• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of aemPermissionsJournals.ASM.orgJournalAEM ArticleJournal InfoAuthorsReviewers
Appl Environ Microbiol. May 2004; 70(5): 3047–3054.
PMCID: PMC404398

Nucleic Acid Amplification Strategies for DNA Microarray-Based Pathogen Detection


DNA microarray-based screening and diagnostic technologies have long promised comprehensive testing capabilities. However, the potential of these powerful tools has been limited by front-end target-specific nucleic acid amplification. Despite the sensitivity and specificity associated with PCR amplification, the inherent bias and limited throughput of this approach constrain the principal benefits of downstream microarray-based applications, especially for pathogen detection. To begin addressing alternative approaches, we investigated four front-end amplification strategies: random primed, isothermal Klenow fragment-based, [var phi]29 DNA polymerase-based, and multiplex PCR. The utility of each amplification strategy was assessed by hybridizing amplicons to microarrays consisting of 70-mer oligonucleotide probes specific for enterohemorrhagic Escherichia coli O157:H7 and by quantitating their sensitivities for the detection of O157:H7 in laboratory and environmental samples. Although nearly identical levels of hybridization specificity were achieved for each method, multiplex PCR was at least 3 orders of magnitude more sensitive than any individual random amplification approach. However, the use of Klenow-plus-Klenow and [var phi]29 polymerase-plus-Klenow tandem random amplification strategies provided better sensitivities than multiplex PCR. In addition, amplification biases among the five genetic loci tested were 2- to 20-fold for the random approaches, in contrast to >4 orders of magnitude for multiplex PCR. The same random amplification strategies were also able to detect all five diagnostic targets in a spiked environmental water sample that contained a 63-fold excess of contaminating DNA. The results presented here underscore the feasibility of using random amplification approaches and begin to systematically address the versatility of these approaches for unbiased pathogen detection from environmental sources.

The ability to accurately detect and identify microorganisms that are capable of causing infectious disease has become increasingly important in environmental surveillance, clinical medicine, and biodefense settings. Currently, the predominant techniques used to identify microbial pathogens rely upon conventional clinical microbiology monitoring approaches that are well established and effective but suffer from a number of considerable drawbacks. Standard culture and susceptibility tests permit pathogen identification and antimicrobial susceptibility profiling, but they are laborious, time-consuming, and expensive and require labile natural products. More importantly, the biochemical and serologic tests that are routinely utilized for pathogen identification type only to the species or serogroup level and do not directly characterize virulence factors. Thus, these assays do not provide any information about the potential pathogenicity or virulence of the organism identified. Conventional techniques also do not lend themselves well to managing large numbers of environmental or clinical samples.

Genotypic identification methods that utilize molecular biology-based techniques offer several potential advantages over conventional microbiological approaches. The direct detection of pathogen-specific DNA or RNA (via PCR or reverse transcription-PCR) addresses the issues of presence and viability without the need for culturing the organism. In this respect, PCR-based methods offer a distinct advantage for the detection of fastidious and noncultivable organisms. Although PCR-based assays are exquisitely sensitive, accurate, relatively rapid, and require minimal amounts of material for identification, they too introduce a new set of drawbacks. As successful identification depends almost entirely on appropriately chosen primer sets, all PCR-based testing requires a priori knowledge pertaining to the identity of the contaminating organism. Additionally, technical challenges make it difficult to simultaneously address the issues of identification, potential pathogenicity, and antimicrobial susceptibility. In an attempt to circumvent these drawbacks for pathogen detection, researchers have developed multiplex PCR assays that amplify up to six unique diagnostic regions for a single organism (11, 14) or identify nine different microorganisms within a single experimental reaction (6). Although they are a marked improvement over traditional PCR-based assays, multiplex assays are still limited in the scope and overall throughput that are now necessary for pathogen identification. As such, there remains a critical need for advanced diagnostic systems that can rapidly screen clinical and environmental samples without bias for the presence of pathogenic organisms. Nucleic acid hybridization to DNA microarrays is an experimental approach that has demonstrated great promise in addressing this need (15).

The advantage of microarray-based detection is that it can combine powerful nucleic acid amplification strategies with the massive screening capability of microarray technology, resulting in a high level of sensitivity, specificity, and throughput capacity. In addition to the previously mentioned caveats, the cost and organizational complexity of performing large numbers of PCRs for downstream microarray applications render this option feasible but unattractive. This limitation has severely reduced the utility of this technique and impeded the continued development of downstream applications.

In an attempt to address the bottleneck created by specific PCR-based amplification techniques, we have investigated the utility of various random DNA amplification methods for use in oligonucleotide microarray applications. For this study, we compared an established multiplex PCR strategy (2, 7), targeting enterohemorrhagic Escherichia coli O157:H7 (EHEC) virulence factor and antigen-encoding genes, with random PCR (rPCR), isothermal Klenow fragment, and [var phi]29 DNA polymerase amplification strategies. Both laboratory and complex environmental samples were used to evaluate specific and random amplification approaches with regard to amplification yield, sensitivity, specificity, and overall suitability for downstream microarray-based pathogen detection. The results of this study indicate that single and tandem random amplification approaches provide an unbiased and highly effective alternative to specific amplification strategies for microarray-based pathogen detection.


