Monitoring in planta bacterial infection at both cellular and whole-plant levels using the green fluorescent protein variant GFPuv


Author for correspondence: Kirankumar S. Mysore Tel: +580 224 6740 Fax: +580 224 6692 Email:


  • • Green fluorescent protein (GFP) labeling of bacteria has been used to study their infection of and localization in plants, but strong autofluorescence from leaves and the relatively weak green fluorescence of GFP-labeled bacteria restrict its broader application to investigations of plant–bacterial interactions.
  • • A stable and broad-host-range plasmid vector (pDSK-GFPuv) that strongly expresses GFPuv protein was constructed not only for in vivo monitoring of bacterial infection, localization, activity, and movement at the cellular level under fluorescence microscopy, but also for monitoring bacterial disease development at the whole-plant level under long-wavelength ultraviolet (UV) light.
  • • The presence of pDSK-GFPuv did not have significant impact on the in vitro or in planta growth and virulence of phytobacteria. A good correlation between bacterial cell number and fluorescence intensity was observed, which allowed us to rapidly estimate the bacterial population in plant leaf tissue. We demonstrated that GFPuv-expressing bacteria can be used to screen plants that are compromised for nonhost disease resistance and Agrobacterium attachment.
  • • The use of GFPuv-labeled bacteria has a wide range of applications in host–bacterial interaction studies and bacterial ecology-related research.


The green fluorescent protein (GFP) from Aequorea victoria is an excellent nondestructive in vivo imaging marker for understanding biological systems and has been extensively used in microbiology and cell biology in both prokaryotes and eukaryotes (Chalfie et al., 1994). As it does not require exogenous cofactors or any substrates for fluorescence, GFP and its variants with improved properties and altered colors have been recently used as prominent reporters for real-time analysis of the spatial and temporal dynamics of gene expression, protein localization, protein–protein interactions, and cell-to-cell communication in living tissues (Zhang et al., 2002; Hu & Kerppola, 2003). In studies involving plant–microbe interactions, GFP labeling of various microorganisms including fungi, bacteria, and viruses has been successfully used to monitor their activities in planta at the cellular level (Oparka et al., 1997; Stuurman et al., 2000; Bendahmane et al., 2002; Lagopodi et al., 2002; Lu et al., 2004). For example, research projects related to plant viral diseases have employed this nondestructive imaging technique to investigate virus invasion and spread in a whole plant under ultraviolet (UV) light (Oparka et al., 1997; Bendahmane et al., 2002). Monitoring of microbe infection will help us to better understand the initial steps of infection (Lagopodi et al., 2002), the localization of bacteria and fungi in the infected plant tissues (Gage et al., 1996), and the movement of viruses in planta (Tamai & Meshi, 2001).

There are two potential problems associated with the use of GFP-labeled bacteria in investigating plant–bacterial interactions. To enable use of GFP as a marker for monitoring bacterial activities in planta, the GFP gene must be stable in the labeled bacterial cells throughout the process of infection, colonization, multiplication and movement in the infected plants. This is the major concern when the GFP gene is introduced using a plasmid vector, as the bacteria could lose the plasmid containing the GFP gene during their multiplication (Gage et al., 1996; Stretton et al., 1998). Although this problem can be overcome by integrating the GFP gene into the bacterial genome using transposons such as Tn5, many biological properties, such as pathogenicity, competition and survival, have to be examined before it can be used as a marker strain for further analyses (Grall & Manceau, 2003; Newman et al., 2003). In addition, Tn5 reporter gene systems are not effective in all Gram-negative bacteria (Sangari & Agüero, 1991). The second concern is the expression levels of GFP required in bacteria to enable its detection in planta, particularly when GFP is expressed under a weak promoter or is present in a single chromosomal copy. These limitations restrict the broad and efficient application of GFP to the investigation of plant–bacterial interactions.

Recently, GFP has been successfully applied to visualize bacterial colonization on the surfaces of plant leaves and flowers as well as in cross-sections of grapevine (Vitis vinifera) petioles using a confocal laser scanning microscope (Newman et al., 2003; Sabaratnam & Beattie, 2003; Spinelli et al., 2005). However, the GFP fluorescence emitted by these bacterial colonies was not strong enough to be visualized by the naked eye. Visualization of bacterial infection in planta would be useful for the screening of plant mutants that are compromised for disease resistance. Real-time monitoring of bacterial infection in living plant tissues requires not only high stability of the GFP protein but also high expression levels of GFP in a given bacterium. Recently, new GFP variants with shifted excitation and emission wavelengths have been developed for different color imaging and enhanced brightness (Zhang et al., 2002; Miyawaki et al., 2003). One of these GFP variants, GFPuv (mutated by in vitro DNA shuffling), produces 45-fold brighter green fluorescence in Escherichia coli cells than the wild-type GFP, while it retains the same excitation and emission maxima (Crameri et al., 1996). GFPuv-labeled bacterial colonies can be easily visualized with the naked eye under long-wavelength UV light without the requirement of additional specialized equipment. In contrast to the wild-type GFP, which largely produces nonfluorescent inclusion bodies, GFPuv is a soluble protein and has lower toxicity to bacteria (Crameri et al., 1996).

