Colonization of Potato Rhizosphere by GFP-Tagged Bacillus subtilis MB73/2, Pseudomonas sp. P482 and Ochrobactrum sp. A44 Shown on Large Sections of Roots Using Enrichment Sample Preparation and Confocal Laser Scanning Microscopy

The ability to colonize the host plants’ rhizospheres is a crucial feature to study in the case of Plant Growth Promoting Rhizobacteria (PGPRs) with potential agricultural applications. In this work, we have created GFP-tagged derivatives of three candidate PGPRs: Bacillus subtilis MB73/2, Pseudomonas sp. P482 and Ochrobactrum sp. A44. The presence of these strains in the rhizosphere of soil-grown potato (Solanum tuberosum L.) was detected with a classical fluorescence microscope and a confocal laser scanning microscope (CLSM). In this work, we have used a broad-field-of-view CLMS device, dedicated to in vivo analysis of macroscopic objects, equipped with an automated optical zoom system and tunable excitation and detection spectra. We show that features of this type of CLSM microscopes make them particularly well suited to study root colonization by microorganisms. To facilitate the detection of small and scattered bacterial populations, we have developed a fast and user-friendly enrichment method for root sample preparation. The described method, thanks to the in situ formation of mini-colonies, enables visualization of bacterial colonization sites on large root fragments. This approach can be easily modified to study colonization patterns of other fluorescently tagged strains. Additionally, dilution plating of the root extracts was performed to estimate the cell number of MB73/2, P482 and A44 in the rhizosphere of the inoculated plants.


Introduction
The plant rhizosphere supports a large microbial population by providing it with nutrients present in the root exudates [1]. In the resident microflora of the rhizosphere, harmless or beneficial commensals can be found, but also plant pathogens. The commensals may provide the roots with a basal level of protection against pathogens simply by occupying the niche [2]. Some of the strains however, known as plant growth-promoting rhizobacteria (PGPR), exhibit additional traits that can directly or indirectly benefit health and yield of the plant. Direct growth stimulation can be mediated by facilitated acquisition of nutrients, mainly phosphorus and nitrogen, or by modulation of the phytohormone equilibrium [2]. The indirect effect stems from increased plant resistance to diseases. This can result either from induction of the defense system of the plant or from antagonism shown by PGRPs towards plant pathogens (for review see [3,4]).
Since the first reports on plant beneficial microbes [5], extensive research has been conducted on their potential to act as bio-fertilizers and/or biological control agents (BCAs). Application of such microorganisms in agriculture is expected to reduce the use of chemical fertilizers and pesticides. Candidate PGPRs are usually obtained by screening large collections of bacterial isolates seeking biochemical or genetic traits known to, or assumed to be involved in plant beneficial activity (for a review see [6]). However, to be effective in planta, the candidate strains need to be able to establish and maintain a sufficient population in the host plant [7]. A variety of cell surface molecules contribute to the colonization process, as does efficient utilization of the major carbon and nitrogen sources available in the root exudates of a certain plant species or even a certain cultivar. Plant growth conditions, including soil type and occurrence of stress factors, also play a part in determining the composition of plant microflora. This complex web of interdependence makes it problematic to study and predict plant-microbe interactions, especially in case of soil-grown plants. Therefore, the colonization potential of each PGRP candidate needs to be experimentally determined.
Tracking the presence of such bacterial strain in an overcrowded environment as the rhizosphere requires selective detection methods. This can be accomplished after modification of the strain to obtain additional antibiotic resistance for selective growth and/or introduction of a reporter gene, such as a fluorescent protein-coding gene. Quantitative techniques for colonization study, providing cell count in colony forming units (CFU) per gram, cm or sq cm of root, include serial dilution plating on selective media and flow cytometry [8]. To visually demonstrate colonization and obtain qualitative data, such as the location of bacteria on plant roots, classical fluorescence and confocal laser scanning microscopy (CLSM) can be applied [9]. CLSM is a powerful investigation tool for studying many aspects of root physiology [10], including interaction of the plant with rhizobacteria and fungi [9,11]. In combination with fluorescently-tagged microorganisms, the method can provide 3D images of the studied objects with minimum invasive manipulation. Noteworthy, apart from modified strains, immunochemistry and fluorescent in situ hybridization (FISH) can also be used for specific staining. However, due to the problem of cross-reactivity, labeled antibodies are often used in combination with gnotobiotic plant growth conditions [12] and FISH is a golden standard when it is necessary to localize organisms representing different taxonomic ranks in one sample [13].
The aim of this study was to investigate the potential of three candidate PGPRs-Bacillus subtilis MB73/2, Ochrobactrum sp. A44 and Pseudomonas sp. P482-to colonize the rhizosphere of potato. Fluorescence microscopy and CLSM were employed, as well as the quantitative dilution plating approach. For the purpose of microscopy, root samples were embedded in an enrichment medium, triggering in situ growth of mini-colonies. The developed method enables visualization of bacterial colonization sites on large root fragments.

