A Whole-Cell Biosensor for Detection of 2,4-Diacetylphloroglucinol (DAPG)-Producing Bacteria from Grassland Soil

The interest in bacterial biocontrol agents as biosustainable alternatives to pesticides to increase crop yields has grown. To date, we have a broad knowledge of antimicrobial compounds, such as DAPG, produced by bacteria growing in the rhizosphere surrounding plant roots.

ABSTRACT

Fluorescent Pseudomonas spp. producing the antibiotic 2,4-diacetylphloroglucinol (DAPG) are ecologically important in the rhizosphere, as they can control phytopathogens and contribute to disease suppression. DAPG can also trigger a systemic resistance response in plants and stimulate root exudation and branching as well as induce plant-beneficial activities in other rhizobacteria. While studies of DAPG-producing Pseudomonas have predominantly focused on rhizosphere niches, the ecological role of DAPG as well as the distribution and dynamics of DAPG-producing bacteria remains less well understood for other environments, such as bulk soil and grassland, where the level of DAPG producers are predicted to be low. In this study, we constructed a whole-cell biosensor for detection of DAPG and DAPG-producing bacteria from environmental samples. The constructed biosensor contains a phlF response module and either lacZ or lux genes as output modules assembled on a pSEVA plasmid backbone for easy transfer to different host species and to enable easy future genetic modifications. We show that the sensor is highly specific toward DAPG, with a sensitivity in the low nanomolar range (>20 nM). This sensitivity is comparable to the DAPG levels identified in rhizosphere samples by chemical analysis. The biosensor enables guided isolation of DAPG-producing Pseudomonas. Using the biosensor, we probed the same grassland soil sampling site to isolate genetically related DAPG-producing Pseudomonas kilonensis strains over a period of 12 months. Next, we used the biosensor to determine the frequency of DAPG-producing pseudomonads within three different grassland soil sites and showed that DAPG producers can constitute part of the Pseudomonas population in the range of 0.35 to 17% at these sites. Finally, we showed that the biosensor enables detection of DAPG produced by non-Pseudomonas species. Our study shows that a whole-cell biosensor for DAPG detection can facilitate isolation of bacteria that produce this important secondary metabolite and provide insight into the population dynamics of DAPG producers in natural grassland soil.

IMPORTANCE The interest in bacterial biocontrol agents as biosustainable alternatives to pesticides to increase crop yields has grown. To date, we have a broad knowledge of antimicrobial compounds, such as DAPG, produced by bacteria growing in the rhizosphere surrounding plant roots. However, compared to the rhizosphere niches, the ecological role of DAPG as well as the distribution and dynamics of DAPG-producing bacteria remains less well understood for other environments, such as bulk and grassland soil. Currently, we are restricted to chemical methods with detection limits and time-consuming PCR-based and probe hybridization approaches to detect DAPG and its respective producer. In this study, we developed a whole-cell biosensor, which can circumvent the labor-intensive screening process as well as increase the sensitivity at which DAPG can be detected. This enables quantification of relative amounts of DAPG producers, which, in turn, increases our understanding of the dynamics and ecology of these producers in natural soil environments.

INTRODUCTION

Secondary metabolites are well known for their potential as drugs in the medical industry. They were initially defined as dispensable compounds, nonvital to their respective producers. However, recent advances in genome mining and microbial ecology are beginning to shed light on the prevalence, role, and importance of these metabolites in natural environments. In soil ecology, species of fluorescent Pseudomonas isolated from naturally suppressive soils have received a great deal of attention due to their production of antimicrobial secondary metabolites, such as 2,4-diacetylphloroglucinol (DAPG). Suppression of wheat take-all disease caused by Gaeumannomyces graminis var. tritici was shown to be induced by years of crop monoculture and was associated with the root colonization of DAPG-producing fluorescent Pseudomonas (1). Moreover, production of DAPG was shown to be involved in disease control against the causative agent of tobacco black root rot, Thielaviopsis basicola (2). It has also been demonstrated that DAPG has antibacterial properties against the pathogen Erwinia carotovora subsp. atroseptica causing soft rot of potatoes (3).

The biosynthetic gene cluster related to DAPG production in Pseudomonas comprises eight genes, phlACBDEFGH (4, 5). Proteins encoded by the operon phlACBD are responsible for the synthesis of DAPG (4). The type III polyketide synthase PhlD initially condenses three malonyl coenzyme A molecules into phloroglucinol (PG), which is further acetylated by the enzyme complex PhlACB into monoacetylphloroglucinol (MAPG) and DAPG (6). PhlE was identified as a putative membrane transporter with similarities to a known efflux pump in Staphylococcus aureus (4). The protein product of phlG has been described as a hydrolase that catalyzes the degradation of DAPG to MAPG, thus controlling the intracellular levels of DAPG (5, 7). Both phlF and phlH encode tetR-like repressors that inhibit transcription of phlACBD and phlG, respectively (5, 8). PhlF and PhlH bind to operator sites located in promoters upstream of the genes they regulate, thereby sterically blocking transcription. DAPG serves as the ligand for both repressors. Thus, in the presence of DAPG repression is relieved, leading to expression of phlACBD and phlG, which, in turn, leads to induced DAPG biosynthesis and posttranslational regulation (5, 8).

