Mechanism and use of the commercially ... - Wiley Online Library

5 downloads 0 Views 147KB Size Report
Aug 30, 2004 - Ethidium bromide, a singly charged relative. FIG. 2. Excitation and emission spectra for 400 nM SYTO9 and 2.4 M. PI in 20 M DNA. For SYTO9 ...
© 2004 Wiley-Liss, Inc.

Cytometry Part A 61A:189 –195 (2004)

Technical Note

Mechanism and Use of the Commercially Available Viability Stain, BacLight S.M. Stocks* Novozymes Fermentation Pilot Plant, Bagsværd, Denmark Received 13 November 2003; Revision Received 19 March 2004; Accepted 22 April 2004

Background: BacLight (Molecular Probes, Eugene, OR, USA) is a popular fluorescence-based two-component stain for determining bacterial cell viability. The main purpose of this work was to fully elucidate the mechanism and to determine why it is sometimes reported that cells stain simultaneously live and dead. Methods: Solutions of DNA were stained with the two components, propidium iodide (PI) and SYTO9, in different combinations, and fluorescence spectra were collected. Results: KPI and KSYTO9 were approximately 3.7 ⫻ 105/M and 1.8 ⫻ 105/M. SYTO9 emissions were stronger and overlapped those of PI. Fluorescence resonance energy transfer from SYTO9 to PI was observed. It was, even under normal conditions, possible for DNA bound SYTO9

The characterization of bacterial viability has relevance across a broad spectrum of commercial and academic areas, including food safety, medicine, biotechnology, and environmental monitoring. Traditionally, bacterial viability has been determined by serial dilution of a sample suspension followed by plating of a known volume onto a solid medium and counting the colonies that form. Necessarily, this involves a significant delay, typically 24 h to 5 days. Such a delay can be problematic, particularly in safety-critical sectors where immediate results would be preferable. Another drawback for this technique is its failure to resolve spatial distributions of viability in multicompartmental systems such as those of filamentous or chain-forming bacteria, namely the Actinomycetes and the Streptococci. BacLight is a fluorescent bacterial viability stain that is becoming more widespread in its use. Manufactured by Molecular Probes (Eugene, OR, USA), it consists of two stains, propidium iodide (PI) and SYTO9, that stain nucleic acid. According to the manufacturer, PI is a red intercalating stain that is membrane impermeant and is therefore excluded by healthy cells. SYTO9 is a green

to have a component in the red region equal to that of DNA bound PI. Potentially confusing emissions were also found to occur when PI was not in sufficient excess to saturate nucleic acid (⬎0.4 M PI to 1 M DNA base pairs). Conclusions: The mechanism is a combination of displacement of SYTO9 by PI and quenching of SYTO9 emissions by fluorescence resonance energy transfer. Confusing results can occur if the relative intensities of the stains or the concentration of PI relative to nucleic acid are not properly accounted for. © 2004 Wiley-Liss, Inc.

Key terms: BacLight; propidium iodide; SYTO9; fluorescence spectrum; viability; actinomycetes; flow cytometry, image analysis

intercalating membrane permeant stain and will stain all cells, provided they contain nucleic acid. The manufacturers argue that PI has a stronger affinity for nucleic acid and that, when the two stains are present within a cell, SYTO9 will be displaced from nucleic acid and the cells will fluoresce in red. Live cells, being exclusive of PI, will stain SYTO9 positive and fluoresce in green. It should be noted that, although the membrane impermeability of PI is claimed by the manufacturer to be the major reason for nonstaining of vital cells, some strains are said to possess efflux pumps that can actively remove PI from the cell. The technique can be applied rapidly because no protracted incubation period is necessary. Depending on the specific case, quantitative results can typically be obtained within 1 h using a fluorescent microscope, fluorometer, or Contract grant sponsor: BBSRC. *Correspondence to: S. M. Stocks, 2JS.012, Novozymes Fermentation Pilot Plant, Smørmosevej 25, 2880 Bagsværd, Denmark. E-mail: [email protected] Published online 30 August 2004 in Wiley InterScience (www. interscience.wiley.com). DOI: 10.1002/cyto.a.20069

