Multiple and Simultaneous Detection of Specific Bacteria in

Multiple and Simultaneous Detection of Specific Bacteria in Enriched Bacterial Communities Using a DNA Microarray Chip with Randomly Generated Genomic...
1 downloads 0 Views 685KB Size
Download PDF
Anal. Chem. 2005, 77, 2311-2317

Multiple and Simultaneous Detection of Specific Bacteria in Enriched Bacterial Communities Using a DNA Microarray Chip with Randomly Generated Genomic DNA Probes Byoung Chan Kim, Ji Hyun Park, and Man Bock Gu*

Advanced Environmental Monitoring Research Center (ADEMRC) and National Research Laboratory on Environmental Biotechnology, Gwangju Institute of Science and Technology (GIST), 1 Oryong-dong, Buk-gu, Gwangju 500-712, Republic of Korea

A DNA microarray chip for detecting the presence of specific bacterial strains was developed using random genomic probes derived from genomic DNA, i.e., without any sequence information. Thirteen bacteria from different genuses were selected as targets. For the fabrication of the random genomic probes, genomic DNA from pure cultures of each bacterium was fractionated using several pairs of restriction endonucleases. After size fractionation of the genomic DNA fragments, random genomic libraries for each bacterium were constructed. From the library, specific probes were amplified by PCR and the probes were affixed to a slide glass to fabricate the DNA microarray chip. The results from tests with pure and mixed cultures of the bacteria used in the fabrication of the chips showed specific responses and only a small portion of cross-hybridization. This DNA microarray chip was also tested to detect the presence of specific bacteria in mixed populations. In these tests, it was demonstrated that this system provided a fast and specific response to the presence of bacterial species in mixed samples, even in activated sludge samples. This indicates that any DNA microarray chip for the detection of specific bacteria can be fabricated using the same protocols as presented in this study without requiring any genus level sequence information from pure isolates.

various bacteria that may be present since most natural bacteria cannot be cultured in vitro.3 Nowadays, biomolecular approaches are more favorable for detecting and identifying specific bacteria in complex samples. One approach for the direct detection of bacteria is the application of specific antibodies based on protein chip platform technology.4,5 Chips that use antibodies offer both specificity and a high affinity, depending on the choice of reagents. However, antibody methods have some limitations, including the difficulty of generating specific antibodies for every target as well as the stability of the antibody during long time storage after chip fabrication. Furthermore, appropriate labeling strategies should be considered when using antibodies for target proteins or cells.6 Recently, Dworzanski et al. discriminated different bacteria using tandem mass spectrometry combined with a proteomics analysis.1 This is more comprehensive and is a direct tool to discriminate bacteria using composed protein profiles. However, this method needs large and expensive analysis instrumentation and a huge reference database for the protein profiles of each bacterium. Another approach uses nucleic acid-based microarray platforms. Initially, microarrays were, on the whole, used for gene expression profiling.7-9 However, a potential existed for microarrays that could be applied to the detection of specific bacteria. Recently, techniques for bacterial determination or identification, via DNA or RNA hybridization with microarray chip platforms, have been developed.10-14 The microarray is a powerful tool

Morphological, physiological, and biochemical approaches have commonly been used for bacterial species detection and identification.1 Morphological identification using microscopy is a fast and intuitive method with no special laboratory equipment. Although morphological identification is easy, cell morphologies have been shown to differ greatly due to different environmental conditions.2 The use of nutritional medium could also be implemented for these purposes, but it is still difficult to differentiate

(3) Amann, R. I.; Ludwig, W.; Schleifer, K. H. Microbiol. Rev. 1995, 59, 143169. (4) Bae, Y. M.; Oh, B. G.; Lee, W. C.; Lee, W. H., Choi, J. W. Anal. Chem. 2004, 76, 1799-1803. (5) Delehanty, J. B.; Ligler, F. S. Anal. Chem. 2002, 74, 5681-5687. (6) Zhu, L.; Ang, S.; Liu, W. T. Appl. Environ. Microbiol. 2004, 70, 597-598. (7) Afshari, C. A.; Nuwaysir, E. F.; Barrett J. C. Cancer Res. 1999, 59, 47594760. (8) Eisen, M. B.; Spellman, P. T.; Brown P. O.; Botstein, D. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 14863-14868. (9) Waring, J. F.; Ciurlionis, R.; Jolly, R. A.; Heindel, M.; Ulrich, R. G. Toxicol. Lett. 2001, 120, 359-368. (10) Adamczyk, J.; Hesselsoe, M.; Iversen, N.; Horn, M.; Lehner, A.; Nielsen, P. H.; Schloter, M.; Roslev, P.; Wagner, M. Appl. Environ. Microbiol. 2003, 69, 6875-6887. (11) Kingsley, M. T.; Straub, T. M.; Call, D. R.; Daly, D. S.; Wunschel, S. C.; Chandler, C. D. Appl. Environ. Microbiol. 2002, 68, 6361-6370. (12) Wu, C. F.; Valdes, J. J.; Bentley, W. E.; Sekowski, J. W. Biosens. Bioelectron. 2003, 19, 1-8.

