Controlled synthesis of the DSF cell-cell signal is required for biofilm formation and virulence in Xanthomonas campestris.

Virulence of the black rot pathogen Xanthomonas campestris pv. campestris (Xcc) is regulated by cell-cell signalling involving the diffusible signal factor DSF. Synthesis and perception of DSF require products of genes within the rpf cluster (for regulation of pathogenicity factors). RpfF directs DSF synthesis whereas RpfC and RpfG are involved in DSF perception. Here we have examined the role of the rpf/DSF system in biofilm formation in minimal medium using confocal laser-scanning microscopy of GFP-labelled bacteria. Wild-type Xcc formed microcolonies that developed into a structured biofilm. In contrast, an rpfF mutant (DSF-minus) and an rpfC mutant (DSF overproducer) formed only unstructured arrangements of bacteria. A gumB mutant, defective in xanthan biosynthesis, was also unable to develop the typical wild-type biofilm. Mixed cultures of gumB and rpfF mutants formed a typical biofilm in vitro. In contrast, in mixed cultures the rpfC mutant prevented the formation of the structured biofilm by the wild-type and did not restore wild-type biofilm phenotypes to gumB or rpfF mutants. These effects on structured biofilm formation were correlated with growth and disease development by Xcc strains in Nicotiana benthamiana leaves. These findings suggest that DSF signalling is finely balanced during both biofilm formation and virulence.

Introduction

Xanthomonas campestris pv. campestris (Xcc) is the causal agent of black rot disease which affects cruciferous crops worldwide (Onsando, 1992). As with many phytopathogenic bacteria, Xcc produces a range of factors that contribute to the ability of the bacterium to parasitize the host (Dow and Daniels, 1994). Among these are extracellular enzymes capable of degrading plant cell components and an extracellular polysaccharide (EPS) called xanthan. These factors may play a number of roles during disease. Extracellular enzymes may be required to overcome plant defence responses, to allow bacteria to move into uncolonized plant tissues and to mobilize plant polymers for nutritional purposes. Xanthan induces susceptibility to Xcc in Nicotiana benthamiana and Arabidopsis thaliana by suppressing basal defences such as callose deposition (Yun ), has a role in biofilm formation (Dow ) and may have further roles in protecting bacteria from stresses of desiccation and host-elaborated defences.

In Xcc the production of extracellular enzymes and EPS is subject to coordinate positive regulation by a cluster of genes, the rpf cluster (for regulation of pathogenicity factors) (Tang ; Dow and Daniels, 1994; Barber ; Slater ). Mutations in rpf genes lead to a reduced virulence in host plants. Several of the rpf genes mediate regulation via a small diffusible molecule named DSF (for diffusible signal factor) (Barber ). DSF has recently been structurally characterized as cis-11-methyl-2-dodecenoic acid (Wang ). The synthesis of DSF is directed by RpfB and RpfF and DSF perception and signal transduction is mediated by the RpfC/RpfG two-component system, which is encoded by the rpfGHC operon (Slater ). Mutation of rpfC, which encodes the sensor component, leads to overproduction of DSF and to lower levels of EPS and extracellular enzymes (Tang ; Slater ). The addition of DSF can phenotypically restore rpfF but not rpfC mutants to wild type for production of extracellular enzymes and EPS (Barber ). Regulation of EPS biosynthesis by DSF occurs at least in part at the level of transcription of the gum operon, which encodes the sugar transferases required for EPS biosynthesis (Vojnov ; 2001; 2002). There is circumstantial evidence for the operation of the DSF regulatory system in planta (Vojnov ).

More recent work has implicated the DSF signalling system in the regulation of biofilm dispersal in Xcc (Dow ). When grown in a rich medium containing glucose, rpfF, rpfG, rpfC and rpfGHC mutants form aggregates in which the bacteria are held together in a polymeric matrix. The integrity of this matrix is dependent on the synthesis of xanthan (Dow ). Aggregates formed by rpfF mutants disperse upon addition of DSF, but those formed by other rpf mutants do not. As the wild type grows in a dispersed planktonic form under these conditions, it was concluded that the role of DSF was in induction of biofilm dispersal in Xcc.

