Novel High-Throughput Detection Method To Assess Bacterial Surfactant Production

Abstract

A novel biosurfactant detection assay was developed for the observation of surfactants on agar plates. By using an airbrush to apply a fine mist of oil droplets, surfactants can be observed instantaneously as halos around biosurfactant-producing colonies. This atomized oil assay can detect a wide range of different synthetic and bacterially produced surfactants. This method could detect much lower concentrations of many surfactants than a commonly used water drop collapse method. It is semiquantitative and therefore has broad applicability for uses such as high-throughput mutagenesis screens of biosurfactant-producing bacterial strains. The atomized oil assay was used to screen for mutants of the plant pathogen Pseudomonas syringae pv. syringae B728a that were altered in the production of biosurfactants. Transposon mutants displaying significantly altered surfactant halos were identified and further analyzed. All mutants identified displayed altered swarming motility, as would be expected of surfactant mutants. Additionally, measurements of the transcription of the syringafactin biosynthetic cluster in the mutants, the principal biosurfactant known to be produced by B728a, revealed novel regulators of this pathway.

Biosurfactant-producing organisms have classically been identified by their ability to emulsify and utilize hydrocarbons as a nutrient source (28). It has been only recently appreciated that biosurfactants are produced by bacteria for many reasons other than access to hydrophobic nutrient sources. Among the numerous functions identified are their use for swarming motility (movement across moist surfaces/low-percentage agar plates), biofilm structure and maintenance, and the delivery of insoluble signals (34, 37). Biosurfactants that can either promote biofilms or disperse them on root and abiotic surfaces have been identified (2, 20). Additionally, some biosurfactants have been noted for their membrane-disrupting and, thus, zoosporicidal or antimicrobial activities (2, 12, 33).

An unexplored arena where biosurfactants may prove particularly important is the colonization of waxy leaf surfaces. In order to survive on leaf surfaces, epiphytes must be able to access limited and spatially heterogeneous nutrient supplies and endure daily fluctuations in moisture availability in forms such as dew and rainfall (15, 24). Continuous water films may not normally form on such waxy surfaces, and surfactants might thus aid in the diffusion of compounds across the plant. If the bacteria have a pathogenic life phase, they must first have a method to enter plant tissue, after which they create a favorable apoplastic environment for growth (39). It is already known that once inside the leaf, bacteria such as Pseudomonas syringae use surfactants to cause plant cell leakage and disease symptoms (33). However, some studies have also implicated biosurfactants in the prepathogenic stages of plant-associated bacteria (16, 24, 36).

Pseudomonas syringae pv. syringae B728a, a sequenced model organism with a prominent epiphytic life-style, produces biosurfactants (3, 14). A study of the genetic regulation of biosurfactant production should provide insight into its function in this species. The identification of mutants altered in surfactant production would be an important first step in this process. However, an effective method of identifying such mutants needed to be found. Many studies have compared various screening methods to identify biosurfactant producers from limited collections of environmental isolates. Some of the most commonly used methods for analyzing biosurfactant production are drop collapse, emulsification, and tensiometric evaluation (4, 9). However, when many strains need to be assessed for surfactant production, the drop collapse assay has been the method of choice (10, 20). While some of the other methods are more sensitive and quantifiable than the drop collapse method, none of them are practical for high-throughput screening. Unfortunately, even the drop collapse assay involves a number of steps, including growing each strain in broth culture and testing the supernatant for its ability to collapse a water drop on a hydrophobic surface; this can be highly labor- and time-intensive and thus not suitable for a truly high-throughput screen in which thousands of strains would need to be tested. Furthermore, this test is generally used as a qualitative assay only, and a measurement of the collapsed water droplet under a microscope or many serial dilutions of each sample is required to get a semiquantitative estimate of surfactant abundance (4, 9). For this reason, the high-throughput use of the drop collapse assay in a mutagenesis screen would not identify strains that have either an increased or incomplete loss of surfactant production.

A novel biosurfactant detection method was developed here in order to quickly screen large numbers of bacteria for surfactant production directly on an agar plate. This atomized oil method is at least as sensitive as the drop collapse assay and was found to be useful for all tested biosurfactant-producing strains as well as synthetic surfactants. Additionally, it is semiquantitative and is capable of identifying intermediate phenotypes. As an illustration of this method, the atomized oil procedure was used in the context of a high-throughput screen of mutants of P. syringae B728a to identify those altered in surfactant production. This method proved very effective, identifying multiple mutations of the gene cluster encoding the nonribosomal peptide synthetase responsible for syringafactin production as well as several genes involved in its regulation.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

P. syringae pv. syringae B728a (25), P. syringae pv. tomato DC3000 (3), and Pseudomonas fluorescens SS101 (10) were maintained on King's medium B (KB) plates with 1.5% technical agar (19a) and grown at 28°C. Escherichia coli strains DH5α, BW20767 (22), and SM10(λpir) (11); Bacillus subtilis 3610 (18); and Pseudomonas aeruginosa PA14 (6) were maintained on Luria agar and cultured at 37°C. Antibiotics were used at the following concentrations: kanamycin at 25 μg/ml for P. syringae and 50 μg/ml for E. coli, rifampin at 100 μg/ml, gentamicin at 75 μg/ml, and spectinomycin at 100 μg/ml.

Biosurfactant detection assays.