Bacterial Strains, Plasmids and Culture Conditions
All bacterial strains and plasmids used in the present study are listed in Table 1. Bacterial cells were grown on Miller's LB agar plates (Novagen, Madison, WI, USA) at 28 °C. When necessary, the medium was supplemented with antibiotics at the following concentrations: gentamicin 20 µg·mL −1 , ampicillin 50 µg·mL −1 , spectinomycin 100 µg·mL −1 , cycloheximide 200 µg·mL −1 and rifampicin 100 µg·mL −1 . For each experiment, bacteria were freshly plated from deep-frozen (−80 °C) glycerol stocks. Insertion of a gfp reporter cassette into the amyE locus in the genome of B. subtilis MB 73/2 was performed by transformation of chromosomal DNA obtained from B. subtilis OMG982 into the recipient strain and homologous recombination. Protoplast electroporation method, described by Romero et al. [19], was applied to introduce DNA into the B. subtilis MB 73/2 cells.

Bacterization of the Potato Seed Tubers and Plant Growth Conditions
Certified potato seed tubers cultivar (cv.) Krasa, caliber 35-45 mm, were obtained from the Plant Breeding and Acclimatization Institute (Instytut Hodowli i Aklimatyzacji Roślin-IHAR, Bonin, Poland). Bacterial cells from an overnight culture on LB agar plates, grown at 28 °C, were harvested and suspended in 1% carboxymethylcellulose (Calbiochem, Merck KGaA, Germany) to obtain an optical density of 6 units in the McFarland scale (approx. log 8 CFU·mL −1 , determined by plating). Carboxymethylcellulose was added to the inoculation mixture to enhance the adhesion of bacterial cells to the tubers, as previously described by Geels et al. [22]. The tubers were coated with the suspension and potted in 12 cm diameter pots filled with potato field-derived soil (source: Lake Golubie area, Kaszuby, Poland; soil type: very light soil, granulometric group: sand). Tubers coated with bacteria-free 1% carboxymethylcellulose were potted to serve as control samples. For extraction and dilution plating, the plants were grown for three weeks. Growth conditions were as follows: temperature 21 ± 2 °C, humidity 80 ± 5%, white light (Cool White, model TLD 58W/84o, photosynthetic photon flux 20 µM·s −1 ·m −2 ; Koninklijke Philips Electronics N.V., Eindhoven, The Netherlands) and a long day photoperiod (16 h light/8 h dark). For the microscopic analysis, four-week old plants were used, reaching potato growth stage II (vegetative growth) to early III (tuber initiation).

Sample Preparation for Microscopy
Plants were removed from the pots and excess soil was removed by shaking. Root samples (sections from stolon to the root tip) were cut, soaked in sterilized ddH 2 O and placed in sterile plastic Petri dishes. Enrichment medium, cooled down to 42-45 °C, was gently poured over the sample to embed the roots. The base of the enrichment medium was composed of: 2 g·L −1 KH 2 PO 4 , 4.66 g·L −1 K 2 HPO 4 , 1.33 g·L −1 (NH 4 ) 2 SO 4 , 50 mg FeSO 4 , 80 mg MgSO 4 , 10 g·L −1 agar, pH 7. For M73/2 GFP, spectinomycin (100 µM·mL −1 ) and glycerol (0.4%) were added. For A44 GFP and P482 GFP, the medium was supplemented with gentamicin (20 µg·mL −1 ) and glucose (0.4%). After settling, the samples were incubated at 28 °C for 36 h. When visualization of plant tissue was difficult using autofluorescence alone, the embedded roots were sprayed with 1 mg·mL −1 water solution of Fluoresent Brightner 28 (also known as calcofluor) (Sigma-Aldrich, St. Louis, MO, USA), 20 min prior to sample analysis.