Probe design and microarray fabrication.

Seventy-mer oligonucleotide probes targeting the eaeA (intimin adherence protein), fliC (flagellin), rfbE (perosamine synthetase), stx1 (Shiga-like toxin I, β subunit), and stx2 (Shiga-like toxin II, α subunit) genes were designed to incorporate previously published 20- to 26-mer oligonucleotides (2) at their 5′ termini and were extended by 44 to 50 bases of downstream gene sequence. In addition, two sets of negative control probes targeting the ipaH (Shigella and enteroinvasive E. coli invasion plasmid antigen) gene and the adenovirus type 7 hexon gene were also included in the analyses. Probe quality was confirmed with Array Designer 2.02 (Premier Biosoft, Palo Alto, Calif.), and minor adjustments were made to ensure a melting temperature minimum of 73°C. Detailed information regarding the 70-mer probes, multiplex primers, and amplicons can be found in Table Table1.1. The probes were synthesized with a 5′ amino modifier and a 12-carbon spacer (Qiagen Operon, Alameda, Calif.) and were spotted onto 3-aminopropyltriethoxysilane (silanization)-plus-1,4-phenylene diisothiocyanate (cross-linker)-modified glass slides for covalent probe immobilization as previously described (1). Oligonucleotides were imprinted in 100 mM carbonate-bicarbonate buffer (pH 9.0) at a concentration of 50 pmol/μl by use of a Virtek ChipWriter Pro contact printer at Kamtek Inc. (Gaithersburg, Md.). The printed slides were stored desiccated at room temperature.

Primer and probe oligonucleotide sequences

DNA templates, amplification, and preparation.

Purified genomic DNAs of wild-type E. coli K-12 (MG1655) and EHEC O157:H7 (EDL 933) were obtained from the American Type Culture Collection (Manassas, Va.). An additional primary human EHEC isolate, E. coli O157:NM (F8775), from an outbreak in Lorain County, Ohio, was obtained from the Centers for Disease Control and Prevention (Atlanta, Ga.). Unless otherwise indicated, 100 ng of EHEC genomic DNA (~1.6 × 107 genome copies) was used as the starting template material for all amplification reactions.

(i) Multiplex PCR amplification.

Biotinylated multiplex PCR amplification was performed as previously described (2, 7), with minor modifications. Amplification reactions (50 μl) consisted of 1× PCR buffer (Qiagen), 2.5 mM MgCl2, a 200 μM concentration (each) of dATP, dGTP, and dTTP, 20 μM dCTP, 20 μM biotin-14-dCTP (Invitrogen Life Technologies, Carlsbad, Calif.), a 200 nM concentration of each primer, and 5 U of Taq DNA polymerase (Qiagen). Each multiplex reaction was subjected to 35 cycles of amplification under the following conditions: 94°C for 30 s, 59°C for 30 s, and 72°C for 30 s.

(ii) Klenow fragment-based amplification.

Isothermal Klenow DNA polymerase-based amplification with random octamers was performed by using the BioPrime DNA labeling system (Invitrogen). Biotinylated amplicons were generated according to the manufacturer's recommended labeling protocol, with minor modifications (50-μl amplification reactions contained 80 U of Klenow fragment and were incubated at 37°C for 4 h).

(iii) rPCR amplification.

Biotinylated rPCR amplicons were generated by using 2.5× random primer solution and 10× deoxynucleoside triphosphate mix from the Invitrogen BioPrime DNA labeling system in a final reaction volume of 50 μl. The reaction components consisted of 1× PCR buffer (Qiagen), 2.5 mM MgCl2, 1× deoxynucleoside triphosphate mix (containing biotin-14-dCTP), 5 μl of 2.5× random octamers, and 5 U of Taq DNA polymerase (Qiagen). The amplification reactions were performed under the following conditions for 35 cycles: 94°C for 30 s, 26°C for 2 min, and 72°C for 1 min.

(iv) [var phi]29 polymerase-based amplification.

Isothermal [var phi]29 DNA polymerase-based amplification with random hexamers was performed with a TempliPhi 100 amplification kit (Amersham Biosciences Corp., Piscataway, N.J.). Biotinylated amplicons were generated according to the manufacturer's recommended protocol, with minor modifications (14.4-μl amplification reactions contained 0.4 μl of enzyme mix and 3 μl of 350 μM biotin-14-dCTP and were incubated at 30°C for 16 h).