The drop collapse assay was performed according to a method described previously by Bodour and Miller-Maier (4). Two microliters of 10W-40 Pennzoil (Pennzoil Products Company, Houston, TX) was applied to delimited wells on the lid of a 96-well plate and allowed to equilibrate at room temperature. Next, 5 μl of either diluted surfactant samples or supernatant from bacterial cultures or resuspended bacterial colonies was pipetted onto the oil surface. Drops that retained a spherical shape were scored as negative for surfactant content, while drops that had a visibly decreased contact angle with the oil and spread (collapsed) were scored as positive for surfactant content.

The atomized oil assay was conducted as follows. Bacteria were spotted onto LB or KB agar plates using sterile toothpicks and grown overnight. For a more uniform inoculation of plates with cells diluted to a common cell concentration, a colony was resuspended in phosphate buffer, the optical density at 600 nm (OD₆₀₀) was determined with a spectrophotometer, and a small volume of suspension containing the desired number of cells was pipetted onto the plate surface and incubated overnight. Alternatively, if visualizing purified surfactant, 5 μl of diluted surfactant was pipetted onto the plate and allowed to equilibrate for 30 min before assaying. An airbrush (type H; Paasche Airbrush Co., Chicago, IL) was used to apply a fine mist of mineral oil (light paraffin oil; Fisher Scientific) onto the plate with an air pressure of between 15 and 20 lb/in². Depending on the airbrush and setup used, experimenters will need to optimize the appropriate settings in order to deposit a constant and controlled stream of oil droplets. Biosurfactant halos were then immediately visualized with an indirect source of bright light. Halo radii were measured with a ruler from the leading edge of the bacterial colony to the edge of the surfactant halo.

Microscopy.

Bright-field microscopy of oil droplets was performed with a Zeiss Lumar V12 microscope using transmitted light at a ×80 magnification. The microscope was fitted with a charge-coupled-device (CCD) camera (QImaging), and images were captured by using Ivision software (BioVision Technologies). Images were processed with Adobe Photoshop (version 6.0).

Extraction of syringafactin, surfactin, and rhamnolipids.

Crude biosurfactant extracts were prepared with a modification of the protocol detailed previously by Berti et al. (3). Instead of broth cultures, agar plates with confluent lawns of P. syringae B728a were grown for 48 h, while P. aeruginosa and B. subtilis strains were grown for 24 h. Cells were harvested from four plates in 90 ml H₂O and centrifuged (5,000 × g for 10 min). This was due to an increased yield of biosurfactant on solid medium, an observation that is being pursued in a separate report. The supernatant was extracted with 150 ml ethyl acetate with 1% (vol/vol) formic acid, and the organic fraction was dried to completion. This material was resuspended in 20 ml H₂O, the pH was increased to 8.0 with dilute NaOH, and the material was again dried to completion. This material was then resuspended in 4 ml of methanol, filtered though a 0.45-μm Nalgene filter (Fisher Scientific), and dried to completion (3). The final product was weighed and diluted with deionized water for further testing.

Production of biosurfactant mutants.

The production of transposon mutants was done by use of a method similar to that described previously by Larsen et al. (22). Briefly, P. syringae B728a and one of the two conjugative E. coli strains were grown overnight on agar plates with appropriate antibiotics. Strain BW20767 harboring plasmid pRL27 (22) has a kanamycin resistance-conferring mini-Tn5 transposon with a hyperactive Tn5 transposase, and strain SM10(λpir) harboring pUT mini-Tn5 Sm/Sp (11) has a spectinomycin resistance transposon. Cells were then harvested with a loop, washed and resuspended in potassium phosphate buffer (10 mM; pH 7.5), and then mixed at a ratio of 1:3 (E. coli to P. syringae) and incubated overnight as a confluent lawn on a KB plate. After incubation, the cells were resuspended in phosphate buffer, and 1/10 of the resuspension was plated onto KB medium containing 100 μg/ml rifampin and either 25 μg/ml kanamycin or 100 μg/ml spectinomycin, as appropriate, and allowed to grow for 3 days.

Screening of mutants.

P. syringae transposon mutants were screened by the following method. Mutants were spotted from selection plates onto KB plates using sterile toothpicks, with spots being separated by at least 2 cm. Colonies were allowed to develop overnight and then sprayed with atomized mineral oil drops as described above. Mutants that displayed substantially larger (over 20%) or smaller halos were retested. Mutants with phenotypes that were consistently different from that of the wild-type (WT) strain were further investigated. The location of the transposon insertion in these mutants was determined by using arbitrarily primed PCR similarly to a method described previously by O'Toole et al. (30). To identify mutations generated by the transposon from plasmid pRL27, primers complementary to the 5′ end of the transposon were designed. Primer pRLext1 (5′-CGAACTAAACCCTCATGGCTAACG) was used in the initial PCR, and primer pRLint1 (5′-AACAAGCCAGGGATGTAACG) was used in the second reaction to amplify sequences 5′ to the insertion site. The PCR product was cleaned (QIAquick PCR purification kit; Qiagen) and submitted for sequencing with primer pRLint1. When working with the transposon from pUT mini-Tn5 Sm/Sp, the identification of the 5′ insertion site was done according to the same protocol except that the initial PCR primer was tn5sm-ext (5′-GCGCGAGCAGGGGAATTG), and the second-round primer was tn5sm-int (5′-CGGTTTACAAGCATAAAGCTTGCTC). The locations of the sequenced fragments were determined directly by a BLAST search of the Pseudomonas genome database (38) and compared to the published sequence of P. syringae B728a (14).

Swarming motility assay.