190

STOCKS

tion of apparently simultaneous PI/SYTO9-positive staining. This was done in a cell-free system of DNA stained with different proportions of PI and SYTO9. MATERIALS AND METHODS Stains The standard BacLight kit (L-7007) contains premixed stain that is ready for use. SYTO9 is present at 1.67 mM, and PI is present at 18.3 mM. Another kit (L-7012) provides the stains as separate solutions in dimethylsulfoxide, with SYTO9 at 3.34 mM and PI at 20 mM. The L-7012 kit allows the user to adjust the staining conditions to suit the application, and this was the kit used in this study. The stains were separately diluted in phosphate buffered saline (10 mM Na2HPO4, 1.8 mM KH2PO4, 170 mM NaCL, and 3.3mM KCl in ultrapure water) before staining. FIG. 1. Spatially resolved viability distribution in Streptomyces murinus. The SYTO9 solution (Molecular Probes) was diluted fourfold and mixed in equal volume with the PI solution provided in the kit. Three microliters was added to a 1-ml sample of approximately 0.2 g DCW/L per sample (1). This image was captured on Kodak Gold 400ASA film with an Olympus BX microscope using a 40⫻ objective. Courtesy of Novozymes A/S.

flow cytometer. More complex analyses such as image analysis may take longer depending on the quality of images and the degree of autonomy that can be achieved. In studying the Actinomycetes or Streptococci, the true value of a fluorescent method for viability determination lies in its ability to spatially resolve heterogeneities in viability between the compartments of hyphae or mycelia (Fig. 1), something a traditional colony-forming unit assay cannot do. As such, BacLight represents the only “routine” practical method for establishing viability in filamentous bacteria (1). The rapidity of the technique has led to its application for evaluating biological threats (2), water cleanliness (3), processed meat hygiene (4), and the efficacy of antiseptics (5) or bacteriocin leucocin (6), to name a but a few. When BacLight is compared with traditional colony-forming unit counts, there tends to be good correlation, at least for single-celled species (2,4,6), although it can advantageously report viable but nonculturable cells as live and therefore correctly identifies them as potentially harmful (7). It is accepted that cells staining positive for PI are nonculturable and nonviable (8 –13), but it has also been noted that apparently simultaneous staining of cells by PI and SYTO9 can occur, leading to results that can be difficult to interpret. For example, it is possible to obtain equal red fluorescence intensity for live and dead populations of Bacillus clausii when observed in a flow cytometer (14) when only the level of green fluorescence is used for determination of viability. Similar observations have been obtained with Listeria monocytogenes (6). Routine application of BacLight in any specific case should therefore be thoroughly investigated and validated. The purpose of this research was to further characterize the PI/SYTO9 staining system and to identify the conditions that can give rise to the potentially confusing situa-

Fluorescence Measurements Fluorescent properties of the cell-free preparations were established in a Perkin-Elmer Fluorometer (Oak Brook, IL, USA). With this particular instrument, a suitable high-tension setting for the photomultiplier was 700 V. Where excitation spectra were collected, excitation and emission slit widths were set at 5 and 15 nm, respectively. These values were reversed when emission spectra were collected. Excitation and Emission Spectra To establish emission– excitation spectra, SYTO9 and PI were added to solutions of DNA. The manufacturer recommends that PI should be used at a concentration six times that of SYTO9 in initial trials with the kit. In accordance with this recommendation, the excitation– emission spectra of SYTO9 and PI were determined separately at concentrations of 400 nM and 2.4 ␮M, respectively. DNA concentration was 20 ␮g/ml in both systems. Excitation spectra were established by recording the intensity of emissions at a fixed wavelength and changing the excitation wavelength, and vice versa for determination of emission spectra. Estimation of Association Constants To estimate the affinities of the stains for DNA, SYTO9 and PI were used independently to stain DNA at different concentrations, and the resulting intensity of fluorescence was recorded. With these data, it was possible to estimate the equilibrium constants for the DNA-SYTO9 and DNA-PI complexes. The genome of Escherichia coli is 4 ⫻ 106 base pairs and weighs 0.0044 pg (15), thus yielding 1.5 nM of base pairs for 1 ␮g of DNA, the total molar concentration of DNA in the system ([DNA] ⫹ [DNAStain]), expressed as base pair concentration, was then known. The fluorescence intensity reflects the amount of DNAStain complex present. Using the maximum fluorescence intensity observed, it was assumed that all of the stain was bound (reasonable because the intensity of emission was rising slowly as a function of DNA concentration). After correcting for the initial, unbound fluorescence intensity,