After amplification and prior to real-time PCR and hybridization analyses, all rPCR and [var phi]29 polymerase-based amplification products were first fragmented with a DNase I-containing buffer to reduce the amplicons to a size range of 50 to 450 bp. rPCR mixtures (50 μl) were digested by the addition of 5.5 μl of 5× fragmentation buffer (200 mM Tris-acetate [pH 8.1], 500 mM potassium acetate, 150 mM magnesium acetate) and 0.75 U of DNase I (NEB, Beverly, Mass.) and by incubation for 6 min at 25°C followed by incubation at 95°C for 10 min. [var phi]29 polymerase amplification reaction mixtures (14.4 μl) were brought to a 50-μl volume in distilled H2O (dH2O) and digested by the addition of 5.5 μl of 5× fragmentation buffer and 1.0 U of DNase I and by incubation for 30 s at 25°C followed by incubation at 95°C for 10 min. All digested and undigested amplification products were analyzed by gel electrophoresis in precast 4% NuSieve 3:1 Plus agarose gels containing ethidium bromide (BioWhittaker Inc., Walkersville, Md.).

The individual amplification reactions for all tandem amplification approaches (Klenow fragment plus Klenow fragment [Klenow + Klenow], rPCR plus Klenow fragment [rPCR + Klenow], and [var phi]29 polymerase plus Klenow fragment [[var phi]29 + Klenow]) were performed as described above, with the exceptions that the amplicons from the first step were not biotin labeled and that rPCR and [var phi]29 polymerase-based amplification products were not fragmented prior to the secondary amplification step. After the primary amplification, each sample was purified with a MasterPure DNA purification kit (Epicentre Technologies, Madison, Wis.) and then subjected to a second Klenow fragment-based amplification by which biotinylated amplicons were generated.

For assessments of EHEC target detection from a relevant environmental sample, 2 liters of water was collected from a community beach on the Chesapeake Bay, at Churchton, Md. Two 800-ml aliquots of the environmental sample, one spiked with 108 CFU of the primary human isolate, EHEC O157:NM, and the other acting as a negative control, were processed by use of an UltraClean Water DNA kit (Mo Bio Laboratories, Carlsbad, Calif.). The total purified DNA was eluted in 3 ml of dH2O for each sample. Samples were then further concentrated (10-fold) to achieve a concentration of 3.3 × 105 genome equivalents/ μl and this material was then used as the template DNA for amplification and hybridization experiments. The amount of contaminating background DNA (non-EHEC) was calculated by use of the following optical density at 260 nm (OD260) spectrophotometric ratio: OD260 of environmental negative-control sample/OD260 of 108 purified EHEC genomic DNA copies = 63.

Quantitative PCR.

All amplification products were spin purified prior to use in real-time analyses by use of an UltraClean PCR clean-up kit (Mo Bio). Real-time PCR assays were conducted with an iCycler machine (Bio-Rad Laboratories, Hercules, Calif.) to determine the number of specific target copies produced by each amplification method. For each test sample, a 10-fold serial dilution of the original amplification reaction was compared to synthesized two-step high-performance liquid chromatography-purified 80-mer templates (Qiagen) of known copy numbers. Real-time PCR mixtures consisted of 1× SYBR Green PCR master mix (Applied Biosystems, Foster City, Calif.), an additional 1 mM MgCl2 , 200 nM sense primer, 600 nM antisense primer, and amplified template DNA at various dilutions and were subjected to 35 cycles of 95°C for 30 s, 55°C (eaeA and fliC), 53°C (stx1), or 51°C (rfbE and stx2) for 30 s, and 72°C for 30 s. Threshold crossings (CT) for each test target were compared to the plotted standard curve. Target sample copy numbers were extrapolated from the CT values of test samples that fell within the range of the linear trend line from the plotted standards. Postamplification gel analyses confirmed an absence of spurious amplicons due to mispriming. Melt curve analyses indicated significant primer dimerization for only the fliC primer pair (CT = 22), and thus only dilutions that exhibited CT values of <17 cycles (at which amplification of the specific product was not inhibited by the primer dimers) were included in the calculations for the fliC standard curve. In addition to their use for determining the number of specific amplicons produced by each method, the quantitative PCR data were also used to determine amplification bias. Bias was calculated for each amplification strategy by the following equation: gene target with largest number of amplicons/gene target with smallest number of amplicons = amplification bias. For example, number of amplicons for multiplex eaeA/number of amplicons for multiplex stx2 = 58,594.

Microarray hybridization and processing.

