococcus4 Streptococcus

BACTEROIDETES5 Sediminibacterium

ACTINOBACTERIA6 Actinomyces7 Actinosynnema8 Lechevalieria9 Lentzea

10 Gordonia11 Quadrisphaera12 Curtobacterium13 Microbacterium14 Actinoplanes15 Mycobacterium16 Aeromicrobium17 Cellulosimicrobium18 Propionibacterium19 Amycolatopsis20 Crossiella21 Kibdelosporangium22 Kutzneria23 Streptacidiphilus24 Unclassified_Acidimicrobineae25 Unclassified_Actinobacteria26 Unclassified_Actinosynnemataceae27 Iamia28 Conexibacter29 Solirubrobacter30 Unclassified_Solirubrobacterales31 Unclassified_Microbacteriaceae

ALPHAPROTEOBACTERIA32 Bosea33 Bradyrhizobium34 Ochrobactrum35 Prosthecomicrobium36 Methylobacterium37 Aminobacter38 Mesorhizobium39 Rhizobium40 Inquilinus41 Phaeospirillum42 Skermanella43 Brevundimonas44 Geminicoccus45 Sphingomonas46 Unclassified_Alphaproteobacteria

BETAPROTEOBACTERIA47 Azohydromonas48 Burkholderia49 Ralstonia50 Aquabacterium51 Ideonella52 Leptothrix53 Acidovorax54 Pelomonas55 Unclassified_Comamonadaceae56 Collimonas57 Herminiimonas58 Massilia59 Unclassified_Oxalobacteraceae60 Unclassified_Betaproteobacteria61 Unclassified_Burkholderiales _incertae_sedis62 Neisseria

DELTAPROTEOBACTERIA63 Bdellovibrio64 Byssovorax65 Sorangium66 Unclassified_Nannocystineae67 Unclassified_Polyangiaceae68 Unclassified_Sorangiineae

GAMAPROTEOBACTERIA69 Enterobacter70 Pantoea71 Serratia72 Unclassified_Enterobacteriaceae73 Aquicella74 Acinetobacter75 Pseudomonas76 Nevskia77 Steroidobacter78 Dyella79 Pseudoxanthomonas80 Rudaea81 Unclassified_Gammaproteobacteria

CYANOBACTERIA82 Streptophyta

TERENICUTES83 Spiroplasma

TM784 TM7_ genera_incertae_sedis

DEINOCOCCUS-THERMUS85 Deinococcus

Unclassified_Bacteria

ScaleNumber of sequences 1 2 3 4 5 6-9 10-19 >20

Fig. 5 Frequencies of the different bacterial taxa from 16SrRNA (a) and nifH (b) mRNA-derived in the root and stem ofthree different elephant grass genotypes. The abbreviations in the

figure represent: root (R) and stem (S) of the genotypes Camer-oon (G1), Roxo (G4), and CNPGL (G5)

Plant Soil (2013) 373:737–754 747

Fig. 6 Phylogenetic tree ofnifH based on analysis ofnucleotide sequences. OUTsfrom the present study areshown in bold and markedwith root (R) and stem (S)from Cameroon genotype(G1), Roxo genotype (G4),and CNPGL genotype (G5)indicating the number ofclones and their origins.Phylogenetic analyses wereconducted in MEGA4 usingneighbor-joining method.The numbers at nodes indi-cate levels of bootstrap sup-port based on analysis of1,000 resampled datasets.There were a total of 315positions in the final datasetand the number of nucleo-tide substitutions per sites isindicated in the scale bars.Methanosarcina mazei wasused as a bacterial outgroup

748 Plant Soil (2013) 373:737–754

fixation, we have used molecular methods to analysethe total bacterial communities (16S rRNA gene) aswell as the diazotrophic communities (nifH gene).Moreover, by using both DNA- and RNA-basedmethods, we were able to differentiate the metabolical-ly active populations present in roots and stems of thisplant. Bacterial community studies involving depen-dent or independent culturable approaches have showna much higher diversity and abundance of microorgan-isms in the roots and/or rhizoplane in comparison to thestems (Brasil et al. 2005; Pariona-Llanos et al. 2010;Hardoim et al. 2011). Our DGGE profiles targeting theoverall diversity of bacterial communities, as well assequencing analyses of the active bacterial members(RNA-based analysis), indeed showed that communi-ties obtained from roots had higher richness and weremore diverse than the communities associated with thestem tissue. This higher bacterial diversity in the rootsof P. purpureumwas expected considering that the rootsample in this study is defined by both the rhizoplaneand the root interior. This is in contrast to the stemsamples that were surface sterilized and peeled, reveal-ing only the endophytic bacterial community for anal-ysis. Additionally, these results may be also due tophysiological differences associated with these twoareas of the plant. Stem tissues have relatively lownutrient concentrations and inhibitory compounds,which may act as selective forces on the structure ofthe bacterial community (James et al. 1994); whereas,roots are known to release organic compounds that canbe consumed by nearby bacteria. Rhizopshere and rhi-zoplane have typically very high bacterial densities (upto 10−9 per gram plant tissue); whereas, the endospherebacterial population densities are typically several or-ders of magnitude lower (Compant et al. 2010).