MECHANISM AND USE OF BACLIGHT

191

[DNAStain], [DNA], and [Stain] were then calculated for each of the other points, and the equilibrium constant was found (equation 1). It is more typical to draw a Scatchard plot to calculate these constants, but the results from doing so are often poor (as is the present case, data are not presented). This is a property of traditional linearization techniques that are not robust in the presence of even very small levels of experimental error. DNA ⫹ Stain % DNAStain

K⫽

[DNAStain] [DNA][Stain]

(1)

Testing the Displacement of SYTO9 by PI To test the displacement mechanism proposed by the manufacturer, a solution of 800 nM SYTO9 was placed in the fluorometer and the fluorescent properties were determined. Having determined the unbound emission spectrum for SYTO9, DNA was added to a final concentration of 1.3 ␮g/ml (i.e., stain is in excess), and the fluorescent spectrum was determined again. PI was added to a final concentration of 4.7 ␮M, and these concentrations represented the concentration ratio of SYTO9 to PI recommended by the manufacturer. To ensure that the order of addition did not affect the resultant spectra, the experiment was repeated, beginning with a solution of PI to which DNA and then SYTO9 were added. To be consistent with flow cytometry or microscopy, an excitation at 488 nm only would have been used. However, due to the band width of the excitation beam, interference with the SYTO9 emission spectrum occurs. For this reason, the emission spectra were also determined with an excitation wavelength of 420 nm. Determining Whether Simultaneous Staining Could Occur To find conditions where the potentially ambiguous situation of simultaneously PI-positive and SYTO9-positive staining could occur in vivo (simultaneously labeled live and dead), fluorescence spectra were gathered for a range of DNA, SYTO9, and PI concentrations. To be consistent with fluorescence microscopic and flow cytometric techniques, experiments were performed with an excitation wavelength of 488 nm. To improve the clarity of the trends, a longpass filter at 530 nm was included before the emission detector to eliminate the interference of the excitation band (similar to filters used in microscopy or flow cytometry). RESULTS Excitation and Emission Spectra Figure 2 shows the excitation and emission spectra for SYTO9 and PI. Notice that different scales for intensity have been used because of the large difference between the intensity of SYTO9 and PI emissions. From the excitation spectra it can be seen that SYTO9 emits most strongly when excited at 480 nm, close to the 488 nm of a typical flow cytometer. In contrast, PI is best excited at 540 nm but nevertheless fluoresces when excited at 488

FIG. 2. Excitation and emission spectra for 400 nM SYTO9 and 2.4 ␮M PI in 20 ␮M DNA. For SYTO9, emission spectra were collected with excitation at 420 nm and the absorption spectrum was measured by emissions at 580 nm. For PI, the emission spectrum was collected with excitation at 488 nm and the absorption spectrum was collected with emissions at 640 nm.

nm. Emissions of SYTO9 are also much stronger than those of PI, having a similar emission strength at approximately 600 nm, a result that might explain apparently simultaneous staining. Another important feature is that the emission spectrum of SYT09 overlies the excitation spectrum of PI, indicating a potential for fluorescence resonance energy transfer (FRET) to occur. Estimation of Association Constants The fluorescence intensities of SYTO9 and PI increased as a function of the DNA concentration until a steady value was achieved, when the DNA was saturated with stain (Fig. 3). At the limits, when DNA was in excess or stain was in excess, the assumptions resulted in less accurate estimations of the association constant; however, for the points on the most curved parts of the graph, the estimates result in values of K in good agreement with each other. KPI was 3.7 ⫻ 105/M and KSYTO9 was 1.8 ⫻ 105/M. Although further work could be undertaken to refine these estimates, they clearly demonstrate the higher affinity of PI for nucleic acid. The estimates are of the correct order. Ethidium bromide, a singly charged relative