* To whom correspondence should be addressed. Phone: +82 62 970 2440. Fax: +82 62 970 2434. E-mail: [email protected]. Present address effective September 1, 2005: Department of Biotechnology, Korea University; E-mail: [email protected]. (1) Dworzanski, J. P.; Snyder, A. P.; Chen. R.; Zhang, H.; Wishart, D.; Li, L. Anal. Chem. 2004, 76, 2355-2366. (2) Ka¨mpfer, P. FEMS Microbiol. Ecol. 1997, 23, 169-181. 10.1021/ac048703c CCC: $30.25 Published on Web 03/12/2005

© 2005 American Chemical Society

Analytical Chemistry, Vol. 77, No. 8, April 15, 2005 2311

offering a parallel, high-throughput detection and quantification of many nucleic acid molecules. The most important aspect when using microarray platform technologies based upon DNA or RNA is the selection of specific target probes, which are then spotted on the chips. Depending on the target probes, a single microarray could theoretically be used to detect thousands of different bacterial strains. In most DNA microarray chips, 16S ribosomal DNA sequences are widely used as the probes. However, resolution among bacteria of the same species is low while the possibility of cross-hybridization with unknown samples also exists.15-17 One study that looked into an alternative means to discriminate among bacteria of the same species with a greater resolution used functional genes.18 All of these approaches are sequence-based and specific. However, to achieve a real-world application, such as when labeled probes from the complex samples are analyzed, cross-hybridization should be considered. Whereas the use of specific genes would eliminate this issue, it is difficult to find specific gene probes that are representative of only a single bacterium. Although the number of bacteria for which the whole genome is sequenced are increasing,19 with information on nearly 100 bacterial strains currently available, the sequences from countless strains are not known. Thus, even though specific probes for target bacteria could be generated to prevent null hybridization results in the cases where the genome is known, this would require a great deal of time and money when the sequencing of the genome is not complete or even begun. Previously, we constructed a random genomic DNA probebased DNA microarray chip for three different bacterial strains.20 In this study, therefore, we report on the development of a DNA microarray chip that contains random genomic DNA probes generated from 13 different bacterial strains, for the purpose of multiple bacterial detection or bacterial community analysis. Individual specificity tests for each bacterium were performed with more a strict analysis to demonstrate fully the ability of the chip. Furthermore, this DNA microarray chip was applied in tests to detect the presence of specific bacterial strains within real samples, i.e., activated sludge. MATERIALS AND METHODS Bacterial Strains and Genomic DNA Isolation. The bacterial strains used in this study were Gordonia amarae America Type Culture Collection (ATCC) 27808T, Zoogloea ramigera ATCC 19623, Ralstonia eutropha ATCC 17699, Sphingomonas terrae Korea Culture Type Collection (KCTC) 2814, Skermania piniformis KCTC 9829, Comamonas denitrificans KCTC 12931, Enterobacter (13) Anthony, R. M.; Brown, T. J.; French, G. L. J. Clin. Microbiol. 2000, 38, 781-788. (14) Yoo, S. M.; Keum, K. C.; Yoo, S. Y.; Choi, J. Y.; Chang, K. H.; Yoo, N. C.; Yoo, W. M.; Kim, J. M.; Lee, D., Lee, S. Y. Biotechnol. Bioprocess Eng. 2004, 9, 93-99. (15) Rudi, K.; Flateland, S. L.; Hanssen, J. F.; Bengtsson, G.; Nissen, H. Appl. Environ. Microbiol. 2002, 68, 1146-1156. (16) Wilson, K. H.; Wilson, W. J.; Radosevich, J. L.; DeSantis, T. Z.; Viswanathan, V. S.; Kuczmarski T. A.; Andersen, G. L. Appl. Environ. Microbiol. 2002, 68, 2535-2541. (17) Wang, R. F.; Beggs, M. L.; Robertson, L. H.; Cerniglia, C. E. FEMS Microbiol. Lett. 2002, 213, 175-182. (18) Bodrossy, L.; Stralis-Pavese, N.; Murrell, J. C.; Radajewski, S.; Weilharter, A.; Sessitsch, A. Environ. Microbiol. 2003, 5, 566-582. (19) Doolitle, R. F. Nature 2002, 416, 697-700. (20) Kim, B. C.; Park, J. H.; Gu, M. B. Environ. Sci. Technol. 2004, 38, 67676774.