Interestingly, whereas our results from DNA-derivedDGGE profiles based on the nifH gene confirmed theresults observed for total bacterial communities, i.e., thatcommunities associated with roots are more diverse thanthose associated with stems, no such trend was foundwhen studying the active diazotrophic communities(mRNA-based extraction). As expected, sequencinganalyses of the nifH clone libraries showed low levelsof diversity when compared to those observed for totalbacterial communities (16S rRNA gene). Moreover, itconfirmed the trend observed for DGGE profiles, byproviding evidence that the diazotrophic communitiesassociated with the stems and roots were equally diverse,as determined by the Shannon index. In fact, a close look

at the rarefaction analysis indicates that except for Cam-eroon genotype (G1), the diazotrophic communities as-sociated with the stems showed a higher species richnessthan those associated with root tissues, as shown by thesteeper curves observed for stems from cultivars Roxoand CNPGL91F06-3. These findings might be due to thehigher competitive ability of diazotrophic communitiesin this N-limited environment. In other words, the abilityfix atmospheric nitrogen might allow diazotrophic bac-teria to occupy a niche inside the plants that is relativelyempty (stems) due to limited nutrients. In metabolic richenvironments such as roots, this functional capabilitymight not be so relevant

Regarding the identity of the bacterial groups,Proteobacteria and Actinobacteriawere the most prom-inent phyla of bacteria detected in our study, corrobo-rating other studies that suggest these two phyla as themost common ones that are found in the rhizosphere ofmany plant species (Chelius and Triplett 2001; Hardoimet al. 2012; Hallmann and Berg 2006; Kaiser et al.2001). In root samples, sequences related to Gamma-proteobacteria were the most abundant, and unclassi-fied Gammaproteobacteria followed by Steroidobactersp predominated in the clone libraries. Steroidobacter-affiliated sequences are commonly found in soil andaquatic ecosystems (Fahrbach et al. 2008; Cébronet al. 2008). This bacterium is only described in theliterature as a steroid degrader (Fahrbach et al. 2008)but appears to also be plant-associated as it has beenpreviously identified in sugarcane rhizosphere (Pisaet al. 2011) and roots of aquatic plants (Tanaka et al.2012), albiet with no known biological function. TheActinobacteria, including Actinobacteridae andRubrobacteridae, were also important root colonizersof elephant grass. Early studies have demonstrated thatsome Actinobacteria can form intimate associationswith plants, such as wheat, rice, sugarcane, carrots,tomato, and citrus (Fischer et al. 2012; Tian et al.2007; Qin et al. 2011), and may be related to planthealth. In some cases, they can act as biological controlagents, plant growth enhancers (Kannan and Sureendar2009; Sun et al. 2009; Nimnoi et al. 2010), and promoterof the plant establishment under adverse conditions(Hasegawa et al. 2006).

Alpha- and Betaproteobacteria were frequently de-tected among the root clone libraries from Cameroonand CNPGL91F06-3 genotypes in contrast to the Roxogenotype library. These classes were also detected in allstem-associated bacterial communities, displaying a

Plant Soil (2013) 373:737–754 749

higher number of sequences than in the root. The mostfrequent clones in the stems were closely related toLeptothrix and Burkholderia (Betaproteobacteria);Bradyrhizobium (Alphaproteobacteria) were also dom-inant genera, yet to a lesser extent as Leptothrix andBurkholderia (Betaproteobacteria). Species of the ge-nus Leptothrix belong to the family Burkholderiales andare commonly found in habitats such as lakes, lagoonsand swamps (Spring et al. 1996). Recently, speciesassociated with this genus were isolated from the rootsof Phaseolus vulgaris (López-López et al. 2010) oridentified by clone library screening (16S rRNAgene)as root endophytes of Typha angustifolia (Li et al. 2011).These microorganisms are recognized by their ability tooxidize high amounts of Fe (II) and/or Mn (II) in thiernatural habitats, but there is no information on thefunction of this genus in plants. In contrast, theBurkholderia genus is a diverse group of bacteria com-monly found, by both culturable and unculturable ap-proaches, in many environments such as soil. It is alsoassociated with crops such as rice, maize, yellow lupine,sugarcane, and elephant grass (Salles et al. 2002; Perinet al. 2006; Elliott et al. 2007; Videira et al. 2012).Burkholderia species have been reported to exhibit di-rect or indirect mechanisms as plant growth promotersand biological control agents (Compant et al. 2005;Sessitsch et al. 2005; Govindarajan et al. 2008). Thethird most abundant group of clones were affiliated tothe genus Bradyrhizobium which are capable ofestablishing symbiotic relationships with a broad rangeof plants belonging to the family Fabaceae (Menna andHungria 2011; Lindström et al. 2010), contributing tobiological nitrogen fixation. Recent studies involvingnon-legume plants, such as rice, sugarcane, and sweetpotato, have also shown sequences related to this genus(Burbano et al. 2011; Fischer et al. 2012; Hardoim et al.2011; Terakado-Tonooka et al. 2008). Moreover, somereports also indicate that these symbiotic bacteria havethe potential to be used as PGPR with non-legumes(Antoun et al. 1998; Chaintreuil et al. 2000), but thishypothesis has not yet been elucidated.