Oligonucleotide-printed slides were blocked with a 3% bovine serum albumin-casein solution (pH 7.4) for 15 min at room temperature, rinsed with dH2O, air dried, and outfitted with Secure-Seal hybridization chambers (Schleicher & Schuell, Keene, N.H.) immediately prior to hybridization. Hybridization buffer was added to the biotinylated target amplicons (entire amplification reaction) to achieve a total volume of 150 μl, with final concentrations of 4× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate), 0.2% sodium dodecyl sulfate, and 30% formamide (8). Target hybridization samples were denatured for 5 min at 98°C and then immediately applied to the microarray. Hybridization was performed for 16 h at 45°C in a GeneChip hybridization oven (model 640; Affymetrix, Santa Clara, Calif.) rotating at 35 rpm. The results were verified by replicate experiments in which independently amplified products were hybridized to identical arrays, with each array receiving a unique amplification mixture. After hybridization, the slides were washed twice with a 4× SSC-0.2% sodium dodecyl sulfate buffer for 5 min at 68°C and once with a 1× SSC buffer for 2 min at room temperature. Hybridization events were detected by the sequential addition of a 1:500 dilution of Cy5-conjugated monoclonal mouse anti-biotin immunoglobulin G (Jackson ImmunoResearch, West Grove, Pa.) and a 1:500 dilution of Cy5-conjugated goat anti-mouse immunoglobulin G (Jackson ImmunoResearch), with an incubation with each antibody for 15 min at room temperature. The slides were then subjected to three 1× SSC rinses for 3 min each at room temperature and a final dH2O rinse. Slides were dried under a nitrogen stream and subsequently scanned with a GSI Lumonics ScanArray Lite confocal laser scanning system (Perkin-Elmer, Torrance, Calif.). Quantitative comparisons based on fluorescence intensities were made with the QuantArray analysis software package. Unless otherwise noted, the microarray images were captured at laser power 100/PMT gain 80 and the signal from each microarray element was considered positive only when its fluorescence intensity was three times the neighboring background or more.


Optimization of random and multiplex EHEC target amplification.

EHEC genomic DNA was amplified by multiplex PCR and three random amplification strategies, namely random primed PCRs using Taq polymerase, Klenow fragment, and [var phi]29 DNA polymerase isothermal amplification (Fig. (Fig.1).1). Two important factors that influence hybridization efficiency, the selection of a detection label and the target product size, were evaluated and optimized for all amplification strategies. Initially, we chose to simultaneously amplify and incorporate the fluorescent reporter molecule Cy5-dCTP for the detection of hybridization to reduce the time and number of steps required per assay. However, preliminary studies showed low amplification yields and poor incorporation of the fluorescent label (2, 3), particularly with the [var phi]29 DNA polymerase (data not shown). In comparison, the incorporation of biotin-14-dCTP did not significantly impact the amplification yield for any of the methods surveyed and was chosen as the method for labeling amplicons in this study. The product sizes for each amplification strategy were then examined by gel electrophoresis (Fig. (Fig.2).2). As expected, multiplex PCR (lane 2) resulted in the generation of five distinct bands (portions of the eaeA, fliC, rfbE, stx1, and stx2 genes) that corresponded with those in previous studies (2, 5, 7, 9). The majority of both the rPCR (lane 3) and [var phi]29 DNA polymerase (lane 5) amplification products were too large to migrate into the gel and remained in the wells. A notable exception was the single high-molecular-weight [var phi]29 DNA polymerase amplification product (lane 5), which represented tightly coiled concatemers (4). Klenow fragment amplification (lane 4) resulted in some high-molecular-weight product, but the vast majority was found in an electrophoretic smear between 50 and 450 bp. No unamplified template DNA (lane 1) was visibly detectable.

FIG. 1.
Schematic depiction of the methods utilized for specific and random amplification of E. coli O157:H7 genomic DNA for subsequent hybridization to oligonucleotide microarrays.
FIG. 2.
Electrophoretic profiles of specific and random amplification products. Lane 1, unamplified EHEC genomic DNA; lane 2, multiplex PCR; lane 3, random PCR; lane 4, Klenow fragment-based isothermal amplification; lane 5, [var phi]29 polymerase-based isothermal ...

Previous work has suggested that large amplification products hybridize poorly to immobilized probes on two-dimensional surfaces due to spatial and steric constraints (10, 12). Thus, the rPCR and [var phi]29 polymerase-based amplification products were digested with a DNase I fragmentation buffer to achieve amplicon sizes that were comparable to the Klenow fragment and multiplex amplicons prior to hybridization. The products of the four amplification strategies were used as templates for real-time PCR quantitation to determine the levels of amplification achieved and the numbers of relevant target amplicons applied to the microarrays (postfragmentation) (Table (Table2).2). Due to the use of specific primers and the exponential amplification nature of PCR, the multiplex PCRs produced the largest target copy increase (average, 209,628-fold increase), whereas each of the random amplification strategies, with amplification based on the Klenow fragment, rPCR, or [var phi]29 polymerase, resulted in average increases of 70-fold, 52-fold, and 14-fold, respectively. Although multiplex PCR resulted in the largest fold increase in specific target sequences, this method also demonstrated the most amplification bias (58,594-fold). In contrast, each of the random amplification approaches, Klenow fragment (8-fold), rPCR (20-fold), or [var phi]29 polymerase (2-fold), demonstrated markedly less bias and provided a more uniform genetic locus representation. Quantitative assessments performed with full-length (unfragmented) rPCR amplicons showed a 138- to 300-fold increase in copy number for all five targets (data not shown). Although the quantitation of the unfragmented rPCR amplicons demonstrated a larger fold increase in copy number than the fragmented rPCR amplicons, an evaluation of unfragmented versus fragmented rPCR products by microarray hybridization demonstrated that fragmentation was necessary for efficient hybridization (data not shown).

