Rapid fluorometric measurement of trace amounts of contaminant DNA in drinking water.

Introduction

Contamination of drinking water by pathological organisms is a critical issue in the field of environmental water quality. Because an enormous number of pathological species exist on earth, it is impractical to implement an initial water quality test to identify each pathogen. Methods for evaluating the bacteriological quality of water are often based on cultivation of bacteria in a sample (e.g., standard plate count or heterotrophic plate count); however, this approach has the disadvantage that a long incubation time is required for colony growth (1). Moreover, the species of bacteria grown depends on the type of isolation media used, and some bacteria species and nonbacterial organisms are difficult to grow in an ordinary medium (2). Demand is growing for a rapid water quality test that covers a broad spectrum of pathological organisms. Because all living organisms contain nucleic acids, detection of DNA and RNA in water may serve as a comprehensive screening test for water quality. In this paper, we introduce a rapid and simple-to-use methodology for measuring DNA content in water and report findings of significant DNA contamination in some bottled drinking water.

Materials and Methods

Materials

The following materials were purchased from suppliers:

* salmon sperm DNA,

* SYBR DX,

* nitrocellulose and nylon membranes,

* RNase-free DNase I,

* rRNA,

* DNase-free RNase, and

* 24-well and 96-well plates.

The suppliers are listed in Table 1. Various brands of bottled water were purchased at a local supermarket in Irvine, California.

Methods

Various concentrations of DNA were suspended in autoclaved double-distilled water (dd[H.sub.2]O) or the following solution: 10 millimoles per liter (mM) Tris - pH 7.6, 1 mM EDTA - pH 8.0, 25 mM NaCl (TEN). The resulting solutions were vacuum-filtered through nylon membranes with the aid of a filter holder. The nylon membranes were 13 millimeters (mm) in diameter and had 0.45-micrometer ([[micro]meter]) pores. In some experiments, small volumes of DNA were simply spotted onto the membranes, then air dried.

The membranes were placed into 24-well plates with the DNA side up and were exposed to ultraviolet light (Stratalinker) for crosslinking. Each membrane was then incubated with 200 [[micro]liter] of 1:1000 dilution of SYBR DX or 0.5 micrograms per milliliter ([[micro]gram]/mL) of ethidium bromide in 0.5X the following solution: 45 mM Tris-borate, i mM EDTA - pH 8.0 (TBE). Incubation was at room temperature for five to 30 minutes. Each membrane, after it was washed with 1 mL of dd[H.sub.2]O three times, was placed DNA side down, and the fluorescent signal of the SYBR DX was determined in a fluorescent plate reader (CytoFluor 2300). For SYBR DX, excitation and emission wavelengths were 485 nanometers (ntu) (bandwidth, 20 nm) and 530 nm (bandwidth, 25 nm), and for ethidium bromide, they were 485 nm (bandwidth, 20 nm) and 630 nm (bandwidth, 20 nm), as described in previous studies (3-5).

In some experiments, immobilized DNA was incubated with 10 units of DNase I in 200 [[micro]liter] of the following solution: 50 [[micro]molar] Tris - pH 8.0, 10 mM Mn[Cl.sub.2], 50 [[micro]gram]/mL bovine serum albumin (BSA). Immobilized RNA was incubated with 0.5 [[micro]gram] of RNase in 200 [[micro]liter] of 1X the following solution: 40 mM Tris-acetate, 1 mM EDTA - pH 8.0 (TAE). The DNA and RNA were incubated at 37 [degrees] C for three to five hours before the SYBR DX assay was conducted.

Results

The first series of experiments was conducted to find suitable membranes for DNA quantification. As shown in Figure 1, two different sources of nitrocellulose membranes failed to demonstrate any positive DNA signals over fresh DNA-free membranes in both SYBR DX and ethidium-bromide staining. Although one nylon membrane (Brand a) expressed positive DNA signals after staining with ethidium bromide, DNA signals were not detected in other membranes [ILLUSTRATION FOR FIGURE 1B OMITTED]. In contrast to ethidium bromide, SYBR DX expressed significant fluorescent signals over plain membranes for two brands (brands a and d), although Brand a exhibited much higher background than Brand d [ILLUSTRATION FOR FIGURE 1A OMITTED]. Interestingly, background SYBR DX fluorescence of two different lots of Brand d (brands d and [d.sup.1]) differed significantly [ILLUSTRATION FOR FIGURE 1A OMITTED].

