Rapid and Efficient Methods to Isolate, Type Strains and Determine Species of Agrobacterium spp. in Pure Culture and Complex Environments

Agrobacterium are Alphaproteobacteria common in most soils that closely interact with plants in two respects. Firstly, and as a general trait of the whole taxon, they are rhizospheric bacteria saprophytically living in the root environment (i.e. rhizosphere) of numerous plants. Rhizospheric interactions are generally considered to be of commensal type with no detrimental effect to the plant, but in most instances they are likely beneficial to plants. For evident agronomic purposes it is worthwhile to explore whether agrobacteria are themselves plant growth-promoting rhizobacteria (PGPR) or not. However, this requires an expert determination of the Agrobacterium taxonomy. Indeed our current investigations suggest that only some agrobacterial species from the abundant soil Agrobacterium guild are selected in the rhizosphere of a given plant. Secondly, but only when they harbor a dispensable Ti plasmid (i.e. tumor inducing plasmid), agrobacteria are plant pathogens able to cause the crown gall disease to most dicots and gymnosperms and some monocots (Pitzscke & Hirt, 2010). Ti plasmids are conjugative and can easily spread in indigenous soil agrobacteria. As a result transconjugant agrobacteria become in turn pathogenic, contributing both to disease spread and perennial soil contamination. An epidemiological survey of crown gall thus also requires expert determination of the Agrobacterium taxonomy. A set of biochemical and molecular methods were thus set up to facilitate the taxonomic assessment of agrobacteria either in pure culture or directly from complex environments such as soils, rhizospheres or tumours. After a presentation of the present state of Agrobacterium taxonomy, this work provides efficient methods to : i) isolate agrobacteria from complex environments thanks to elective media; ii) determine the Agrobacterium genus status of newly isolated strains on the basis of a minimal set of biochemical tests; iii) determine species and type novel isolates of Agrobacterium by sequence analysis of relevant marker genes; iv) determine amplicon content and genome architecture; v) detect the presence of Agrobacterium and Ti or Ri plasmid directly in complex environments by PCR using selected primers and metagenomic DNA extracted from whole bacterial communities.


Introduction
Agrobacterium are Alphaproteobacteria common in most soils that closely interact with plants in two respects. Firstly, and as a general trait of the whole taxon, they are rhizospheric bacteria saprophytically living in the root environment (i.e. rhizosphere) of numerous plants. Rhizospheric interactions are generally considered to be of commensal type with no detrimental effect to the plant, but in most instances they are likely beneficial to plants. For evident agronomic purposes it is worthwhile to explore whether agrobacteria are themselves plant growth-promoting rhizobacteria (PGPR) or not. However, this requires an expert determination of the Agrobacterium taxonomy. Indeed our current investigations suggest that only some agrobacterial species from the abundant soil Agrobacterium guild are selected in the rhizosphere of a given plant. Secondly, but only when they harbor a dispensable Ti plasmid (i.e. tumor inducing plasmid), agrobacteria are plant pathogens able to cause the crown gall disease to most dicots and gymnosperms and some monocots (Pitzscke & Hirt, 2010). Ti plasmids are conjugative and can easily spread in indigenous soil agrobacteria. As a result transconjugant agrobacteria become in turn pathogenic, contributing both to disease spread and perennial soil contamination. An epidemiological survey of crown gall thus also requires expert determination of the Agrobacterium taxonomy. A set of biochemical and molecular methods were thus set up to facilitate the taxonomic assessment of agrobacteria either in pure culture or directly from complex environments such as soils, rhizospheres or tumours. After a presentation of the present state of Agrobacterium taxonomy, this work provides efficient methods to : i) isolate agrobacteria from complex environments thanks to elective media; ii) determine the Agrobacterium genus status of newly isolated strains on the basis of a minimal set of biochemical tests; iii) determine species and type novel isolates of Agrobacterium by sequence analysis of relevant marker genes; iv) determine amplicon content and genome architecture; v) detect the presence of Agrobacterium and Ti or Ri plasmid directly in complex environments by PCR using selected primers and metagenomic DNA extracted from whole bacterial communities.