Following the trends observed for 16S rRNA gene,Proteobacteria-related transcripts dominated the clonelibraries based on the nifH gene. As previously shown,Alpha- and Betaproteobacteria are common membersof diazotrophic communities, contributing substantial-ly to the nifH gene pool (Fischer et al. 2012; Hardoimet al. 2012), as well as to the part of the diazotrophiccommunity that is actively expressing nifH (Reinhold-

Hurek and Hurek 2011). It is worth noting thatsequences-affiliated to Bradyrhizobium predominatedthe diazotrophic community assessed by transcribednifH pools in roots and stems of elephant grass geno-types, whereas some other bacterial groups randomlydistributed among the libraries (Fig. 6). The activediazotrophic community in root samples showed highersimilarity with the phototrophic rhizobial strains(IRBG228 e BTAi1) that are able to fix nitrogen underfree living conditions (Ladha and So 1994; Giraud et al.2007) and the non-phototrophic strain (MAFF210318)that was isolated from nodule of Aeschynomene grownin Thailand (Cantera et al. 2004). The strains ofBradyrhizobium sp (ORS331, ORS278, and ORS285)obtained from Aeschynomene grown in Senegal (Nzoueet al. 2009) contributed with the major part of nifHtranscripts in stem samples. This may be explained bythe observation that some rhizobial strains could beefficient colonizers of gramineous plants grown underfield conditions (Wu et al. 2009; Burbano et al. 2011;Fischer et al. 2012).

Within the Alphaproteobacteria clones, sequencesclosely related to Xhantobacter autotrophicus were alsodetected in both tissues. The nitrogen-fixation capabilitiesof Xhantobacter species have been demonstrated by thepresence of both nifH gene fragments and nitrogenaseactivity (Malik andClaus 1979; Arzumanyan et al. 1997).On the other hand, Azospirillum and Mesorhizobiumwere detected only in stem of theRoxo genotype (Figs. 5band 6). Azospirillum strains are known as plant growth-promoting bacteria rhizobacteria, but they have also beenfound within plant tissue as endophytes of diverse non-legume plants (Bashan et al. 2004). Some studies haveshown that Mesorhizobium, a natural endophyte in legu-minous plants, was also detected in various non-leguminous plants (Mirza et al. 2001; Bhattacharjeeet al. 2008; Pisa et al. 2011). The Betaproteobacteriaclones were closely related to Azohydromonas andBurkholderia genera, while Gammaproteobacteriaclones include Klebsiella, Enterobacter, Azotobacter,and Pseudomonas.

It is important to notice that most metabolicallyactive diazotrophs, except Bradyrhizobium spp, werespecific for a given library. This might be due to thePCR based methods used in the study, which mayunderestimate the diversity of dominating diazotrophicphylotypes (Mårtensson et al. 2009). Demba Dialloet al. (2008), in a study of nifH pools in roots of rice,found that Zehr primers (Zehr and Mc Reynolds 1989)

750 Plant Soil (2013) 373:737–754

amplified communities of greater complexity than thePoly primers, indicating possible bias of Poly primers.As tested previously (data not shown), different sets ofnifH primers were analyzed for our samples, but onlyPoly primers (Poly et al. 2001) gave a single andcorrect product size from sample tissues.

Analyses of clone libraries constructed for threeselected cultivars, two of which produced greater bio-mass than the third, revealed that root samples from thetwo more productive cultivars selected some bacterianaturally associated with plant such as Burkholderia,Bradyrhizobium , Mesorhizobium , Pelomonas ,Leptothrix, and Bdellovibrio genera. Interestingly, theliterature has shown that Roxo genotype has beenbenefited less from the BNF contribution as comparedto the Cameroon and CNPGL91F06-3 genotypes(Xavier et al. 1998; Morais et al. 2012). However, itis important to mention that our results are based onone sampling time, thus long-term studies during thegrowing season are necessary in order to confirm theplant and tissue effect on the bacterial communitycomposition. Host plant selectivity for its bacterialcommunity, especially plant growth promoting, hasalso been observed in previous studies with gramine-ous plant (Baldani et al. 1997; Perin et al. 2003), andselectivity has been strongly related to root exudationpatterns (Bürgmann et al. 2005). In this context, it ispossible to speculate that the root bacterial communitymight be directly involved in plant growth promotion.

