Design and evaluation of oligonucleotide-microarray method for the detection of human intestinal bacteria in fecal samples


*Corresponding author. Tel.: +1 (870) 543-7084; Fax: +1 (870) 543-7307, E-mail address:


An oligonucleotide-microarray method was developed for the detection of intestinal bacteria in fecal samples collected from human subjects. The 16S rDNA sequences of 20 predominant human intestinal bacterial species were used to design oligonucleotide probes. Three 40-mer oligonucleotides specific for each bacterial species (total 60 probes) were synthesized and applied to glass slides. Cyanine5 (CY5)-labeled 16S rDNAs were amplified by polymerase chain reaction (PCR) from human fecal samples or bacterial DNA using two universal primers and were hybridized to the oligo-microarray. The 20 intestinal bacterial species tested were Bacteroides thetaiotaomicron, Bacteroides vulgatus, Bacteroides fragilis, Bacteroides distasonis, Clostridium clostridiiforme, Clostridium leptum, Fusobacterium prausnitzii, Peptostreptococcus productus, Ruminococcus obeum, Ruminococcus bromii, Ruminococcus callidus, Ruminococcus albus, Bifidobacterium longum, Bifidobacterium adolescentis, Bifidobacterium infantis, Eubacterium biforme, Eubacterium aerofaciens, Lactobacillus acidophilus, Escherichia coli, and Enterococcus faecium. The two universal primers were able to amplify full size 16S rDNA from all of the 20 bacterial species tested. The hybridization results indicated that the oligo-microarray method developed in this study is a reliable method for the detection of predominant human intestinal bacteria in the fecal samples.


The human gastrointestinal tract is colonized by a diverse microbial community of at least 400 bacterial species, with 30–40 species accounting for 99% of the total microflora [1–3]. Interest in the intestinal microflora has been stimulated by concern over the safety implications of antibiotic resistant bacteria in foods [4,5], fecal contamination of foods[6], effects of diets[7], food additives[7] and veterinary drug residues [8,9] on the intestinal ecosystem, and the use of probiotics in the treatment and prevention of gastrointestinal disorders[10]. Indigenous intestinal microfloras play important roles because they aid in the digestion of food, metabolize endogenous and exogenous compounds, produce vitamins and other essential nutrients, and help prevent pathogens from colonizing the gastrointestinal tract [11–14].

Traditionally, the population of anaerobic bacteria in the human gastrointestinal tract was characterized by microscopic, biochemical, physiological, and selective culture plating methods of fecal samples from human subjects [1,3]. In recent years, a variety of molecular techniques have been used to analyze the bacterial community in human fecal samples [10,15–19]. Molecular analysis can detect perturbations in the human intestinal microflora in a rapid and precise manner [6,20]. Previously, we developed polymerase chain reaction (PCR) methods for the detection and quantitation of predominant anaerobic bacteria in human and animal feces [21–23]. Although these investigations provided useful data, one of the limitations of our methodology was that we had to test for each bacterial species separately. Microarray technology is a powerful tool that can be used for simultaneous detection of thousands of genes or target DNA sequences on one glass slide [24,25]. Most studies using microarray methods are on gene expression. However, microarray method can also be used for the detection of bacteria and DNA-based typing of specific pathogenic bacterial strains [25,26]. The purpose of this study was to design and evaluate a microarray method for the detection of predominant bacterial species from human fecal samples.

2Materials and methods

2.1Source of bacterial strains and culture conditions

Reference strains for 20 predominant human intestinal bacteria (listed in Fig. 1) were obtained from the American Type Culture Collection (ATCC). Anaerobic bacteria were cultured at 37°C either in prereduced anaerobically sterilized (PRAS) brain heart infusion broth supplemented with vitamin K and hemin (BHI; Carr-Scarborough Microbiologicals, Stone Mountain, GA, USA), inoculated under an oxygen-free cannula using 85% nitrogen, 10% hydrogen and 5% carbon dioxide, or on PRAS brucella blood agar plates supplemented with vitamin K and hemin (Anaerobic Systems, San Jose, CA, USA). The cultures were inoculated in an anaerobic gas chamber (Coy Laboratory Products, Inc., Ann Arbor, MI, USA) under 85% N2, 10% H2, and 5% CO2. Culture dilutions were made under an oxygen-free gas cannula in dilution blanks (Anaerobe Laboratory Manual, Virginia Polytechnic Institute and State University, Blacksburg, VA, USA).