Growth conditions are known to influence many aspects of bacterial behaviour including the formation of biofilms. The work in this article was prompted by the observation that when grown in minimal medium (conditions that may more closely mimic those found in planta), the wild-type Xcc formed a structured biofilm on glass slides. Here we have examined the role of DSF signalling in biofilm formation in minimal medium with confocal laser-scanning microscopy (CLSM) of GFP-tagged Xcc strains, used both singly and in pairwise combinations. We provide evidence that biofilm formation requires tight control of the level of the DSF, with both DSF overproduction and non-production adversely affecting the formation of a structured biofilm. Furthermore, examination of the effects of inoculation of combinations of strains in the model plant N. benthamiana provided evidence for the action of DSF within plants and suggested that balanced DSF levels were required for optimal virulence.

Results

The development of a structured Xcc biofilm is under the control of the rpf/DSF system

The ability of Xcc to form biofilms was examined in the minimal Y medium (see Experimental procedures). In preliminary experiments, bacterial adhesion to polystyrene 96-well plates was analysed by crystal violet staining. Interestingly, the rpfF, rpfC and xanthan-deficient gumB mutants showed significantly less adherence than the wild-type strain 8004 after 6 h of growth in Y-minimal medium (Fig. 1). This is in contrast to effects seen in rich L medium, where levels of attachment of rpfG, rpfGHC and gum mutants were higher than the wild type (Crossman and Dow, 2004).

Fig. 1

Attachment of Xcc to polystyrene is reduced by mutations affecting the rpf/DSF signalling system and xanthan synthesis. Bacteria attached to the surfaces of wells in microtitre plates were stained with crystal violet, and the absorbance at 600 nm was measured during 6 h. The strains used were 8004 (wild type), 8523 (rpfF), 8557 (rpfC) and 8397 (gumB). Data are presented as the mean absorbance values plus standard errors for four replicate wells from a representative experiment.

The characteristics of the Xcc biofilm formed in vitro in minimal medium were analysed by CLSM over a 4-day time-course experiment of static cultures in chambered cover slides (Russo ). In the formation of a typical Xcc biofilm, the bacteria contacted the glass surface via the lateral cell surface and also predominantly attached to each other through lateral interactions forming microcolonies (Fig. 2, day 2). This phase was followed by the formation of compact aggregates of bacteria with a characteristic three-dimensional structure separated by extensive water spaces (Fig. 2, day 4). A z-projection of the x–y stacks (optical sections) showed a mushroom-type biofilm structures (Fig. 2). Bacteria in these structures were mostly interacting laterally (Fig. 2, day 4).

Fig. 2

Development of biofilms in strains of Xcc. Strains 8004 (wild type), 8523 (rpfF), 8557 (rpfC) and 8397 (gumB) containing the GFP-expressing plasmid pRU1319 were grown for 4 days in chambered cover slides and observed daily by CLSM. The panels (left to right) show day 2, day 4 in a layer at 6, 4 μm from the base glass and the day 4 z-axis projected images. Arrows indicate microcolonies. Scale bars = 5 μm.

With the rpfF mutant (DSF-minus), microcolonies were seen after 2 days, but these did not develop into a structured biofilm, so that after 4 days only unstructured layers of bacteria were observed. With the rpfC mutant (DSF overproducer), although the bacteria showed some aggregation at day 2, only unstructured layers of bacteria were observed at day 4 (Fig. 2). The relative levels of DSF of the different strains, which were predicted from behaviour in other media, were confirmed experimentally for growth in minimal medium in static culture (see Fig. S1). Overall these results showed that DSF-mediated signalling is required for the formation of a structured biofilm in minimal medium.

Role of xanthan in the formation of the structured biofilm in minimal medium

To evaluate the importance of xanthan in the formation of the structured biofilm, the behaviour of the gumB mutant grown in Y medium was analysed by CLSM. We observed that the gumB mutant (strain 8397) was severely affected in the microcolony formation and did not form more complex structures (Fig. 2). After 4 days, no evident biofilm architecture was observed on the base of the chamber (Fig. 2, z-stage). The gum cluster of genes cloned in pIZD261-15 restored normal levels of EPS and a typical structured biofilm to gumB strain 8397 (not shown). These observations confirmed that xanthan synthesis in Xcc is crucial for the development of the structured biofilm in Y medium. Measurement of the relative levels of xanthan in the static cultures showed that after 2 days of growth, both rpfF and rpfC mutants had significantly lower levels than the wild type (see Fig. S2), consistent with previous observations of the role of the rpf/DSF system in regulating synthesis of this polysaccharide (Tang ; Slater ; Vojnov ).