The complex of species belonging to the group Pseudomonas fluorescens has been well characterized over the past 3 decades, due to the potential of several species to act as biocontrol agents in agriculture. A recent study surveyed the phylogenetic relationship between 166 type strains of Pseudomonas (among which 66 belonged to the P. fluorescens group) based on amino acid sequences of 100 gene orthologues, which further proposed the existence of 10 subgroups within the P. fluorescens clade (9). However, despite the vast diversity among P. fluorescens group members, only a few species belonging to two subgroups, Pseudomonas protegens and Pseudomonas corrugata, are known to produce DAPG (1, 8). With the advances in genome mining and the increased availability of complete genomes, the biosynthetic gene cluster phlACBDE was recently identified in Pseudomonas species outside the P. fluorescens group, as well as in two genera of Betaproteobacteria (10). However, in these cases production of DAPG has not yet been demonstrated.

Identification and enumeration of DAPG-producing microorganisms have, to our knowledge, exclusively relied on DNA probe hybridization and PCR-based techniques. One of the most commonly employed techniques uses colony hybridization combined with a confirmatory PCR to verify the presence of phlD (11). A more recent method involves culture-independent real-time PCR to quantify populations of DAPG-producing Pseudomonas in the plant rhizosphere (12). A different approach is to quantify the amount of DAPG produced in situ by chemical analysis. Bonsall et al. demonstrated that an optimized extraction protocol enabled quantification of DAPG isolated directly from the plant rhizosphere (13). While these techniques have clear advantages, there are several drawbacks that also exist. PCR-based methods are limited by DNA binding of specific primers, and measures have to be taken to address the quantity of diverse genotypes of DAPG producers. Chemical identification, on the other hand, is restricted by detection limits, which are directly correlated with the size of the bacterial population.

In recent years, the synthetic biology toolbox has expanded rapidly and the use of genetically engineered molecular circuits to sense molecules and conditions of interest has gained increased attention. These developments have given rise to whole-cell biosensors that utilize natural regulatory systems engineered to detect metabolites and small molecules (14, 15). Whole-cell biosensors rely on molecule recognition to either activate transcription or lift repression of a reporter gene and are thus often highly sensitive, with detection limits in the nano- to micromolar range (16–18). Furthermore, whole-cell biosensors are tunable by addition or alteration of genetic parts, which allows for higher sensitivity and increased specificity (15, 19). Lastly, biosensors may also be implemented as biological detectors for uncovering metabolic activities in situ (20, 21).

Studies of DAPG-producing Pseudomonas species have predominantly focused on rhizosphere niches, whereas the ecological role of DAPG as well as the distribution and dynamics of DAPG-producing bacteria is not well understood for other environments, such as bulk soil and grassland. In this study, we constructed a whole-cell biosensor as an alternative and efficient approach for detection of DAPG and directed isolation of DAPG-producing bacteria from environmental samples.

RESULTS

Construction of whole-cell DAPG biosensors with high sensitivity and specificity.

Two whole-cell biosensors were constructed to enable specific detection of DAPG and identification of DAPG-producing bacteria. The two sensors contain identical modules for DAPG sensing in combination with either the lux operon or the lacZ gene as a reporter (Fig. 1A). The biosensor plasmids were constructed as repressor-mediated modules in an Escherichia coli K-12 ΔlacIZYA host. In the absence of DAPG, the TetR-like repressor protein PhlF binds as a dimer to the phlO operator site (8) in the promoter upstream of the reporter gene. As bioavailable DAPG diffuses into the cytoplasm and binds to PhlF, the repression on the target P_phlF promoter is relieved. Two reporter modules were chosen. The lux operon was used as the output reporter to obtain a highly sensitive response measured in bioluminescence units. To enable agar plate screenings for investigating the distribution and dynamics of DAPG-producing bacteria in natural microbial communities, the lacZ gene was used as the second output reporter. To ensure stable inhibition of the P_phlF promoter under noninduced conditions, the phlF gene is constitutively expressed from the P_lacIQ promoter. Both genetic circuits were introduced into a pSEVA plasmid background (Fig. 1B) to allow for rapid and simple cloning, as well as efficient mobilization into distinct hosts by triparental mating.

FIG 1.

Design of highly specific and sensitive whole-cell biosensors for DAPG detection. (A) Schematic illustration of the genetic circuit representing the DAPG biosensors with various reporter modules. (B) Map of the biosensor present on a pSEVA plasmid background with an RK2 replicon, origin of transfer (oriT), and a kanamycin resistance gene. Terminators (T0, T1, and IOT) are also indicated. (C) The response of the whole-cell biosensor harboring pSEVA226-DAPG_lux to DAPG and similar molecules (PG and MAPG) measured in luminescence per OD₆₀₀. (D) The response of the whole-cell biosensor harboring pSEVA225-DAPG_lacZ to DAPG and similar molecules (PG and MAPG) measured in Miller units.