In modern bacterial taxonomy, the criteria to delineate taxa now include genomic information. This has brought considerable modifications to the nomenclature. According to Wayne et al. (1987) and Stackebrandt et al. (2002), homogenous genomospecies is the ultimate criterion to validly delineate a bona fide species. Amongst agrobacteria lato sensu, genomospecies generally encompass the same set of strains than biovars. Upon this criterion, Agrobacterium vitis (i.e. biovar 3), A. rubi and A. larrymoorei have been validated as bona fide species (Ophel & Kerr, 1990;Holmes & Roberts, 1981;Bouzar & Jones, 2001). For its part biovar 1 is heterogeneous, including ten genomospecies (De Ley, 1973Popoff et al., 1984) currently called genomovar G1 to G9  and G13 (Portier et al., 2006 ). Biovar 1 is thus a species complex and not a single bacterial species. Noticeably, G4 contains both type-strains of the formerly described species "A. tumefaciens" and "A. radiobacter" that should be both renamed A. radiobacter for antecedence reasons (Young et al., 2006). However, even though the name "A. tumefaciens" is not valid to designate a bona fide genomic species, we proposed that the group of closely related species corresponding to biovar 1 should be collectively called "A. tumefaciens complex" in order to avoid confusion with the genomic species G4, which must be validly named A. radiobacter (Costechareyre et al., 2010). For the ICSP subcommittee on the taxonomy of Rhizobium and Agrobacterium, this seems to be a good interim solution until genomovars can be formally named (Lindström & Young, 2011). In addition to the epithet radiobacter for genomovar G4, fabrum and pusense have recently been proposed for homogenous species corresponding to genomovar G8 and to genomovar G2, respectively (Lassalle et al., 2011;Panday et al., 2011).
Modifications of the nomenclature also happen at the genus level. Although, biovar 2 is a homogenous genomospecies Popoff et al., 1984), this species clearly appears phylogenetically more related to Rhizobium species such as R. etli or R. tropici, than to other Agrobacterium species. Therefore, biovar 2 was proposed to be classified in the genus Rhizobium, while all other species were kept in the genus Agrobacterium (Costechareyre et al., 2010). This proposal had the great advantage to solve the paraphyletic problem of the old classification pointed out by Young et al. (2001). In agreement with taxonomy rule of name antecedence, biovar 2 should be named Rhizobium rhizogenes. The epithet for this species is rhizogenes because the type-strain ATCC 113225 T received this epithet at the time of its first description by Riker (1930). However, most members of this taxon are nonpathogenic or induce crown gall but not hairy root symptoms. This is particularly the case for the famous crown gall biocontrol agent K84 (Kerr, 1974). Frequently named "A. radiobacter" , it is actually a true member of the species Rhizobium rhizogenes in its present definition. There are likely numerous other species of Agrobacterium either within or out the A. tumefaciens complex, such as the unnamed species that includes strain NCPPB 1650, or the novel species that got the epithet skierniewicense (Puławska et al., 2011) and-at least provisionallynepotum for the novel A. tumefaciens genomovar G14 found by Puławska et al. (unpublished). In addition, it is also likely that bacteria able to induce the formation of nodules with some plants are true members of the genus Agrobacterium in its novel definition. It could be the case of the nitrogen fixing bacteria initially called Allorhizobium undicola (Costechareyre et al., 2010). A reasonably good view of the phylogeny of the Agrobacterium genus is given by the recA gene phylogeny (Figure 1) adapted from Costechareyre et al. (2010), completed with sequences from type-strains of novel species A. skierniewicense and A. nepotum (i.e. genomovar G14). The recent denomination A. fabrum and A. pusense are used for genomovars G8 and G2, respectively. Table 1 lists type-strains and recA accession numbers used to construct the phylogeny.