The present study, via a culture-independent ap-proach, has pioneered the exploration of bacterial diver-sity in elephant grass. In summary, it revealed that thediversity of the elephant grass bacterial community, bothin terms of total bacteria and nitrogen-fixing bacteria, ismuch higher than previously detected by culture-dependent techniques (Videira et al. 2012). Moreover,it described for the first time certain plant growth-promoting species and dominant groups of metabolical-ly active diazotrophs that were yet to be found in asso-ciation with elephant grass. These findings may help toidentify possible ecological drivers shaping this bacteri-al community. Interestingly, it revealed that sequencesrelated to Bradyrhizobium sp. dominated the active bac-terial community associated with both tissues. However,their contribution to BNF remains unknown, as geneexpression does not necessarily coincide with in situcontribution and nitrogen fixation activity might varyamong diazotrophic bacteria. In this context, the appli-cation of molecular techniques based on gene transcripts

associated with 15N isotope dilution techniques may bea useful tool to evaluate the BNF contribution by thenative and inoculated diazotrophic bacteria to elephantgrass N nutrition. Furthermore, more detailed spatialand temporal studies of gene transcripts may help clarifythe dynamics of the diazotrophic assemblage.

Acknowledgments This work was part of the activities carriedout at Department ofMicrobial Ecology, Center for Ecological andEvolutionary Studies, University of Groningen. The authors thankthe support in the form of fellowships to SSV by Foundation forResearch Support in the State of Rio de Janeiro (FAPERJ) andthankfully acknowledged Embrapa and CNPq/INCT-FBN for par-tial financial support.

References

Ando S,GotoM,Meunchang S, Thongra-Ar P, Fujiwara T, HayashiH, Yoneyama T (2005) Detection of nifH sequences in sugar-cane (Saccharum officinarum L.) and pineapple (Ananascomosus [L.] Merr.). J Soil Sci Plant Nutr 51:303–308

Antoun H, Beauchamp CJ, Goussard N, Chabot R, Lalande R(1998) Potential of Rhizobium and Bradyrhizobium speciesas plant growth promoting rhizobacteria on nonlegumes: ef-fect on radishes (Raphanus sativus L.). Plant Soil 204:57–67

Arzumanyan VG, Sakharova ZV, Panikov NS, Vshivtsev VS(1997) Growth and nitrogen-fixing activity of the batchculture of Xanthobacter autotrophicus at various concen-trations of dissolved oxygen. Microbiology 66:627–630

Baldani JI, Caruso L, Baldani VLD, Goi SR, Döbereiner J (1997)Recent advances in BNF with non-legume plants. Soil BiolBiochem 29:922–928

Bashan Y, Holguin G, De-Bashan LE (2004) Azospirillum–plantrelationships: physiological, molecular, agricultural, andenvironmental advances (1997–2003). Can J Microbiol50:521–577

Bhattacharjee S, Lee LY, Oltmanns H, Cao H, Veena Cuperus J,Gelvin SB (2008) IMPa-4, an Arabidopsis importin α iso-form, is preferentially involved in Agrobacterium-mediatedplant transformation. Plant Cell 20:2661–2680

Brasil MS, Baldani JI, Baldani VLD (2005) Occurrence anddiversity of diazotrophic bacteria associated to forage grassesof the Pantanal in the state of Mato Grosso do Sul. Rev BrasCienc Solo 29:179–190

Brons JK, Van Elsas JD (2008) Analysis of bacterial communi-ties in soil by use of denaturing gradient gel electrophoresisand clone libraries, as influenced by different reverseprimers. Appl Environ Microbiol 74:2717–2727

Burbano CS, Liu Y, Roesner KL, Reis VM, Caballero-Mellado J,Reinhold-Hurek B, Hurek T (2011) Predominant nifH tran-script phylotypes related to Rhizobium rosettiformans infield-grown sugarcane plants and in Norway spruce.Environ Microbiol Rep 3:383–389

Bürgmann H, Meier S, Bunge M, Widmer F, Zeyer J (2005)Effects of model root exudates on structure and activity of asoil diazotroph community. Environ Microbiol 7:1711–1724

Plant Soil (2013) 373:737–754 751

Byrt CS, Grof CPL, Furbank RT (2011) C4 plants as biofuelfeedstocks: optimising biomass production and feedstockquality from a lignocellulosic perspective. J Integr PlantBiol 53:120–135

Cantera JJL, Kawasaki H, Seki T (2004) The nitrogen-fixinggene (nifH) of Rodopseudomonas palustris: a case of lateralgene transfer. Microbiology 150:2237–2246