Real-time PCR quantitation of specific target copy number increase as a result of various amplification methodsa

Hybridization specificity.

Random amplification approaches inherently generate nontargeted amplicons. For this particular study, in which entire random amplification reactions with microgram quantities of total amplified DNA were applied to 70-mer microarrays, hybridization specificity was of critical importance. As is evident in the representative microarrays presented in Fig. Fig.3,3, the combination of long oligonucleotide probes, formamide-based hybridization buffers, and high-temperature posthybridization washes provided an acceptable means to control hybridization specificity. When EHEC O157:H7 genomic DNA was used as the template for each amplification approach and subsequent hybridization (Fig. (Fig.3,3, left panels), both unique 70-mer probes/gene target provided positive fluorescent hybridization signals. It is important that despite the overwhelming abundance of nonspecific amplified DNA, we did not observe any false-positive hybridization reactions based on the nonreactivity (or signal above background) of the amplified targets to the negative-control oligonucleotide probes targeting Shigella (F1 and F2) and adenovirus (G1 and G2). The success of this approach was further supported by the observation that when the nonpathogenic wild-type strain, E. coli K-12, was used as the template for each amplification approach (Fig. (Fig.3,3, right panels), only the fliC probes demonstrated positive fluorescent hybridization signals. This result was to be expected, as E. coli K-12 not only contains the fliC gene, but as BLAST searches revealed, also contains target sequences complementary to the immobilized fliC1-oli70 and fliC2-oli70 probes that have 97 and 93% match identities, respectively. Finally, there appeared to be little correlation between the number of specific target copies amplified (as determined by real-time PCR) and the hybridized microarray element fluorescence intensities within a particular amplification scheme. This fact suggested that probe design, the formation of target or probe secondary structures, the location of probe complementary sequences within a particular amplicon, and amplicon length probably all played roles in altering the results expected based on specific target copy number alone.

FIG. 3.
Hybridization of specific and random amplification products to low-density EHEC oligonucleotide microarrays. (Top) Amplification products from both E. coli O157:H7 and E. coli K-12 genomic DNA templates were hybridized to a low-density oligonucleotide ...

Hybridization sensitivity.

Our successful demonstration of hybridization specificity among the four amplification methods led us to address the second critical parameter of microarray-based approaches, assay sensitivity. Testing of a range of copy numbers spanning 8 orders of magnitude revealed that multiplex PCR provided the most sensitive detection of specific target DNA, as positive hybridization reactions for all probes were seen at the level of 100 genomic copies (Table (Table3).3). Not surprisingly, the random amplification methods were markedly less sensitive, demonstrating detection levels at 105, 106, and 107 genomic copies for the Klenow fragment-, rPCR-, and [var phi]29-based amplification methods, respectively. However, when each random amplification approach was sequentially combined with a secondary Klenow fragment-based amplification, the detection levels became considerably more sensitive. For the Klenow + Klenow tandem random amplification approach, the detection sensitivity was 102 genomic copies, which was a 1,000-fold increase in the detection sensitivity of a single Klenow fragment-based amplification. In addition, both probes for fliC and stx1 were detectable when only a single EHEC genome was used as the amplification template. Both rPCR + Klenow and [var phi]29 + Klenow methods showed similarly enhanced sensitivities. The [var phi]29 + Klenow method also allowed for the detection of both eaeA, fliC, and rfbE probes when only a single EHEC genome was used as the amplification template. It was surprising that although a single [var phi]29 polymerase-based amplification resulted in the least sensitive detection method (10-probe detection at 107 genomic copies), the use of the [var phi]29 polymerase method in tandem proved to be an excellent strategy, as it enhanced the sensitivity of detection for all probes 100,000-fold. In addition, in a manner identical to the single amplification strategy product quantitation (Table (Table2),2), real-time PCR quantitation was also performed with the tandem random amplification reactions (data not shown) to determine the levels of amplification bias. In comparison to what was seen with the single random amplification approaches, quantitative PCR calculations revealed that the tandem random amplification approaches (Klenow + Klenow [129-fold], [var phi]29 + Klenow [104-fold], and rPCR + Klenow [122-fold]) introduced more amplification bias. Despite the increased amplification bias, each of the three tandem random amplification methods still demonstrated >450 times less bias than a single multiplex reaction. Thus, the results clearly demonstrated that two of the three tandem random amplification approaches (Klenow + Klenow and [var phi]29 + Klenow) provided detection levels that were as sensitive as or more sensitive than the specifically primed multiplex PCR amplification method while introducing markedly less amplification bias.