Extracellular complementation of biofilm-defective phenotypes in mixed bacterial cultures

Our CLSM analysis indicated that mutations in rpfF and gumB genes resulted in the absence of a typical and structured biofilm. To confirm that this phenotype is due to the contribution of extracellular DSF and secreted xanthan, mixed cultures of gumB and rpfF strains, both GFP-labelled, were analysed over a time-course experiment by CLSM. The mixed culture of the gumB and the rpfF mutants was able to induce cell clustering and to develop a structured biofilm (Fig. 3D). These results suggested that reciprocal complementation had taken place, where the lack of DSF in the rpfF mutant had been restored by DSF produced by the gumB mutant and the xanthan produced by the rpfF mutant was substituting for the lack of xanthan in the gumB mutant. Structured biofilms formed in this way comprised a mixture of the two bacteria as demonstrated by the use of a GFP-labelled rpfF mutant and an EYFP-labelled gumB mutant (Fig. 3E).

Fig. 3

Biofilms formed by mixed Xcc cultures. A–E. Biofilms formed after 4 days by mixed cultures are shown for the following. Wild type-gfp + rpfFgfp (A), wild type-gfp + rpfCgfp (B), rpfFgfp + rpfCgfp (C) and rpfFgfp + gumBgfp (D). All the inoculations were at a 1:1 ratio of the two strains. (E) shows dual-colour confocal images acquired from mixed cultures of rpfF and gumB carrying the GFP-expressing plasmid (pRU1319) or the EYFP-expressing plasmid (pMP4518) respectively. F and G. Biofilms formed after 4 days by the rpfFgfp strain supplemented with DSF extracted from 4-day static culture of the wild type (F) or rpfF mutant (G). Scale bars = 5 μm.

To investigate whether regulated production of DSF was needed for biofilm development, the wild-type strain 8004 was co-inoculated with the rpfC mutant (Fig. 3). At a 1:1 ratio the ability of 8004 to form the wild-type structure was abolished (Fig. 3B). At a 1:4 ratio of rpfC mutant to wild-type strain, only slightly modifications of the biofilm were observed (not shown). Similar results were obtained in mixed cultures when both rpfC and wild-type strains were GFP-labelled or when only the wild-type strain was GFP labelled (not shown). This eliminated the possibility that the rpfC mutant simply out-competes the wild-type strain for attachment, thereby always giving an rpfC-like pattern. Co-inoculation of the rpfC mutant with the rpfF mutant strain 8523 (DSF-defective) in a 1:1 ratio did not restore the formation of the typical wild-type structure (Fig. 3C). In contrast to the effects caused by the rpfC mutant, the rpfF mutant did not alter the wild-type 8004 biofilm when the two strains were mixed (Fig. 3A).

Taken together with the results of reciprocal complementation of rpfF and gumB mutants, these findings suggest that the amount of DSF produced has to be tightly controlled for the development of the biofilm and increased levels of DSF interfere with this process. These conclusions were supported by the results of experiments in which exogenous DSF was added to cultures. Addition of DSF extracted from a 4-day culture of the wild type to a culture of the rpfF/GFP mutant allowed the production of a structured biofilm at 4 days (Fig. 3F). Extracts from cultures of the rpF mutant (DSF-minus) by contrast had no effect on biofilm structure (Fig. 3G). Furthermore, addition of DSF extracted from the rpfC mutant (DSF overproducer) to cultures of the wild-type strain 8004 culture inhibited structured biofilm development (not shown).

Phenotypic characterization of in planta behaviour of rpf mutants and in vivo complementation studies

On the basis of the above findings, we aimed to investigate the possible biological implication of altering DSF levels on the interaction between Xcc and plants. Our model pathosystem was the interaction of Xcc with N. benthamiana (Yun ). Nicotiana benthamiana has become a useful model plant, primarily because it shows an unusual susceptibility to a variety of pathogens. In initial experiments, we analysed the virulence phenotype of the rpf mutants, the gumB mutant and wild-type strain 8004 in N. benthamiana (Fig. 4). Each strain was inoculated in a leaf of N. benthamiana and symptoms and bacterial growth were monitored (see Experimental procedures). In contrast to the wild type, all mutant strains produced almost no symptoms in N. benthamiana (Fig. 4A–D). With the wild-type strain the number of colony-forming units (cfu) recovered from leaf disks cut from the inoculated area increased more than three orders of magnitude over 4 days of infection (Fig. 4E). In contrast, both the rpfF and rpfC strains showed significantly less growth after the same infection period (Fig. 4F and G). As previously reported (Yun ), the xanthan-defective gumB mutant was completely asymptomatic on the N. benthamiana leaves (Fig. 4D) and was severely compromised in growth in the plant tissue (Fig. 4H).

