A short overview of available auto fluorescent proteins and their properties

The first application of green fluorescent protein (GFP) isolated from the jellyfish Aequorea victoria as a reporter (Chalfie et al. 1994) has become a hallmark in modern biology and is used throughout a wide range of different biology and biotechnology research areas including many fields of microbiology and cell biology. AFPs have become the most important reporters providing new tools to mark whole cells, study protein localization, and monitor gene expression and molecule interactions in vivo in a non-invasive way thereby preserving the integrity of the cell. Advantages of the use of GFP in comparison with other reporters or dyes is that GFP is present within the cell as a product of gene expression and that the visualization does not require any fixation or preparation protocols, which are time-consuming and might result in artifacts or influence cellular properties. Furthermore it does not require substrates or additional energy such as often is the case in bioluminescence. Additional advantages are that GFP due to its tertiary barrel structure is very stable and can be applied in many different species. In addition, GFP labelled cells can be used for flow cytometry analysis and quantitative analysis by PCR (Utermark and Karlovsky 2006). Disadvantages of GFP are that its structure and fluorescence is dependent on pH and the presence of oxygen (Heim et al. 1994). However, studies on Rhizobium tagged with a GFP-derivative showed that GFP was well visualized in bacteroids present in root nodules, which is an oxygen-limiting environment (Gage et al. 1996; Stuurman et al. 2000). Since GFP is not an enzyme it does not have amplifying reporter ability, such as, for example, LacZ. Successful GFP visualization is therefore a balance between sufficient production and prevention of overproduction that could cause toxicity or a metabolic burden to the cell resulting in growth inhibition. The latter should be tested for each bacterial species. Cloning of e-gfp under control of the lac promoter on a high copy number plasmid in E.coli results, for instance, in a severe growth inhibition when grown on Luria-Bertani agar medium (unpublished results). After 2 days of incubation white colonies grow out of the bacterial streak representing spontaneous gfp mutants with a growth advantage. Another point of consideration before applying GFP is the autofluorescence background or noise from the environment in which the bacteria are to be analysed. For example, sand and other soil particles can severely hamper GFP visualization as well as certain plant structures or organelles such as chloroplasts. Such problems might be solved by using other auto fluorescent proteins with different excitation and emission wavelength spectra.

Modifications of GFP (often by gene shuffling experiments) have resulted in the isolation of mutants that have shifted emission and excitation wavelengths, which give the opportunity to use multiple auto fluorescent proteins in one system in order to differentiate between different cells or to visualize different processes within one cell. In addition, such variants are also brighter and more stable. Important GFP derivatives are Enhanced GFP (EGFP), Enhanced Cyan Fluorescent Protein (ECFP) and Enhanced Yellow Fluorescent Protein (YFP) (Yang et al. 1998; Tsien 1998; Matus 1999; Ellenberg et al. 1999). Blue fluorescent protein (BFP) has also been developed but is less used due to its low brightness. In addition many other gfp derivatives have been isolated or developed with enhanced brightness and optimized codon usage for optimal expression in different organisms. A wide range of auto fluorescent proteins with useful background information is available on the website of the Clontech company (http://www.clontech.com/). Although the stability of GFP is very advantageous for tracking studies, it hampers transient gene expression studies. Andersen et al. (1999) have developed a set of GFP derivatives with reduced half-lifetimes by the addition of short amino acid tags to the C-terminus, recognized by specific proteases widely present in bacterial cells, which usually break down partially produced proteins.

Although many efforts were made to isolate a red fluorescent derivative of GFP, this has never been achieved and was bypassed by the discovery and application of Red Fluorescent Protein (RFP or DsRed) isolated from the coral Discosoma striata (Matz et al. 1999). Since the rfp sequence is not homologous to gfp the use of both genes in one cell will not result in unwanted recombinations. An efficient use of DsRed is hampered by its slow maturation due to its tetramerization, which is required for its fluorescent properties, and its toxic properties when overproduced. Recently, several improved DsRed derivatives have been constructed to overcome these problems. One of these new derivatives DsRed.T3_S4T, which maturates faster (Sorensen et al. 2003) was successfully applied in Pseudomonas spp. for rhizosphere studies in being brighter and without causing loss of competitive colonization ability (Dandie et al. 2005). A more recent paper by Shaner et al. (2004) reports on the construction of improved monomeric red, orange and yellow fluorescent proteins derived from DsRed, which mature more efficiently, are more tolerant to N-terminal fusions and have an improved photostability. These forms have not been reported for studies of PGPR.

Genetic tools to mark microorganisms with auto fluorescent proteins

A requirement of the application of AFPs is that the genes encoding these proteins have to be transformed into the bacteria. This requirement can prevent the use of GFP as a marker when a certain bacterial species or strain is not genetically accessible with the available transformation protocols. AFP genes are usually delivered on plasmids or transposons (Bloemberg et al. 2004). The advantage of a plasmid is that it is present in multiple copies, which can improve the production of AFPs and does not disrupt host genes by chromosomal integration.