The swarming motility of P. syringae B728a was assessed with semisolid KB plates containing 0.4% technical agar as described previously (32). Cells were grown for 2 days on KB medium and then harvested and washed in potassium phosphate buffer (10 mM; pH 7.5). Cells were resuspended in buffer to an OD₆₀₀ of 0.27, and 5 μl (approximately 2.5 × 10⁶ cells) of the appropriate bacterial strain was pipetted onto each plate and incubated for 24 h at room temperature. The swarming distance was calculated as the average diameter of swarming fronts chosen randomly from two perpendicular vectors for each colony.

Construction of a PsyfA-gfp transcriptional fusion.

The upstream promoter region of the P. syringae B728a syfA gene was amplified by PCR from genomic DNA with primers syf5-HindIII (5′-TAAGCTTCTTGAGCTTTCCTGATTCCGACCGC) (HindIII site is underlined) and syf3-EcoRI (5′-TGAATTCGGCTCAAGGTCCTTCTTGGCGGG) (EcoRI site is underlined) to generate a 289-bp promoter region. PCR conditions were as follows: 28 cycles of 95°C, 59°C, and 72°C for 1 min each, with a final extension time of 10 min at 72°C. The PCR product was first cloned into pTOPO Blunt (Invitrogen) to generate pTOPO-PsyfA and then transformed into E. coli DH5α. The insert was sequenced to verify its identity. pTOPO-PsyfA was digested with HindIII and EcoRI, and the resulting fragment was cloned into pPROBE-OT (26), which contains a promoterless green fluorescent protein (GFP) gene to generate pPsyfA-gfp.

pPsyfA-gfp was electroporated into P. syringae B728a as well as mutant strains altered in biosurfactant production (see Table 3). The appropriate transformed strains were grown overnight on KB plates and then resuspended in phosphate buffer (10 mM; pH 7.5) to an OD₆₀₀ of approximately 0.2. GFP fluorescence intensity was determined by using a TD-700 fluorometer (Turner Designs, Sunnyvale, CA) with a 486-nm-band-pass excitation filter and a 510- to 700-nm combination emission filter. A relative fluorescence unit (RFU) was defined as the fluorescence of the suspensions normalized for the suspension turbidity measured as the OD₆₀₀.

TABLE 3.

Identification and characteristics of a variety of mutants of Pseudomonas syringae strain B728a with altered biosurfactant production identified by using an atomized oil assay

Locus of Tn5 insertion	Predicted function	No. of individual transposon hitsa	Avg surfactant halo radiusb (mm) ± SD	Drop collapsec	Avg swarming diamd (mm) ± SD	Avg SyfA expression (arbitrary units)e ± SD
WT (no insertion)			8.7 ± 0.2	Yes	24.1 ± 0.5	1,716 ± 41
Psyr_1233 (suhB)	Inositol monophosphatase	1	5.0 ± 0.2	No	6.7 ± 0.2**	110 ± 12**
Psyr_4094 (secA)	Protein secretion	1	5.3 ± 0.5	Yes	12.5 ± 3.2**	2,232 ± 116**
Psyr_2576 (syfA)	Syringafactin biosynthesis	4	5.3 ± 0.3	No	10.8 ± 0.3**	NA
Psyr_2577 (syfB)	Syringafactin biosynthesis	5	5.8 ± 0.4	No	10.17 ± 0.3**	NA
Psyr_4346	Unknown	1	6.0 ± 0.4	Yes	14.3 ± 0.7**	900 ± 27**
Psyr_3619	RNA processing and degradation	3	6.2 ± 0.2	No	13.0 ± 0.6**	53 ± 1**
Psyr_3958 (algT)	Alginate regulation	1	11.0 ± 0.6	Yes	32.3 ± 0.8**	2,691 ± 69**
Psyr_1407 (pmpR)	Virulence factor regulation	1	12.0 ± 0.5	No	32.3 ± 3.8**	87 ± 3**
Psyr_1747 (clpP)	Posttranslational modification	1	12.3 ± 0.4	Yes	19.3 ± 0.7**	626 ± 64**
Psyr_3957 (mucA)f	Alginate regulation	1	12.3 ± 0.4	Yes	20.3 ± 1.4**	2,378 ± 63**
Psyr_1748 (clpX)	Posttranslational modification	5	12.2 ± 0.9	Yes	21.5 ± 1.3*	1,595 ± 54
Psyr_1350 (rseP)	Alginate regulation	4	12.7 ± 0.3	Yes	32.8 ± 1.7**	2,655 ± 71**

^aNumber of times that independent mutants were identified as insertions in the same gene.

^bHalos with significantly smaller or larger radii than those of the WT (P < 0.01 for all by t test).

^cAbility to cause drop collapse of a water droplet on an oil surface.

^dBacterial motility over semisolid agar plates. Motility was significantly different from that of the wild type at a P value of <0.05 (*) or a P value of <0.01 (**), as determined by a t test.

^eArbitrary units of relative fluorescence of the PsyfA-gfp reporter in mutant strains. Expression was significantly different from that of the wild type at a P value of <0.01 (**), as determined by a t test.

^fThe original mutant strain had a decreased halo size. This strain saved for testing most likely possesses a secondary mutation that has reversed the phenotype of the original mutation.

Statistical analysis.

Most data and regression analyses were carried out by using Statistica (StatSoft, Tulsa, OK). Graphs were constructed with CoPlot (CoHort Software, Berkeley, CA).

RESULTS

Detection of biosurfactants with an atomized oil method.