2312

Analytical Chemistry, Vol. 77, No. 8, April 15, 2005

sp. KCTC 2701, Thiobacillus thioparus ATCC 23645, Paracoccus thiocyanatus KCTC 2848, Nitrobacter winogradskyi Korea Culture Center of Microorganisms (KCCM) 41770T, Beggiatoa alba KCTC 2738, Acinetobacter calcoaceticus KCCM 40204, Aeromonas hydrophila KCTC 2358, and Escherichia coli RFM443. All strains were cultivated following type culture guidelines. The activated sludge used in this study was from a laboratory-scale operated system. The seed of the sludge was from a Gwangju wastewater treatment plant and was acclimated to degrade glucose-based artificial wastewater in the laboratory. Genomic DNA from all samples was extracted using a soil DNA extraction kit (Mobio). The concentration of the DNA extracted was determined using a UV spectrophotometer (Ultrospec 3100 pro, Amersham Bioscience), and its quality was checked using the 260 nm/280 nm ratio and by gel electrophoresis. The DNA was stored at -20 °C until being used. Fabrication of DNA Microarray Chip. The genomic DNA from each strain was individually purified from the pure culture stocks and each was fractionated by initially digesting it with EcoRI and BamHI, and part of this sample was further digested with HindIII and XhoI or HindIII, and SacII restriction enzymes (NEB, MA). To select for fragments having sizes between 200 and 1500 bp, size fractionation was performed using gel electrophoresis. The QIAquick gel extraction kit (Qiagen) was used to elute and purify the DNA fragments from the gel. To construct a random genomic library for each strain, the purified DNA fragment pools were ligated into the pPCR-Script Amp vector (Stratagene) and transformed into E. coli DH5R (Invitrogen). Several colonies were obtained carrying DNA fragments randomly digested by EcoRI/ BamHI, HindIII/XhoI, or HindIII/SacII. Several of these colonies were selected randomly and their plasmids were extracted, the cloned region was amplified through PCR with the T3 and T7 primer pairs, and then these amplified fragments were digested with the same restriction enzymes as used in their cloning to remove the plasmid-borne flanking regions. Of the clones carrying fragments having the same size, only one of them was selected and the other clones were rejected. After purification, the digested products were denatured and then spotted on a slide glass (Genomictree, Daejeon, Korea). Different sized PCR fragments were selected for each of the probes. The bacteriophage lambda PR promoter, which was amplified from plasmid pPL45021 using the primers 5′-aaa aac agg gta ctc ata c- 3′ and 5′-cca tac aac ctc ctt agt a-3′, was used as a control. Hybridization. Before hybridization, the chips were incubated in a prehybridization solution (1.75× SSC, 0.1% SDS, 10 mg/mL of bovine serum albumin) for 30 min at 65 °C. After washing the chips with distilled water for 1 min, followed by 100% 2-propanol for 1 min at room temperature, they were dried by centrifugation (600 rpm, 5 min, at room temperature). For all hybridizations, 390 ng of genomic DNA from the 13 strains (i.e., 30 ng each in a 30-µL random priming reaction) were labeled with Cy5-dCTP (Amersham Bioscience) for the reference signals while the test genomic DNA was labeled with Cy3-dCTP (Amersham Bioscience). Random priming was performed as suggested by the supplier (Roche) with 3 ng of positive control DNA in a 30-µL reaction. After 1 h of random priming at 37 °C, the reaction pools labeled with Cy3-dCTP and Cy5-dCTP were mixed together and then purified using a PCR purification kit (Qiagen). The purified (21) Love, C. A.; Lilley, P. E.; Dixon. N. E. Gene 1996, 176, 49-53.