Isolation of agrobacteria from complex environments
Agrobacteria usually form ecological guilds in soils consisting of several genomovars with several strains within genomovars. There are also several strains even within a single tumor (Nesme et al., 1987;Vogel et al., 2003;Costechareyre et al., 2010). As a consequence, it is necessary to take extreme care during strain isolation to be sure of the genotypic purity of isolates. Efficient purification of agrobacteria requires both selective media and the strict respect of the procedure described below. Elective media 1A and 2E, proposed by Brisbane and Kerr (1983) to isolate different biovars, are based on the particular ability of agrobacteria to use certain compounds, such as carbon and energy sources. In addition, these media contain selenite salt (Na 2 SeO 3 ) because agrobacteria lato sensu has the general ability to reduce this salt. However, as the ability to reduce selenite also determines a general ability to resist to toxic amounts of tellurite (K 2 TeO 3 ), this latter compound was substituted for selenite to improve the medium selectivity, especially when isolating bacteria from complex environments such as soil and decaying tumors (Mougel et al., 2001). Members of the A. tumefaciens species complex (i.e. biovar 1) and A. rubi can be isolated with 1A-Te containing arabitol as an elective carbon source and 80 mg/l K 2 TeO 3 . R. Rhizogenes (i.e. biovar 2) can be isolated using 2E-Te containing erythritol and 320 mg/l K 2 TeO 3 . Agrobacteria lato sensu (i.e. Agrobacterium spp. and Rhizobium rhizogenes) as well as most Rhizobiaceae can be isolated using MG-Te containing mannitol and 200 mg/l K 2 TeO 3 (Table 2). On tellurite-amended media, colonies have the typical circular glistening morphologies of agrobacteria plus a characteristic black color with a metallic shine ( Figure 2B).
Agrobacterium vitis (i.e. biovar 3) and Agrobacterium larrymoorei can be isolated on 3DG medium (Brisbane and Kerr, 1983) or Roy and Sasser medium (Roy and Sasser, 1983). On this latter medium, colonies have a typical pinkish-white (red centre) color. Both media are based on the utilization of sodium L-tartrate as the carbon source. It is likely that the selectivity of these media could be improved by addition of tellurite. This is because A. vitis members are resistant to this compound as well as most Rhizobiaceae (unpublished results).
For the first purification step, ca. 0.5 g of soil, root, shoot or tumor tissues are crushed with a micropestle in microtube containing 500 µl of sterile distilled water and then indispensably let to macerate for at least 30 min. This maceration is thought to be necessary to allow efficient cell separation from polysaccharides produced by agrobacteria. Next, ca. 25 µl of macerate (100 µl for soil samples) are streaked on appropriate selective media and then incubated at 28°C. Agrobacteria usually require between 3 and 4 days to form colonies on tellurite amended media because this compound tends to slow down bacterial growth. For the second purification step, clear individual colonies with the typical morphology ( Figure 2B) are picked and suspended separately in 100 µl of sterile distilled water. These Table 2. Improved selective media for agrobacteria members (adapted from Mougel et al., 2001).
individual colony suspensions are left to rehydrate overnight with shaking at 28°C before they are streaked onto appropriate media that must NOT contain tellurite. A third purification step following the same procedure is absolutely necessary before introduction of the novel isolate into the lab collection.

Presumptive genus determination by minimal biochemical tests
The INCO-DC European program ERBIC18CT970198, "Integrated Control of Crown Gall in Mediterranean Countries" has delivered a simple and efficient identification scheme for agrobacteria. By a simple and fast urease and/or esculinase biochemical test, any strains isolated from soil, tumor, or root that posseses the typical circulate glistening colony morphology on tellurite amended media can be conviently and confidently classified as part of the genus Agrobacterium (Table 3). To test the presence of these activities, fresh cells are suspended in urease solution or esculin solution (Table 3). After incubation at 28°C for 1 h, a colorimetric change is observed for agrobacteria strains: pink and black coloration, respectively, for urease and esculinase tests.
The A. tumefaciens species complex (biovar 1) has the enzymatic ability to aerobically convert lactose to 3-ketolactose. This is tested by streaking on medium containing lactose (Bernaerts and De Ley, 1963). After 2 days of growth at 28°C, the plates are flooded with a layer of Benedict's reagents. The presence of 3-ketolactose in the medium is indicated by the formation of a yellow ring around the growth of a positive strain (Table 3 and Figure 3).