Cébron A, Norini MP, Beguiristain T, Leyval C (2008) Real-Time PCR quantification of PAH-ring hydroxylatingdioxygenase (PAH-RHDalpha) genes from Gram positiveand Gram negative bacteria in soil and sediment samples. JMicrobiol Methods 73:148–159

Chaintreuil C, Giraud E, Prin Y, Lorquin J, Ba A, Gillis M, deLajudie P, Dreyfus B (2000) Photosynthetic bradyrhizobiaare natural endophytes of the African wild rice Oryzabreviligulata. Appl Environ Microbiol 66:5437–5447

Chelius MK, Triplett EW (2001) The diversity of Archaea andBacteria in association with the roots of Zea mays L.MicrobEcol 41:252–263

Chen DH, Ronald PC (1999) A rapid DNA minipreparationmethod suitable for AFLP and other PCR applications.Plant Mol Biol Rep 17:53–57

Clarke KR, Gorley RN (2006) PRIMER (v6): user manual/tutorial. PRIMER-E, Plymouth, 192pp

Cole JR, Wang Q, Cardenas E, Fish J, Chai B, Farris RJ, Kulam-Syed-Mohideen AS, Mcgarrell DM, Marsh T, Garrity GM,Tiedje JM (2009) The ribosomal database project: improvedalignments and new tools for rRNA analysis. Nucleic AcidsRes 37:141–145

Compant S, Reiter B, Sessitsch A, Nowak J, Clement C, AitBarka E (2005) Endophytic colonization of Vitis vinifera L.by a plant growth-promoting bacterium, Burkholderia sp.strain PsJN. Appl Environ Microbiol 71:1685–1693

Compant S, Clément C, Sessitsch A (2010) Plant growth-promoting bacteria in the rhizo- and endosphere of plants:their role, colonization, mechanisms involved and prospectsfor utilization. Soil Biol Biochem 42:669–678

Debroas D, Humbert JF, Enault F, Bronner G, Faubladier M,Cornillot E (2009) Metagenomic approach studying thetaxonomic and functional diversity of the bacterial commu-nity in a mesotrophic lake (Lac du Bourget–France).Environ Microbiol 11:2412–2424

Demba Diallo M, Reinhold-Hurek B, Hurek T (2008) Evaluationof PCR primers for universal nifH gene targeting and forassessment of transcribed nifH pools in roots of Oryzalongistaminata with and without low nitrogen input. FEMSMicrobiol Ecol 65:220–228

Elliott GN, Chen WM, Chou JH, Wang HC, Sheu SY, Perin L,Reis VM, Moulin L, Simon MF, Bontemps C, SutherlandJM, Bessi R, De Faria SM, Trinick MJ, Prescott AR, SprentJI, James EK (2007) Burkholderia phymatum is a highlyeffective nitrogen-fixing symbiont of Mimosa spp. andfixes nitrogen ex planta. New Phytol 173:168–180

FahrbachM, Kuever J, RemeschM, Huber BE, Kämpfer P, Dott W,Hollender J (2008) Steroidobacter denitrificans gen. nov., sp.nov., a steroidal hormone-degrading gammaproteobacterium.Int J Syst Evol Microbiol 58:2215–2223

Fischer D, Pfitzner B, Schmid M, Simões-Araújo JL, Reis VM,PereiraW, Ormeño-Orrillo E, Hofmann A,Martinez-RomeroE, Baldani JI, Hartmann A (2012) Molecular characterisationof the diazotrophic bacterial community in uninoculated and

inoculated field-grown sugarcane (Saccharum sp.). Plant Soil356:83–99

Giraud E, Moulin L, Vallenet D, Barbe V, Cytryn E, Avarre JC,Jaubert M, Simon D, Cartieaux F, Prin Y, Bena G, HannibalL, Fardoux J, Kojadinovic M, Vuillet L, Lajus A, CruveillerS, Rouy Z, Mangenot S, Segurens B, Dossat C, Franck WL,ChangWS, Saunders E, Bruce D, Richardson P, Normand P,Dreyfus B, Pignol D, Stacey G, Emerich D, Verméglio A,Médigue C, Sadowsky M (2007) Legumes symbioses: ab-sence of nod genes in photosynthetic Bradyrhizobia. Science316:1307–1312

Govindarajan M, Balandreau J, Kwon SW, Weon HY,Lakshminarasimhan C (2008) Effects of the inoculation ofBurkholderia vietnamensis and related endophytic diazotrophicbacteria on grain yield of rice. Microb Ecol 55:21–37

Hallmann J, Berg G (2006) Spectrum and population dynamicsof bacterial root endophytes. Soil Biol 9:15–31

Hardoim PR, Andreote FD, Reinhold-Hurek B, Sessitsch A, VanOverbeek LS, Van Elsas JD (2011) Rice root-associatedbacteria: insights into community structures across 10 cul-tivars. FEMS Microbiol Ecol 77:154–164