192

STOCKS

thetical displacement mechanism predicts that the subsequent addition of SYTO9 would have had no effect on the PI emission; however, addition of SYTO9 caused an enhancement of the PI emission. This is consistent with the emission enhancement typical of FRET. The excitation– emission spectra of the two fluorophores, the dimensions of DNA (0.34 nm between base pairs) (15), the proposed frequency of intercalation of PI (every four to five base pairs for the closely related ethidium bromide) (16), and the fact that fluorophores will be absorbed with dipoles in parallel planes while only slightly rotated are highly consistent with a FRET mechanism, indicating that the stains can be simultaneously present on the DNA molecule. Determining Whether Simultaneous Staining Could Occur

FIG. 3. Fluorescence intensity as a function of DNA concentration: 400 nM SYTO9, ␭ex ⫽ 420, ␭em ⫽ 500 nm; 2.4 ␮M PI, ␭ex ⫽ 488nm, ␭em ⫽ 617 nm.

of the doubly charged PI, has an equilibrium constant of 1.5 ⫻ 105/M under similar conditions of pH and ionic strength and calculated in the same way (16). Testing the Displacement of SYTO9 by PI The resultant spectra from sequential additions of SYTO9, DNA, and PI and of PI, DNA, and SYTO9 are show in Figure 4. Where spectra were collected with initially unbound SYTO9, addition of DNA enhanced fluorescence in excess of 100-fold, indicating that binding had occurred. Subsequent addition of PI completely suppressed SYTO9 fluorescence to an intensity less than the unbound value, and although much less intense than its SYTO9 predecessor (hence, the use of a logarithmic scale), a peak for the PI emission was observed. Similar observations were made when an excitation of 488 nm was used, except that the band width of the excitation source was observed to overlap the SYTO9 emission spectrum. These observations are consistent with the hypothesis that PI displaces SYTO9 from DNA. Where spectra were collected with initially unbound PI (Fig. 4), addition of DNA resulted in an enhancement of fluorescence by a factor of approximately 2. The hypo-

Results for a wide range of PI, SYTO9, and DNA concentrations are presented in Figure 5. For each of the conditions tested, it was clear that the relative concentrations of PI, SYTO9, and DNA were of crucial importance. At low DNA concentrations, PI was almost exclusively responsible for fluorescent emissions. As DNA concentration was increased, the intensities of the PI and SYT09 emissions increased. Under the right conditions (DNA in excess of PI), the SYTO9 emission surpassed the PI emission, presumably because sufficient sites for SYTO9 and PI intercalation were present, with these being sufficiently separate to avoid suppression of the SYTO9 emission by FRET (i.e., spacing ⬎ 10 nm, or 30 base pairs). Increasing the concentration of PI had the effect of reducing the contribution of SYTO9 to the spectra emitted, although the PI emission itself was relatively unaffected. This may arise as the result of a tradeoff between the affinity displacement of SYTO9 by PI and a reduction in FRET due to desorption of SYTO9. Increasing the SYTO9 concentration at low PI concentration enhances the SYTO9 emission. At the high PI concentration, increasing the SYTO9 concentration enhances the PI emission, presumably because more SYTO9 is forced into the DNA enhancing FRET to PI. Conditions where SYTO9 and PI had similar emission intensities (i.e., giving rise to potentially ambiguous viability staining in vivo) were identified at PI-to-DNA base pair ratios (PI:DNA) of approximately 0.4:1. DISCUSSION Changing the DNA concentration in the presence of a constant quantity of stain permitted an approximation of the two equilibrium constants for the two stains. These were 1.8 ⫻ 105/M and 3.7 ⫻ 105/M for SYTO9 and PI, respectively. This confirmed the potential for an affinity displacement of SYTO9 by PI as the manufacturer suggested. The excitation and emission spectra of the two stains showed that SYTO9 and PI have distinctly different emission spectra and, most interestingly, that the SYTO9 emission spectrum almost completely covers the same range as the PI excitation spectrum. This finding and consideration of the dimensions and architecture of nucleic acid indi-