As shown in Figure 2, SYBR DX fluorescence increased in proportion to the amount of DNA in the solution in a dose-dependent fashion. The detection limit for DNA in this assay was approximately 50 ng. Although background SYBR DX fluorescence was much higher in the membranes than in the solutions, SYBR DX fluorescence also increased on DNA-spotted membranes in a dose-dependent fashion with a similar detection limit of 50 ng of DNA [ILLUSTRATION FOR FIGURE 2 OMITTED].

To test the recovery of applied DNA on membranes, the same amount of DNA (500 ng) was dissolved in different volumes of dd[H.sub.2]O and filtered through a nylon membrane (Brand d). As a positive control, the same amount of DNA dissolved in 50 [[micro]liter] of dd[H.sub.2]O was simply spotted onto the membranes and air dried. The 500-ng amount was selected for this experiment because this value is within the linear range of the dose-response curve as shown in Figure 2. As a result, DNA was captured by [TABULAR DATA FOR TABLE 1 OMITTED] nylon membranes even when diluted to 1 ng/mL, although for the 50-to-500-mL range, some DNA was lost during filtration [ILLUSTRATION FOR FIGURE 3 OMITTED].

DNA contamination was then tested in various bottled drinking waters. When SYBR DX was diluted in tested water, no significant fluorescence was detected [ILLUSTRATION FOR FIGURE 4A OMITTED]. When 100 mL of tested water was passed through the nylon membranes, and stained with SYBR DX, however, significant fluorescence was detected in two different bottles of Brand E, corresponding to that of 500 ng of DNA in 100 mL of dd[H.sub.2]O [ILLUSTRATION FOR FIGURE 4B OMITTED]. High SYBR DX fluorescence in Brand E water was confirmed for five different bottles (data not shown).

If bottled water contains some components that inhibit the reaction of SYBR DX with DNA or interfere with fluorescent signals, the SYBR DX reading may not reflect the actual amount of contaminant DNA in the solution. Because DNA was trapped on a membrane and resuspended in a 0.5X TBE buffer, however, such components may have been removed before SYBR DX staining. To further analyze the effect on SYBR DX staining of unknown factors in drinking water, 500 ng of DNA was added to various bottled waters, and the solutions were immediately filtered through nylon membranes. As shown in Figure 4C, the SYBR DX reading was reduced in brands C and D water, but not in Brand A, Brand B, or dd[H.sub.2]O. This suggests that some factor(s) in brands C and D water may influence SYI3R DX reading, although precise characteristics are unknown at this time.

In the experiment described in Figure 3, it was found that some portion of applied DNA passed through the nylon membranes when a large volume of diluted solution was applied. The passed-through fraction was further trapped by a second line of nylon membranes, which were then stained with SYBR DX. As shown in Figure 5, SYBR DX fluorescence on second membranes was much less than that on first membranes for 500 ng of dd[H.sub.2]O and Brand E water. This result suggests the efficient capture capacity of the first membrane. Furthermore, slightly high SYBR DX fluorescence on the second membranes indicates that DNA passed through the first membranes was successfully captured by the second membranes. This also indicates that nylon membranes are useful not only for detection of DNA, but also for removal of DNA from solution.

Various volumes of DNA solution (5 ng/ mL and 1 ng/mL in dd[H.sub.2]O) and Brand E water were filtered through nylon membranes and were stained with SYBR DX. As shown in Figure 6, SYBR DX fluorescence of Brand E water was volume dependent, which is a pattern similar to that shown by solutions with between 1 and 5 ng/mL of DNA.

To further confirm that SYBR DX fluorescence on membranes was due to immobilized DNA or RNA, membranes were digested with DNase or RNase. As shown in Figure 7, SYBR DX fluorescence of DNA-immobilized membranes was significantly reduced by DNase treatment, but not by RNase treatment, whereas SYBR DX fluorescence of RNA-immobilized membranes was significantly reduced by RNase, but not by DNase. Moreover, SYBR DX fluorescence of Brand E water was reduced by DNase, but not by RNase, suggesting DNA contamination in Brand E water.

Discussion

Nitrocellulose and nylon membranes are frequently used in molecular biology to detect or quantify the amount of specific DNA, RNA, and polymerase chain reaction (PCR) products, because the membranes have a high capacity for absorption of nucleic acids. In traditional Southern and Northern blotting experiments, DNA and RNA, respectively, are first separated by electrophoresis in agarose gels, then transferred to membranes (6,7). Alternatively, samples are simply blotted onto membranes for rapid quantification of the target gene (dot blot or slot blot) (8). Immobilized DNA or RNA is then hybridized with labeled probes, and this step is followed by various visualization procedures such as an exposure of radioisotopes to X-ray films, chemiluminescence or fluorescence detection, or colorimetric reaction. Because of the technical difficulty, however, the total amount of DNA or RNA immobilized on membranes is not usually quantified in each experiment.