mixture was concentrated through membrane filtration using a Microcon YM-30 (Millipore) for 5 min at 14000g and room temperature. The concentrated, labeled DNA pools were then adjusted to 50 µL with hybridization buffer, with a final concentration of 6× SSC, 0.2% sodium dodecyl sulfate (SDS), 5× Denherdt’s solution, and 0.1 mg/mL of denatured salmon sperm DNA. This hybridization mixture was boiled for 2 min at 100 °C and then injected into the gap between the chip and slide cover glass, and the hybridization was allowed to proceed for 12 h at 65 °C. Afterward, the chips were washed once with buffer 1 (2× SSC) at room temperature to remove the cover glass, then for 10 min with prewarmed buffer 2 (2 X SSC, 0.2% SDS, 65 °C), and then twice with buffer 3 (0.05× SSC) for 5 min each. Afterward, the chips were dried by centrifugation (1200 rpm, 3 min, at room temperature). Scanning and Data Analysis. The microarray chips were scanned with a Genepix 4000B laser scanner (Axon), and the images were captured as a 16-bit TIFF file format with a 10-µm resolution while intensity analysis was performed with GenepixPro 3.0 software (Axon). The intensity values of the test and reference DNA were derived from a median value of the pixel-by-pixel intensity while the ratio between the two was calculated and corrected, with the correction factor derived from the positive control. The real intensity values were calculated by subtracting the background intensity values from the pixel-by-pixel intensities obtained with the software, giving the spot intensity. The Cy3 intensity of an individual spot was corrected by multiplying by the positive control’s Cy5 intensity/Cy3 intensity ratio. This ratio, as calculated using 10 spots on the slide glass, was 1.01 ( 0.03 for all experiments. Almost all the spots satisfied the threshold value, which was 2 × standard deviation (SD) + background intensity signals, when the specific binding occurred. Positive binding was determined by selecting only those spots that had signal intensity values of greater than 2 × SD + background signal in the Cy3 channel after hybridization. The negative control showed no Cy3 or Cy5 intensity and was always below the threshold value. Tests were performed using a Dataplot package22 to validate whether the ordered intensities from the scanned image analyses contained random or nonrandom distributions.23 Microsoft Excel and SigmaPlot (SPSS) were used for all statistical calculations and data analyses. Safety Consideration. The bacteria used in this study were not reported to have pathogenic characteristics. However, all the experiments dealing with bacteria were performed in a hood and wearing presterilized gloves and protective clothes. The remaining unused cell suspensions were autoclaved after treatment with 25% sodium hypochlorite. Likewise, all the tubes, gloves, and containers that came in contact with the cell suspensions were autoclaved after treatment with 25% sodium hypochlorite. RESULT AND DISCUSSION Fabrication of the DNA Chip Using the Randomly Generated Probes. To fabricate the DNA microarray chip, the PCR probes from randomly selected colonies were spotted as shown in Figure 1A with following number of probes: G. amarae (45), Z. ramigera (45), R. eutropha (55), S. terrae (54), S. piniformis (22) http://www.itl.nist.gov/div898/software/dataplot/. (23) Baader, S. L.; Badder, K. L.; Schilling, K. Brain Res. Protoc. 1998, 3, 173182.

Figure 1. Arrangement of the probes on the DNA microarray chip and individual detection of genomic DNA. (A) The probes were deposited at discrete locations within 13 rows and immobilized on the glass surface. Each row represents the probes for a single bacterial species and the number of genomic probes used are in parentheses. (B) Detection of individual strains. The number to the left corresponds to specific bacteria: 1. G. amarae; 2, Z. ramigera; 3, R. eutropha; 4, S. terrae; 5, S. piniformis; 6, C. denitrificans; 7, Enterobacter sp.; 8, T. thioparus; 9, P. thiocyanatus; 10, N. winogradskyi; 11, B. alba; 12, A. calcoaceticus; and 13, A. hydrophila.