MECHANISM AND USE OF BACLIGHT

193

FIG. 4. Sequential additions of 800 nM SYTO9, 1.3 ␮g/ml DNA, and 4.7 ␮M PI for excitation at 420 and 480 nm. The emission for PI in the initially PI-stained system (right column) is enhanced after SYTO9 addition, indicating FRET, which requires simultaneous binding of PI and SYTO9 in close proximity. The SYTO9 emission is completely quenched when PI is bound.

cate that potential for FRET exists, and this was subsequently confirmed in experimental observations. This demonstrates that it is possible for the stains to be simultaneously intercalated. Changing the relative quantities of DNA, SYTO9, and PI showed that it was possible, especially in the presence of excess DNA, to obtain DNA labeled simultaneously by SYTO9 and PI. This indicates that the choice of concentrations in determining bacterial viability is of crucial importance. Most specifically, it must be ensured that PI is in excess of the DNA concentration (by a molar ratio of greater than 0.4:1 under the conditions used here), whereas the concentration of SYTO9 is not as critical and can be adjusted if attenuation of brightness is required. This was done for Streptomyces murinus (Fig. 1), which was grown in a complex production medium in a pilot scale vessel at Novozymes A/S, where the organism is used for production of glucose isomerase. To check that PI was in excess of the nucleic acid concentration by the determined ratio, the molar ratios of PI and nucleic acid were estimated: the DNA and RNA contents of Streptomyces coelicolor (a close relative of S. murinus) were previously measured at 0.2 to 0.1 g/g and 0.1 to 0.22 g/g for growth rates of 0.024 to 0.303 per hour (17). These values yielded a total

nucleic acid content of approximately 0.31 g/g, regardless of the growth rate. For the diluted 0.2 g DCW per sample, the final molar ratio of PI:DNA was approximately 300:1, which is considerably higher than the ratio for ambiguous staining of 0.4:1 for PI:DNA. This indicates that the conditions used for staining in vivo were suitable. Investigators reporting apparently equal red fluorescence in live and dead populations in the flow cytometer correctly identified the potential problems associated with relative dye concentrations and the possibility of PI quenching the SYTO9 emissions and correctly interpreted the apparently dual (PI and SYTO9) stained population of Bacilli as only SYTO9 positive (14). The present results indicate that the apparent dual staining results from the fact that the 600-nm emission of SYTO9 bound to DNA is equal in intensity to that expected from PI bound to DNA. The user of the flow cytometer cannot see the sample directly to determine its coloration and thus does not readily detect this effect. In addition, there is a natural tendency for photomultiplier tubes to be more sensitive to the higher energy of photons at shorter wavelengths of light, i.e., they tend to be more sensitive to green than to red emissions, unless they have been specifically optimized for detec-

194

STOCKS

FIG. 5. In vitro emission spectra as a function of different DNA, SYTO9, and PI concentrations. DNA concentrations were 0.5, 1.3, 3.2, 8, and 20 ␮g/ml (equally spaced on a logarithmic scale), corresponding to base pair concentrations in the range 0.8 to 30 ␮M. Approximately equal SYTO9 (left peak) and PI (right peak) emissions can give rise to ambiguous viability determination in vivo.

tion of one wavelength over another. A simple check with a fluorescence microscope is therefore strongly recommended before interpreting any data from the flow cytometer (6). CONCLUSION When staining bacteria to routinely determine viability, these findings are of crucial importance. It has been demonstrated that consideration of the relative intensities of the emission spectra of PI and SYTO9 is needed to correctly interpret quantitative data from image analysis or flow cytometry. Moreover, poor choice of the relative concentrations of nucleic acid and PI could result in confusing results for cell viability. For BacLight to produce good results, PI must be in sufficient excess to stain all

nucleic acid present (including extracellular material), whereas the SYTO9 concentration can be decreased to attenuate the green emission and need not be in excess. The molar PI:DNA ratio resulting in a SYTO9 emission lower than the PI emission was 0.4:1. Provided that appropriate control or validation experiments are performed, BacLight can provide a robust and previously unavailable method for determining viability with a range of applications. It is perhaps the only routinely applied method for determination of hyphal viability in many commercially important Actinomycete species, and its application as a research tool at the research and design level may result in process improvements through strain improvement or better physiologic understanding of fermentation processes (18).