A novel surfactant detection assay was developed by using P. syringae B728a, which produces the lipopeptide surfactant syringafactin as a test organism. Syringafactin was previously demonstrated to be a surfactant by use of the drop collapse assay; supernatant from P. syringae DC3000 collapses on a hydrophobic surface, demonstrating the presence of a surfactant, while supernatants from mutant strains that do not produce syringafactin do not. Although they focused on characterizing the syringafactin extract from P. syringae DC3000, the authors of that previous study also confirmed that syringafactin is produced in strain B728a (3). We developed a method of surfactant detection involving the misting of oil droplets onto agar plates, hypothesizing that the presence of surfactants would alter the interaction of the oil with the agar surface. When a fine mist of mineral oil was sprayed over the surface of a KB agar plate on which bacterial colonies of P. syringae B728a had grown, a light-diffractive halo was seen around the colonies (Fig. 1 B). In contrast, no such halo was observed around E. coli DH5α (Fig. 1A), a strain that is not predicted to produce a biosurfactant.

FIG. 1.

Comparison of the atomized mineral oil droplets deposited on agar plates around a growing colony of E. coli DH5α that does not produce biosurfactant (A, E, and I), a growing colony of P. syringae B728a that produces biosurfactant (B, F, and J), Silwet L-77 at a 500-fold dilution (C, G, and K), and Tween 20 at a 5-fold dilution (D, H, and L). (A to D) Overviews of halos seen with this assay. Bars represent 1 cm. (E to G) Microscopic closeups of the oil droplets observed within the halos viewed from the top. (H to L) Droplets viewed from the side. Bars represent 0.2 mm for microscopic images in E to L.

Upon microscopic inspection, it was seen that oil droplets on an uninoculated agar surface and near DH5α were in energetically unfavorable distorted shapes (Fig. 1E). This was presumably due to random heterogeneity in the hydrophobicity of the agar surface. However, when surfactants spread over the agar surface, such as in the vicinity of P. syringae B728a, the droplets assumed a more uniform, energetically favorable, hemispherical shape (Fig. 1F). Furthermore, the light-diffractive halo observed macroscopically was actually caused by the dewetting, or beading, of the oil droplets near the surfactant-producing bacteria. The oil droplets, which presumably were in contact with the biosurfactant, stood higher on the plate and appeared more spherical than droplets on the agar surface away from surfactant-producing colonies (Fig. 1J). These raised droplets reflected light at a different angle, making them appear brighter under an indirect source of light.

In order to show that this atomized oil assay was indeed detecting biosurfactants, we obtained a variety of strains with characterized biosurfactant production and for which isogenic strains blocked in biosurfactant production are available. In addition to P. syringae DC3000, which produces syringafactin, we tested Pseudomonas fluorescens SS101, Bacillus subtilis 3610, and Pseudomonas aeruginosa PA14. All of the tested biosurfactant-producing bacterial strains produced easily detectable bright halos when sprayed with atomized mineral oil, while none of the biosurfactant mutants exhibited halos in this assay (Table 1). Thus, all biosurfactants tested were readily detected with the atomized oil assay, and no evidence of false-positive indications of surfactant activity was obtained.

TABLE 1.

Surfactant production by characterized biosurfactant-producing bacterial strains detected with an atomized oil assay

Organism (reference)	Surfactant produced	Type of surfactant	Avg halo radiusa (mm) ± SD
Bacillus subtilis 3610	Surfactin	Lipopeptide	9.5 ± 0.5
B. subtilis srfAA mutant (18)			0
Pseudomonas aeruginosa PA14	Rhamnolipid	Glycolipid	2.4 ± 0.2
P. aeruginosa rhlA mutant (6)			0
Pseudomonas fluorescens SS101	Massetolide A	Lipopeptide	8.3 ± 0.3
P. fluorescens massA mutant (10)			0
Pseudomonas syringae DC3000	Syringafactin	Lipopeptide	3.6 ± 0.2
P. syringae syfA mutant (3)			0

^aValues are average measured atomized oil halos ± standard deviations from triplicate samples.

The atomized oil assay can detect a wide variety of surfactants.

While this new assay readily detected a variety of both lipopeptides and glycolipids of bacterial origin, we tested the behaviors of other types of surfactants with this procedure. A variety of commercially available surfactants were detectable by this assay (Table 2). Many of the surfactants behaved similarly to the biosurfactants, causing the oil droplets to assume raised hemispherical shapes that appeared bright when illuminated (Fig. 1C, J, and K). However, a few of the surfactants created a less obvious “dark halo” in which the oil droplets still assumed a circular form but were less hemispherical and had increased contact with the water-agar surface (Fig. 1D, H, and L). These “dark-halo” droplets, in contrast to the raised droplets in “bright halos,” were flat and appeared less bright than the surrounding surfactant-free droplets at certain angles. Interestingly, when the surfactants were ranked by their hydrophilic-lipophilic balance (HLB) values, a common value used to describe surfactants in industry, it was found that surfactants with low HLB values all yielded bright halos, while those with higher HLB values resulted in dark halos (Table 2).

TABLE 2.