Hardoim PR, Hardoim CCP, Van Overbeek LS, Van Elsas JD(2012) Dynamics of seed-borne rice endophytes on earlyplant growth stages. PLoS One 7:e30438

Hasegawa S, Meguro A, Shimizu M, Nishimura T, Kunoh H(2006) Endophytic actinomycetes and their interactionswith host plants. Actinomycetologica 20:72–81

Hayden HL, Drake J, Imhof M, Oxley APA, Norng S, Mele PM(2010) The abundance of nitrogen cycle genes amoA andnifH depends on land-uses and soil types in South-EasternAustralia. Soil Biol Biochem 42:1774–1783

James EK, Reis VM, Olivares FL, Baldani JI, Dobereiner J(1994) Infection of sugar cane by the nitrogen-fixing bac-terium Acetobacter diazotrophicus. J Exp Bot 45:757–766

Kaiser O, Puhler A, Selbitschka W (2001) Phylogenetic analysis ofmicrobial diversity in the rhizoplane of oilseed rape (Brassicanapus cv Westar) employing cultivation dependent andcultivation-independent approaches. Microb Ecol 42:136–149

Kannan V, Sureendar R (2009) Synergistic effect of beneficialrhizosphere microflora in biocontrol and plant growth pro-motion. J Basic Microbiol 49:158–164

Kirchhof G, Reis VM, Baldani JI, Eckert B, Döbereiner J,Hartmann A (1997a) Occurrence, physiological and molec-ular analysis of endophytic diazotrophic bacteria in gramin-eous energy plants. Plant Soil 104:45–55

Kirchhof G, Schloter M, Abmus B, Hartmann A (1997b)Molecular microbial ecology approaches applied todiazotrophs associated with non-legumes. Soil BiolBiochem 29:853–862

Kirchhof G, Eckert B, Stoffels M, Baldani JI, Reis VM,Hartmann A (2001) Herbaspirillum frisingense sp. nov., anew nitrogen fixing bacterial species occurring in C4-energy plants. Int J Syst Evol Microbiol 51:157–168

Knauth S, Hurek T, Brar D, Reinhold-Hurek B (2005) Influenceof different Oryza cultivars on expression of nifH genepools in roots of rice. Environ Microbiol 7:1725–1733

Ladha JK, So RB (1994) Numerical taxonomy of photosyntheticrhizobia nodulating Aeschynomene species. Int J SystBacteriol 44:62–73

Li YH, Liu QF, Liu Y, Zhu JN, Zhang Q (2011) Endophyticbacterial diversity in roots of Typha angustifolia L. in the

752 Plant Soil (2013) 373:737–754

constructed Beijing Cuihu Wetland (China). Res Microbiol162:124–131

Lindström K, Murwira M, Willems A, Altier N (2010) Thebiodiversity of beneficial microbe-host mutualism: the caseof rhizobia. Res Microbiol 161:453–463

López-López A, Rogel MA, Ormeno-Orrillo E, Martinez-RomeroJ, Martinez-Romero E (2010) Phaseolus vulgaris seed-borneendophytic community with novel bacterial species such asRhizobium endophyticum sp. nov. Syst Appl Microbiol33:322–327

Lozupone C, Knight R (2005) UniFrac: a new phylogeneticmethod for comparing microbial communities. ApplEnviron Microbiol 71:8228–8235

Malik KA, Claus D (1979) Xanthobacter flavus, a new species ofnitrogen-fixing hydrogen bacteria. Int J Syst Bacteriol29:283–287

Mårtensson L, Díez B, Wartiainen I, Zheng WW, El-Shehawy R,Rasmussen U (2009) Diazotrophic diversity, nifH gene ex-pression and nitrogenase activity in a rice paddy field inFujian, China. Plant Soil 325:207–218

Menna P, Hungria M (2011) Phylogeny of nodulation andnitrogen-fixation genes in Bradyrhizobium: supporting ev-idence for the theory of monophyletic origin, and spreadand maintenance by both horizontal and vertical transfer. IntJ Syst Evol Microbiol 61:3052–3067

Mirza MS, Ahmad W, Latif F, Haurat J, Bally R, NormandP, Malik KA (2001) Isolation, partial characterization,and effect of plant growth promoting bacteria (PGPB)on micro-propagated sugarcane in vitro. Plant Soil237:47–54

Morais RF, Souza BJ, Leite JM, Soares LH, Alves BJR, BoddeyRM, Urquiaga S (2009) Elephant grass genotypes forbioenergy production by direct biomass combustion. PesquiAgropecu Bras 44:133–140

Morais RF, Quesada DM, Reis VM, Urquiaga S, Alves BJR,Boddey RM (2012) Contribution of biological nitrogen fix-ation to Elephant grass (Pennisetum purpureum Schum.).Plant Soil 356:23–34