Here we report the development of a stable and broad-host-range plasmid vector, pDSK-GFPuv, which expresses GFPuv (driven by a constitutive chloroplast promoter, psbA) at high levels. The GFPuv-labeled plant pathogenic bacteria not only can be easily visualized at the cellular level under a fluorescence microscope but also are clearly visible, as bacterial colonies, to the naked eye at the whole-plant level under long-wavelength UV light. We demonstrated that pDSK-GFPuv can be used for real-time monitoring of bacterial colonization patterns and for screening mutant or gene-silenced plants compromised for nonhost disease resistance and Agrobacterium attachment.

Materials and Methods

Bacterial strains, plasmids, and growth conditions

The bacterial strains and plasmids used in this study are described in Table S1 in the Supplementary Material. Escherichia coli DH5α was used as the host strain for all plasmids, except for pUTgfp which was maintained in E. coli CC118 (λpir). Escherichia coli strains and Agrobacterium tumefaciens GV2260 were routinely grown in Luria–Bertani (LB) medium at 37 and 28°C, respectively. Pseudomonas syringae pv. tomato strains T1 and DC3000 and P. syringae pv. tabaci were grown in King's B (KB) medium at 28°C. The following concentrations of antibiotics (Sigma, St Louis, MI, USA) were used individually or in combinations for selection: ampicillin, 100 µg ml−1; rifampicin, 100 µg ml−1; kanamycin, 50 µg ml−1. Bacterial growth was monitored by measuring optical density at 600 nm (OD600) and the number of viable cells was determined by counting colonies grown on appropriate plates supplemented with appropriate antibiotics.

Construction of plasmids

Plasmids pUC19, pUTgfp (Tombolini et al., 1997) and pGFPuv (ClonTech, Mountain View, CA, USA) were extracted using the QIAprep Spin Miniprep Kit (Qiagen Inc., Valencia, CA, USA). To enable use of the NdeI restriction site for convenient cloning of the GFPuv gene from its start codon, the NdeI restriction site at the position 185 in pUC19 was removed by filling in the overhanging sequences of the NdeI fragment with a Klenow fragment and carried out blunt-end ligation that resulted in plasmid pUC19a. The DNA fragment containing the constitutive psbA promoter and an efficient ribosome binding site from T7 gene10 was excised from pUTgfp (Tombolini et al., 1997) with BamHI and EcoRI and cloned into pUC19a, resulting in pUC19b. The BamHI restriction site internal to the GFPuv gene in pGFPuv was removed by polymerase chain reaction (PCR) point mutation without changing the amino acid sequence, resulting in pGFPuv1, which was used as a template to amplify the GFPuv1 fragment by PCR using pfu DNA polymerase (Stratagene, La Jolla, CA, USA) with the following primers: GFPuv1, 5′-TGAGTAAAGGAGAAGAACTTTCAC-3′ and GFPuv2, 5′-CCCGGATCCTTATTTGTAGAGCTCATCCATG-3′ with a BamHI restriction site (underlined and italicized). The amplified GFPuv1 fragment was digested by BamHI and cloned into pUC19b cut with NdeI and blunted with a Klenow fragment following by cutting with BamHI to make pUC19-GFPuv1. The sequence of GFPuv1 was confirmed by DNA sequencing. The final cassette containing the GFPuv1 gene downstream of the constitutive psbA promoter and ribosomal binding site (RBS) was removed from pUC19-GFPuv1 as a PstI-EcoRI fragment and inserted into pDSK519, resulting in pDSK-GFPuv (Fig. 1). The plasmid was transferred into various bacteria by electroporation using standard protocols or triparental mating (Figurski & Helinski, 1979) using pRK2013 as a helper plasmid. Fluorescent transformants, containing the plasmid pDSK-GFPuv, were selected on appropriate agar plates supplemented with kanamycin and identified under long-wavelength UV light.

Figure 1.

Schematic diagram of the plasmid pDSK-GFPuv which can express the green fluorescent protein variant GFPuv at high levels under the constitutive chloroplast promoter psbA (PpsbA) and a ribosomal binding site (RBS) from T7 gene10.