Since antibiotics cannot usually be applied for in vivo studies of PGPR on the plant the development of a set of broad host range plasmids that are stably maintained (in Gram-negative bacteria) without antibiotic pressure has been extremely valuable (Heeb et al. 2000). We have used such plasmids to construct a set of marker plasmids carrying egfp (green), ecfp (cyan), eyfp (yellow) ebfp (blue) and rfp (red), which were successfully used for the visualisation of PGPR such as Pseudomonas biocontrol strains and Rhizobium strains respectively (Stuurman et al. 2000; Bloemberg et al. 2000). The value of these vectors was also shown for the analysis of the improved DsRed form (Dandie et al. 2005). Another point of attention for an efficient marking is the use of an appropriate promoter. Both the use of a tac and lac promoter were shown successful for the constitutive expression of AFPs in Pseudomonas and Rhizobium and many other Gram-negative bacterial strains (Bloemberg et al. 1997, 2000) (Fig. 1). During the past 10 years valuable transposon constructs have also become available, that carry mainly gfp derivatives for marking and/or gene expression studies (Burlage et al. 1995; Tombolini et al. 1997; Unge et al. 1997; Xi et al. 1999).

Fig. 1
figure 1

Confocal laser scanning microscopy (CLSM) analyses of Pseudomonas biocontrol strains and the pathogen Fusarium oxysporum f.sp. radicis lycopersici marker with auto fluorescent proteins in the tomato rhizosphere. Panels (A) P. fluorescens WCS36 marked with Green Fluorescent Protein (GFP) on the tomato root surface; (B) Colonies of P. fluorescens WCS365 after inoculation of a mixed culture of GFP and cyan fluorescent protein (CFP) marked derivative strains; (C and D) Simultaneous imaging of P. fluorescens WCS365 marked with red fluorescent protein (DsRed) and Fusarium oxysporum f.sp. radicis lycopserisci marked with GFP. (images were produced by A. Wijfjes, A. Lagopodi and A. Bolwerk)

Full size image

Since fungi are frequently part of the endogenous microflora and can even be the direct target for the PGPR effect as in case of biological control, it is of great relevance to tag fungi and study their interaction with the PGPR. However, genetic transformation of fungi is usually more difficult than transformation of bacterial cells, due to the presence of the rigid cell wall and need for stable integration of genetic material in the chromosome. A classic method consists of the preparation of protoplasts, which will subsequently take up genetic material, mostly in the form of plasmids, which will integrate into the chromosome. Consequently, fungal transformants have to be tested for the conservation of phenotypic properties such as morphology, growth and pathogenicity before using these transformants in confrontation studies. Using an optimized protoplast transformation protocol we have tagged Fusarium oxysporum f.sp. radicis lycopersici with different autofluorescent proteins by co-transformation with two plasmids of which one contained a hygromycin resistance gene and the second an afp gene (Lagopodi et al. 2002; Bolwerk et al. 2003). In addition to protoplastation methods, fungi can be transformed by ballistic bombardment and more recently by the transformation ability of Agrobacterium (de Groot et al. 1998), which has been shown to be a simple, efficient and successful transformation method for many fungi (Michielse et al. 2005).

Some examples of the use of auto fluorescent proteins for the visualization of PGPR in the plant environment

Initial studies using auto fluorescent proteins (in specific GFP) as markers for PGPR were used for localization studies. Most of these studies showed that PGPR and other microorganisms such as phytopathogenic fungi preferentially colonize the junctions between the root cells (Bloemberg et al. 1997, 2000; Tombolini et al. 1999; Lagopodi et al. 2002; Bolwerk et al. 2003; Gamalero et al. 2005) (Fig. 1). Most steps in the symbiosis process between Rhizobiacea and leguminous plants occur inside the root after entrance into the infection thread. The use of GFP has allowed us to visualize the process of attachment, entrance and nodule occupancy in great detail (Gage et al. 1996) making it even possible to determine the growth rate of the cells in the infection thread (Gage et al. 1996). Stuurman et al. (2000) showed that GFP tagged Rhizobium bacteroids move in the root nodule. Since the root is sometimes too thick for successful imaging due to loss of light in the deeper root parts, sectioning of the plant material can solve this. This is also used for the study of endophytes as was shown by several publications on Herbaspirillum spp. (Elbeltagy et al. 2001) and the pathogen invasion of Xylella fastidiosa (Newman et al. 2003). When necessary plant material can also be stored before visualization by fixation with paraformaldehyde, which leaves GFP intact for fluorescent studies (Stuurman et al. 2000; Elbeltagy et al. 2001). Confocal laser scanning microscopy (CLSM) analysis of the colonization behaviour of afp-tagged antagonistic strains can also provide important information on the sampling strategy required for monitoring inoculant strains and combine these with data on the kinetics of the endogenous microflora (Gotz et al. 2006).