Fig. 4

Growth and symptom production by Xcc strains in Nicotiana benthamiana. Symptoms and growth were determined after inoculation into N. benthamiana leaves of Xcc strains 8004 (wild type) (A and E), 8523 (rpfF) (B and F), 8557(rpfC) (C and G) and 8397 (gumB) (D and H).

To examine the role of DSF cell–cell signalling during Xcc pathogenesis, co-inoculations of N. benthamiana with mixed cultures of different mutants were performed. Although leaves of N. benthamiana inoculated with gumB and rpfF separately showed no symptoms associated with limited bacterial growth (Fig. 4), when the two strains were inoculated together a different outcome was seen. In this case, the growth of both the gumB and the rpfF in N. benthamiana increased significantly (Fig. 4G) and symptoms induced by the mixed culture were similar to those induced by the wild type (Fig. 5C). These observations strongly suggested that reciprocal complementation was occurring. That is, DSF produced by the gumB mutant was complementing the DSF deficiency of rpfF and that xanthan secreted by the rpfF strain complemented the defect in the gumB strain. From these findings we inferred that DSF cell–cell signalling occurred in planta.

Fig. 5

Growth and symptom production by co-inoculated Xcc strains in Nicotiana benthamiana. Symptoms and growth were determined after inoculation into N. benthamiana leaves of a 1:1 mixture of Xcc strains. These mixed inocula were strains 8004 (wild type) and 8523 (rpfF) (A and E), strains 8004 (wild type) and 8557 (rpfC) (B and F), strains 8523 (rpfF) and 8397(gumB) (C and G) and strains 8523 (rpfF) and 8557(rpfC) (D and H). For all mixed inoculations, the numbers of the each bacterial strain were quantified after 4 days. In all cases the ratio remained close to 1:1.

To examine whether regulated production of DSF was required for virulence, N. benthamiana leaves were inoculated with the wild-type 8004 and the rpfC mutant in a 1:1 ratio. The rpfC mutant interfered with symptoms caused by the wild-type 8004 and the total bacterial population was two orders of magnitude lower than that of the wild-type strain (Fig. 5B and F). In contrast, the presence of the DSF-defective rpfF mutant did not modify the symptoms and growth of wild-type strain 8004 in N. benthamiana leaves (Fig. 5A and E). Furthermore, co-inoculation of the rpfC and rpfF mutants allowed very limited bacterial growth and no symptoms were observed (Fig. 5D and H).

One possible interpretation of experiments with the rpfC mutant is that the elevated levels of DSF trigger plant defence responses that are responsible for the restriction of bacterial growth and symptom production. To address this point we examined the effects of inoculation of leaves of N. benthamiana with a DSF preparation on expression of a number of defence-related responses. DSF did not induce expression of the defence-related PR1 gene or callose synthesis (data not shown), suggesting that the adverse effects of elevated levels of DSF on virulence are through direct effects on the co-inoculated wild-type bacteria.

Discussion

Cell–cell signalling and biofilm formation in different environments

The work in this article offers further insight into the role of the rpf/DSF signalling system in the biology of Xcc. By examination of bacterial behaviour in static cultures in minimal medium we have demonstrated that DSF signalling has a role in the formation of structured biofilms and that an excess of DSF prevents such biofilm formation. This is a substantially different picture from that obtained from the study of aggregation/biofilm formation in shaken rich nutrient medium (Dow ). In this latter case, rpf mutants form matrix-enclosed aggregates whereas the wild-type strain does not. Furthermore, DSF causes aggregate dispersal in rpfF but not other rpf mutants. These experiments in rich nutrient medium suggested an effect of DSF on biofilm dispersal requiring the RpfC/RpfG two-component system, but no influence on biofilm formation.