Quantification of green fluorescence intensity

Overnight cultures were used to inoculate fresh LB or KB media to obtain 0.1–0.15 OD600. The fluorescence and OD600 of the cultures were measured at 1-h intervals with four replicates. The fluorescence of the cultures was quantified using a Triad LT Multimode Reader (Dynex Technologies, Inc., Chantilly, VA, USA) with a black 96-well Fluorolux HB microplate (Dynex Technologies, Inc.) containing 200-µl culture aliquots with an excitation filter at 485 nm and an emission filter at 535 nm. To quantify the fluorescence of bacteria containing pDSK-GFPuv in plant leaves, plants were inoculated by vacuum infiltration (Kang et al., 2004). Leaf samples (two 0.5-cm2 leaf discs) were taken at 24-h intervals and ground in 1 ml of 1 × phosphate-buffered saline (PBS; 140 mm NaCl, 2.7 mm KCl, 10 mm Na2HPO4 and 1.8 mm KH2PO4, pH 7.3). Bacterial numbers in plant leaves were determined by dilution plating on KB agar plates supplemented with rifampicin, while the fluorescence of leaf extracts containing bacteria was quantified with four replicates using the same method as fluorescence quantification of bacterial cultures as described above. Data for the bacterial population of each sample were transformed onto the logarithmic scale and subjected to analysis of variance using Student's t-test for significance at P = 0.05.

Plasmid stability test

A single colony of the bacterial strain containing pDSK-GFPuv was used to inoculate appropriate liquid media without any antibiotics and was incubated for 24 h at 28°C with shaking. The 1-d-old bacterial culture was then transferred to fresh liquid media without antibiotics at a ratio of 1 : 1024 and was cultured under the same conditions. After a further 1 d of incubation, it was again transferred to fresh liquid media without any antibiotics at the same ratio and was cultured under the same conditions. This process was repeated for 10 consecutive days. Bacterial growth was measured before each transfer using OD600. Bacteria were considered to multiply through 10 generations between each transfer. The bacterial culture was plated on appropriate agar plates without antibiotics at 10-fold serial dilutions every 2 d and incubated at 28°C. The bacterial colonies were examined under long-wavelength UV light for the loss or presence of green fluorescence.

Suppressor of G2 allele of SKP1 (SGT1) gene silencing in Nicotiana benthamiana by virus-induced gene silencing (VIGS)

Nicotiana benthamiana L. seeds were germinated in Metro Mix 300 (Sungro Horticulture Distribution Inc., Bellevue, WA, USA) in a growth chamber at 26°C with 16 h daylight. Fertilizer along with a soluble trace element mix (The Scotts Co., Marysville, OH, USA) was applied to 2-wk-old seedlings with water in the tray. Three-week-old seedlings were transplanted to 4-inch pots containing BM7 (Berger Co., Quebec, Canada) and grown in the glasshouse under the following conditions: 24 ± 2°C, 70% humidity, and 16 h daylight supplemented with 50–100 µE s−1 m−2 light intensity. Agrobacterium tumefaciens strain GV2260 containing TRV1 and TRV2::SGT1 vectors was used for VIGS experiments as described previously (Kang et al., 2004; Ryu et al., 2004). TRV2::SGT1 was obtained from Dr Dinesh-Kumar, Yale University.

Inoculation of plants

Bacterial cells were collected by centrifugation of overnight culture at 2600 g for 10 min, washed three times and resuspended in 10 mm MgCl2 and diluted to a final concentration of 2–5 × 104 colony-forming units (cfu) ml−1 in 10 mm MgCl2 with 0.01% (volume/volume (v/v)) Silwet L-77 (Osi Specialties, Friendship, WV, USA) to facilitate infiltration. Plants were inoculated with P. syringae by vacuum infiltration. The inoculated plants were kept in a growth chamber at 22–24°C for symptom development. Bacterial growth in planta was examined by plating serial dilutions of leaf extracts in 10 mm MgCl2 on KB plates with appropriate antibiotics and calculated as cfu per square centimeter. At least four replicates of the experiments were performed.

Detection of green fluorescence and confocal microscopy

Bacterial colonies harboring plasmid pDSK-GFPuv were identified under long-wavelength UV light (Ultraviolet Products, Cambridge, UK). To visualize bacterial infection in leaves, the whole plants were directly examined under UV light in a dark room after plants had been inoculated with GFPuv-labeled bacteria. The photographs were taken using a regular digital camera. Fluorescent bacteria in leaf tissues were visualized at the cellular level using a Leica TCS SP2 AOBS confocal system (Leica Microsystems, Wetzlar, Germany). A small piece of plant leaf inoculated with bacteria was placed on a microscope slide, submerged in a deionized water droplet, covered with a glass slip and sealed with grease. Image acquisitions were carried out using excitation at 488 nm and by collecting emitted light from 500 to 600 nm with either a ×20 dry objective (numerical apperture (NA) 0.7) or a ×63 water immersion objective (NA 1.2).