Detection of a variety of surfactants with a droplet collapse assay and an atomized oil assay

Surfactantd	HLBa	Type of halo	Limit of detectionb (g/liter)
			Drop collapse	Atomized oil
Crude syringafactinc	NA	Bright	0.5	0.01
Crude surfactinc	NA	Bright	15	0.25
Crude rhamnolipidc	9.5	Bright	7.5	0.25
Silwet L-77	8-10	Bright	0.25	0.0125
CTAB	NA	Bright	0.5	0.001
Tergitol-7	NA	Bright	2.5	0.025
Triton X-100	13.5	Dark	0.25	0.125
Tween 80	15.0	Dark	NA	0.25
Tween 20	16.7	Dark	10	0.5
SDS	40.0	Dark	2	0.1

^aHLB values are as given in McCutcheon's Emulsifiers & Detergents, North American Edition (25a), with the exception of rhamnolipid (29) and Silwet (17). NA indicates surfactants for which an HLB value has not been conclusively determined.

^bThese values are the lowest dilutions of surfactant that still yielded visual detection by the respective assays. NA indicates samples that were undetectable by the assay at any concentration.

^cSample represents ethyl acetate extract of culture supernatant from biosurfactant-producing strains.

^dCTAB, cetyltrimethylammonium bromide.

Sensitivity of the atomized oil assay.

The sensitivities of the atomized oil and drop collapse assays to detect a variety of surfactants were compared. Using a range of dilutions of a given surfactant, we determined the lowest concentration of that surfactant that was still detectable by a given assay. Additionally, crude extracts of surfactin, rhamnolipid, and syringafactin were prepared, and their limits of detection by the two assays were compared. For all tested surfactants and biosurfactants, the atomized oil assay was found to be more sensitive than the drop collapse assay (Table 2). In general, the atomized oil assay detected surfactant at concentrations more than 10-fold lower than the concentrations detected by the drop collapse assay.

In order to relate the size of the observed halo around a source of surfactant to the amount of that surfactant, different dilutions of a syringafactin-containing extract were tested with the atomized oil assay, and halo diameters were measured. A log-linear relationship between the amount of surfactant applied to plates and the diameter of the halo was observed (Fig. 2 A). Thus, a quantitative estimate of the relative difference in the amounts of surfactant in different prepared samples can be readily estimated. For each 10-fold increase in the concentration of the spotted surfactant, the radius of the oil drop alteration increased by about 1.7 mm. Because halo sizes were very consistent for a given amount of surfactant, with standard deviations rarely above 0.25 mm, careful replicate measurements of halos should easily enable the distinction of amounts of surfactant that differ by 3-fold or more. However, it must be emphasized that such semiquantitative estimates are relevant only when comparing samples of the same surfactant on a single medium, since different surfactants will diffuse at different rates.

FIG. 2.

Effects of syringafactin concentration (A) and diffusion time (B) on size of halos produced in an atomized oil assay. Vertical bars represent the standard deviations of the means of four replicate measurements for each point. The line drawn in A represents the following linear relationship: y = 1.67x − 0.106 (R² = 0.97; P < 0.0001). The results are representative of three independent experiments.

While it may be possible to quantify the surfactant in a prepared sample by measuring halos, calculation of the surfactant produced by a bacterial colony is confounded by the additional parameter of time. The prepared samples discussed above were applied at a distinct time and measured 1 h later, but bacterial colonies could produce surfactant over many hours of growth. Given that the distance over which a specified amount of surfactant will spread across an agar surface would be expected to be somewhat dependent on time, we determined the extent to which this factor would influence estimates of surfactant concentration using the atomized oil assay. A fixed concentration of a syringafactin-containing extract from P. syringae B728a was applied onto agar plates and destructively analyzed by the atomized oil assay at various times after application. Halo radii continued to increase with time, although the rate slowed considerably after about 2 h (Fig. 2B). Because of this, in addition to the fact that the bacteria continue to multiply and that the production of many biosurfactants is regulated by cell density (33), we concluded that halo measurements could not be used to calculate the absolute amount of surfactant produced by a colony without further investigation. Fortunately, for screening purposes, relative amounts of surfactant production should be readily assessed by using the atomized oil method unless the growth rates of the strains being compared differ greatly.

Given that a consistent estimate of surfactant production from a given bacterial strain would be needed to compare strains in a high-throughput survey, we estimated the variance in estimates of syringafactin production in replicate cultures of wild-type P. syringae B728a. Replicate cultures of P. syringae were established on plates by toothpick inoculation. On average, about 2.3 × 10⁶ ± 0.6 × 10⁶ bacteria were applied onto a plate by using this technique. Radii of halos from the resulting syringafactin production after colony formation were 8.8 ± 0.8 mm. To determine if variations in the numbers of cells initially deposited to establish spots (colonies) affected the apparent surfactant production, a defined number of cells (10⁷) was applied in replicate spots onto the plate, and oil was sprayed onto the plates after incubation overnight, as for the toothpick-inoculated plates. The radii of oil drop halos around these replicate spots (8.9 ± 0.6 mm) exhibited a similarly small variation compared to those around colonies established by toothpick inoculation. The application of cells by toothpick therefore results in inconsequential variations in eventual surfactant production measured by this assay. Due to this limited variation, any strains displaying a halo that differed in radius by 20% or more than a reference strain would likely be significantly different in surfactant production. However, it is important to later confirm the regulation transcriptionally in the event that a smaller halo is the result of a slower growth rate of a mutant strain.

Analysis of surfactant production by P. syringae B728a transposon mutants on plates.

The atomized oil assay was used to individually screen a library of about 7,700 transposon mutants of P. syringae for surfactant production. Mutants with a halo radius that differed by more than 1.5 mm from that of wild-type colonies were identified in an initial assessment; this should correspond to an approximate 10-fold increase or decrease in surfactant production. Mutants with large growth defects were discarded based on the logic that fewer cells will produce less total surfactant, although three mutants with slight growth defects were saved for further testing, which includes a cell-normalized measurement of surfactant production. These mutants with visible growth defects were later determined to have insertions in the suhB homolog Psyr_1233, the secA homolog Psyr_4094, and a PhoH-like protein, Psyr_4346. No mutations were observed to cause visible increases in the growth rates.