Localization studies are also valuable for the analysis of the properties of mutant strains, for instance for their adhering abilities. For example Biancotto et al. (2001) showed by using mucoid mutant strains of P. fluorescens CHA0, that acidic extracellular polysaccharides (EPS) are an important factor in adhesion to roots and fungi.

The ability to mark cells with different AFPs makes it possible to visualize simultaneously different species and populations. We have shown that it is possible to distinguish three different populations of P. fluorescens WCS365 cells tagged with ecfp, egfp and rfp simultaneously in the rhizosphere (Bloemberg et al. 2000). By looking at the ratios of differentially tagged cells present within the microcolonies it is suggested that most colonies are initiated by a single cell after which other cells from outside can still join the colony later in time as they will also leave colonies to colonize other parts of the root (Bloemberg et al. 2000). When visualizing two different organisms or populations a combination of GFP and DsRed (preferentially the improved forms) is most deficient since their spectra hardly overlap. Studies using mixed populations of two species, e.g., P. chlororaphis PCL1391 and P. fluorescens WCS365 showed that mixed colonies were formed, which were mostly present on the upper root part and that P. chlororaphis had a preference for colonizing the root hairs (Dekkers et al. 2000). The study of bacterial communities is also important to understand collaborations between bacteria as shown for instance for P. putida strains that are able to degrade polyaromatic hydrocarbons, which is stimulated by the presence of polyaromatic hydrocarbons suggesting that sensing processes, such as chemotaxis, are involved (Kuiper et al. 2002). Studies on phase variation of Pseudomonas brassicacearum, in which Phase I cells were labelled with GFP and Phase II cells with DsRed showed that Phase I and II cells colonize different niches on the root of Arabidopsis thaliana (Achouak et al. 2004). Whereas Phase I cells were mostly observed at the basal part of the root, Phase II cells were observed at secondary roots and root tips, which can be explained by the fact that Phase II cells make flagellin and are more motile. Also competition studies have been performed other than for Pseudomonas spp. such as for S. meliloti populations for which it was shown that mixed populations can be present in infection threads, which subsequently can result in mixed populations in the root nodule (Gage 2002).

Another valuable possibility for dual imaging is the visualization between biocontrol agents and phytopathogenic fungi that they control. After visualization of the infection process of Fusarium oxysporum f.sp. radicis lycopersici marked with GFP on tomato resulting in tomato foot and root rot (Lagopodi et al. 2002) interactions between Pseudomonas biocontrol species (tagged with rfp) and Fusarium were performed to obtain a better fundamental understanding of their interactions in the rhizosphere (Bolwerk et al. 2003) (Fig. 1, panels C and D). These studies showed that Pseudomonas and Fusarium compete for the colonization of same niches (intercellular junctions) and directly interact with each other. At sites where bacteria were present, infection of the root by penetration of Fusarium was not observed. In addition, Pseudomonas attached to the Fusarium hyphae and were able to extensively colonize those hyphae similar to what was observed during in vitro confrontation assays (Bolwerk et al. 2003). Molecular mechanisms involved in the attachment and colonization of the hyphae are hardly known. In the presence of P. chlororaphis PCL1391 many stress responses of the Fusarium were observed such as loss of growth directionality, increased vacuole formation, curly growth, swollen bodies and increased branching. It was shown that the antifungal compound produced by P. chlororaphis strain PCL1391 is (partially) responsible for these stress effects. We have also used GFP and CFP-expressing Fusarium strains to visualize the effect of biocontrol strains on F. oxysporum. Studies using gfp labelled Trichoderma atroviride and the (non-labelled) phytopathogenic fungi Pythium ultimum and Rhizoctonia solani on cucumber seeds showed that T. atroviride reacted to the presence of these pathogens by increased branching and the formation of morphological changes similar to hooks, appressoria, and papillae (Lu et al. 2004). AFPs will be an important tool for further elucidation of the molecular mechanisms involved in bacterium-fungus interactions.

Auto fluorescent proteins applied as tools for the visualization of ecological processes

Besides the purpose of localization of bacterial cells in their environment, auto fluorescent proteins are progressively used for other applications such as gene expression analysis, biosensor reporter systems, identification of specific environmental expressed genes and analysis of horizontal gene transfer in the environment.