Agrobacterium tumefaciens attachment assay

The axenic leaf discs of N. benthamiana were prepared by washing whole leaf explants with 10% Clorox for 10 min, followed by rinsing with sterile distilled water three times before making discs using a cork borer (0.5 cm). Ten to 15 aseptic leaf discs were cocultivated with either the disarmed strain A. tumefaciens GV2260 or the mutant strain chromosomal virulence (chv)A (A191) or chvB (A192) harboring the plasmid pDSK-GFPuv. Bacteria from a fresh overnight culture grown in LB broth were induced in 10 mm 2-(4-morpholino)-ethane sulfonic acid (MES) (pH 5.8) with 100 µm acetosyringone (final concentration) for 3 h at room temperature and diluted in saline water (0.9% NaCl), and made up to a final concentration of OD600 = 0.1 (1 × 108 bacteria ml−1). The leaf discs were incubated individually with different A. tumefaciens strains for 15 min, blotted on sterile filter paper to remove excess bacteria, and then transferred to Murashige and Skoog (MS) basal medium (4.32 g l−1 MS minimal salts (Gibco-BRL, Gaithersberg, MD, USA)). The leaf discs were sampled at 2, 12 and 24 h post inoculation, washed gently 3–4 times with PBS and briefly vortexed at very low speed for 5–10 min to remove unbound bacteria before observation under a microscope.

Results and Discussion

Development of a broad-host-range plasmid vector for expression of GFPuv in bacteria

Although GFP has been widely used as a prominent reporter for labeling various bacteria in order to monitor their localization and activities in infected plant tissues, to our knowledge, all applications of GFP-labeled bacteria have been examined in plant tissues under a fluorescent microscope (Newman et al., 2003). As the GFPuv protein, one of the GFP variants, produces strong green fluorescence and can be visualized under long-wavelength UV light, we proposed that GFPuv-expressing bacterial colonies could be visualized in planta with the naked eye under long-wavelength UV light.

We selected a broad-host-range plasmid, pDSK519 (a derivative of RSF1010), for expression of GFPuv, as this plasmid has been shown to be very stable in many Gram-negative bacteria and has a relatively high copy number (Keen et al., 1988). The genes cloned into the pDSK519 vector were expressed by the lac promoter. However, it has been reported that a GFP gene driven by the lac promoter is usually expressed at a very low level or even not expressed at all in Pseudomonas fluorescens and Psychrobacter sp. (Bloemberg et al., 1997; Stretton et al., 1998). To achieve strong and stable expression of GFPuv, we constructed a GFPuv expression cassette under a constitutive plastid promoter (psbA) from Amaranthus hybridus (Wolk et al., 1991; Kragelund et al., 1995) containing an efficient ribosomal binding site (RBS) from T7 gene10 in pDSK519 (see the Materials and Methods section). The psbA promoter was shown to be 18-fold more efficient than the T7 promoter for overexpression of a foreign gene in E. coli (Brixey et al., 1997). The resulting plasmid, pDSK-GFPuv (Fig. 1), produced extremely strong and uniform green fluorescence, which was easily visible to the naked eye under long-wavelength UV light, when expressed in E. coli (DH5α). pDSK-GFPuv also produced green fluorescence when transferred into economically important bacterial phytopathogens such as A. tumefaciens, P. syringae pv. tabaci, P. syringae pv. tomato, P. syringae pv. glycinea, P. syringae pv. syringae, Xanthomonas axonopodis pv. vesicatoria and X. campestris pv. campestris. Our data suggest that GFPuv in pDSK-GFPuv can be expressed at high levels in a wide range of bacterial species and can be used to monitor bacterial disease development in plants.

Stuurman et al. (2000) found that some plasmid vectors used for GFP expression were lost in half of the re-isolated bacteria (originally carrying the plasmid vector) from the infected plant tissues or after continuous culturing on plates. In some cases, expression levels of fluorescent proteins became relatively low or fluorescence was not uniform, although the vectors were maintained in the bacteria (Stretton et al., 1998; Stuurman et al., 2000). We therefore examined the stability of the plasmid pDSK-GFPuv in several bacterial species by consecutive transferring of bacterial cultures to a fresh liquid medium under nonselective conditions. Each bacterial culture transfer was diluted 1024-fold in the fresh liquid medium, i.e. 20 µl of bacterial culture to 20.48 ml of fresh medium, which corresponded to 10 generations of bacterial multiplication when bacteria grew to the same population size, i.e. the same OD600 value. After 100 generations, i.e. 10 consecutive transfers (approx. 10 d), all cells of P. syringae pv. tomato strains T1 and DC3000 and > 95% of P. syringae pv. tabaci and A. tumefaciens strain GV2260 retained the plasmid and exhibited uniform green fluorescence. The stability of the plasmid was also assessed in planta by infecting the bacterial strains expressing GFPuv on their respective hosts and estimating the bacterial population that expressed GFPuv in the leaves by serial-dilution plating. The results revealed that 95% of P. syringae pv. tomato T1 and DC3000 still carried the plasmid, while > 90% of bacterial cells of P. syringae pv. tabaci showed green fluorescence at 3 d post inoculation (dpi), suggesting that GFPuv-expressing strains could be reliably used to monitor bacterial population dynamics.