Work on other bacteria has established that the environment has an impact on the contribution of cell–cell signalling or quorum sensing to the development of bacterial biofilms and that quorum sensing may be integral to biofilm formation only under certain conditions (Kjelleberg and Molin, 2002; Kirisits and Parsek, 2006). The same considerations appear to apply to the role of DSF signalling in biofilm formation in Xcc. In Xcc the perception of DSF is linked to the degradation of the intracellular signalling molecule cyclic di-GMP by the response regulator RpfG, an HD-GYP domain protein (Ryan ). Cellular levels of cyclic di-GMP are controlled through synthesis, catalysed by the GGDEF protein domain, and degradation by EAL and HD-GYP domains. The genome of Xcc encodes 37 proteins with potential roles in cyclic di-GMP turnover. Many of these proteins contain additional signal transduction and sensory domains suggesting that their activities in cyclic di-GMP turnover are responsive to environmental cues (Ryan ). Xcc may thus integrate information from a number of environmental inputs, including cell–cell signalling, to modulate cellular cyclic di-GMP levels with consequent effects for biofilm formation and virulence factor synthesis (Ryan ). Consequently cell–cell signalling may not have a primary role in biofilm formation under all growth conditions. A further consideration is that the synthesis of xanthan, which is required for biofilm formation, is considerably enhanced in rich medium in the presence of glucose, so that mutation of rpf genes may reduce xanthan production below a critical level for biofilm formation only in minimal medium.

Cell–cell signalling and virulence in Xcc

The results of our experiments with mixed inoculations of bacteria indicate that DSF cell–cell signalling occurs in planta, but that an optimal concentration of DSF is required for virulence. A model of the production of DSF during the growth of Xcc in planta and its relationship to the production of virulence factors and the development of structured biofilms is shown in Fig. 6. Although a close correlation was observed between the effects of DSF levels on structured biofilm formation in minimal medium and on virulence, we cannot conclude a direct cause and effect relationship. Although there is no evidence to suggest that addition of excess DSF negatively influences the synthesis of extracellular enzymes or xanthan, we cannot exclude effects on the synthesis of other virulence determinants. Our findings are similar to those of Lindow and colleagues (2006) who have reported the effects of interference in rpf signalling in the virulence of Xylella fastidiosa the causal agent of Pierce's disease of grape. Xylella fastidiosa is closely related to Xcc and synthesizes a DSF-like signal molecule that is recognized by the Xcc rpf system but which is probably slightly different from DSF. Inoculation of bacteria able to degrade DSF or bacteria able to synthesize DSF (including an rpfC mutant of X. fastidiosa) can reduce virulence and symptom production by X. fastidiosa in grape. Lindow and colleagues concluded that DSF signalling was normally finely balanced during the disease process and that such a fine balance might therefore be readily disrupted. This may have substantial consequences for development of measures for the control of Pierce's disease. The role of rpf/DSF signalling in disease is somewhat different for Xylella and Xanthomonas. In particular, mutation of rpfF leads to enhancement of virulence in X. fastidiosa but reduced virulence in Xcc. Nevertheless, our findings suggest that interference with DSF signalling may also have a role in the control of diseases caused by Xcc and perhaps other Xanthomonas spp.

Fig. 6

Model of DSF production and its effect in biofilm formation and virulence factor production. At low bacteria cell density (A), where the production of DSF (diamonds) is limited, bacteria attach to the surfaces of the xylem vessels (XV). As the microcolony forms (B), DSF levels rise and the bacteria start to produce virulence factors (VF) including extracellular enzymes. Extracellular enzymes can promote disease through interference with plant defences (PD), provide nutrition through degradation of the xylem walls and allow passage of bacteria between xylem elements through degradation of bordered pit membranes (PM). In addition, the structured biofilm begins to form; bacteria within these structures may have increased resistance to host defences. At later stages (C), further elevation of DSF levels promotes biofilm dispersal, so that the bacteria can be released to colonize new tissue. The presence of elevated levels of DSF at early phases prevents the formation of the structured biofilm.