Twenty-eight total mutants with significantly altered surfactant production were identified after replicate tests (Table 3). The identification of the sites of transposon insertion revealed that over half of the identified mutants harbored distinct insertions in genes found to be disrupted in at least one other mutant, yielding a total of 12 different genes found to significantly influence surfactant production in strain B728a. The largest number of mutants (nine) harbored insertions in the large gene cluster encoding the nonribosomal peptide synthetase for syringafactin (3). Given that the disruption of this locus in strain B728a greatly decreased surfactant production (Fig. 3), it appears that syringafactin is a major component of the observed surfactant halo in strain B728a. However, the remaining halo suggests that B728a produces a second surfactant in addition to syringafactin, which is in contrast to P. syringae DC3000, where the disruption of the gene cluster encoding the nonribosomal peptide synthetase for syringafactin completely blocks all surfactant production detectable by the atomized oil assay (Fig. 3). If the remaining bright halo corresponds to a biosurfactant with diffusional properties similar to those of syringafactin, then the observed halo radius of approximately 5.5 mm in a syringafactin knockout (compared to 8.7 mm in the wild type) corresponds to a 2-log decrease in the total surfactant concentration, implying that the second surfactant is produced only at approximately 1% of the levels of syringafactin production.

FIG. 3.

Comparison of surfactant halos (left) and extent of swarming (right) by wild-type P. syringae strains DC3000 and B728a and their respective syringafactin mutants (syfA mutant). The bars represent 1 cm.

In addition to insertions in the syringafactin biosynthetic cluster, a number of other insertions were found to significantly affect surfactant production. In total, 19 additional insertions in a total of 10 genes resulted in strains that consistently produced smaller or larger halos than those of the wild type (Table 3). All of these mutations were within the structural genes noted, with the exception of Psyr_3958, the sigma factor AlgT, in which the transposon was inserted less than 30 bp upstream of the structural gene, presumably disrupting the transcription of the gene. All disrupted genes were under 1,000 amino acids in length, with the exception of the syfA and syfB homologs, which are about 3,000 and 6,000 amino acids in length, respectively. The relatively higher frequency of mini-Tn5 transposon insertions into the syringafactin biosynthetic cluster reflects the increased probability of a random insertion event into such a large target, although a few of the smaller genes also had multiple insertions (Table 3).

As a further assessment of surfactant production in the mutants obtained in the screen, their ability to cause drop collapse was also evaluated (Table 3). Of the four possible permutations of relative halo size and drop collapse activity, most mutants were found to fall into one of the three following categories: (i) mutants including the syringafactin knockouts, which had smaller surfactant halos and no drop collapse activity; (ii) mutants with smaller surfactant halos but that still conferred drop collapse, suggesting that syringafactin production has been reduced but not completely blocked; and (iii) mutants with larger surfactant halos (which still produced a drop collapse). The 12 mutants belonging to the third category were all found to harbor insertions in multiple components of the AlgT extracellular stress pathway: AlgT, MucA, RseP, ClpX, and ClpP (7, 19). All of the mutants were originally identified as producing a larger surfactant halo, except for one with a disruption of the anti-sigma factor MucA. This mutant was initially noted to confer a smaller halo than the WT strain and to have a highly mucoid phenotype. However, upon the retesting of this mutant after passage in culture for several generations, it switched to having a large halo and a nonmucoid phenotype (Table 3), a phenomenon that will not be further addressed here.

The most surprising result was the identification of a mutant that fell into the fourth category, having a larger surfactant halo but which did not produce a drop collapse (PmpR). Most likely, this PmpR mutant no longer produces syringafactin but overproduces a second surfactant (Table 3). The discovery of this mutant suggests that the second surfactant is much weaker than syringafactin, such that even when produced in large quantities, it does not lower water surface tension enough to cause a droplet of water to collapse on an oily surface. Alternatively, it may suggest that it has low water solubility (a very low HLB value). Another possibility is that the atomized oil assay could be responding to a substance other than a surfactant, although it is unclear what that substance could be. A surfactant is, by definition, a surface-active agent, and the most probable reason for a change in the contact angle of an oil droplet on an aqueous surface would be a change in surface tension.

Because surfactant production is generally required for bacterial swarming ability, the movement of the surfactant mutants was measured. Unlike in DC3000, where a mutant blocked in syringafactin production could no longer swarm (2), the syringafactin mutants of B728a were still able to swarm slowly (Fig. 3). This is consistent with the reduced but not eliminated surfactant production in these mutants. Mutants blocked in each of the 12 genes found to alter biosurfactant production each also had altered swarming phenotypes (Table 3). In general, strains with apparently higher levels of surfactant production as evidenced by larger halos in the atomized oil assay swarmed faster, while those with smaller halos swarmed slower. Regression analysis of the influence of halo size on swarming distance was highly significant (Fig. 4 A). Even the PmpR mutant, having a large halo but not conferring a drop collapse, followed this relationship and swarmed significantly farther than the wild-type strain. In general, the atomized oil assay was much more indicative of swarming ability than the drop collapse assay. The mutants for which swarming was not predicted based on halo size had insertions in the genes encoding MucA, ClpX, and ClpP; all produced large halos but had slightly lower swarming abilities. The swarming distance for these mutants is apparently confounded by the phenotypic changes in these strains; these mutants initially swarmed as fast as the other mutants with large halos (first 16 h) but subsequently had a dry appearance, which seemed to suppress their rate of swarming as the colonies aged.