GPFuv did not affect the growth of phytopathogenic bacteria both in media and in planta

To examine whether the growth of GFPuv-labeled P. syringae pathovars and A. tumefaciens is impaired by the presence of the GFPuv-expressing plasmid or the expression of the GFPuv, we first assessed the growth of these strains with and without the plasmid pDSK-GFPuv in KB or LB liquid medium, respectively. Under the experimental conditions used in this study, expression of the GFPuv gene in these strains had no significant effect on the in vitro bacterial growth rate in liquid media (Supplementary Fig. S1). We then compared the in planta growth rate of the GFPuv-labeled P. syringae pathovars with that of the wild-type strains in the host plants. All inoculations were performed by vacuum infiltration of whole plants to achieve uniform infection. Similar disease symptoms developed 3 dpi for the GFPuv-labeled and the wild-type P. syringae pathovars (data not shown). The bacterial populations in plant leaves were examined for five successive days after inoculation by serial-dilution plating. No significant differences between the GFPuv-labeled and wild-type P. syringae pathovars were observed (Supplementary Fig. S2), suggesting that the expression of the GFPuv gene in the tested bacterial strains had no significant effects on their in planta growth and pathogenicity. We also compared the pathogenicity of GFPuv-labeled bacterial strains with that of the parental strains by spray inoculation and observed no significant differences in pathogenicity (data not shown). These findings suggest that GFPuv can be successfully used to study plant–pathogen interactions.

The green fluorescence intensity of P. syringae and A. tumefaciens containing pDSK-GFPuv was quantified using a Triad LT Multimode Reader (see the Materials and Methods section). Fluorescence intensity and bacterial cell number had a linear relationship (Fig. 2), with a minimum detectable cell number of ∼5 × 105 cfu ml−1 for P. syringae and ∼5 × 104 cfu ml−1 for A. tumefaciens. A linear relationship was also observed between fluorescence intensity and OD600 reading before the stationary phase of bacterial growth (Supplementary Fig. S3), indicating that GFPuv could be used to estimate bacterial populations.

Figure 2.

Correlation of fluorescence intensity and bacterial cell numbers during in vitro growth. Bacterial cells expressing the green fluorescent protein variant GFPuv were collected by centrifugation and resuspended in 10 mm MgCl2 buffer at OD600 = 1.0. Bacterial cell numbers (x-axis) were determined by serial-dilution plating. The fluorescence intensity (y-axis) of the cell suspension with 4-fold dilutions was quantified in 200-µL aliquots using a Triad LT Multimode Reader. cfu, colony-forming units. Error bars represent standard deviations of four replicates.

Use of GFPuv for in vivo real-time monitoring of bacterial infection in whole plants

Since the first report of the use of GFP as a marker in both eukaryotic and prokaryotic cells in the mid 1990s (Chalfie et al., 1994), GFP has proved to be an invaluable tool for monitoring gene expression, protein localization, cell-to-cell communication (Steidle et al., 2001), viral infection, and virus movement in plant systems. In the past 10 yr, many plasmids (Gage et al., 1996; Bloemberg et al., 2000) and transposons (Tombolini et al., 1997; Stretton et al., 1998) expressing the GFP gene have been used to label a broad variety of bacteria and fungi to investigate their infection and localization in planta. However, to our knowledge, visualization of pathogen infection of whole plants under UV light has been reported only for viral pathogens. Therefore, we tested the potential utilization of the plasmid pDSK-GFPuv for monitoring bacterial infection of whole plants under long-wavelength UV light.

Pseudomonas syringae pv. tabaci containing the plasmid pDSK-GFPuv was used to infect its hosts N. benthamiana and N. tabacum, while P. syringae pv. tomato strains T1 and DC3000 containing pDSK-GFPuv were used to infect their host plant tomato (Solanum lycopersicum L.) by vacuum infiltration at a low concentration (2–5 × 104 cfu ml−1). Bacterial infection was examined under long-wavelength UV light (365 nm). The infection sites of P. syringae pv. tabaci in N. benthamiana and N. tabacum leaves were clearly visible to the naked eye as numerous bright green spots under long-wavelength UV light at 3 dpi (Supplementary Fig. S4). Green fluorescence emitted from P. syringae pv. tomato strains on tomato leaves was also visible under long-wavelength UV light, but was not as strong as that seen on N. benthamiana and N. tabacum with P. syringae pv. tabaci infection (Supplementary Fig. S4). The relatively weak fluorescence intensity of P. syringae infections on tomato leaves when compared with infections on N. benthamiana leaves was probably a result of smaller bacterial populations in tomato leaves (less than 107 cfu cm−2 at 3 dpi) than in N. benthamiana leaves (approx. 108 cfu cm−2 at 3 dpi). Interestingly, the infection patterns of P. syringae pv. tomato strains T1 and DC3000 on tomato leaves were different, although they both caused bacterial speck disease of tomato. Uniform green spots were observed on tomato leaves infected by P. syringae pv. tomato strain DC3000, while clusters of green fluorescence were seen along veins in plants infected with P. syringae pv. tomato strain T1 (Supplementary Fig. S4).