(54), C. denitrificans (57), Enterobacter sp. (56), T. thioparus (55), P. thiocyanatus (56), N. winogradskyi (46), B. alba (55), A. calcoaceticus (56), and A. hydrophila (56). The chips consist of 13 rows, one for each bacterial species, and each of the rows was Analytical Chemistry, Vol. 77, No. 8, April 15, 2005

2313

divided into four parallel blocks. The random probes were generated as described in the Materials and Methods. Using this procedure, the target probes for each bacterium were fabricated and are assumed to be random. The concentration of the probes for spotting was between 30 and 100 ng/µL. Detection of Individual Bacterial Strains. To examine the performance of the microarray, its ability to detect DNA from one specific bacterium is of interest. To address this issue, initial experiments were employed with a hybridization method that uses all genomic DNA involved in the fabrication of the chip, i.e., a mixed Cy5-dCTP-labeled sample including DNA from all 13 bacterial strains, and the test genomic sample, which was labeled with Cy3-dCTP. Figure 1B shows a representative response obtained when such hybridizations were performed on the microarray for each of the bacterial strains. The image shows the resulting fluorescence patterns generated by Cy5 and Cy3. Detection of the hybridized signals, an indication of target genomic DNA being present, was achieved after the test DNA was labeled with Cy3-dCTP during the random priming reaction. This is clearly shown for sample 1, where 30 ng of G. amarae genomic DNA was labeled and hybridized with the microarray, leading to positive results from the first row of the microarray (Figure 1B-1). In a similar fashion, the remaining 12 analyses were performed and their hybridization results shown (Figure 1B-2-12). The Cy3 signal intensities from each of these tests are also shown in Figure 2A. The results from tests with individual strains show that the specific responses from the region on the chip corresponding to the strain are distinct and much stronger when compared with those from the other strains, which consist of nonspecific or cross-hybridization signals. The variations in the Cy3 intensities from one probe to the next resulted from differences in the probe’s hybridization efficiencies since each differs in their size and, presumably, nucleotide sequence. However, when the average Cy3/Cy5 fluorescent ratios from individual spots within the same row were investigated, similar ratios were seen, as demonstrated in Figure 2B. For these tests, the concentration of the target DNA in both the reference and test samples was identical, and thus, the expected ratio would be a value of 1. However, the average values were always larger than 1. This is most likely due to the labeling conditions where a total of 390 ng of genomic DNA from all the bacteria was used for Cy5 labeling, while only 30 ng of specific genomic DNA for the specific detection were labeled with Cy3. The larger quantity of DNA in the mixed sample could potentially lead to a poorer degree of labeling within the sample, as well a competitive environment, which would result in a lower amount of a specific DNA being labeled and, therefore, a higher ratio. Furthermore, the fluorescent signals and ratios from replicate spots present on the same slide showed an average variation from each other of less than 2%. Although both the chip-bound probes and the labeled probes were generated randomly, Figure 2A shows that strong but specific signals were obtained, demonstrating that the random DNA probe approaches employed in this study gave specific, i.e., nonrandom, results. To determine whether the fluorescence intensities generated from such random hybridization is done so in a random or nonrandom fashion, the runs test23 was performed. This test was selected since it analyzes one-dimensionally arranged data to determine whether it is random while also estimating 2314 Analytical Chemistry, Vol. 77, No. 8, April 15, 2005

Figure 2. Representative microarray of Cy3 signal intensity. (A) Cy3 signal intensity at individual genomic DNA hybridization test. (B) Cy3/Cy5 signal ratios from individual genomic DNA hybridization tests. The standard deviations from the results in the same row are shown as error bars.

whether the probability of a given datum is a function of the outcome of a previous one. For the runs test, a Dataplot package was used. 22 The Zi score, after i runs, represents the randomness of the data and is defined as Zi ) (Yi - Y h )/s, where Y h is the sample mean and s is the sample standard deviation. To ensure the reliability of the runs test procedure, a huge amount of data over several hundreds of elements must be collected.24 In this study, a total of 690 data points, i.e., the Cy3 intensity, from individual tests was used. With a 5% significance level, the Z-score should have an absolute value that is greater than 1.96 to demonstrate nonrandomness. The Z-scores for each of the strains after 10 runs were 2521.7 (AC), 180.12 (AH), 1982.81 (BA), 721.02 (CD), 1.75 (ES), 1440.97 (GA), 1081.53 (NW), 180.12 (PT), 54.52 (RE), 1440.97 (SP), 260.23 (ST), 360.51 (TT), and 900.6 (ZR). Only Enterobacter sp. showed a value below the threshold value of 1.96 but is included if the significance level is increased to 8%. The high Z-scores for the other strains, however, showed that the Cy3 signal intensities generated were not random and that the (24) Sokal, R. R.; Rohlf, F. J. Tests of randomness of nominal data: runs tests. In Biometry; Sokal, R. R., Rohlf, F. J., Eds.; W. H. Freeman: New York, 1995; pp 797-803.