To address the sensitivity and specificity of the whole-cell biosensors, microtiter bioassays were conducted. For characterization of the lux version (Fig. 1C), the E. coli host with pSEVA226-DAPG_lux was grown in Luria-Bertani (LB) broth containing various concentrations of PG, MAPG, or DAPG with continuous measurements of luminescence and cell density. PG and MAPG are both precursors of DAPG and thus similar compounds, allowing determination of biosensor specificity toward DAPG. For each concentration of inducer, the average luminescence was normalized to cell density over a period of 35 min (8 data points, time [t] = 172 to 207 min). This corresponds to late exponential growth phase (see Fig. S1 in the supplemental material). The whole-cell biosensor exhibited excellent sensitivity toward DAPG, with a response to 20 nM being statistically significantly higher than the negative control without added DAPG (Student’s t test, P = 0.01). The biosensor did not respond to the concentrations of PG tested, but a minor response to MAPG at >1.25 μM was observed. For characterization of the lacZ version (Fig. 1D), the E. coli host with pSEVA225::DAPG_lacZ was grown in LB broth containing various concentrations of either PG, MAPG, or DAPG to allow for enzymatic expression. Subsequently, the biosensor response was determined in a β-galactosidase microtiter assay. Enzymatic activity of transcribed lacZ was estimated by continuously measuring the increase in o-nitrophenol concentration over time (Fig. S2). The output is displayed in Miller units [MU; determined by the equation MU = (5,000 × OD₄₂₀/min)/OD₆₀₀, where OD₄₂₀ is optical density at 420 nm]. Exposing the biosensor to 0.625 μM DAPG yielded a statistically significantly higher response than for the negative control without added DAPG (Student’s t test, P = 0.0032).

Detection of DAPG production from bacterial colonies during growth on agar surfaces.

After addressing the sensitivity and specificity of the biosensor, we proceeded to utilize it in the identification of DAPG-producing bacteria. We investigated the response of the whole-cell biosensor when coinoculated with known DAPG-producing bacteria commonly found in soil (Fig. 2). The biosensor harboring pSEVA225::DAPG_lacZ was grown as a lawn on KBmalt agar (22) supplemented with 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal). Inoculation of DAPG-producing cultures of P. protegens CHA0 and P. protegens DTU9.1 on these agar plates resulted in induction of the biosensor and production of clear blue halos surrounding the colonies after 24 h (Fig. 2A). We also constructed a ΔphlACBD mutant strain of P. protegens DTU9.1 by allelic replacement, in which the DAPG biosynthesis genes were deleted (Materials and Methods). As expected, the mutant strain did not elicit a response from the biosensor (Fig. 2A) and did not produce DAPG detectable by liquid chromatography-mass spectrometry (LC-MS) analysis (Fig. 2B). We note that DAPG production in Pseudomonas species has been shown to be high under growth conditions including maltose, such as the conditions used in this study (8, 23). In concordance with these findings, we did not observe clear blue halos when LB agar was used (data not shown). We also tested Pseudomonas putida KT2440, which does not contain the DAPG biosynthetic gene cluster. Similar to the P. protegens DTU9.1 ΔphlACBD mutant, P. putida did not elicit a response in the biosensor after 24 h of growth. However, after prolonged incubation (>72 h) of P. putida KT2440, a slight blue coloring was detected surrounding the bacterial colony (Fig. S3).

FIG 2.

Detection of DAPG produced by bacterial colonies grown on agar surfaces. (A) Four fluorescent Pseudomonas isolates were grown on a lawn of the biosensor with pSEVA225::DAPG_lacZ on KBmalt medium supplemented with X-Gal. Wild-type P. protegens CHA0 and DTU9.1, known to produce DAPG, elicited a biosensor response after 24 h, whereas the negative controls, P. putida KT2440 and a ΔphlACBD mutant of P. protegens, DTU9.1, did not. (B) Extracted ion chromatograms (EIC) for DAPG (m/z 211.0601 ± 5 ppm) of the four Pseudomonas extracts confirmed the production of DAPG after 24 h by P. protegens CHA0 and DTU9.1.

Finally, we also explored if our setup can be used to identify DAPG production in species other than Pseudomonas. Recently, the genome of Chromobacterium vaccinii MWU328 was shown to contain genes with high similarity to the essential genes required for DAPG biosynthesis (phlACBDE) (10). We found that C. vaccinii MWU328 produced molecules that induced a response in the biosensor, resulting in a blue halo around the colony. A small amount of DAPG was subsequently confirmed by LC-MS (Fig. S4).

Biosensor-guided identification of DAPG-producing Pseudomonas.

Next, we used our biosensor to guide the identification of DAPG-producing pseudomonads from environmental samples. To this end, we collected soil samples from the same grassland soil site (labeled P5) in both August 2018 and August 2019 and randomly isolated 30 fluorescent Pseudomonas strains at both time points. This site is located in Dyrehaven, which is a Danish natural reserve, thus representing a relatively unaffected, natural soil niche. Using the approach described above, all 60 isolates were screened on KBmalt agar plates supplemented with X-Gal and a lawn of the biosensor harboring pSEVA225::DAPG_lacZ. A blue halo indicative of DAPG production was observed for one isolate from 2018 (isolate P5.21) and two isolates from 2019 (isolates P5.52 and P5.53) (Fig. 3A). Subsequent LC-MS analysis confirmed DAPG production in all three isolates (Fig. S5). For taxonomic identification of the 60 isolates, part of the housekeeping gene rpoD was sequenced for each isolate. The rpoD sequences were aligned to a database of 165 Pseudomonas type strains (9). Species identification of each isolate was determined based on the highest match to the type strains using nucleotide BLAST on the NCBI website. The diversity of cultivable fluorescent Pseudomonas remained similar, as species of P. jessenii, P. koreensis, and P. corrugata subgroups (as well as species of P. putida) were identified in the two samplings (Fig. 3B). However, isolates belonging to the P. fluorescens subgroup were found only in 2018.