Nimnoi P, Pongsilp N, Lumyong S (2010) Endophytic actino-mycetes isolated from Aquilaria crassna Pierre ex Lec andscreening of plant growth promoters production. World JMicrobiol Biotechnol 26:193–203

Nzoue A, Miche L, Klonowska A, Laguerre G, Delajudie P, MoulinL (2009) Multilocus sequence analysis of bradyrhizobia iso-lated from Aeschynomene species in Senegal. Syst ApplMicrobiol 32:400–412

Pariona-Llanos R, Ibañez De Santi FFF, Soto Gonzales HH,Barbosa HR (2010) Influence of organic fertilization onthe number of culturable diazotrophic endophytic bacteriaisolated from sugarcane. Eur J Soil Biol 46:387–393

Pereira e Silva MC, Semenov AV, Van Elsas JD, Salles JF (2011)Seasonal variations in the diversity and abundance ofdiazotrophic communities across soils. FEMS MicrobiolEcol 77:57–68

Perin L, Silva MF, Ferreira JS, Canuto EL, Medeiros AFA,Olivares FL, Reis VM (2003) Avaliação da capacidade deestabelecimento endofítico de estirpes de Azospirillum eHerbaspirillum em milho e arroz. Agronomia 37:47–53

Perin L, Martinez-Aguilar L, Paredes VG, Baldani JI, Estrada-De Los Santos P, Reis VM, Caballero-Mellado J (2006)Burkholderia silvatlantica sp. nov., a diazotrophic

bacterium associated with sugarcane and maize. Int J SystEvol Microbiol 56:1931–1937

Pisa G, Magnani GS, Weber H, Souza EM, Faoro H, MonteiroRA, Daros E, Baura V, Bespalhok JP, Pedrosa FO, Cruz LM(2011) Diversity of 16S rRNA genes from bacteria of sugar-cane rhizosphere soil. Braz J Med Biol Res 12:1215–1221

Poly F, Monrozier LJ, Bally R (2001) Improvement in the RFLPprocedure for studying the diversity of nifH genes in com-munities of nitrogen fixers in soil. ResMicrobiol 152:95–103

Qin S, Xing K, Fei SM, Lin Q, Chen XM, Cao CL, Sun Y,Wang Y,Li WJ, Jiang JH (2011) Pseudonocardia sichuanensis sp.nov., a novel endophytic actinomycete isolated from the rootof Jatropha curcasL. Antonie Van Leeuwenhoek 99:395–401

Rademaker JLW, De Bruijn FJ (1997) Characterization and clas-sification of microbes by REP-PCR genomic fingerprintingand computer-assisted pattern analysis. In: Caetano-AnollésG, Gresshoff PM (eds) DNAmarkers: protocols, applicationsand overviews. JohnWiley and Sons, NewYork, pp 151–171

Reinhold-Hurek B, Hurek T (2011) Living inside plants: bacte-rial endophytes. Curr Opin Plant Biol 14:435–443

Reis VM, Reis Junior FB, Salles JF, Schloter M (2000)Characterisation of different polyclonal antisera to quantifyHerbaspirillum spp. in Elephant grass (Pennisetum purpureumSchum.). Symbiosis 29:139–150

Reis VM, Reis Junior FB, Quesada DM, Oliveira OCA, AlvesBJR, Urquiaga S, Boddey RM (2001) Biological nitrogenfixation associated with tropical pasture grasses. Aust JPlant Physiol 28:837–844

Roesch LFW, Camargo FAO, Bento FM, Triplett EW (2008)Biodiversity of diazotrophic bacteria within the soil, rootand stem of field-grown maize. Plant Soil 302:91–104

Salles JF, De Souza FA, Van Elsas JD (2002) Molecular methodto assess the diversity of Burkholderia species in environ-mental samples. Appl Environ Microbiol 68:1595–1603

Samson R, Mani S, Boddey R, Sokhansanj S, Quesada D,Urquiaga S, Reis V, Ho Lem C (2005) The potential of C4

perennial grasses for developing a global BIOHEAT indus-try. CRC Crit Rev Plant Sci 24:1–35

Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M,Hollister EB, Lesniewski RA, Oakley BB, Parks DH,Robinson CJ, Sahl JW, Stres B, Thallinger GG, Van HornDJ, Weber CF (2009) Introducing MOTHUR: open-source,platform-independent, community-supported software fordescribing and comparing microbial communities. ApplEnviron Microbiol 75:7537–7541

Sessitsch A, Reiter B, Pfeifer U, Wilhelm E (2002) Cultivation-independent population analysis of bacterial endophytes in threepotato varieties based on eubacterial and Actinomycetes-specificPCR of 16S rRNA genes. FEMS Microbiol Ecol 39:23–32

Sessitsch A, Coenye T, Sturz AV, Vandamme P, Ait Barka E, SallesJF, Van Elsas JD, Faure D, Reiter B, Glick BR, Wang-PruskiG, Nowak J (2005) Burkholderia phytofirmans sp. nov., anovel plant-associated bacterium with plant-beneficial prop-erties. Int J Syst Evol Microbiol 55:1187–1192

Simonet P, Grosjean M, Arvind KM, Nazaret S, Cournoyer B,Normand P (1991) Frankia genus-specific characterizationby polymerase chain reaction. Appl Environ Microbiol57:3278–3286

Spring S, Kämpfer P, Ludwig W, Schleifer KH (1996) Polyphasiccharacterization of the genus Leptothrix: new descriptions ofLeptothrix mobilis sp. nov. and Leptothrix discophora sp.