Previous studies demonstrated that oligonucleotides, PCR products, and mRNA can be quantified by the fluorescent indicator dye Yoyo-1 in microtiter plates (3-5,9). This study expanded the Yoyo-1 experiments to develop a simple method for quantification of membrane-bound DNA. The first attempts lysed DNA-bound membranes with various solvents, then the Yoyo-1 fluorescence of DNA was measured in a fluorometer. It was found that nitrocellulose could be lysed with methanol or dimethylsulfoxide; however, Yoyo-1 fluorescence was diminished significantly in these solvents. When solvents were diluted with buffer to recover Yoyo-1 fluorescence, nitrocellulose became insoluble and produced opaque materials, making the assay cumbersome.

A second series of experiments focused on nitrocellulose membranes, because these membranes become transparent after mixing with immersion oil (10). It was initially thought that all the trapped DNA could be stained with Yoyo-1 and measured by an ordinary fluorescent plate reader after immersion oil had made the nitrocellulose membranes transparent. Commercially available 96-well filter plates with nitrocellulose membranes attached at the bottom (Millipore) made the experiments easy to carry out. When DNA was first mixed with Yoyo-1 and then applied to membranes, which were next washed and treated with immersion oil, the amount of DNA could be measured in a fluorescent plate reader in a dose-dependent manner (data not shown). When DNA was first filtered through nitrocellulose and stained with Yoyo-1, however, the high background fluorescence of Yoyo-1 on nitrocellulose membranes prevented detection of significant DNA signals (data not shown).

The third trial used SYBR DX, a recently available fluorescent dye for DNA on membranes. The manufacturer's protocol recommends using nylon membranes, but not nitrocellulose. The experiments presented in Figure 1 confirmed that recommendation. Although the membranes are not transparent, fluorescent signals can be measured in an ordinary fluorescent plate reader, probably because applied DNA is immobilized on the surface of the membranes. When large amounts of DNA are applied, some portion of DNA may be immobilized within the interior o f the membranes, where excitation and emission fluorescence are hindered by opaque membrane fibers. Interestingly, the SYBR DX staining performance of nylon membranes varied widely among membranes from different sources and even among different lots from the same manufacturer [ILLUSTRATION FOR FIGURE 1 OMITTED]. Therefore, one of the key elements of the successful experiments in this study was finding appropriate nylon membranes.

Similar results have been reported by Wilkins and Kearney in a study that deposited DNA onto black membranes and then stained the membranes with ethidium bromide (11). The major advantage of the assay described in this paper is that it uses well-characterized nylon membranes to immobilize nucleic acids. In the authors' preliminary experiments, the fluorescent plate reader did not detect fluorescent signals of stained DNA on black membranes (data not shown). Furthermore, SYBR DX can stain both single- and double-stranded DNA and RNA, whereas ethidium bromide does not stain RNA or single-stranded DNA very well. Therefore, SYBR DX is a better choice than ethidium bromide for screening nucleic-acid contamination in various solutions. The method described in this paper allows the amount of DNA or RNA in a diluted solution to be measured simply by passing a large volume of solution through nylon membranes until a detectable level of DNA or RNA is captured.

Drinking water has been shown to contain a natural bacterial content, and no relation has been found between human disease such as gastroenteritis and the natural bacteria found in drinking water (12, 13). Edberg et al. estimated health risk from these natural bacteria by the determination of cytotoxicity and invasiveness in a human enterocyte cell line (14). The results of that study, however, do not necessarily reflect in vivo conditions.

The experiments presented in this paper found DNA contamination in one brand of bottled drinking water when 100 mL was filtered. If a much larger volume of water were filtered, DNA contamination might be detected in other brands. The presence of DNA does not directly indicate any risk of infection or hazard for human health; however, because the presence of DNA or RNA in solution may suggest a present or past history of contamination with living organisms such as virus, bacteria, fungi, moss, or animal debris, DNA-free drinking water may be, in general, preferable to DNA-contaminated water.

There is no consensus about a DNA threshold level in drinking water that could cause infection in human beings. Further studies are required to set a drinking water threshold level for DNA or bacterial contamination that might constitute a hazard to human health. The method described in this paper may be applicable not only to water quality testing, but also to clean-room air quality testing. Such a procedure would simply filter a large volume of air.