Figure 1.

PCR amplification of 16S rDNA by primers Amp-F and Amp-R from the DNA of 20 bacterial species. Lanes 1–20 are PCR products from: Bacteroides thetaiotaomicron ATCC29148 (BT), B. vulgatus ATCC8482 (BV), B. fragilis ATCC23745 (BF), B. distasonis ATCC8503 (BD), Clostridium clostridiiforme ATCC29084 (CC), C. leptum ATCC29065 (CL), Fusobacterium prausnitzii ATCC27768 (FPR), Peptostreptococcus productus ATCC27340 (PSP), Ruminococcus obeum ATCC29174 (ROB), R. bromii ATCC27255 (RBR), R. callidus ATCC27760 (RCA), R. albus ATCC27210 (RAL), Bifidobacterium longum ATCC15707 (BIL), B. adolescentis ATCC15703 (BIA), B. infantis ATCC15697 (BIN), Eubacterium biforme ATCC27806 (EBI), E. aerofaciens ATCC25986 (EAE), Lactobacillus acidophilus ATCC332 (LAA), Escherichia coli ATCC25922 (ECO), Enterococcus faecium ATCC19434 (EFA), respectively.

Fecal samples were obtained from three healthy adult volunteers (ages 30–45 years old) and a person with long-term diarrhea. The fecal DNA and bacterial DNA preparations used in this investigation are described in Section 2.4.

2.2Design of oligonucleotide probes and preparation of oligonucleotide-microarray slides

Three 40-mer oligonucleotides specific for each bacterial species (total 60 probes) were designed based on comparison of the 16S rDNA sequences in the GenBank data library. First consideration was to choose the regions specific to each species, then consider the hairpins and dimers of the oligonucleotides. The 60 oligonucleotide sequences are listed in Table 1. The oligos were ordered from MWG-Biotech (High Point, NC, USA) and used for making microarray by the custom service of the company. Briefly, the oligos were 5′-amino-modified and suspended in 20% dimethyl sulfoxide (DMSO), 0.05% Tween 20 at a final concentration of 50 μM. Using the Affymetrix Arrayer, oligos were printed on aldehyde slides (CEL Associates, Houston, TX, USA).

Table 1.  The sequences and sources of the 60 oligo probes
  1. Each bacterial species has three probes and the 20 bacterial species listed here are the same order as listed in Fig. 1. The numbers at the end show the probe start locations at the 16S rDNA sequences, and if the numbers are smaller than 750 (half of 1.5 kb), the reverse sequences were used for the probes. This enabled the probes to hybridize to the location nearest the 3′-end of the target DNA.