Fluorescence and Confocal Laser Scanning Microscopy
Root preparations were analyzed without removing the roots from the surrounding medium (directly in the Petri dish). For initial visualization of bacterial mini-colonies on the roots, classical fluorescence microscopy analysis was performed with the Nikon Eclipse TE 300 inverted microscope (Nikon, Tokyo, Japan), equipped with a GFP-suitable filter system (excitation filter: 450-490 nm, barrier filter: 520 nm). Images were collected using a high resolution color digital camera (Hammamatsu Photonics C4742-95, Hammamatsu, Japan) and processed with LUCIA Image Software. The 10× objective was used. Due to the limitations of classical fluorescence microscopy when it comes to observing thick objects, it is difficult to obtain a sharp image of the whole root section. Thus, the images of the adjacent root surfaces were juxtaposed manually using graphic-processing programs (Corel Photo Paint and Microsoft Power Point). The confocal images were acquired using a Leica TCS LSI Macro Confocal (Leica Microsystems, Wetzlar, Germany). This CLSM device, dedicated to the analysis of macroscopic objects, is equipped with a 5× macro-objective, automated optical zoom system and tunable excitation and detection spectra. Leica Application Suite software was used for confocal images processing. This software enables automatic assembly of mosaic images from multiple adjacent fields of view.

Estimation of Bacterial Cell Number by Dilution Plating
Rifampicin-resistant clones of all tested strains were selected under antibiotic pressure [23]. To confirm their identity, REP-PCR was performed as described by Versalovic et al. [24] and the profiles obtained for the mutants were compared with the ones for the wild type (wt) strains. Potato seed tubers were inoculated with A44 Rif, P482 Rif and MB73/2 Rif and potted (see Section 2.4 for details). Three weeks after, root samples were collected (1-2 g), placed in extraction bags (Bioreba) and weighed. Sterile saline (0.9% NaCl) was added to the each sample-containing bag in 1:9 (w/v) ratio and the material was homogenized. Serial dilutions of the suspension were plated on LB agar supplemented with cycloheximide (200 µg·mL −1 ) and rifampicin (100 µg·mL −1 ). Following 48 h of incubation at 28 °C, bacterial colonies were counted and the mean value of CFU per gram of root fresh weight (CFU·gfw −1 ) was determined. A total of 6-8 plants were used per treatment in two replicates.