A. fabrum (G8 genomovar of A. tumefaciens species complex), has the biovar 1 characteristics, plus a specific ability to degrade ferulic acid and caffeic acid. These properties are due to a gene cluster present only in the A. fabrum strains (Lassalle et al., 2011). A minimal medium supplemented with caffeic acid turns the color brown in five days. As caffeic acid degradation by A. fabrum inhibits this browning, inoculation of an isolate of agrobacteria in a minimal medium supplemented with caffeic acid is an easy test to detect if a strain belong to the A. fabrum species (Figure 3).   supplemented with caffeic acid 0.1 mg.ml -1 is browning after five days of incubation except when the medium is inoculated with A. fabrum (here A. fabrum C58 or J07). Table 3. Recipe for esculin, urease, and 3-ketolactose tests (adapted from Bernaerts and De Ley, 1963).

Infraspecific diversity of Agrobacterium spp. at the genome level
Multi-locus sequence analysis (MLSA) performed with housekeeping or ecologically relevant genes recA, mutS, gyrB, glgC, chvA, ampC, reveals the occurrence of several strains with different alleles in most species (Table 1). As already showed with chvA and recA (Costechareyre et al., 2009(Costechareyre et al., , 2010, these genes allow the identification of novel isolates of Agrobacterium spp. at both species and strain levels. Nevertheless, even if MLSA is better than single gene analysis, we found that recA alone is definitively a good proxy for the identification of Agrobacterium species, since we have so far not found any indication that this gene has been laterally transferred between different species. Identification can be easily done by a nucleotide Blast in databases. Thus, in order to facilitate the identification of A. tumefaciens genomovars, we have clearly indicated the genomovar and allele variant status of sequences deposited at EMBL in the description section, as shown below.
Accession#: FM164286 Description: Agrobacterium tumefaciens partial recA gene for recombinase A, genomovar G1, strain TT111, allele recA-G1-1 Extensive analysis of the infraspecific diversity is now conducted at the whole genome level by comparative genomics. The first analyses were done with DNA microarrays containing probes covering the whole genome of C58. This procedure permitted us to find species specific genes of A. tumefaciens genomovar G8 that were used to find typical phenotypic traits of this species (such as the ability to degrade the caffeic acid, Fig. 3). It was then possible to propose this genomovar as a novel species: A. fabrum (Lassalle et al., 2011). Comparative genomics of Agrobacterium spp. is now done with 22 complete genomes using the Agrobacterscope plateform at Genoscope: https://www.genoscope.cns.fr/agc/microscope/about/collabprojects.php?P_id=51 However, while MLSA and genome sequencing are suited to type a limited number of strains for taxonomy purposes, these methods are not easily applicable to analyze large agrobacterial populations, as is required for epidemiological investigations. For this reason, we developed a PCR-RFLP approach consisting of the restriction of PCR products obtained from selected genome regions (Figure 4). The intergenic region spanning between the 16S and 23S rRNA genes (16S+ITS) was found to allow the distinction of strains within both the R. rhizogenes and A. tumefaciens complex (Ponsonnet & Nesme, 1994), with an accuracy we now know to be equivalent to those of recA (unpublished results). and C indicate primer locations in the T-DNA region (regions between the tmr and nos genes), the virA and virB 2 genes, and the ribosomal operon (16S and 23S rRNA genes), respectively.