FIG 3.

Examination of fluorescent Pseudomonas from grassland soil for DAPG production. (A) Thirty fluorescent Pseudomonas isolates from grassland soil in 2018/2019 were grown on a lawn of the biosensor with pSEVA225::DAPG_lacZ on KBmalt medium supplemented with X-Gal. The biosensor responded to one isolate from 2018 and two from 2019. (B) Part of the rpoD gene was sequenced and aligned to a database of 165 type strains of Pseudomonas. The three isolates eliciting a response from the biosensor were identified as P. kilonensis, which is part of the P. corrugata subgroup (9). (C) An rpoD-based neighbor-joining tree representing the phylogenetic relationship of the three P. kilonensis isolates to the P. corrugata subgroup. P. protegens CHA0 was included as an outlier. A bootstrap consensus tree (500 replicates) of the rpoD PCR products was constructed via the neighbor-joining method. The bootstrap percentage values are depicted next to each branching point. The P. corrugata subgroup is shown in a green area, whereas the species of this subgroup known to produce DAPG are in a blue box, along with the three soil isolates.

The three DAPG-producing Pseudomonas isolates were identified as P. kilonensis. This species is part of the P. corrugata subgroup of fluorescent Pseudomonas (Fig. 3C, green area). Members of this subgroup are known to harbor the biosynthetic gene cluster required for DAPG production (10). To determine the phylogenetic relationship of the three P. kilonensis isolates compared to the P. corrugata subgroup, we constructed an rpoD-based bootstrap consensus tree (500 replicates) with the neighbor-joining method (Fig. 3C). As expected, the three P. kilonensis isolates cluster together with the members of the P. corrugata subgroup known to produce DAPG (marked with a blue box). Taken together, these results show that genetically highly related DAPG-producing Pseudomonas organisms can be isolated from the same grassland soil site over a 12-month period.

Measuring the frequency of DAPG-producing pseudomonads in grassland soils.

To further explore the populations and frequencies of DAPG-producing pseudomonads in grassland soils, we sampled soil from three additional grassland sites (an area close to P5 labeled P5′, P8, and P9) (see Materials and Methods). In total, we isolated 288 Pseudomonas strains as libraries in 96-well microplates from each of the three sites (n = 864). The three libraries were then screened for potential DAPG producers by replica plating them onto KBmalt agar supplemented with X-Gal and a lawn of the whole-cell biosensor harboring pSEVA225::DAPG_lacZ. Simultaneously, the libraries were screened on separate plates for production of natural β-galactosidases, and isolates displaying a response were discarded from further analyses. Note that in this experimental setup (using replica plating), the development of blue halos around DAPG-producing colonies took longer than when larger aliquots of cultures were spotted onto the agar plates. After 48 h of incubation, the biosensor elicited a response to six isolates from P5′, five isolates from P8, and 49 isolates from P9 (Table 1). Subsequently, we analyzed the colonies displaying a blue halo for the presence of phlD, as well as their taxonomy by sequencing part of rpoD and aligning it to the database of Pseudomonas type strains (9). For P5′ and P8, one isolate from each site was confirmed to encode the polyketide synthase responsible for DAPG biosynthesis (Table 1), and both isolates were identified as P. protegens (Fig. 4). For P9, the 49 isolates with a surrounding blue halo were confirmed to have phlD, where 38 of those isolates belonged to P. kilonensis and the remaining 11 were identified as P. protegens (Fig. 4). The false positives from P5′ and P8 were subsequently analyzed with LC-MS (see Materials and Methods) to confirm the absence of DAPG production. DAPG was not detected in any of the nine isolates.

TABLE 1.

Frequencies of DAPG producers in natural soil microbiomes

Site	Total CFU/g of soila	Blue halo		phlD PCR verified
		Count (no./total)	%b	Count (no./total)	CFU/g of soilc
P5′	1.7 × 104	6/288	2.08	1/6	60
P8	5.7 × 103	5/288	1.74	1/5	20
P9	6.9 × 103	49/288	17.01	49/49	1.2 × 103

^aTotal CFU of bacteria cultivable on ¼ KB⁺⁺⁺ medium, which is predominantly Pseudomonas.

^bProportion of isolates displaying a blue halo in relation to the 288 colonies tested.

^cCFU per gram of soil of potential DAPG producers was calculated using the ratio of PCR-confirmed isolates compared to the total number of tested isolates times CFU per gram of soil of cultivable Pseudomonas.

FIG 4.

Taxonomic identification of isolates with a surrounding blue halo in the high-throughput screening assay. For each isolate, part of the rpoD gene was sequenced and aligned to a database of 165 type strains of Pseudomonas. For both P5′ and P8, one isolate was identified as P. protegens, whereas the remaining isolates belonged to either the P. putida group or the P. koreensis and P. mandelii subgroups of P. fluorescens, according to Hesse et al. (9). For P9, the rpoD gene of 49 isolates displaying a blue halo was sequenced. Only species of P. kilonensis and P. protegens were identified.

