Nile red detection of bacterial hydrocarbons and ketones in a high-throughput format.

A method for use in high-throughput screening of bacteria for the production of long-chain hydrocarbons and ketones by monitoring fluorescent light emission in the presence of Nile red is described. Nile red has previously been used to screen for polyhydroxybutyrate (PHB) and fatty acid esters, but this is the first report of screening for recombinant bacteria making hydrocarbons or ketones. The microtiter plate assay was evaluated using wild-type and recombinant strains of Shewanella oneidensis and Escherichia coli expressing the enzyme OleA, previously shown to initiate hydrocarbon biosynthesis. The strains expressing exogenous Stenotrophomonas maltophilia oleA, with increased levels of ketone production as determined by gas chromatography-mass spectrometry, were distinguished with Nile red fluorescence. Confocal microscopy images of S. oneidensis oleA-expressing strains stained with Nile red were consistent with a membrane localization of the ketones. This differed from Nile red staining of bacterial PHB or algal lipid droplets that showed intracellular inclusion bodies. These results demonstrated the applicability of Nile red in a high-throughput technique for the detection of bacterial hydrocarbons and ketones.

Observation

There is intense interest in using biotechnology for the production of fuels and chemicals from renewable resources (1, 2). Current fuels and chemicals derive largely from petroleum hydrocarbon feedstocks that are becoming scarcer and more expensive. Thus, a major goal of commodity biotechnology is to produce hydrocarbons via bacterial metabolism (3). Recent studies have identified the genes involved in cyanobacterial medium-chain alkane biosynthesis (4). Other studies have described bacterial polyolefin biosynthesis via the head-to-head condensation of fatty acyl groups (5–7). While these studies lay the groundwork for future bacterial hydrocarbon production, obtaining bacterial strains suitable for mass production will require rapid screening tools for facilitating the iterative processes of directed microbial engineering and random mutagenesis.

Bioinformatics has revealed approximately 70 bacterial genomes that contain ole genes, and a number of those organisms have been demonstrated to produce long-chain hydrocarbons (8). Hydrocarbon biosynthesis has been studied with bacteria of the genus Shewanella (7, 9). Shewanella oneidensis strain MR-1 produces a single polyolefinic hydrocarbon, 3,6,9,12,15,19,22,25,28-hentriacontanonaene, and contains the oleABCD genes. The ole gene cluster was shown to be essential for polyolefin biosynthesis (10), and the isolated expression of oleA, which encodes a fatty acyl-coenzyme A (CoA) condensase, produces a long-chain ketone in S. oneidensis. Additionally, experiments with a strain of S. oneidensis MR-1 showing a deletion of the ole genes suggested that the polyolefin may be involved in adaption to lowered temperature (7). Foreign oleA genes expressed in S. oneidensis result in the production of a variety of ketones, depending on the source. For example, the expression of oleA from Stenotrophomonas maltophilia resulted in the production of more than 15 ketones derived from the condensation of different fatty acids (7). The long-chain ketones, ranging from 26 to 30 carbons in length, are of interest as fuel precursors or novel chemicals.

Chen et al. (11) previously described a high-throughput method using Nile red to detect neutral lipids in microalgae. The present study describes a similar high-throughput method for identification of mutant strains of S. oneidensis and Escherichia coli that produce higher levels of hydrocarbons or ketones. While nonfluorescent in water, Nile red fluoresces in a hydrophobic environment (12). Nile red has previously been used to screen bacteria, cyanobacteria, haloarchaea, and microalgae for those producing fatty acids and esters (11, 13–20). Nile red has also been used to screen for isoprenoid-derived hydrocarbon production in microalgae (21) and for triacylglyceride production in Saccharomyces cerevisiae (22). Previous studies typically used agar plates with Nile red in the medium (14–18, 23). In those examples, strains making more fatty acids or esters could be identified using fluorescence microscopic techniques. In initial agar plate studies, both wild-type S. oneidensis and a derivative expressing S. maltophilia oleA were compared to the non-hydrocarbon-producing oleABCD deletion (∆oleABCD) strain (see Table S1 in the supplemental material). On plates, the three strains showed a low level of Nile red fluorescence that could not be differentiated (data not shown). Unlike for the fatty esters, which comprise up to 80% of the cell mass and are typically deposited in cellular granules, it is possible that the level of hydrocarbon production in these Shewanella strains may be too low to differentiate on plates.