FIG. 4.

Relationship between surfactant halo sizes produced by mutants of Pseudomonas syringae B728a and swarming distance (A) and syringafactin transcription estimated by GFP fluorescence of a PsyfA-gfp fusion (B). The wild type is included in both figures and is indicated as “WT.” Coordinates are taken directly from the measurements presented in Table 3. The lines drawn represent the following linear relationships: y = 2.275x − 0.8244 (R² = 0.68; P < 0.001) (A) and y = 115.27x + 282.6 (R² = 0.12; P < 0.3) (B).

Surfactant production in mutants compared to transcription of the syringafactin locus.

Syringafactin appears to be a major surfactant produced by strain B728a, since mutants in its biosynthesis exhibit greatly reduced surfactant halos, drop collapse ability, and swarming. Because the atomized oil assay was highly predictive of the effect of surfactant production on swarming but not drop collapse, we determined how predictive halo measurements would be of syringafactin production. It seemed likely that many of the mutants with altered surfactant production identified in the atomized oil assay would exhibit an altered expression of the genes required for syringafactin production. To test this, a vector in which the promoter-containing region of the syringafactin biosynthetic locus syfA was fused to a GFP reporter gene was introduced into each of the surfactant mutants, and GFP fluorescence was measured. Furthermore, this calculation was cell normalized and would therefore identify any mutants with smaller halos resulting from an altered growth rate if they had smaller surfactant halos than the wild type but similar SyfA transcription levels.

All putative surfactant mutants identified by the atomized oil assay had an altered expression of syfA compared to the wild type with the exception of the ClpX mutant (Table 3). However, when considering all of the mutants, no direct correlation between halo size and syringafactin gene expression was found (Fig. 4B), most likely because of the confounding effect of the production of a second surfactant. For instance, the PmpR mutant has a very large surfactant halo and swarms well, but it does not produce syringafactin, suggesting that the putative second surfactant is highly upregulated in this strain. It appears that in different genetic backgrounds, the two surfactants contribute differentially to the observed halo and swarming phenotypes. Therefore, the apparent presence of a second surfactant readily explains why the sizes of the aggregate surfactant halos are not correlated with the production of just one of the surfactants. The atomized oil assay has thus enabled the identification of promising regulatory genes for biosurfactant production.

DISCUSSION

The discovery of novel biosurfactants and the exploration of the genomics of biosurfactant production would greatly benefit from a quantitative and high-throughput screening method. The features of the atomized oil assay demonstrated here should make it valuable for these purposes. Multiple strains can be simultaneously assayed within a few seconds, thus enabling thousands of strains to be screened for surfactant production in a reasonable time. Although all of our measurements were taken on KB and LB agar plates, we have found that this assay works well on any solid medium that is conducive to bacterial surfactant production. Additionally, given a standard, this assay can provide estimates of surfactant concentrations.

It may seem counterintuitive that biosurfactants cause the oil to bead in our assay, whereas surfactants normally cause water droplets on a hydrophobic surface to collapse. However, the shape of an oil droplet on an aqueous surface is determined not just by the change of surface tension at the oil-water interface but also by the counteracting force from the tension change at the air-water interface (13). If an added surfactant lowers the surface tension at the air-water interface more than at the oil-water interface, thermodynamics will favor a decrease of the relative contribution of the oil-water tension, seen as an increased contact angle between the oil and water and, hence, a beading of the oil droplet (Fig. 5). In this manner the shape of the oil droplet is determined by the action of the surfactant at the two different interfaces. In general, bright halos such as those conferred by all of the biosurfactants tested result when the predominant effect of the surfactant is on the air-water interface. Although we have arbitrarily classified surfactants as causing either a “bright” or a “dark” halo in oil drops surrounding a surfactant source, it is most probable that there is a spectrum of contact angles for the oil droplet that is dictated by the expected range of change of the various tensions by various surfactants, as discussed above. Similarly, while all of our obtained mutants displayed bright halos, there is a possibility that the contact angles of the oil droplets could be slightly different, especially near mutants unable to produce syringafactin. However, we have not yet found a reliable method for measuring the contact angles of the atomized oil droplets observed with our assay, and no obvious differences in droplet shape were detected during microscopic observations of the droplets.

FIG. 5.

Relationship between interfacial tensions and the contact angle of the oil droplet on the agar-water surface (θ). The following equation was used to keep forces in balance: (tension at air-water interface) = (tension at oil-water interface) + (tension at oil-air interface × cosine θ). Tension at the oil-air interface is a fixed value because it is rarely influenced by surfactants (27). Therefore, as the air-water tension decreases from added surfactant, as does the tension at the oil-water interface, the contact angle, θ, of the oil droplet will change to compensate for the inequal effect of the surfactant on those two interfaces.