Sensitivity of microarray-based detection of EHEC targets postamplification

To assess the utility of this combined platform to detect pathogen DNAs in a complex environmental sample, we examined the same seven amplification strategies using a spiked environmental water sample from the Chesapeake Bay in Maryland (Table (Table4).4). As expected under these experimental conditions, amplification and microarray-based detection sensitivities decreased compared to those for assays done with purified isogenic genomic DNA. Compared to the detection sensitivity obtained with purified genomic DNA, the detection sensitivities of the multiplex PCR, Klenow fragment, and [var phi]29 polymerase-based amplification methods with the environmental sample had a 10-fold decrease (less sensitive), whereas those of the tandem random amplification approaches had ~10 to 1,000-fold decreases. The single rPCR amplification method proved to be the exception, as the level of detection sensitivity remained unchanged (106 genomic copies) regardless of the template DNA source or nonspecific genetic background. Although the specific and random amplification approaches proved less sensitive in the spiked environmental matrix, the results indicated that the tandem random amplification approaches could provide enough target DNA for microarray-based pathogen detection and genotyping when 104 to 105 organisms were present.

Sensitivity of microarray-based detection of EHEC targets from a complex samplea


Pathogen detection and typing methods have become increasingly reliant upon molecular characterization. Current methods, such as pulse-field gel electrophoresis, amplified restriction fragment length polymorphism analysis, repetitive element-based PCR, ribotyping, and PCR, are all effective means of detecting and typing organisms. However, each method is limited by (i) a dependence upon prior selective enrichment or the use of pure organism cultures, (ii) an inability to provide direct pathogenicity or antimicrobial susceptibility profiles, and/or (iii) needing requisite a priori knowledge of the organism to be identified. Individually or combined, these limitations increase the time required for pathogen identification, decrease the number of total samples that can be processed simultaneously, and restrict the information content generated per assay. With this study, we addressed these limitations by examining the utility of several random amplification approaches coupled to oligonucleotide microarray screening for the purpose of unbiased pathogen detection and characterization.

Although random amplification has been used successfully for a variety of applications, such as genomic analysis, target-specific multiplex PCR amplification has been the standard method used for pathogen detection assays. The obvious advantage of multiplex PCR is that the amplification of specific target products enhances both the sensitivity and specificity of detection (Table (Table2).2). Although high levels of amplification are desirable for PCR assays, this feature becomes less critical for assays that combine PCR and microarray screening because the vast majority of amplicons generated from multiplex PCR remain unbound as probe availability on the surface of the microarray becomes limiting. Thus, for microarray-based assays, high levels of amplification do not necessarily translate into enormous differences in detection capability (Fig. (Fig.3).3). Nevertheless, the sensitivities of microarray-based assays that utilize single random amplification approaches were, as expected, lower than that observed for targeted multiplex amplification of both laboratory and environmental samples (Tables (Tables33 and and44).

Of the three single random amplification methods tested, front-end Klenow fragment-based amplification yielded the greatest microarray sensitivity for both laboratory (105 genomic copies) and environmental (106 genomic copies) samples with the least amount of manipulation. Klenow fragment-based amplification is an isothermal strand displacement technique that generates relatively small (50 to 450 bp) amplicons (Fig. (Fig.2,2, lane 4). When used in conjunction with microarray hybridization, this feature proves advantageous, as targets similar in size to the immobilized 70-mer oligonucleotide probes produce less steric hindrance and better hybridization kinetics (8, 12). Thus, when target microbial genomic DNA is not a limiting resource (such as when an organism has been pure cultured or enriched), a single Klenow fragment-based random amplification would be an efficient and unbiased front-end amplification strategy (13). Since the sensitivity of a single Klenow fragment-based amplification was 1,000-fold less (for the detection of all 10 probes) than that of targeted multiplex amplification, we sought to enhance the sensitivity by performing a second Klenow fragment amplification (Klenow + Klenow). The relatively small amplicon size generated by Klenow fragment isothermal amplification obviates the need for product fragmentation, and thus this approach is well suited as a second step in a tandem amplification strategy. Klenow + Klenow random amplification showed a marked improvement in sensitivity for both laboratory (Table (Table3)3) and environmental (Table (Table4)4) samples. An enhancement of the detection sensitivity was also seen when the addition of a secondary Klenow fragment amplification step was added to the rPCR and [var phi]29 polymerase-based amplification strategies.