DISCUSSION

In this study, we constructed a highly sensitive whole-cell biosensor for detection and guided isolation of DAPG-producing microorganisms. Utilization of genetic circuits to detect and report on the presence of small molecules is associated with advantages and disadvantages. Two apparent caveats associated with the in vitro agar-based biosensor-guided identification used in this study are that it is viable only for cultivable organisms and it requires conditions that allow cocultivation of the isolate of interest with the biosensor. One advantage of the biosensor is that it is not restricted to a narrow range of genotypes of DAPG producers, which is a limiting factor of PCR-based approaches. Additionally, with a detection limit of >20 nM in vitro (Fig. 1C), the biosensor may have potential for use in situ to identify DAPG production in specific soil niches and thus serve as a promising alternative to chemical identification of DAPG, where the detection limit is in the low micromolar range (13).

We show that the biosensor is specific toward detection of DAPG. The biosensor did not respond to PG and elicited a minor response toward MAPG. This minor response was absent from its lacZ counterpart, which further demonstrates the sensitivity of the lux variant. We realize that we have used only two molecules (PG and MAPG) to represent natural DAPG analogues in our specificity assessment. It remains a possibility that other molecules can elicit a biosensor response. In a study by Yan et al. on the PhlH transcriptional regulator, it was demonstrated that multiple molecules with structural similarities to DAPG could bind to PhlH and induce a response, albeit significantly lower than the response induced by DAPG (5). Likewise, it was found that MAPG induced a minor response in the same study (5).

In order to demonstrate the biosensor response to bioavailable DAPG on agar surfaces, we inoculated DAPG producers and nonproducers on top of the whole-cell biosensor on agar plates. As expected, a blue halo was observed around the DAPG producers. Additionally, a blue coloring was absent around the nonproducers after 24 h. These findings also correlated with the LC-MS analysis. However, a slight blue halo was observed surrounding P. putida after prolonged incubation (>72 h), suggesting that one or more compounds are secreted by this strain during late stationary phase which interact with PhlF, thus relieving repression of the reporter gene.

Subsequently, the biosensor was utilized for guided identification of DAPG-producing fluorescent Pseudomonas. The isolates were selected on ¼ KB⁺⁺⁺ medium (see Materials and Methods), which has been shown to be optimal for isolation of fluorescent Pseudomonas, including DAPG producers (24). We screened 30 pseudomonads randomly isolated from the P5 site in both 2018 and 2019. DAPG producers were detected in both samplings based on a clear blue halo surrounding their colonies and were identified as P. kilonensis species, which are known to produce DAPG (10). Part of the rpoD gene was sequenced for all isolates and aligned to a database of Pseudomonas type strains (9), which revealed a remarkably similar diversity over a 12-month period (i.e., the same Pseudomonas subgroups were sampled at both time points). However, despite the low sampling depth, the species abundance appeared to shift from P. jessenii to P. koreensis. From an ecological point of view, it is of interest that the DAPG producers seem to persist over time in relatively similar quantities.

Lastly, we sought to optimize the screening assay to a high-throughput 96-well microplate format. We isolated 288 Pseudomonas species from each of three soil sites (P5′, P8, and P9). The sites are located in a Danish natural reserve (Dyrehaven); thus, we argued that they represent pristine grassland soil niches. It is worth noting that bulk soil is an extremely harsh environment with low nutrient availability, which might explain the small amount of CFU per gram of Pseudomonas compared to rhizosphere environments (11–13). We identified 6 and 5 isolates with blue halos around their colonies in P5′ and P8, respectively. Yet only one isolate from each site was confirmed to encode the polyketide synthase responsible for DAPG production. Picard et al. isolated 156 Pseudomonas strains from bulk soil, but no DAPG producers were identified, although DAPG producers were isolated at a later stage from roots of maize plants grown in the same soil (25). It was speculated that DAPG-producing Pseudomonas organisms are present in bulk soil in quantities of <2.6 × 10² CFU g⁻¹, which is comparable to the findings obtained in our study of grassland soil at P5′ (6 × 10¹ g⁻¹) and P8 (2 × 10¹ g⁻¹) (Table 1). In P9, on the other hand, 17% of the isolates displayed a blue halo around their colonies, and the presence of phlD was confirmed for all isolates. However, during the course of our study, it became apparent that a deer-feeding site was located near the sampling site, with wheat being the main feed. This could potentially explain the high frequency of DAPG producers in P9, as DAPG-producing Pseudomonas bacteria, which are known to be associated with the roots of wheat (1), might have translocated into the soil surrounding the feeding site due to animal activities.

The false-positive isolates from the high-throughput screening (i.e., isolates that resulted in a biosensor response without the presence of DAPG biosynthesis genes) could potentially produce compounds similar to those made by P. putida (as described above), which interfere with PhlF. The phlF gene in the biosensor was cloned from P. protegens CHA0 (26), where it naturally functions as a transcriptional repressor of phlACBD (8). The false positives identified in our screen may secrete yet-unknown secondary metabolites that interact with the PhlF repressor, thus inducing biosynthesis of DAPG. This finding highlights the possibility for microbe-microbe interactions in situ leading to induced DAPG production by adjacent nonproducers. Interestingly, we found two isolates belonging to the P. putida group in the high-throughput screening, which may indicate that certain species of this group produce molecules that can induce expression of DAPG. Surprisingly, we also identified isolates of P. koreensis and P. mandelii that elicit a response from the biosensor, which further enhances the potential of yet-unexplored microbe-microbe interactions that could be addressed in future studies.