It is not clear if there is an invariant correlation between a surfactant's hydrophilic-lipophilic balance and the shape that it imparts to oil droplets on an agar surface. However, it is tempting to speculate on the utility of this assay in predicting important characteristics of novel surfactants. HLB values are a scalar factor that reflects the degree to which a surfactant is hydrophilic or lipophilic, with a value of zero reflecting a completely lipophilic (hydrophobic) molecule, a value of 10 corresponding to a compound with equivalent hydrophobic and hydrophilic groups, and values over 10 reflecting predominantly hydrophilic molecules. This value is of great significance commercially, since it is used to determine appropriate functions of surfactants. For example, common surfactants such as SDS and Tween 20 have high HLB values and are therefore best suited for emulsifying a hydrophobic substance into a water phase (oil into water). On the other hand, surfactants such as Silwet L-77 with HLB values near 10 are more suited for wetting, or the spreading of a water phase over surfaces such as leaves (1, 40). These surfactants with balanced water- and oil-loving groups can be very effective as spreading agents capable of lowering the surface tension of water below 30 mN/m (21). Rhamnolipid, with a predicted HLB of 9.5, which can lower the surface tension of water to 28 mN/m, is a highly effective spreading agent involved in bacterial motility (29). Although there is no consensus on the HLB of surfactin, it is also capable of lowering the surface tension of water to 27 mN/m, suggesting that it may also have an HLB near 10 (21, 35). Surfactants like Silwet L-77, which had lower HLB values, conferred bright halos in our assay. The surfactants with HLB values over 13, which are most ideal for the emulsification of oil into water, did not cause the oil droplets to bead, resulting in dark halos when tested by the atomized oil assay. It is interesting that none of the biosurfactants tested conferred dark halos, suggesting that their primary roles are not as emulsifiers.

It is noteworthy that the measurements of biosurfactant production using the halo method were strongly correlated with swarming capability in mutants of P. syringae strain B728a. This suggests that the area covered by surfactants at the air-water interface as measured by our assay reflects a similar distance where swarming movement of bacteria across an aqueous agar surface is facilitated. Moreover, it is significant that drop collapse activity was not a good indicator of the swarming ability of a strain, which raises the question of what specific properties make a surfactant a good lubricant that facilitates bacterial motility. Because the drop collapse assay detects only surfactants that are able to greatly lower the surface tension of water, this property appears unnecessary for functions such as swarming. In addition, the use of the drop collapse assay in biological screens may cause a wide array of biologically active surfactants to be overlooked. In view of that, it is interesting that a syringafactin mutant of P. syringae strain B728a appears to produce a second surfactant that can promote swarming but not cause drop collapse. This is in contrast to a syringafactin mutant in P. syringae strain DC3000 that does not appear to produce this second surfactant (Fig. 3). It is also striking that no mutants that exhibited a total absence of surfactant halo were identified for strain B728a, pointing to a differential regulation of syringafactin and the remaining expressed surfactant (which explains the poor correlation between biosurfactant halos and the transcription of the biosynthetic gene cluster for syringafactin production). Furthermore, the disruption of pmpR apparently causes the downregulation of syringafactin while conferring an upregulation of the other surfactant, suggesting its role in regulating (inversely) both surfactants. While both P. syringae strains are pathogenic to plants, strain B728a is a much better epiphyte than DC3000 (5). Perhaps this second surfactant is particularly useful for the life-style of epiphytes such as strain B728a on waxy leaf surfaces. We are actively pursuing the identity and specific properties of this second surfactant. The phytotoxins syringomycin and syringopeptin were previously suggested to possess surfactant activities (16), although preliminary results have not yet provided support for the identity of either of these surfactants as the second surfactant (data not shown). It is possible that combining one of the mutations found from this screen with a syfA or syfB mutation could reveal the identity of the second surfactant.

Some but not all of the genes found to regulate both biosurfactant production and swarming ability in P. syringae have homologs that influence swarming in Pseudomonas aeruginosa. The disruption of Psyr_3619, encoding an RNA helicase, conferred a reduction in swarming similar to that seen with the blockage of its homolog, PA2840, in P. aeruginosa (31). Likewise, the disruption of pmpR (PA0964) in P. aeruginosa, a homolog of Psyr_1407, resulted in enhanced swarming in both species (Table 3) (23). It is significant that P. syringae B728a mutations were not identified in homologs of any of the many other genes found to alter swarming in P. aeruginosa (31) despite the near completeness of the mutant library, emphasizing that the surfactants that contribute to swarming in these strains differ and/or that many factors other than biosurfactant production contribute to swarming ability. It is also noteworthy that relatively few different genes apparently contribute to biosurfactant production in P. syringae B728a. The disruptions of only 12 unique genes, identified from over 7,000 screened mutants, were found to alter biosurfactant production. Assuming random transposon insertion, we predict that we have screened a library of approximately 77% of the P. syringae B728a genes. Although we have identified many of the mutations that have an effect on measured surfactant halos, we may have missed a number of mutations that negatively affected syringafactin production but were masked by a compensatory increase in the production of the second surfactant.

For life on the leaf surface, pseudomonads have been shown to employ a variety of traits to grow and survive despite fluctuating water availability (24). In response to desiccation stress, pseudomonads produce alginate in order to maintain a hydrated microenvironment (8). Our finding of multiple components of the AlgT regulatory pathway among mutants of strain B728a with altered biosurfactant production suggests an intimate relationship between water availability and biosurfactant production. This potential relationship warrants a further exploration of either the AlgT pathway or, perhaps, alginate production itself as a regulator of surfactant production. The role of biosurfactants on the leaf surface is most likely complex and, as such, may likely prove to have very complex regulatory networks. The atomized oil assay has revealed a likely diversity of biosurfactants that are produced by strain B728a and their complex patterns of expression, details that would have been difficult to discern using other assays for biosurfactant production. The tools and genetic resources developed here should prove useful in further studies of the roles of surfactants in the interaction of P. syringae with plants.