Although the sensitivities of the Klenow + Klenow and [var phi]29 + Klenow tandem amplification approaches were equal to or greater than that of multiplex PCR with a purified genomic DNA template, there are several reasons that would explain why the single random amplification strategies proved less sensitive. Firstly, as a group, the five target genes examined in this study constitute a small fraction (0.12%) of the EHEC genome. Furthermore, if only the exact 70-mer probe complementary sequences are considered, then 0.01% of each EHEC genome is targeted in each assay. Thus, it is probable that the vast majority of amplicons produced by random amplification do not contain the target sequences of interest. The ratio of nonspecific to specific target amplicons has the potential of reducing hybridization specificity and detection sensitivity. However, unlike the hybridization specificity (Fig. (Fig.3),3), the detection sensitivity is markedly affected by the presence of non-EHEC DNA, as was demonstrated by the comparison of purified EHEC genomic DNA and a Chesapeake Bay water sample (Table (Table33 versus Table Table4).4). Secondly, the rPCR and [var phi]29 polymerase-based amplification strategies generated large amplicons that required size fragmentation prior to hybridization. Although size fragmentation resulted in a smaller total nucleic acid yield, it was necessary for hybridization, as attempts with the full-length rPCR and [var phi]29 polymerase-based amplicons resulted in poor (or absent) hybridization to complementary immobilized probes (10; also data not shown). In an attempt to circumvent this loss in yield, we chose secondary Klenow fragment-based amplification exclusively for the tandem approaches to reduce amplicon sizes without the need for fragmentation. Thirdly, for this study, the 70-mer probes were optimized to hybridize to the target amplicons generated by multiplex PCR amplification. The probes were designed to hybridize to the target amplicons' termini, thus reducing steric and spatial constraints at the probe-target interface. Although fragmentation of the amplicons produced sizes that were comparable to those generated by multiplex amplification (<625 bp), there was no guarantee that the probe-complementary regions present in the target amplicons were in positions for efficient hybridization. For example, after fragmentation, the location of a probe-complementary region may be susceptible to folding or secondary structural effects, resulting in steric constraints that ultimately reduce or eliminate the hybridization efficiency (12).

Despite these drawbacks, random amplification approaches also presented some clear advantages when compared to specific amplification methods for microarray applications. The most notable advantage, which is completely lacking in multiplex approaches, is the potential for random amplification-based microarray assays to detect unknown or unanticipated organisms. All of the random amplification strategies tested provide for a relatively unbiased and highly parallel analysis that is only limited in throughput by the number of unique microarray elements. This unbiased approach has the added benefit of reducing overall organizational complexity and cost, as the detection of species variants and emerging pathogens would not require novel selection and optimization or reoptimization of appropriate specific primers and assay conditions for molecular characterization.

Additionally, our results indicate that the random amplification approaches also provide a more uniform genetic locus representation (or limited target gene amplification bias) than does multiplex PCR (Table (Table2).2). The circular EHEC O157:H7 chromosome coordinates of each single-copy target gene demonstrated a sampling from physically distant loci on the chromosome, suggesting that all three random amplification approaches were more effective at uniform whole-genome amplification than the specific amplification method. Large amplification bias skews for PCR-based methods can only be addressed by optimizing the parameters of each individual reaction. Multiplex PCR amplification reactions that result in large amplification skews, such as the example provided in this study, increase the risk of false-negative results, especially at lower concentrations of template DNA, and consequently decrease the confidence of a correct identification. Thus, methods that amplify with minimal bias provide an effective and more favorable alternative for the purpose of genotyping pathogenic microorganisms by using DNA microarrays.

The selective pressures acting upon bacterial pathogens often necessitate their ability to mutate and acquire foreign genetic elements containing antimicrobial resistance-, immune evasion-, novel metabolic pathway-, and toxin-encoding genes. This level of genomic plasticity requires us to have the ability to simultaneously detect and genotype pathogenic organisms to elucidate entire virulence and antimicrobial susceptibility profiles in an unbiased and highly parallel manner. In this study, we have demonstrated that the use of random amplification methods combined with 70-mer oligonucleotide microarrays may provide a sensitive and specific means to fulfill these requirements. This combined platform permits the use of random primers to amplify the total DNA content of a sample while allowing the specificity of hybridization to detect target sequences in complex genetic backgrounds. Furthermore, the use of multiple unique probes per target gene elevates the confidence of identification and discrimination among related species while reducing the chances of improper identification as a result of genomic plasticity. Thus, combining specificity, sensitivity, and the ability to test in an unbiased manner provides a more versatile and powerful approach for pathogen detection and typing which cannot be matched by any single biochemical or serological test for information content generated per assay. This experimental approach may have an immediate impact on the environmental surveillance of pathogenic microorganisms, the detection of novel genetic variants, and microorganism ecology.


We thank Paul Charles and Jermaine Rangassammy for their preparation of the surface-modified slides, Kamtek Inc. for microarray printing support, Cheryl Bopp for the E. coli O157:NM primary isolate, and Ellen Goldman and James Delehanty for critical evaluation of the manuscript.