In conclusion, this study demonstrates the use of an engineered whole-cell biosensor for guided identification of DAPG-producing microorganisms. This approach surpasses the limits of previous PCR-based and chemical identification methods, although future optimization to further increase sensitivity and reduce unexpected response to false positives might be required.

MATERIALS AND METHODS

Strains, media, and growth conditions.

Plasmid cloning and genetic circuit characterization were performed in Escherichia coli K-12 ΔlacIZYA or E. coli CC118-λpir. Cells were cultured in Luria-Bertani (LB) broth (Lennox; Merck, St. Louis, MO) with appropriate antibiotics. The antibiotic concentrations used were 25 μg ml⁻¹ for kanamycin, 10 μg ml⁻¹ for chloramphenicol, and 8 μg ml⁻¹ for tetracycline. The engineered whole-cell biosensor was cultured by inoculating a single colony in 5 ml of LB broth supplemented with kanamycin and incubating it overnight at 37°C with shaking (200 rpm). For characterization of the biosensor response to DAPG producers, control strains were routinely cultured by inoculating a single colony in 5 ml of LB broth and incubating it overnight at 30°C with shaking (200 rpm). Control strains included Pseudomonas putida KT2440, Pseudomonas protegens DTU9.1 (previously isolated by our group), Pseudomonas protegens DTU9.1 ΔphlACBD (see below), and Chromobacterium vaccinii MWU328.

Plasmid circuit construction.

Plasmid construction and DNA manipulation were performed following standard molecular biology techniques. The strain E. coli K-12 ΔlacIZYA was transformed with all plasmid constructs by chemical transformation. The plasmid pAJM847 (GenBank accession number MH101727.1), comprising the P_lacIQ promoter, phlF, the induction operon terminator (IOT), and the P_phlF promoter, was a kind gift from Christopher Voigt (26). The genetic circuit from pAJM847 was reorganized into pSEVA225T (GenBank accession number KC847299.1) (27) to obtain the DAPG biosensor. To this end, the fragment containing the P_lacIQ promoter, phlF, and the induction operon terminator was PCR amplified with AvrII and EcoRI overhangs (PlacI_AvrII_fw, 5′-TAAGCACCTAGGCGTTTTGGTCCAATGG; Term847_EcoRI_rev, 5′-TGCTTAGAATTCAGCGAGGAAGCACC). The purified PCR product was digested with appropriate restriction enzymes and inserted in pSEVA225T to yield pSEVA225::P_lacIQ-phlF. Subsequently, the fragment containing the P_phlF promoter was PCR amplified with EcoRI and HindIII overhangs (PphlF_EcoRI_fw, 5′-TAAGCAGAATTCCGACGTACGGTGG; PphlF_HindIII_rev, 5′-TGCTTAAAGCTTTATTTCCCCTCTTTCTCTAG). Purified PCR product was restriction digested and inserted in pSEVA225::P_lacIQ-phlF to yield the lacZ version of the DAPG biosensor (pSEVA225::DAPG_lacZ). The lux operon (luxCDABE from Photorhabdus luminescens) was PCR amplified from pUC18-mini-TN7T-Gm-lux with HindIII and SpeI overhangs (Lux_HindIII_fw, 5′-GAATTCAAGCTTATGACTAAAAAAATTTCATTC; Lux_SpeI_rev, 5′-GAATTCACTAGTAGGATATCAACTATCAAAC). The purified PCR product was restriction digested and inserted in pSEVA225::DAPG_lacZ to yield the lux version (pSEVA226::DAPG_lux).

Deletion of phlACBD by allelic replacement.

To abolish DAPG production in P. protegens DTU9.1, the biosynthesis genes, phlACBD, were deleted by allelic replacement according to the method of Hmelo et al. (28). In short, DNA fragments directly upstream of phlA (Up_F, 5′-ATCCCGTCTAGACAGAGATTTCGCAGTAAAAAG; Up_R, 5′-GGCGGGACAGGCACAGGCAGTCACATTTCCTCTTGATTCCATTCTTTTC) and directly downstream of phlD (Down_F, 5′-TGTGACTGCCTGTGCCTG; Down_R, 5′-ATCCGGGAGCTCTTGAAAGCCGCCAGCC) were PCR amplified. The fragments were joined by splicing-by-overlap extension PCR with XbaI and SacI overhangs using primers Up_F and Down_R. The purified PCR product was restriction digested and inserted into pNJ1 (29). The resulting plasmid was mobilized into P. protegens DTU9.1 via triparental mating with E. coli HB101 harboring the helper plasmid pRK600. Merodiploid transconjugants were initially selected on Pseudomonas isolation agar (PIA; Merck, St. Louis, MO) supplemented with 50 μg ml⁻¹ of tetracycline. A second selection was performed on NSLB agar (10 g liter⁻¹ of tryptone, 5 g liter⁻¹ of yeast extract, 15 g liter⁻¹ of Bacto agar) with 15% sucrose. Candidates for successful deletion were confirmed by PCR and verified by Sanger sequencing at Eurofins Genomics.