The instable GFP variants made by Andersen et al. (1999) facilitate the analysis of transient gene expression in the rhizosphere. For instance, they were used to monitor the ribosomal activity of P. putida cells (Ramos et al. 2000). More recently, they were applied for the monitoring of anti-fungal metabolite production by the biocontrol agent P. fluorescens CHA0 (Baehler et al. 2005). Promoter regions of the biosynthetic genes of the three antifungal metabolites produced by CHA0, e.g., 2,4-diacetylphloroglucinol (DAPG), pyoluteorin (PLT) and pyrrolnitrin (PRN), were fused to GFP and their expression kinetics analysed in vitro. It will be exciting to apply these constructs in the rhizosphere and in the presence of fungi to analyze the influence of the fungus on the expression of the anti-fungal metabolite genes.

Stable GFP can also be used for expression studies in order to show that a gene of interest is expressed under certain conditions, without being able to analyze the transient expression profile. For example, it was shown by Rothballer et al. (2005) that the ipdC gene of Azospirillum brasilense, which is involved in indole-3-acetic acid production, is indeed expressed in the wheat rhizoplane. Other GFP-based expression systems were constructed to learn more about the direct interaction between the plant and the bacterium and the conditions that bacteria encounter in the rhizo- and phyllosphere. For example Leveau and Lindow (2001) applied an expression system that showed that the use of sugars, e.g., fructose and/or sucrose is mainly responsible for the growth of Erwinia herbicola on bean leaves. The construction and application of two bacterial sensors for the detection of nitrate availability indicated that roots compete for nitrate with the microbial rhizosphere population (DeAngelis et al. 2005). Studies on thiamine synthesis gfp reporter systems in Rhizobium leguminosarum bv. viciae showed that the rhizosphere of vetch and pea is poor in thiamine and that thiamine production is induced in the rhizosphere (Karunakaran et al. 2006). Another GFP-based study by Aldon et al. (2000) showed that physical contact between the bacterial cell and the plant strongly induces the expression of the hrp genes. The use of bioreporters has significantly contributed to the fundamental understanding of how bacteria interact with the plant (Leveau and Lindow 2002).

The development of GFP-based biosensor systems for the detection of AHL production were used to show that AHLs are produced in the microcolonies present in the rhizosphere, which will enable cross-talk between species in the rhizosphere (Andersen et al. 2001; Steidle et al. 2001).

A real challenge for researchers is to identify genes that are specifically expressed in the natural environment in order to understand the functioning of bacteria in their natural environment and to elucidate novel processes. In vivo expression technology is an elegant and powerful example of how environmentally-expressed genes can be identified (Rediers et al. 2005). In addition, other systems have been developed such as differential fluorescence induction (DFI) and optical trapping microscopy (Allaway et al. 2001), which for example allowed the identification of a rhizosphere-specific putative ABC transporter of putrescine (Allaway et al. 2001). Further development of systems for promoter trapping will facilitate the identification of other plant-associated genes (Izallalen et al. 2002).

When dual fluorescence imaging systems are not available, reporter systems based on two different techniques were shown to be of great value. For example, constructs carrying gfp and gusA in mini-Tn5 transposons (Xi et al. 1999) or plasmids (Ramos et al. 2002) were applied for the imaging of Azospirillum brasilense on wheat roots. These analyses showed that high concentrations of Azospirillum were present in the intercellular root cell spaces and at points of lateral root emergence, which are niches where nutrients presumably leak out of the root (Xi et al. 1999; Ramos et al. 2002). Other examples of combined reports are the use of immunofluorescence and a rRNA-targeting probe, which was used to analyze P. fluorescens DR54 in the sugar beet rhizosphere, showing that most metabolically active cells are found at the root tip and that endogenous soil microorganisms start colonizing the rhizosphere 2 days after seed inoculation (Lübeck et al. 2000). Another way to analyze metabolic activity is the use of luciferase, the activity of which is dependent on the cellular energy level. In combination with GFP, a construct containing luxAB genes was applied for P. fluorescens SBW25 showing that SBW25 was metabolically active in the whole rhizosphere (Unge et al. 1999; Unge and Jansson 2001), which can explain why the rhizosphere of wheat is more extensively colonized than, for example, the tomato rhizosphere.

As a last example I want to draw attention to recent work published by Molbak et al. (2006), in which horizontal gene transfer in the rhizosphere of pea and barley was analysed by CLSM studies. For that purpose they used P. putida KT2442 carrying a plasmid expressing gfp, which could be received by P. putida LM24 expressing dsRed. Their studies showed that plasmid transfer in the rhizosphere is determined by cell density and cell distribution, which are determined by exudation and root growth.

Future prospects

The application of auto fluorescent proteins will make a continuous contribution to the understanding of how PGPR function during their beneficial interactions with the plant and the endogenous microflora. Studying the molecular basis of the interactions between fungi and bacteria is an emerging field with great relevance in plant microbiology. Future studies will highly benefit from the developed tools for visualization of these organisms.