Results and Discussion
Bacillus subtilis MB73/2, Ochrobactrum sp. A44 and Pseudomonas sp. P482 isolates obtained from different plant-associated environments (Table 1) [15,16] showed antagonism towards pectynolytic bacteria of genera Dickeya and Pectobacterium, important pathogens of potato [25]. P482 and MB 73/2 inhibit the growth of the pathogens by production and secretion of the unidentified antibiotic compounds, whereas A44 disturbs quorum sensing-dependant regulation of virulence factors production by the pathogens cells [16,17]. For this study, inoculation of tubers was chosen since seed tuber treatment is broadly used in potato production to prevent seed-borne and soil-borne diseases.
To selectively re-isolate the A44, P482 and MB73/2 cells from the rhizosphere and to estimate their number, a classical microbiology approach was applied, involving the use of spontaneous rifampicin resistant mutants and dilution plating. Potato root samples (cv. Krasa) were analyzed three weeks after tuber inoculation and potting. All three strains were recovered from the root samples. The following cell numbers were detected: A44 Rif log 4.29 ± 0.58 CFU·gfw −1 , P482 Rif log 3.75 ± 0.29 CFU·gfw −1 , MB 73/2 Rif log 3.2 ± 0.08 CFU·gfw −1 . Population size of an introduced PGPR usually drops rapidly with time [26]. With regard to the time of sampling, we could consider A44 an efficient potato root colonizer and P482 and MB73/2 as moderate to weak ones. This is, however, rather arbitrary. Despite the extensive research conducted on the biology of PGPRs, the concept of 'efficient colonization' is not uniformly defined. The studied plant-microbe systems, as well as the experimental setups employed by different research groups, often vary to such an extent that the results cannot be compared. Apart from the specificity of each plant-microbe system, the following variables can be distinguished: the size of the initial inoculum, the inoculation method, the conditions of plant growth, the time of sample harvesting post inoculation and the unit adopted to express the results (CFU·gfw −1 , CFU·cm −1 or CFU·sq·cm −1 ). Probably the best way to define 'efficient colonization' is to judge it by the lack or presence of a desired effect that colonization has on the host plant (if known). The efficient colonization 'threshold' can differ depending on the effect of interest (i.e., induction of systemic resistance, competition with pathogens, bioremediation of toxic compounds).
The main aim of this work was to visually demonstrate the presence of A44, P482 and MB73/2 on the roots of soil-grown potato plants. For this purpose, fluorescence and CLSM were employed. To provide the fluorescence signal, tested strains were GFP-tagged. The Gram negative Ochrobactrum sp. A44 and Pseudomonas sp. P482 were labeled with gfp borne on the pPROBE-GTkan vector under a constitutive nptII promoter [18]. The pVS1/p15a ori present on the plasmid enables its stable propagation in bacterial cells without selection [27]. Plasmids from the pPROBE series were already successfully applied for tagging bacterial cells for flow cytometry and microscopy (i.e., [8,27,28]).
Cells of B. subtilis MB73/2 proved to be very difficult to transform. This is not an uncommon problem. Environmental isolates are more problematic to genetically modify than the laboratory strains, especially when it comes to Gram positive bacteria. To overcome this obstacle, we have applied the protoplast electroporation method, described previously by Romero et al. [19]. It enabled us to introduce the λ R -gfp fusion from OMG 982 [14] into the amyE locus of the MB73/2 genome and to obtain constitutive GFP expression in the modified cells.
Preparation of samples for microscopic analysis involved embedding of the harvested roots in a specially formulated enrichment medium. Enrichment was already proved useful for showing bacterial infection of plant tissue [29]. In this study, the composition of the medium enabled in situ growth of mini-colonies of the studied PGPR candidate while inhibiting the growth of most other microorganisms. The samples were not removed from the embedding matrix for microscopic observation, in this way enabling preservation of the original colonization sites. Enrichment approach applied in this study enables fast screening for bacterial colonization sites along the roots even in case of sparse target microbial populations. It is important as certain strains can prefer to colonize different segments along the root (i.e., the root base or the root tip) [30]. Using this method, in combination with   (Figure 2(A)), induced with a 405 nm laser line, although other stains can also be used, i.e., propidium iodide [33]. The employed enrichment method, by triggering in situ growth of mini-colonies, enabled visualization of bacterial colonization sites on relatively large root fragments (Figures 1(A) and 2(A)). As a point of reference, magnifications used in other colonization studies are typically five to ten times greater, resulting in a smaller field of view (i.e., [34]). The few exceptions concern observation of high-density bacterial populations, forming relatively large bacterial patches on the root surface [9]. The enrichment approach removes the limitations encountered when tracking individual bacterial cells on a relatively large surface area of the root, making it a helpful tool in studying colonization pattern of small and scattered bacterial populations.
The method described in this study can be easily applied to other bacterial strains by adjusting the composition of the enrichment medium. Regarding the possibilities offered by the modern CLSM devices (multiple fluorescence channels available), one could modify the method described above to serve co-colonization study. This would require tagging the second strain with a different fluorescent protein reporter in concert with the same plasmid-or genome-borne antibiotic resistance marker for both strains.

Conclusions
In this study, we have evaluated the potential of three candidate PGPRs-B. subtilis MB73/2, Ochrobactrum sp. A44 and Pseudomonas sp. P482-to colonize the potato rhizosphere following seed tuber bacterization. Rifampicin-resistant mutants and dilution plating were used for a quantitative study. Regarding the inoculation conditions and the time of sampling, we can consider A44 an efficient root colonizer, and both P482 and MB 73/2 a moderate ones.
To detect bacterial colonization sites on the potato roots, we have created GFP-tagged derivatives of the tested strains and designed an enrichment sample preparation method for classical fluorescence microscopy and CLSM. Both methods proved suitable for observation of bacterial colonization sites. However, although fluorescence microscopy enables fast overview of the preparations, it is not very suitable to handle thick preparations (i.e., roots). Obtaining a sharp image of a larger root section required manual juxtaposition of many adjacent fields of view. Confocal microscopes designed to observe large structures (i.e., embryos, tissue fragments) have the advantage of a large field of view, uncommon in most confocal microscopes, dedicated to research at the cell level. Combining this feature with a spectral detector and a wide range of available fluorescent proteins, stains and probes makes it a powerful investigation tool for studying the physiology of plant roots. This also includes research on the ability of PGPRs with potential agricultural application to colonize the rhizosphere of the target plant species. The enrichment method described in this study enables in situ growth of minicolonies on the root surface. This removes limitations resulting from the small size of bacterial cells in comparison to relatively big size of the plant root. In concert with wide field microscopy, the described approach enables fast visualization of bacterial colonization sites on large root fragments, even in case of small and scattered bacterial populations. Considering that the composition of the enrichment medium can be easily modified, the proposed method can also be used for selective enrichment and detection of other fluorescently-tagged bacterial strains.