Experimental procedures

Microbiological techniques

Xanthomonas campestris pv. campestris strains 8004 (wild type), 8523 (rpfF::Tn5lac) and 8557 (rpfC::pUIRM504) are previously described (Daniels ; Tang ; Slater ) and were grown at 28°C in PYM medium (Cadmus ) or in Y minimal medium containing glucose (1%, w/v) as the carbon source (Sherwood, 1970). Escherichia coli was grown at 37°C in L medium (Sambrook ). Bacterial growth was monitored at 600 nm using MSE Spectroplus spectrophotometer. Plasmids were mobilized into Xanthomonas by triparental mating using a helper plasmid. For analysis of biofilm growth, bacteria were grown in PYM medium for 1 day [optical density at 600 nm (OD600), about 1.5], and then the culture was used as an inoculum at a 1:1000 dilution in Y medium. Biofilm growth on glass was monitored in static cultures by confocal microscopy (see below). In some experiments, bacterial attachment to the side and at the bottom of the glass tubes and the wells of polystyrene plates was assayed by first growing the bacteria in shake flasks with Y medium to an OD600 of 0.8–1.0 and then pipetting 5 ml of this culture into 10 ml glass tubes or 2 ml into the wells of polystyrene 24-well flat-bottom tissue culture plates (Corning Incorporated, Corning, NY), which were then allowed to stand at 28°C for 48 h. Unbound bacteria were removed by gently washing the tubes or the wells three times with fresh growth medium, and attached bacteria were quantified by staining them with 0.01% (w/v) crystal violet (Acros Organics, Geel, Belgium), as described previously (O'Toole ).

Extraction and quantification of DSF from static cultures

DSF was extracted from culture supernatants using ethyl acetate as previously reported (Barber ). DSF was estimated by a bioassay in which restoration of endoglucanase production to an rpfF mutant is assessed (Barber ) Endoglucanase activity was measured by a radial diffusion assay and units were established by using a cellulase I enzyme (Sigma) as standard.

Nicotiana benthamiana growth conditions and inoculations

Nicotiana benthamiana seed germination and growth in soil were performed as previously reported (Yun ). All plant inoculations involved a minimum of three leaves from each of the three plants, and each experiment was carried out at least three times. Inoculation was performed according to published methods (Newman ). Bacteria were hand infiltrated into plant leaves at the abaxial surface by using a 1 ml syringe without needle, with Xcc strains (107 cfu ml−1 in H2O) or H2O, and bacterial development was assessed as reported (Yun ).

Confocal laser-scanning microscopy (CLSM)

A confocal laser-scanning microscope (Carl Zeiss LSM510-Axiovert 100 M) was used to visualize the different events of biofilm formation in a 4/5-day time-course experiment using chambered cover glass slides containing a borosilicate glass base 1 μm thick (Laboratory-Tek Nunc; No. 155411) and GFP-labelled bacteria as described previously (Russo ). Confocal images were acquired from bacterial cultures carrying the plasmid pRU1319, which expresses the green fluorescent protein (GFPuv) (Allaway ) or the plasmid pMP4518 expressing the enhanced yellow fluorescent protein (EYFP) (Stuurman ). GFP- or EYFP-labelled bacterial cultures were diluted 1:1000 and grown in the chambers for up to at least 10 days at 28°C. Such static cultures typically reached an OD600 of about 1.7, as determined by re-suspending the biofilm bacteria and measuring their optical density. To prevent desiccation, the chambers were incubated in a humid sterile Petri dish. A typical mature biofilm was developed by the 8004 wild-type strain in static cultures in Y minimal medium containing glucose after 4/5 days at 28°C, when the OD600 was about 1.1. Three-dimensional images were reconstructed using the Zeiss LSM Image Browser version 3.2.0. Dual-colour confocal images were acquired from mixed cultures carrying the GFP-expressing plasmid (pRU1319) or the EYFP-expressing plasmid (pMP4518) (Stuurman ). GFP-expressing bacteria appeared green, and EYFP-expressing bacteria appeared in yellow. The detection of the emitted light was performed as described previously (Stuurman ). Dual-colour images were acquired by sequentially scanning with settings optimal for GFP (488 nm excitation with argon laser line and 505-nm-long pass emission) or EYFP (488 nm excitation with argon laser line and detection of emitted light between 530 and 600 nm). Rates of biofilm formation by bacteria expressing both constructs were similar, and no difference in growth or biofilm formation could be detected using (non-fluorescence) microscopy of biofilms formed by bacteria containing or lacking the GFP or EYFP construct.