Figure 3. Hybridization of E. coli RFM443 genomic DNA showing the lack of nonspecific binding. (A) Scanned image. (B) Cy3 signal intensity (Y-axis) after hybridization.

Figure 4. Microarray hybridization of a DNA mixture containing DNA from G. amarae, C. denitrificans, and A. hydrophila. The concentration of each genomic DNA was 30 ng, giving a total of 90 ng of genomic DNA labeled with Cy3. (A) Scanned image. (B) Cy3 signal intensity (Y-axis) after hybridization.

unsystematic selection of the probes provides results that are specific. Furthermore, to check for cross-hybridization with other bacteria not involved in the DNA chip fabrication, E. coli was selected as a representative bacterium and its genomic DNA was hybridized onto the chip. Panels A and B in Figure 3 show the scanned image after hybridization and the Cy3 intensity, respectively. Although several spots showed Cy3 signals, most of which can be found in the Enterobacter sp. region, the intensities are not high. For most of the spots, however, no observable Cy3 intensity is seen. This is surprising since, due to highly conserved and similar sequences among bacteria, most DNA chips use specific probes.25 However, the results of this study show that, although it does occur, cross-hybridization is not a serious problem. To generate the random fragments, three different restriction enzyme pairs were used. The fragments generated from the restriction endonuclease treatments are generally assumed to be different from one bacterial genome to the next. For example, (25) Zhou, J.; Thompson, D. K. Curr. Opin. Biotechnol. 2002, 13, 204-207.

RFLP is a technique in which organisms may be differentiated by analysis of patterns derived from cleavage of their DNA.26,27 In other words, it discriminates between bacterial genomes using the bands as a type of “fingerprinting”. Therefore, consider it probable that this random generation of the fragments would lead to the selection of fragments representative of each bacterium. Furthermore, taking into account the size of the probes, the portion of genome represented by all 40 probes would be less than 1%, assuming a genome of 4-5 mega base pairs per bacterial strain. This small selection from the genomic sequence may aid in limiting the number of highly conserved genes or intergenic regions that will be spotted on the microarrays. Discrimination of Specific Bacteria in a Mixed Population. After demonstrating that DNA microarray chips containing randomly generated DNA probes could detect genomic DNA from (26) Papadopoulos, D.; Schneider, D.; Meier-Eiss, J.; Arber, W.; Lenski, R. E.; Blot, M. Proc. Natl. Acad. Sci. U.S.A. 1999, 30, 3807-3812. (27) Jamann, S.; Fernandez, M. P.; Normand P. Mol. Ecol. 1993, 2, 17-26.

Analytical Chemistry, Vol. 77, No. 8, April 15, 2005

2315

Figure 5. Microarray hybridization of genomic DNA extracted from cultures containing both G. amarae and T. thioparus. A total of 60 ng of genomic DNA was labeled with Cy3. (A) Scanned image. (B) Cy3 signal intensity (Y-axis) after hybridization.

Figure 6. Microarray hybridization of genomic DNA extracted from sludge alone or sludge samples containing G. amarae. A total of 390 ng of DNA was labeled with Cy3. (A) Scanned image after hybridization with only sludge DNA. (B) Cy3 signal intensity (Y-axis) within (A) after hybridization. (C) Scanned image after hybridization with sludge DNA. The sludge was first spiked with G. amarae before the DNA was purified. (D) Cy3 signal intensity (Y-axis) within (C) after hybridization.

individual bacterial species, it was of interest to characterize this chip’s ability to detect the DNA from multiple bacterial strains present in mixed samples. Therefore, the genomic DNA from three different bacterial strains was isolated mixed, labeled, and hybridized. Therefore, if the chip is capable of simultaneously discriminating a specific bacterium among different bacterial genomes when present within a single sample, three clear lines, 2316 Analytical Chemistry, Vol. 77, No. 8, April 15, 2005

representing the Cy3 fluorescence, should be obvious on the scanned image. Figure 4A shows the fluorescence pattern that resulted when 30 ng of genomic DNA each from G. amarae, C. denitrificans, and A. hydrophila was used. Each row corresponding to the specific bacterial species, i.e., rows 1, 6, and 13, showed distinct Cy3 fluorescence signals. Figure 4B shows the Cy3 signal intensity after hybridization for each spot and strain. The most