Luminescence dose/response microplate assay.

Overnight cultures of the whole-cell biosensor harboring pSEVA226-DAPG_lux were prepared in six biological replicates as described above. Ninety-six-well black, clear-bottom microplates (In Vitro, Denmark) were prepared with LB broth supplemented with kanamycin and various concentrations of PG, MAPG, or DAPG (0, 0.005, 0.01, 0.02, 0.039, 0.078, 0.156, 0.3125, 0.625, 1.25, 2.5, 5, 7.5, 10, and 15 μM). The overnight cultures were diluted to inoculate the microplates to an initial OD₆₀₀ of 0.01. The plates were sealed with a semipermeable membrane (Breathe-Easy; Merck, St. Louis, MO) and incubated in a Cytation 5 microplate reader for 5 h at 37°C with shaking (600 rpm), with continuous measurements of luminescence and absorbance at OD₆₀₀. These data and any other microplate assay data were collected with Gen5 2.07 software and exported to Excel 2016 and GraphPad for data analysis.

β-Galactosidase dose/response microplate assay.

Overnight cultures of the whole-cell biosensor harboring pSEVA225-DAPG_lacZ were prepared in triplicate as described above. Transparent 96-well microplates (TPP; Merck, St. Louis, MO) were prepared with LB broth supplemented with kanamycin and various concentrations of PG, MAPG, or DAPG (0, 0.625, 1.25, 2.5, 5, 7.5, 10, and 15 μM). The overnight cultures were diluted to inoculate the microplates to an initial OD₆₀₀ of 0.01. Cultures were grown for 3 h at 37°C with shaking (600 rpm), followed by measurement of the endpoint OD₆₀₀. Subsequently, 20 μl from each well was transferred to new transparent 96-well microplates and mixed with 80 μl of permeabilization buffer (0.1 M Na₂HPO₄, 0.02 M KCl, 0.002 M MgSO₄, 0.8 mg ml⁻¹ of hexadecyltrimethylammonium bromide, 0.4 mg ml⁻¹ of sodium deoxycholate, 5.4 μl ml⁻¹ of β-mercaptoethanol) (30). The plates were incubated at 30°C with shaking (600 rpm) for 30 min to facilitate cell lysis. Next, 28 μl of lysed cell culture from each well was transferred to 96-well black microplates and mixed with 172 μl of substrate solution (0.06 M Na₂HPO₄, 0.04 M NaH₂PO₄, 1 mg ml⁻¹ of o-nitrophenyl-β-d-galactopyranoside, 2.7 μl ml⁻¹ of β-mercaptoethanol) (30). The plates were sealed with a semipermeable membrane (Breathe-Easy; Merck, St. Louis, MO) and incubated in a Cytation 5 microplate reader for 16 h at 37°C with shaking (600 rpm), with continuous measurements of OD₄₂₀. Data were collected and analyzed as described above.

Detecting DAPG from bacterial cultures grown on agar surfaces.

Overnight cultures of the whole-cell biosensor harboring pSEVA225-DAPG_lacZ and the control strains P. putida KT2440, P. protegens DTU9.1, P. protegens DTU9.1 ΔphlACBD, and C. vaccinii MWU328 were prepared as described above. The biosensor was normalized to an OD₆₀₀ of 1.0 and spread on King’s agar B supplemented with malt extract (KBmalt) (20 g liter⁻¹ of Proteose peptone no. 3, 1.5 g liter⁻¹ of K₂HPO₄, 1.5 g liter⁻¹ of MgSO₄, 7.5 g liter⁻¹ of malt extract, 10 ml liter⁻¹ of glycerol, and 20 g liter⁻¹ of Bacto agar) supplemented with 25 μg ml⁻¹ of 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal; Thermo Fisher Scientific). Overnight cultures of the control strains were normalized to an OD₆₀₀ of 1.0 and inoculated as 20-μl spots on the agar plates containing the biosensor. Agar plates were incubated at 30°C for 24 to 96 h. Plates were inspected for blue halos surrounding the bacterial spots every 24 h.

Detection of DAPG by LC-MS.

Overnight cultures of the control strains P. putida KT2440, P. protegens DTU9.1, P. protegens DTU9.1 ΔphlACBD, and C. vaccinii MWU328 were prepared as described above. The cultures were normalized to an OD₆₀₀ of 1.0, inoculated as 20-μl spots on KBmalt and malt agar plates, and incubated at 30°C for 24 h. An agar plug (6-mm diameter) of the bacterial culture was transferred to a vial and extracted with 1 ml of isopropanol-ethyl acetate (1:3, vol/vol), containing 1% formic acid, under ultrasonication for 60 min. The extracts were then transferred to new vials, evaporated under N₂, and redissolved in 200 μl of methanol for further sonication over 15 min. After centrifugation at 13,400 rpm for 3 min, the supernatants were transferred to high-performance liquid chromatography (HPLC) vials and subjected to ultrahigh-performance liquid chromatography-high resolution electrospray ionization mass spectrometry (UHPLC-HRESIMS) analysis. UHPLC-HRESIMS was performed on an Agilent Infinity 1290 UHPLC system equipped with a diode array detector (DAD). UV-visible spectra were recorded from 190 to 640 nm. Liquid chromatography of 1 μl of extract was carried out using an Agilent Poroshell 120 phenyl-hexyl column (2.1 by 150 mm, 1.9 μm) at 60°C using acetonitrile and H₂O, both containing 0.02 M formic acid, as mobile phases. Initially, a linear gradient of 10% acetonitrile/H₂O to 100% acetonitrile over 10 min was employed, followed by an isocratic wash of 100% acetonitrile for 2 min. The gradient was returned to 10% acetonitrile/H₂O in 0.1 min and, finally, an isocratic condition of 10% acetonitrile/H₂O for 1.9 min, all at a flow rate of 0.35 ml min⁻¹. Mass spectrometry detection was performed in positive ionization on an Agilent 6545 quadrupole time of flight (QTOF) MS equipped with an Agilent dual-jet stream electrospray ion source with a drying gas temperature of 250°C, drying gas flow of 8 liters min⁻¹, sheath gas temperature of 300°C, and sheath gas flow of 12 liters min⁻¹. The capillary voltage was set to 4,000 V and nozzle voltage to 500 V. Mass spectrometry data analysis and processing were performed using Agilent MassHunter Qualitative Analysis B.07.00.