We further tested whether GFPuv-labeled P. syringae in the infected plant leaves can be monitored in vivo, using fluorescence microscopes. Strong green spots in N. benthamiana leaves infected with P. syringae pv. tabaci containing pDSK-GFPuv were observed under ×3.2 magnification in a fluorescent stereo microscope, while single bacterial cells were clearly observed under ×63 magnification in a confocal laser scanning microscope (Fig. 3). Bacterial colonization was seen mainly in the plant apoplast and some bacterial cells were also observed on the leaf surface. Our data demonstrate that the plasmid pDSK-GFPuv is a suitable vector with which to label phytopathogenic bacteria for their real-time monitoring in planta.

Figure 3.

Visualization of bacteria labeled with the green fluorescent protein variant GFPuv in the leaves of Nicotiana benthamiana plants at the cellular level. Plants were inoculated with Pseudomonas syringae pv. tabaci containing pDSK-GFPuv at a low concentration (2 × 104 colony-forming units (cfu) ml−1) by vacuum infiltration. At 2 d post inoculation, a whole leaf was examined under a Nikon SMZ 1500 fluorescent stereo microscope at ×3.2 magnification (a). A small piece of the leaf sample was examined under a Leica confocal laser scanning microscope with excitation at 488 nm and emitted light collected from 500 to 600 nm. Images for green fluorescence and transmission were taken under ×10 (b), ×20 (c) or ×63 (d) magnification and merged together using Leica confocal software.

Use of GFPuv to screen plants that are compromised for disease resistance

Current screening methods for plants (mutants or gene-silenced plants) that are compromised for disease resistance mainly rely either on GUS or luc reporter genes or disease symptom development. Use of GUS or luc as a reporter requires laborious procedures that are often destructive to plants, and this limits the application of GUS or luc to large-scale screening. Symptom development may not always be correlated with the bacterial population, as plants may tolerate high bacterial populations under suitable environmental conditions without developing any symptoms (Lund et al., 1998). Therefore, development of a simple, rapid, nondestructive and accurate screening method will greatly aid the identification of disease resistant or susceptible plants.

To demonstrate that our approach of monitoring GFPuv-labeled bacterial infection in plants can be applied to the study of nonhost disease resistance, we silenced the SGT1 gene in N. benthamiana by VIGS (Liu et al., 2002). SGT1 is a component of the Skp1-Cullin-F-box (SCF) ubiquitin ligase complex mediating protein degradation (Kitagawa et al., 1999). Recently, SGT1 has been shown to be involved in certain cases of plant R gene-mediated disease resistance in Arabidopsis thaliana (Austin et al., 2002; Holt et al., 2002; Tör et al., 2002) and barley (Hordeum vulgare) (Azevedo et al., 2002). SGT1 has also been shown to be required for nonhost resistance against X. axonopodis pv. vesicatoria and P. syringae pv. maculicola, but not against X. campestris pv. campestris, in N. benthamiana (Peart et al., 2002). The SGT1-silenced and control plants were inoculated with a nonhost pathogen, P. syringae pv. tomato strain T1, containing pDSK-GFPuv, by vacuum or leaf infiltration. Dramatic accumulation of bacteria in the leaves of SGT1-silenced plants was observed as numerous bright green fluorescent spots under long-wavelength UV light, while no green spots were visible on the leaves of control plants (Fig. 4a,b). Serial-dilution plating confirmed that the accumulation of the nonhost pathogen P. syringae pv. tomato strain T1 in SGT1-silenced plant was at least 100-fold greater than that in control plants at 3 dpi (data not shown). The infected leaves were also directly examined under a confocal laser scanning microscope. Many individual fluorescent bacterial cells were clearly visible in the intercellular spaces of SGT1-silenced plants, while the control plants had very few bacterial cells (Fig. 4c). These results demonstrate that, by using GFPuv-labeled nonhost bacterial pathogens, one can identify plant mutants or gene-silenced plants that are compromised for nonhost disease resistance.

Figure 4.

Suppressor of G2 allele of SKP1 (SGT1)-silenced Nicotiana benthamiana plants were compromised in terms of nonhost resistance against Pseudomonas syringae pv. tomato strain T1. SGT1-silenced and control N. benthamiana plants were infected with the nonhost pathogen P. syringae pv. tomato T1 expressing a green fluorescent protein variant (GFPuv). Agrobacterium tumefaciens strains containing TRV1 and TRV2 (empty vector or with SGT1) vectors were mixed and infiltrated into the leaves of 2–3-wk-old N. benthamiana plants. TRV::00 (empty vector) was used as a control. Two to three weeks after A. tumefaciens inoculation, plants were challenged with P. syringae pv. tomato T1 containing pDSK-GFPuv at a concentration of 2 × 104 colony-forming units (cfu) ml−1 by leaf infiltration using a 1-ml needleless syringe (a) or vacuum infiltration (b) and kept at 20–22°C. Photographs were taken 4 d post inoculation. (c) Visualization of P. syringae pv. tomato T1 containing the plasmid pDSK-GFPuv in the intercellular spaces of a SGT1-silenced N. benthamiana plant leaf. A leaf sample was taken from control (TRV::00) and SGT1-silenced N. benthamiana plants 2 d post inoculation and examined under a confocal microscope with excitation at 488 nm and emitted light collected from 500 to 600 nm. Images for green fluorescence and transmission were taken under ×63 magnification and merged together using Leica confocal software. Bars, 10 µm.