Ti plasmid diversity
Although agrobacteria are primarily natural soil inhabitants and also commensal rhizospheric bacteria, they become a plant pathogen when they acquire a Ti (Tumor inducing) plasmid (size ~200 kb). Thus it becomes important to characterize agrobacterium plasmid content. Generally agrobacteria of the A. tumefaciens complex harbor, besides their chromosome and chromid, another large plasmid called the At plasmid (~400 kb). Large Ti and At plasmids can be visualized on an agarose gel using a method modified from Eckhardt (1978) by Wheatcroft et al. (1990), which is based on the action of gentle bacterium lysis directly in agarose gel wells in order to avoid DNA breakages. Briefly, agrobacterial strains are grown in standard medium to an optical density of 0.2 at 600 nm. After which a total of 1 ml of cell culture is centrifuged at 13,000g for 1 min at 4°C. The pellet is then washed once in 1 ml Na N-lauroyl sarcosinate 0.3% and resuspended in Ficoll buffer ( While pathogenic agrobacteria differ from the non-pathogenic strains by the presence of Ti or Ri plasmids, there is also a large diversity within this plasmid class that must be considered in epidemiological investigations. Briefly, the eco-pathology of crown gall and hairy root diseases is centered upon the fact that Ti and Ri plasmids both incite the diseased plant to produce a particular class of compounds called 'opines', which give the agrobacteria a unique ability to use these particular compounds for their growth. There are several kind of opines produced in tumors and consumed by agrobacteria. The most common types of Ti plasmids produce nopaline, octopine or agropine and mannopine (for a review about opines and opine concept see Dessaux et al., 1992). While genes encoding the biosynthesis and the consumption of opines are very different according to opine types, genes involved in the pathogenic process are conserved amongst Ti and Ri plasmids and can be used to define the primers to amplify Ti-plasmid specific regions. Conserved regions are the region of virulence or vir region involved in the processing of the DNA transferred from agrobacterium to plant, (i.e. the T-DNA) and T-DNA genes determining the tumoral morphology such as tmr or tms. Thus, the PCR-RFLP approach is also applicable to study the Ti-plasmid diversity using T-DNA and vir regions (Figure 4 and table 4).
There are however a diversity of Ti plasmid conserved genes that can be used to rapidly assess the opine-type of pathogenic agrobacteria. With this idea we developed a set of nopaline-type specific primers in one of the pioneering works using PCR to detect plant pathogenic bacteria (Nesme et al., 1989). Primers F14 and F44 (Table 4) allowed the amplification of a 247 bp-long fragment only with nopaline-type Ti plasmids such as pTiC58 or pTiT37, while primer set F14-F750 permits us to specifically amplify a 217 bp-long amplicon of octopine-type Ti plasmids, such as pTiB6 or pTi15955. The 'universal' primer set F14-F749 was designed based on alignments of vir regions of the most common Ti-plasmid opine type ( Figure 6). As these primers match different genes separated by intergenic regions of different lengths, amplicon sizes can be used to determine Ti-plasmid opine types.  Table 4. PCR primers used to type agrobacteria by PCR-RLFP or for the detection of Ti and Ri plasmids (adapted from Bruce et al., 1992;Ponsonnet et al., 1994 andKawaguchi et al., 2005).
Remarkably, a given Ti plasmid type could be found in different strains while a given strains could harbor different Ti plasmids, likely as the results of Ti plasmid conjugal transfers. In addition, none of the indigenous agrobacteria isolated from a non-contaminated plot harbored a Ti plasmid. Indigenous agrobacteria were found to be different of pathogenic strains isolated from tumors, except for some isolates of ribotype 2520 that harbored a pTi2516. This strongly suggests that indigenous soil agrobacteria -that were initially Ti plasmid free and consequently not pathogenic, such as 2520-were able to receive a Ti plasmid from introduced pathogenic strains, and then in turn become pathogenic. This may facilitate the establishment of Ti plasmids in indigenous soil agrobacteria and could be at the basis of perennial soil contaminations recorded for decades in several instances (Krimi et al., 2002;Costechareyre et al., 2010).
From an epidemiological point of view, the present chromosome and Ti plasmid diversities were found even in single tumors, illustrating the fact that a single isolate is not enough to characterize pathogenic populations involved in crown gall epidemics. However, by studying numerous isolates, we found that pathogenic agrobacteria were almost the same in Orléans and Peyrat-le-Château (data not shown), demonstrating that the crown gall outbreak of the latter nursery was caused by the introduction of apparently healthy (i.e. tumor free) -but in fact contaminated-plant material from Orléans. The same procedure involving both chromosome and Ti plasmid typing of numerous isolates was used to discover the source of crown gall contamination of rose bushes (Pionnat et al., 1999) and is currently used for routine epidemiological investigations.