Because colony fluorescence could not be differentiated, subsequent studies were conducted with liquid culture to test Nile red for use as a high-throughput screening vehicle. Stationary-phase liquid cultures (Luria broth [LB]) of S. oneidensis expressing S. maltophilia oleA on a pBBAD vector, S. oneidensis with an empty pBBAD vector, and the ∆oleABCD strain with an empty pBBAD vector were grown in microtiter plates with LB-kanamycin (50 µg/ml). Nile red in dimethyl sulfoxide (DMSO) (1 mg/ml) was added to give a final concentration of 1 µg/ml. Expression of oleA under the control of the araBAD promoter was induced with 0.2% arabinose at an optical density at 600 nm (OD600) of 0.5 to 0.8, and strains harboring empty vectors were mock induced. The plates were incubated overnight at 30°C in a microtiter plate orbital shaker with 70% humidity (optimization of dye contact time is described in Fig. S1 in the supplemental material). To minimize photobleaching of the dye, the plates were incubated in the dark. Fluorescence intensity was measured with a fluorescence spectrophotometer (Molecular Devices SpectraMax M2) using an excitation wavelength of 580 nm and an emission wavelength of 640 nm. Percentages of the Nile red fluorescence signal relative to that of S. oneidensis with the empty pBBAD vector, and corrected for the optical density of the cultures (relative fluorescence units [RFU]/OD600 unit), are reported for each strain, and standard deviations are shown for triplicate samples (Fig. 1A). The Nile red signal for S. oneidensis that is expressing S. maltophilia oleA is significantly increased over the background signal for both S. oneidensis and the ∆oleABCD strain (analysis of variance [ANOVA]; P = 0.0164). These results are in agreement with results for gas chromatography with flame ionization detection (GC-FID) analysis of hydrocarbon and ketone production performed using a modified Bligh and Dyer method as described in the supplemental material (Fig. 1B; see also Fig. S3 in the supplemental material) (24). These results suggested that Nile red fluorescence is a good indicator of increased long-chain hydrocarbon or ketone production, detectable above the background stain of membrane components.

FIG 1

(A) Percentages of MR-1 Nile red fluorescence normalized to optical density (OD600) for S. oneidensis MR-1 with empty pBBAD (blue), the ∆ole strain with empty pBBAD (red), and S. oneidensis with S. maltophilia oleA in pBBAD (gray). An asterisk denotes a significantly different strain (ANOVA; P = 0.0164). (B) GC-FID amounts for strains listed in panel A relative to the level for a 12-tricosanone standard normalized to optical density (OD600) and to the amount produced by S. oneidensis with empty pBBAD. (C) Percentages of Nile red fluorescence normalized to optical density (OD600) for E. coli with empty pBBAD (blue) and E. coli with S. maltophilia oleA in pBBAD (gray). An asterisk denotes a significantly different strain (ANOVA; P = 0.0004). (D) GC-FID amounts for strains listed in panel C relative to the level for a 12-tricosanone standard normalized to optical density (OD600) and to the amount produced by S. oneidensis with empty pBBAD. (E) Relative Nile red fluorescence normalized to the optical densities (RFU/OD600 unit) of 96 colonies of a 1:1:1 mixture of the S. oneidensis strains listed in panel A. The five highest and the five lowest Nile red signals are denoted by green and red bars, respectively. The GC-FID chromatograms from the colonies producing the highest (green) and lowest (red) Nile red signals are inset.

Analogous experiments were carried out using BODIPY 493/503 (Invitrogen), as similar dyes have also been used for hydrophobic-molecule screening (25). The methods and results are reported in the supplemental material (see Fig. S2 in the supplemental material). BODIPY was found not to be as sensitive to changes in hydrocarbon and ketone production as Nile red. This may be a result of the smaller separation between the excitation and emission maximum wavelengths than for Nile red. Furthermore, BODIPY 493/503 is considerably more expensive and more sensitive to photobleaching than Nile red (26). For these reasons, Nile red was chosen for further study.

To examine whether Nile red fluorescence could also be used as a method for detecting increased hydrocarbon production in E. coli, we analyzed Nile red fluorescence of E. coli WM3064 harboring either S. maltophilia oleA on pBBAD or an empty pBBAD vector. The E. coli cultures were induced and assayed as described above for Shewanella, with the exception that the cells were grown at 37°C in the presence of diaminopimelic acid (DAP) since WM3064 is a DAP auxotroph. The Nile red signal was significantly (ANOVA; P = 0.0004) higher in E. coli expressing oleA (Fig. 1C). These Nile red data are in agreement with GC-FID analysis in which ketone production was observed in the strain expressing oleA but not in the strain carrying an empty vector (Fig. 1D; see also Fig. S4 in the supplemental material). Together with the results for Shewanella, these data demonstrated that the fluorescent dye Nile red could be used to identify bacterial strains producing long-chain hydrocarbons or ketones in a microtiter plate format.