Use of GFPuv to monitor intercellular bacterial movement in planta

Disease development depends on the rapid growth and spread of pathogens after they have colonized the initial infection site. Plant defense against pathogens can be divided into preformed and induced defenses. In the case of induced defense, the most significant defense strategy that plants use against the invading pathogens is the rapid development of cell death, called the hypersensitive response (HR), to restrict pathogen growth and spread (Greenberg, 1997; Morel & Dangl, 1997). Therefore, investigation of bacterial activities, especially how bacteria get into and move inside plant tissues, will provide a fundamental understanding of plant disease resistance and pathogenesis.

Here we examined the possibility of using GFPuv-labeled pathogenic bacteria to investigate intercellular bacterial movement in plant leaves using a confocal microscope. Four- to five-wk-old N. benthamiana plants were inoculated with P. syringae pv. tabaci containing pDSK-GFPuv at a concentration of 2 × 104 cfu ml−1 by vacuum infiltration and were incubated at 20–22°C. Bacterial movement in plant leaves was directly monitored by examining a piece of the infected plant leaf under a Leica confocal microscope. Movement of GFP-labeled P. syringae pv. tabaci was clearly observed in the intercellular space (see Movies S1 and S2 in the Supplementary Material, available online). These results demonstrated that the plasmid vector pDSK-GFPuv can be used to study the dynamics of bacterial infection and can therefore be used as a powerful tool to study plant–bacterial interactions.

Use of GFPuv for rapid estimation of the bacterial population in planta

The resistance or susceptibility of a plant to bacterial pathogens is often measured as the number of bacteria present in leaf tissues at different times after infection. Traditional methods of estimating bacterial number require serial dilution and plating of leaf sample extracts, a process that is time-consuming and labor-intensive. A simple method for quantifying the growth of bacteria in A. thaliana has been reported; this method involves simply shaking the infected plant leaves in buffer followed by serial-dilution plating (Tornero & Dangl, 2001). Here, we developed a rapid method of estimating the approximate bacterial population by quantifying green fluorescence intensity in plant leaves infected with pathogenic bacteria containing pDSK-GFPuv. Wild-type and SGT1-silenced N. benthamiana plants were inoculated with the host pathogen P. syringae pv. tabaci and the nonhost pathogen P. syringae pv. tomato strain T1 containing pDSK-GFPuv, respectively, by vacuum infiltration. The infected plant leaf and control (buffer-infiltrated) samples were ground in 1 ml of 1 × PBS and were directly subjected to fluorescence quantification without dilution (see the Materials and Methods section). To correlate the quantity determined by fluorescence intensity with that determined using the traditional approach, we also performed serial-dilution plating with the same samples that were used to measure fluorescence intensity. The progression of bacterial infection in plant leaves was clearly observed as an increase in green fluorescence of inoculated leaves (Fig. 5a). The increase of green fluorescence intensity was proportional to the bacterial growth, determined by the conventional serial-dilution plating, for both host and nonhost pathogens tested (Fig. 5b). The correlation of green fluorescence intensity and bacterial cell number was linear, with R2 > 0.93 for P. syringae pv. tabaci on wild-type N. benthamiana and P. syringae pv. tomato strain T1 on SGT1-silenced N. benthamiana at 3 dpi (Fig. 5c). These results indicated that estimation of the bacterial population in the infected plant leaves could be achieved by measuring green fluorescence intensity from the leaf extracts. Alternatively, a nondestructive assay can be performed by directly measuring the fluorescence of a single leaf infected with GFPuv-labeled bacteria by using a fluorescence scanner to monitor the in planta growth of the bacteria over a period of time.

Figure 5.

Visualization and quantification of in planta bacterial growth in Nicotiana benthamiana leaves. (a) Progression of Pseudomonas syringae pv. tomato strain T1 infection in the leaves of Suppressor of G2 allele of SKP1 (SGT1)-silenced N. benthamiana. Plants were kept in a growth chamber at 20–22°C after vacuum inoculation at a concentration of 104 colony-forming units (cfu) ml−1. Photographs of the same leaf were taken under long-wavelength UV light at 1, 2, 3, and 4 d post inoculation. (b) Examination of fluorescence intensity (FI) and bacterial growth (cfu) in the infected leaves. Wild-type and SGT1-silenced N. benthamiana plants were inoculated with the pathogen P. syringae pv. tabaci and the nonhost pathogen P. syringae pv. tomato T1 containing pDSK-GFPuv, respectively, in the same way as described in (a). Bacterial cell numbers in planta were counted by serial-dilution plating, while the fluorescence intensity of undiluted leaf extracts were measured using a Triad LT Multimode Reader. (c) The correlation of fluorescence intensity and bacterial cell number in the infected leaves at 3 d post inoculation (dpi). Wild-type and SGT1-silenced N. benthamiana plants were inoculated with P. syringae pv. tabaci and P. syringae pv. tomato strain T1, respectively, containing pDSK-GFPuv at various concentrations (5 × 102, 2 × 103, 8 × 103 and 3.2 × 104 cfu ml−1) by vacuum infiltration to obtain different bacterial concentrations in planta for correlation analysis. Bacterial cell number and fluorescence intensity were examined at 3 dpi in the same way as described in (b). Regression analysis was performed using sigmaplot 9.0 (Systat Software, Inc., San Jose, CA, USA). Error bars represent standard deviations of four replicates.