Plant Soil (2013) 373:737–754 753

nov. nom. rev. and emended description of Leptothrixcholodnii emend. Syst Appl Microbiol 19:634–643

Sun Y, Cheng Z, Glick BR (2009) The presence of a 1-aminocyclopropane-1-carboxylate (ACC) deaminase dele-tion mutation alters the physiology of the endophytic plantgrowth-promoting bacterium Burkholderia phytofirmansPsJN. FEMS Microbiol Lett 296:131–136

Tamura K, Peterson D, Peterson N, Stecher G, Nei M, KumarS (2011) MEGA5: molecular evolutionary genetics anal-ysis using maximum likelihood, evolutionary distance,and maximum parsimony methods. Mol Biol Evol28:2731–2739

TanakaY, Tamaki H,Matsuzawa H, NigayaM,Mori K, KamagataY (2012) Microbial community analysis in the roots ofaquatic plants and isolation of novel microbes including anorganism of the Candidate phylum OP10. Microbes Environ27:149–157

Taylor SH, Hulme SP, ReesM, Ripley BS,Woodward FI, OsborneCP (2010) Ecophysiological traits in C3 and C4 grasses: aphylogenetically controlled screening experiment. NewPhytol 185:780–791

Terakado-Tonooka J, Ohwaki Y, Yamakawa H, Tanaka F,Yoneyama T, Fujihara S (2008) Expressed nifH genes ofendophytic bacteria detected in field-grown sweet potatoes(Ipomoea batatas L.). Microbes Environ 23:89–93

Thaweenut N, Hachisuka Y, Ando S, Yanagisawa S,Yoneyama T (2011) Two seasons’ study on nifH geneexpression and nitrogen fixation by diazotrophic endo-phytes in sugarcane (Saccharum spp. hybrids): expres-sion of nifH genes similar to those of rhizobia. Plant Soil338:435–449

Tian J, Venkatachalam P, Liao H, Yan X, Raghothama KG (2007)Molecular cloning and characterization of phosphorus star-vation responsive genes in common bean (Phaseolusvulgaris L.). Planta 227:151–165

Van Elsas JD, Boersma FGH (2011) A review of molecularmethods to study the microbiota of soil and the mycosphere.Eur J Soil Biol 47:77–87

Videira SS, Araújo JLS, Rodrigues LS, Baldani VLD, Baldani JI(2009) Occurrence and diversity of nitrogen-fixingSphingomonas bacteria associated with rice plants grownin Brazil. FEMS Microbiol Lett 293:11–19

Videira SS, Oliveira DM, Morais RF, BorgesWL, Baldani VLD,Baldani JI (2012) Genetic diversity and plant growth pro-moting traits of diazotrophic bacteria isolated from twoPennisetum purpureum Schum. genotypes grown in thefield. Plant Soil 356:51–66

Wagner R (1994) The regulation of ribosomal RNA synthesisand bacterial cell growth. Arch Microbiol 161:100–109

Wang Q, Garrity GM, Tiedje JM, Cole JR (2007) NaiveBayesian classifier for rapid assignment of rRNA sequencesinto the new bacterial taxonomy. Appl Environ Microbiol73:5261–5267

Wartiainen I, Eriksson T, Zheng W, Rasmussen U (2008)Variation in the active diazotrophic community in ricepaddy-nifH PCR-DGGE analysis of rhizosphere and bulksoil. Appl Soil Ecol 39:65–75

Wu L, Ma K, Lu Y (2009) Prevalence of betaproteobacterialsequences in nifH gene pools associated with roots ofmodern rice cultivars. Microb Ecol 57:58–68

Xavier DF, Botrel MA, Verneque RS, Freitas VP, Boddey RM(1998) Estabilidade da produção de forragem de cultivaresde capim-elefante em solo com baixa disponibilidade denitrogênio. Pasturas Tropicales 20:35–40

Zehr JP (2011) Nitrogen fixation by marine cyanobacteria.Trends Microbiol 19:162–732

Zehr JP, Mc Reynolds LA (1989) Use of degenerate oligonucle-otides for amplification of the nifH gene from the marinecyanobacterium Trichodesmium thiebautii. Appl EnvironMicrobiol 55:2522–2526

754 Plant Soil (2013) 373:737–754