Detection in complex environments
The detection of pathogenic Agrobacterium in complex environments without prior strain isolation can be done easily by PCR using primers designed in the conserved vir region (Table 4). Kawaguchi et al. (2005) also reported primers VCF3 and VCR3 amplifying DNA fragments of 414 bp in the virC1-virC2 regions of both Ti and Ri plasmids. Ti plasmid detection can be conducted with metagenomic DNAs extracted from complex environments such as bulk soil, rhizosphere and tumors ( Figure 5C). To date, in terms of accuracy and reproducibility the best results are obtained with DNA extracted with the PowerSoil ® DNA isolation kit (http://www.mobio.com/soil-dna-isolation/powersoil-dna-isolation-kit.html).
In a similar manner, it is also possible to detect Agrobacterium and Rhizobium spp. in complex environments by using PCR probes designed in the 16S rDNA gene (Mougel et al., 2001). However, the 16S is not diverse enough to allow the distinction of closely related species such as A. tumefaciens genomovars. Instead, Nucleotide sequences of protein encoding genes have a much more resolutive power at both species and infraspecies levels. Nevertheless, the present challenge is to find gene regions well conserved within a species but with marked differences between species in order to design species specific primers. Only a small number of genes fulfill this requirement in A. tumefaciens, because genomovars are very closely related but also have a huge infraspecies diversity. Primary results suggest however that recA could be a convenient gene to find species specific primers suited for in situ detection of Agrobacterium species and genomovars (unpublished results). Fig. 7. Chromosome and Ti-plasmid diversity of agrobacteria isolated from poplar tumors in a single crown gall epidemic involving two distant nurseries. Ribotypes and Ti-plasmid types were determined by PCR-RFLP of the 16S+ITS region using primers F667-PA and F38 (Table 4), and digestion with CfoI, HaeIII, NdeII and TaqI; and primers F9095 and F140, and F9096 and F9097 (Table 4) and digestion with TaqI, DdeI, MspI and RsaI, respectively as described in Ponsonnet & Nesme (1994). 2520 and 2178 are ribotypes identical to those of R. rhizogenes strains CFBP 2520 and CFBP 2178, respectively. 2517, 2177 and 2516: ribotypes identical to those of A. tumefaciens genomovar-G1 strains CFBP 2517, CFBP 2177 and CFBP 2516, respectively. pTi2516, pTiM80, pTi1903, pTi2177: nopaline-type Ti-plasmids with PCR-RFLP patterns identical to those of strains M80, CFBP 1903 (i.e. C58) and CFBP 2177, respectively. pTi292: pTi2177 with the insertion element IS292 (Ponsonnet et al., 1995). Tumor indicates total numbers of isolates for each conjugated chromosome and Ti-plasmid patterns obtained in four different plots in Orléans and Peyrat-le-Château. Soil indicates agrobacteria isolated directly from a non-contaminated soil plot in the vicinity of www.intechopen.com contaminated plots planted with poplars at Peyrat-le-Château; none of which harboring a Ti plasmid. The dendrogram was built with UPGMA using number of substitutions per nucleotidic sites calculated from RFLP patterns as genetic distance.

Conclusion
The genus Agrobacterium encompasses several species and several strains within species that are now clearly named using a novel nomenclature. A set of efficient procedures is available to study either pure culture isolates or agrobacteria directly in soil microbiomes. These methodologies are necessary to unmask the ecologies of Agrobacterium species that usually live sympatrically in soils. By this knowledge we expect to differentially manage agrobacterial species directly in soils to improve the crown gall as well as to stimulate their putative PGPR effects.

Acknowledgements
The author wishes to thank J. Puławska that kindly provided strains of novel species before the definitive publication of their results and Cyrus Mallon for reading and correcting the English. This review prepared within the framework of COST Action 873 was supported by the EcoGenome project of Agence Nationale de la Recherche (grant number ANR-08-BLAN-0090). Biochemical testing necessitates the determination of different parameters, and the identification of the main biological chemical compounds, by using molecular and biochemical tools. The purpose of this book is to introduce a variety of methods and tools to isolate and identify unknown bacteria through biochemical and molecular differences, based on characteristic gene sequences. Furthermore, molecular tools involving DNA sequencing, and biochemical tools based in enzymatic reactions and proteins reactivity, will serve to identify genetically modified organisms in agriculture, as well as for food preservation and healthcare, and improvement through natural products utilization, vaccination and prophylactic treatments, and drugs testing in medical trials.