Use of GFPuv to study A. tumefaciens attachment

Agrobacterium tumefaciens is widely used for plant transformation. The binding efficiency of A. tumefaciens to plant surfaces is associated with its virulence. Traditional methods of determining binding efficiency depend on viable cell counts of free bacteria and attached bacteria using sonication and light microscopy (Matthysse, 1987). Here we describe the development of an assay of A. tumefaciens attachment to the cut leaf surfaces using expression of the reporter gene GFPuv in both wild-type and attachment-deficient (mutant) strains of A. tumefaciens. The T-DNA transfer efficient strain A. tumefaciens GV2260 and the attachment-deficient mutants chvA (A191) and chvB (A192) containing the plasmid pDSK-GFPuv were used for attachment assays (Matthysse, 1987; Iñón de Iannino & Ugalde, 1989). The binding of agrobacteria to leaf segments was analyzed under a confocal laser scanning microscope by visualizing the fluorescent bacteria along the cut ends of the leaf segments. The attachment plus strain, A. tumefaciens GV2260, bound to cut ends of the leaf segments within 2 h of incubation at room temperature and the attachment was more distinct at 12 and 24 h of incubation. No significant differences were seen in attachment between 12 and 24 h of incubation, probably because saturation had been reached by 12 h (Fig. 6). Binding of the attachment-deficient mutants A191 and A192 was barely detectable even after 24 h post inoculation, while the fluorescent spots were more intense in the attachment plus strain A. tumefaciens GV2260 during the same period (Fig. 6). These results were consistent with previous reports on attachment mutants chvA and chvB (Iñón de Iannino & Ugalde, 1989). Our results suggest that the GFPuv reporter gene can be used for easy visualization of A. tumefaciens attachment to plant cells and will be useful in screening plant mutants that are deficient in A. tumefaciens attachment.

Figure 6.

Visualization of Agrobacterium tumefaciens attachment to the cut leaf surface of Nicotiana benthamiana. Cut leaf segments of N. benthamiana were incubated with A. tumefaciens for 15 min, and excess bacteria were removed and then transferred to Murashige and Skoog (MS) basal medium for cocultivation. The leaf segments were sampled at 2, 12 and 24 h post inoculation, washed gently 3–4 times with phosphate-buffered saline and briefly vortexed at very low speed for 5–10 min. The washed leaf segments were examined and photographed using a confocal laser scanning microscope (Bio-Rad, Hemel Hempstead, UK) at ×20 magnification. Merged images of GFPuv reporter gene expression along with the image showing autofluorescence of cut leaf surfaces of N. benthamiana are shown. (a and b) Green fluorescent protein (GFPuv) fluorescence was observed with the attachment plus strain A. tumefaciens GV2260 at 12 and 24 h post inoculation (hpi), respectively. (c and d) Negligible GFPuv fluorescence was seen in the leaf segments inoculated with the attachment-deficient A. tumefaciens mutants A191 (chromosomal virulence (chv)A) and A192 (chvB), respectively, at 24 hpi. Bars, 35 µm.


The plasmid pDSK-GFPuv provides the capability to track the infection and development of bacterial disease in whole plants. This plasmid has great potential for further elucidation of plant–bacterial interactions through real-time monitoring of bacterial movement and multiplication in plant leaf tissues. There is a good correlation between bacterial cell number and green fluorescence from the GFPuv-labeled bacteria, so this method provides a simple and rapid means of estimating the bacterial population in planta. With the ability to monitor bacteria in vivo, we were able to identify that SGT1-silenced N. benthamiana plants were also susceptible to the nonhost pathogen P. syringae pv. tomato strain T1, and to directly monitor bacterial movement in living plant tissues under a confocal microscope. We also demonstrated that GFPuv-expressing A. tumefaciens can be used to study A. tumefaciens–plant interactions. With the development of better techniques and advances in detection technologies, we expect that pDSK-GFPuv will be used in a wide range of applications in the near future and will become a very popular research tool.


We thank E. Blancaflor and A. Valster for help with the confocal microscope, and A. Valster and S. R. Uppalapati for critical reading of the manuscript. This work was supported by the Samuel Roberts Noble Foundation and a NSF award (IOB-0445799) to KSM. The Leica AOBS confocal system used in this study was purchased using a NSF equipment grant (DBI-0400580).