MECHANISM AND USE OF BACLIGHT

ACKNOWLEDGMENTS The kind assistance of David Hopwood, Eric Cundliffe, Imanuel Sebastine, and Colin R. Thomas is acknowledged. LITERATURE CITED 1. Sebastine IM, Stocks SM, Cox PW, Thomas CR. Characterisation of percentage viability. Biotechniques 1996;13:419 – 423. 2. Teska JD, Coyne SR, Henchal EA, Hadfield TL, Hilyard EJ, Ezzell JW. Rapid viability assessment of biological threat agents. Biomed Lett 1998;230:155–162. 3. Endo H, Nakayama J, Hayashi T, Watanabe E. Application of flow cytometry for rapid determination of cell number of viable bacteria. Fisheries Sci 1997;63:1024 –1029. 4. Duffy G, Sheridan JJ. Viability staining in a direct count rapid method for the determination of total viable counts on processed meats. J Microbiol Methods 1998;31:167–174. 5. Langsrud S, Sundelheim G. Flow cytometry for rapid assessment of viability after exposure to a quaternary ammonium compound. J Appl Bacteriol 1996;81:411– 418. 6. Swarts AJ, Hastings JW, Roberts RF, von-Holy A. Flow cytometry demonstrates bacteriocin induced injury to Listeria monocytogenes. Curr Microbiol 1998;36:266 –270. 7. Rigsbee W, Simpson LM, Oliver JD. 1997. Detection of the viable but non-culturable state in Escherichia coli O157:H7. J Food Safety 1997; 16:255–262. 8. Davey HM, Kell DB, 1996. Flow cytometry and cell sorting of heterologous microbial populations: the importance of single cell analyses. Microbiol Rev 1996;60:641– 695. 9. Nebe von-Caron, G. Analysis of naturally occurring microbial populations from diverse environments [PhD thesis]. Coventry, UK: University of Coventry; 1998.

195

10. Hewitt CJ, Boon LA, McFarlane CM, Nienow, AW. The use of flow cytometry to study the impact of fluid mechanical stress on Escherichia coli W3110 during continuous cultivation in an agitated bioreactor. Biotech Bioeng 1998;59:612– 620. 11. Hewitt CJ, Nebe von-Caron G, Nienow AW, McFarlane CM Use of multi-staining flow cytometry to characterise the physiological state of Escherichia coli W3110 in high cell density fed-batch cultures. Biotech Bioeng 1999;63:705–711. 12. Hewitt CJ, Nebe von-Caron G, Nienow AW, McFarlane CM. The use of multi parameter flow cytometry to compare the physiological response of Escherichia coli W3110 to glucose limitation during batch, fed batch and continuous culture cultivations. J Biotech 1999; 75:251–264. 13. Boswell CD, Hewitt CJ, Macaskie LE. An application of bacterial flow cytometry: Evaluation of the toxic effects of four heavy metals on Acinetobacter sp. with potential for bioremediation of contaminated waste waters. Biotech Lett 1998;20:857– 863. 14. Christiansen T, Michaelsen S, Wumpelmann M, Nielsen J. Production of savinase and population viability of Bacillus clausii during high cell density fed batch cultivations. Biotech Bioeng 2003;83:344 –352. 15. Darnell J, Lodish H, Baltimore D. Molecular cell biology. New York: Scientific American Books; 1986. 16. Gaugain B, Barbet J, Capelle N, Roques BP, Le Pecq JB. DNA bifunctional intercalaters. 2. Fluorescence properties and DNA binding interaction of an ethidium homodimer and an acridine ethidium heterodimer. Biochemistry 1978;17:5078 –5088. 17. Shahab N. Modulation of macromolecular composition and morphology of Streptomyces coelicolor A3(2) on growth rate [PhD thesis]. Manchester, UK: University of Manchester Institute of Science and Technology; 1995. 18. Stocks SM, Thomas CR. Viability, strength, and fragmentation of Saccharopolyspora erythraea in submerged fermentation. Biotech Bioeng 2001;75:702–709.