G.J.V. is a National Research Council postdoctoral fellow. This work was supported by the Office of Naval Research.

The opinions and assertions contained herein are those of the authors and are not to be construed as those of the U.S. Navy or military service at large.


1. Charles, P. T., G. J. Vora, J. D. Andreadis, A. J. Fortney, C. E. Meador, C. S. Dulcey, and D. A. Stenger. 2003. Fabrication and surface characterization of DNA microarrays using amine- and thiol-terminated oligonucleotide probes. Langmuir 19:1586-1591.
2. Chizhikov, V., A. Rasooly, K. Chumakov, and D. D. Levy. 2001. Microarray analysis of microbial virulence factors. Appl. Environ. Microbiol. 67:3258-3263. [PMC free article] [PubMed]
3. Chizhikov, V., M. Wagner, A. Ivshina, Y. Hoshino, A. Z. Kapikian, and K. Chumakov. 2002. Detection and genotyping of human group A rotaviruses by oligonucleotide microarray hybridization. J. Clin. Microbiol. 40:2398-2407. [PMC free article] [PubMed]
4. Detter, J. C., J. M. Jett, S. M. Lucas, E. Dalin, A. R. Arellano, M. Wang, J. R. Nelson, J. Chapman, Y. Lou, D. Rokhsar, T. L. Hawkins, and P. M. Richardson. 2002. Isothermal strand-displacement amplification applications for high-throughput genomics. Genomics 80:691-698. [PubMed]
5. Gannon, V. P., S. D'Souza, T. Graham, R. K. King, K. Rahn, and S. Read. 1997. Use of the flagellar H7 gene as a target in multiplex PCR assays and improved specificity in identification of enterohemorrhagic Escherichia coli strains. J. Clin. Microbiol. 35:656-662. [PMC free article] [PubMed]
6. Grondahl, B., W. Puppe, A. Hoppe, I. Kuhne, J. A. Weigl, and H. J. Schmitt. 1999. Rapid identification of nine microorganisms causing acute respiratory tract infections by single-tube multiplex reverse transcription-PCR: feasibility study. J. Clin. Microbiol. 37:1-7. [PMC free article] [PubMed]
7. Hu, Y., Q. Zhang, and J. C. Meitzler. 1999. Rapid and sensitive detection of Escherichia coli O157:H7 in bovine faeces by a multiplex PCR. J. Appl. Microbiol. 87:867-876. [PubMed]
8. Hughes, T. R., M. Mao, A. R. Jones, J. Burchard, M. J. Marton, K. W. Shannon, S. M. Lefkowitz, M. Ziman, J. M. Schelter, M. R. Meyer, S. Kobayashi, C. Davis, H. Dai, Y. D. He, S. B. Stephaniants, G. Cavet, W. L. Walker, A. West, E. Coffey, D. D. Shoemaker, R. Stoughton, A. P. Blanchard, S. H. Friend, and P. S. Linsley. 2001. Expression profiling using microarrays fabricated by an ink-jet oligonucleotide synthesizer. Nat. Biotechnol. 19:342-347. [PubMed]
9. Meng, J., S. Zhao, M. P. Doyle, S. E. Mitchell, and S. Kresovich. 1996. Polymerase chain reaction for detecting Escherichia coli O157:H7. Int. J. Food Microbiol. 32:103-113. [PubMed]
10. Shchepinov, M. S., S. C. Case-Green, and E. M. Southern. 1997. Steric factors influencing hybridisation of nucleic acids to oligonucleotide arrays. Nucleic Acids Res. 25:1155-1161. [PMC free article] [PubMed]
11. Singh, D. V., S. R. Isac, and R. R. Colwell. 2002. Development of a hexaplex PCR assay for rapid detection of virulence and regulatory genes in Vibrio cholerae and Vibrio mimicus. J. Clin. Microbiol. 40:4321-4324. [PMC free article] [PubMed]
12. Southern, E., K. Mir, and M. Shchepinov. 1999. Molecular interactions on microarrays. Nat. Genet. 21:5-9. [PubMed]
13. van Ijperen, C., P. Kuhnert, J. Frey, and J. P. Clewley. 2002. Virulence typing of Escherichia coli using microarrays. Mol. Cell. Probes 16:371-378. [PubMed]
14. Volokhov, D., A. Rasooly, K. Chumakov, and V. Chizhikov. 2002. Identification of Listeria species by microarray-based assay. J. Clin. Microbiol. 40:4720-4728. [PMC free article] [PubMed]
15. Wang, D., L. Coscoy, M. Zylberberg, P. C. Avila, H. A. Boushey, D. Ganem, and J. L. DeRisi. 2002. Microarray-based detection and genotyping of viral pathogens. Proc. Natl. Acad. Sci. USA. 12:15687-15692. [PMC free article] [PubMed]

Articles from Applied and Environmental Microbiology are provided here courtesy of American Society for Microbiology (ASM)
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...