To test the feasibility of using Nile red in a high-throughput screen, 96 colonies of a 1:1:1 mixed population of wild-type S. oneidensis, S. oneidensis expressing S. maltophilia oleA, and the ∆oleABCD strain were screened with Nile red as described above. A distribution of Nile red fluorescence signals ranging from 800 to 150 RFU/OD600 unit was observed (Fig. 1E). Extracts from the five colonies with the highest and lowest Nile red fluorescence levels were analyzed by GC-FID to identify each strain by its previously determined hydrocarbon or ketone profile. Of the colonies giving the five highest Nile red signals, all five produced increased amounts of ketones, indicating that all five were the strains expressing the S. maltophilia oleA gene. The five colonies with the lowest Nile red fluorescence levels either produced the single hydrocarbon signature of MR-1 or did not produce the hydrocarbon, as previously observed in the ∆oleABCD strain (Fig. 1E, insets; see also Fig. S5 in the supplemental material). These results are consistent with the previous observation that both the wild-type and the ∆oleABCD strains gave background levels of Nile red fluorescence.

Unlike polyhydroxyalkanoates (PHAs), which accumulate in granules inside the cells, the hydrocarbon and ketone products of the ole genes are thought to partition into the membrane; however, this has not been investigated. With evidence that Nile red is capable of detecting increased ketone production in vivo, Nile red staining of ketone-producing strains was used to gain information regarding the cellular localization of the ketones using fluorescence microscopy. A ketone producer strain of S. oneidensis expressing S. maltophilia oleA was incubated with 2 µg/ml Nile red and 1 µg/ml 4ˈ,6-diamidino-2-phenylindole (DAPI), a DNA dye, for at least 15 minutes. These cells were adhered to microscope coverslips, mounted on glass slides, and visualized with a Nikon C1si laser scanning confocal microscope (Fig. 2). Cells that have been imaged as a cross section appeared as a ring of Nile red (Fig. 2A). These same cells showed concentrated blue foci of the DAPI stain (Fig. 2B). In the merged image, the Nile red membrane stain surrounded the DAPI nuclear stain (Fig. 2C). These results are consistent with a localization of the ketones and hydrocarbons in cellular membranes and perhaps in some extracellular hydrophobic material.

FIG 2

Confocal fluorescence micrographs of a ketone producer strain of S. oneidensis expressing S. maltophilia oleA. (A) Nile red fluorescence; (B) DAPI fluorescence; and (C) merged image of Nile red and DAPI.

Implications.

These results showed that Nile red fluorescence could be used to identify bacterial strains with increased levels of olefinic hydrocarbons or ketones that derive from the head-to-head hydrocarbon biosynthetic pathways. Academic and industrial investigators are researching the production of alkanes or sesquiterpenoid hydrocarbons. It is likely that the methods described here will prove generally useful for high-throughput screening of bacteria to identify strains that are useful for mass production of hydrocarbons or ketones.

Supplemental information. Download Text S1, DOC file, 0.032 MB.

Text S1, DOC file, 0.032 MB

Strains and plasmids used in this study.

Table S1, DOC file, 0.044 MB.

(A) Percentages of MR-1 Nile red fluorescence normalized to optical density (OD600) for S. oneidensis with empty pBBAD and S. oneidensis with S. maltophilia oleA in pBBAD with the addition of Nile red before (red) or after (blue) growth. An asterisk denotes a significantly different strain (ANOVA; P = 0.0444). Download Figure S1, DOC file, 0.032 MB.

Figure S1, DOC file, 0.032 MB

(A) Percentages of MR-1 BODIPY 493/503 fluorescence normalized to optical density (OD600) for S. oneidensis with empty pBBAD (blue), the Δole strain with empty pBBAD (red), and S. oneidensis with S. maltophilia oleA in pBBAD (gray). Download Figure S2, DOC file, 0.033 MB.

Figure S2, DOC file, 0.033 MB

Representative GC-FID chromatogram for S. oneidensis wild-type MR-1 with empty pBBAD (A), the Δole strain with empty pBBAD (B), and S. oneidensis with S. maltophilia oleA on pBBAD (C). The peak at 18.5 minutes is an internal standard. Download Figure S3, DOC file, 0.125 MB.

Figure S3, DOC file, 0.125 MB

Representative GC-FID chromatogram for E. coli with empty pBBAD (A) and E. coli with S. maltophilia oleA on pBBAD (B). The peak at 18.5 minutes is the internal standard, and the peak at 20 minutes does not represent a ketone, as determined by mass spectrometry (MS). Download Figure S4, DOC file, 0.621 MB.

Figure S4, DOC file, 0.621 MB

Gas chromatograms from Shewanella strains producing the five highest (A) and the five lowest (B) Nile red fluorescent signals reported in Fig. 1E. Download Figure S5, PDF file, 0.419 MB.

Figure S5, PDF file, 0.419 MB