intense Cy3 fluorescence values were seen with the spots corresponding to G. amarae (GA), C. denitrificans (CD) and A. hydrophila (AH). These results show this chip can identify a bacterium when present in a mixture containing other bacterial strains without a significant degree of cross-hybridization. In this case, for the random labeling, the genomic DNA was already extracted and mixed at a very high concentration and, therefore, does not truly present the results that would be seen with a mixed bacterial population since differences in the genome extraction efficiency for each bacterium would likely exist. Therefore, two bacterial cultures, i.e., 1 mL of 109 colony forming units (cfu) T. thioparus with and 1 mL of 106 cfu G. amarae, were mixed and the total genomic DNA was extracted from this mixed population. A total of 60 ng of genomic DNA was labeled with Cy3 in a 30-µL random priming reaction, and the exact concentration of genomic DNA from each bacterium was not known. Figure 5A shows the hybridization results. The most intense Cy3 fluorescence values were seen in rows 1 and 8, which correspond to the G. amarae (GA) and T. thioparus (TT) target probes, respectively. Although the intensity varied from one spot to the next, the signal ratios generated from the G. amarae and T. thioparus probes were 0.31 ( 0.13 and 0.75 ( 0.21, respectively. To offer further proof of this array’s ability to discriminate and report the presence of a specific bacterium in complex microbial communities, 107 cfu of G. amarae were spiked into 1-mL sludge samples. G. amarae was selected since it is a well-known perpetrator of sludge foaming and bulking when present in sludge at a significant number.28 Initially, 390 ng of DNA purified from pure sludge, i.e., not spiked, was labeled with Cy3 and hybridized with DNA chips (Figure 6A and B) along with 390 ng of Cy5labeled DNA prepared from the genomes of the strains used in the fabrication of the microarray chip. No distinct Cy3 fluorescence intensity was observed. However, when the genomic DNA was extracted from the sludge that was spiked with G. amarae and 390 ng of this DNA was labeled, distinct Cy3 fluorescent signals were seen from the G. amarae target probes (Figure 6C), while the strongest fluorescent signals were seen within the same row (Figure 6D). The average signal ratio for all of the G. amarae target probes was 0.21 ( 0.11. Furthermore, Figure 6D clearly shows that the extent of nonspecific binding was minimal. Based upon these results, DNA microarray chips constructed using principles similar to that used in this study would be useful for discerning the presence of specific genomic DNA, which would (28) Cha, D. K.; Jenkins, D.; Lewis, W. P.; Kido, W. H. Water Environ. Res. 1992, 64, 37-43.

be indicative of a certain bacterial strains or species being present, in mixed samples, such as activated sludge. This work also details several novel points related to DNA chip assays. First, to detect genomic DNA samples, a PCR amplification step appears to be negligible. This also holds promise for detection via the direct labeling of the purified DNA, thus avoiding PCR and its inherent bias while improving the potential for quantification of the DNA through the fluorescent signal. Second, the use of nonsequenced probes at an appropriate number provides for a specific detection. This would lead to a decrease in the timeconsuming and expensive process of sequencing, as is done with other DNA chip applications, especially when addition of new strains to the array is warranted. Finally, random selection protocol used in this study to generate the probes gave quite nonrandom results. Finally, this study has broad applications for the future development of DNA microarray chips, especially those intended for species detection within the mixed samples, such as those taken from the environment or present within bacterial consortiums. CONCLUSIONS A DNA microarray chip using randomly generated genomic fragments for 13 different microorganisms was developed for the purpose of detecting specific bacterial species. The focus of this study was to use random genomic probes, i.e., probes generated without any sequence information, for the simultaneous analysis of specific bacterial strains present in mixed samples. Random genomic fragments from 13 pure bacterial cultures were used to generate the specific target signals on the microarray chips. Tests with both pure and mixed cultures showed that the array was specific in its detection of individual genomes, with a minimal amount of nonspecific or cross-hybridization. Furthermore, consistent and stable signals were generated although chip fabrication involved a completely random selection of the target probes. ACKNOWLEDGMENT This work was supported by KOSEF through the Advanced Environmental Monitoring Research center (ADEMRC) at the Gwangju Institute of Science and Technology (GIST) and in the part of 2001 National Research Laboratory (NRL) program of KISTEP. The authors express their gratitude for this support. Received for review August 31, 2004. Accepted January 17, 2005. AC048703C

Analytical Chemistry, Vol. 77, No. 8, April 15, 2005

2317