Recent models of fluorescent plate readers can read fluorescence either from the top or from the bottom. The fluorescent plate reader used in this study could read from bottom only. It was necessary, therefore, to first place the membranes in 24-well plates with the DNA side up for ultraviolet light cross-linking and SYBR DX staining. The membranes were then inverted with the DNA side down for reading of its fluorescence. When membranes were placed with the DNA side up, fluorescent signals were significantly diminished in the fluorometer used for this study (data not shown).

Another difficulty was that the round-shaped nylon membranes, which were 13 mm in diameter and fit perfectly into the 24-well plates and the filtration manifold, were not commercially available in Brand d. For this study, each membrane was cut with sterile scissors to the appropriate size. When similar experiments are conducted on a large scale in the future, Brand d membranes should be attached at the bottom of 96-well plates, which are suitable for vacuum filtration. Because of the simplicity of this assay, automated devices can also be developed. The development of such devices may make the present assay a part of standard water quality tests in various environmental fields in the future.

REFERENCES

1. Reasoner, D.J. (1990), "Monitoring Heterotropic Bacteria in Potable Water," Drinking Water Microbiology, New York, N.Y.: Springer-Verlag, pp. 452-477.

2. Reasoner, D.J., and E.E. Geldreich (1985), "A New Medium for the Enumeration and Subculture of Bacteria from Potable Water," Applied Environmental Microbiology, 49(1):1-7.

3. Ogura, M., C. Keller, K. Koo, and M. Mitsuhashi (1994), "Use of the Fluorescent Dye Yoyo-1 to Quantify Oligonucleotides Immobilized on Plastic Plates," Bio Techniques, 16(6):1032-1034.

4. Ogura, M. and M. Mitsuhashi (1994), "Screening Method for a Large Quantity of Polymerase Chain Reaction Products by Measuring Yoyo-1 Fluorescence on 96-Well Polypropylene Plates," Analytical Biochemistry, 218(2):458-459.

5. Miura, Y., R. de Fries, Y. Ichikawa, T. Ishikawa, M. Mitsuhashi, M. Ogura, and H. Shimada (1996), "Fluorometric Deermination of Total mRNA with Oligo-dT Immobilized on Microtiter Plates," Clinical Chemistry, 42(11):1758-1764.

6. Sambrook, J., E.F. Fritsch, and T. Maniatis (1989), "Extraction, Purification and Analysis of Messenger RNA from Eukaryotic Cells," Molecular Cloning, a Laboratory Manual, 2nd ed., Cold Spring Harbor, N.Y.: Cold Spring Harbor Laboratory Press, pp. 7.28-7.52.

8. Ausubel, F.M., R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl (1995), "Dot and Slot Blotting of DNA," Current Protocol in Molecular Biology, Brooklyn, N.Y.: John Wiley & Sons, pp. 2.9.15-2.9.20.

9. Glazer, A.N. and H.S. Rye (1992), "Stable Dye-DNA Interaction Complexed as Reagents for High Sensitivity Fluorescence Detection," Nature, 359(6,398):259-861.

10. Vera, J.C. and C. Rivas (1988), "Fluorescent Labeling of Nitrocellulose-Bound Proteins at the Nanogram Level Without Changes in Immunoreactivity," Analytical Biochemistry, 173(2):399-404.

11. Wilkins, R.J. and J.T. Kearney (1984), "Fluorometric Assays for DNA Deposited on Filters," Analytical Biochemistry, 136(1):301-308.

12. Leclerc, H., D.A. Mossel, and C. Savage (1985), "Monitoring Noncarbonated ('Still') Mineral Waters for Aerobic Colonization," International Journal of Food Microbiology, 2(4):341-347.

13. Duncan, H.E. and S.C. Edberg (1995), "Host-Microbe Interaction in the Gastrointestinal Tract," Critical Review of Microbiology, 21(2):85-100.

14. Edberg, S.C., M. Escarzaga, C. Kontnick, and S. Kops (1997), "Analysis of Cytotoxicity and Invasiveness of Heterotrophic Plate Count Bacteria (HPC) Isolated from Drinking Water on Blood Media," Journal of Applied Microbiology, 82(4):455-461.

Corresponding author: Dr. Masato Mitsuhashi, Division of Medical Sciences, Hitachi Chemical Research Center, Inc., 1003 Health Sciences Road West, Irvine, CA 92612.

E-mail: [e-mail].