Isolation of fluorescent Pseudomonas from grassland soil.

Three sites of undisturbed grassland were chosen (P5, 55°78′88″N, 12°55′83″E; P8, 55°79′52″N, 12°58′06″E; and P9, 55°79′12″N, 12°57′51″E). Soil was collected approximately 10 centimeters below the grass surface. Five grams of soil was suspended in 30 ml of sterile water and shaken vigorously for 1 min on a Vortex mixer. In order to isolate fluorescent Pseudomonas, the samples were serially diluted and plated onto ¼ KB⁺⁺⁺ (7.5 g liter⁻¹ of King’s agar B, 10 ml liter⁻¹ of glycerol, 7.5 g liter⁻¹ of Bacto agar) supplemented with 100 μg ml⁻¹ of cycloheximide, 13 μg ml⁻¹ of chloramphenicol, and 40 μg ml⁻¹ of ampicillin, as previously described (24). Agar plates were incubated at 30°C for 48 h. Fluorescent colonies were identified under UV light and restreaked on LB agar plates. Species identification of the soil isolates was performed by PCR, amplifying part of the rpoD gene with primers (PsEG30F, 5′-ATYGAAATCGCCAARCG; PsEG790R, 5′-CGGTTGATKTCCTTGA) (31). PCR products were purified, sequenced, and aligned to a database of 166 known type strains of Pseudomonas (9).

Biosensor-guided identification of DAPG producers from grassland soil.

Thirty fluorescent Pseudomonas isolates were randomly selected from sample site P5 both in 2018 and in 2019 as described above. Isolates were cultured overnight in LB broth at 30°C with shaking (200 rpm). An overnight culture of the biosensor harboring pSEVA225-DAPG_lacZ was normalized to an OD₆₀₀ of 1.0 and spread onto KBmalt plates supplemented with 25 μg ml⁻¹ of X-Gal. Overnight cultures of the Pseudomonas isolates were inoculated onto the agar plates as 20-μl spots. Plates were incubated at 30°C for 24 to 48 h. Plates were inspected for blue halos surrounding the bacterial spots every 24 h. The DAPG-producing isolates were identified by PCR-based species identification, as described above. The phylogenetic relationship between the DAPG-producing isolates was determined by analyzing a phylogenetic tree with representatives of each P. fluorescens subgroup (9). In short, the PCR-amplified parts of the rpoD genes were Sanger sequenced and aligned using the MUSCLE algorithm, followed by construction of a bootstrap consensus tree (500 replicates) by the neighbor-joining method in MEGA X (32).

High-throughput screening for DAPG producers in grassland soil.

In 2019, 288 fluorescent Pseudomonas isolates were randomly selected from three sample sites (P5′, 55°78′78″N, 12°56′07″E; P8 and P9, see coordinates above). Fluorescent colonies were streaked on LB agar OmniTray (Nunc; Nalge Nunc International, Rochester, NY) and incubated for 24 h at 30°C. Isolates were cultured in transparent 96-well microplates in terrific broth (TB; 12 g liter⁻¹ of tryptone, 24 g liter⁻¹ of yeast extract, 0.17 M KH₂PO₄, 0.72 M K₂HPO₄, and 5 ml liter⁻¹ of glycerol). An overnight culture of the biosensor harboring pSEVA225-DAPG_lacZ was normalized to an OD₆₀₀ of 0.5 and spread onto KBmalt OmniTrays supplemented with 50 μg ml⁻¹ of X-Gal. The Pseudomonas isolates were inoculated on the OmniTrays with sterile replicators. The OmniTrays were incubated at 30°C for 48 h and inspected for blue halos surrounding the Pseudomonas colonies. Candidate isolates exhibiting a blue halo were screened by PCR for the presence of phlD with primers (B2BF, 5′-ACCCACCGCAGCATCGTTTATGAGC; BPR4, 5′-CCGCCGGTATGGAAGATGAAAAAGTC) (33). Moreover, rpoD was amplified from candidate colonies for species identification with primers PsEG30F and PsEG790R.

Data availability.

All data are included here and in the supplemental material. Plasmid constructs and bacterial strains, including Pseudomonas isolates, will be made available by the corresponding author upon request.