Probe no.Probe nameGenBank no. (GI)Probe sequenceLocation
2BT-2 cattaggcagttggtgaggtaacggctcaccaaaccttcg235
3BT-3 catttgaatatattggaaacagtatagccgcaaggcaaatg984
5BV-2 agatgaattacggtgaaagccgtaagccgcaaggcatctg1001
6BV-3 tgttgtcagttactaacaggttatgctgaggactctgaca1113
8BF-2 tcgtaaacttcttttatataagaataaagtgcagtatgta443
9BF-3 gaaggcagctagcgggtgaccgtatgctaatcccaaaat1249
11BD-2 ttcggaccgaggtggaaacaccttttctagcaatagccgt1004
12BD-3 aggccacctggcgacagggagcgaatccccaaaccacg1260
14CC-2 gaagcaagtctgaagtgaaaacccagggctcaaccctggc581
15CC-3 cccctgacggccggtaacgcggccnttcttcgggacaggg989
17CL-2 tctatgggcttaacccataaactgcgcttgaaactgtctt591
18CL-3 caaagccgcgaggtggagcaaaaccctaaaagcagtcc1238
20FPR-2 cctgcgacgcgcatagaaatatgtgtttcttcgggaccag969
21FPR-3 gagaagcaagaccgcgaggcgagcaaaactcagaaacttcg1215
23PSP-2 ggaagagcaagtctgatgtgaaaggctggggcttaacccca547
24PSP-3 cctctgaccgtc ccgtaacggg ganttccctt cggggcaga958
26ROB-2 gtcccttaaccggatctttccttcgggacaggggagacag959
27ROB-3 cctatccccagtagccagcagtccggctgggcactctgag1065
29RBR-2 cttcttttattaaggacgaaaaatgacggtacttaatga412
30RBR-3 taatacccgaagtcagtagtccaacctcgtgaggacgctg1375
32RCA-2 tgaagaggacgataatgacggtactcttttagaaagctc401
33RCA-3 aaagccggtcgtctaaccttcgggaggatgccgtctaagg1362
35RAL-2 agagggaagcaaaacagtgatgtggagcaaaacccta1182
36RAL-3 cctgtgttctaaccgcaaggaggaagcagtcgaaggtgg1367
38BIL-2 ggcttgacatgttcccgacgatcccagagatggggtttcc959
39BIL-3 agccggtggcctaaccccttgcgggagggagccgtctaatg1407
41BIA-2 ctccagttggatgcatgtccttctgggaaagattctatcggt156
42BIA-3 caacgggatgcgacctcgtgagggggagcggatccctt1207
44BIN-2 ggcttgacatgtgccggatcgccgtggagacacggtttcc967
45BIN-3 ggtagacacccgaagccggttggcccgacccttgttggg1404
47EBI-2 gtgatatgttactaacattgagttgaggactcatatcaga1115
48EBI-3 agagcggcaagcctgtgaaggcaagcgaatctcataaagga1248
50EAE-2 cggcaggccgggggtcgaagcggggggctcaaccccccgaa553
51EAE-3 atgggtgaagcgggggagacccgtggccgagaggagcccata956
53LAA-2 gcaatccgtagagatacggagttcccttcggggacacta1028
54LAA-3 acagtacaacgaggagcaagcctgcgaaggcaagcg1267
56ECO-2 ggaagggagtaaagttaatacctttgctcattgacgttac429
57ECO-3 catccacggaagttttcagagatgagaatgtgccttcgg975
59EFA-2 gaagaacaaggatgagagtaactgttcatcccttgacgg436
60EFA-3 gaagtacaacgagttgcgaagtcgcgaggctaagctaat1233

2.3PCR amplification of cyanine5 (CY5)-labeled 16S rDNA

Twenty five μl of PCR mixture was made by combining 15.6 μl of water, 2.5 μl of 10×BSA buffer (1 ml 10×buffer is composed of 0.5 ml 1 M Tris–HCl, pH 8.5, 0.2 ml 1 M KCl, 30 μl of 1 M MgCl2, 0.27 ml of water, BSA 5 mg), 2.3 μl of dNTP (2.5 mM each of dATP, dTTP, dGTP, but 1.7 mM of dCTP), 1.2 μl of 1 mM of CY5-dCTP (Perkin-Elmer Life Science, Boston, MA, USA), 1.2 μl of primers Amp-F and Amp-R (50 ng μl−1 each), 0.2 μl of Taq DNA polymerase (5 U μl−1), 2 μl of the bacterial DNA or fecal DNA (1–10 ng μl−1). The Amp-F and Amp-R primer sequences are GAGAGTTTGATYCTGGCTCAG and AAGGAGGTGATCCARCCGCA, respectively (Y is C or T; R is A or G).

PCR was performed in a 9700 Gene-Amp PCR System (Perkin-Elmer, Norwalk, CT, USA). Thin walled 0.2 ml tubes were used for the amplification. The amplification conditions were one cycle of 95°C for 3 min, then 35 cycles of 95°C for 10 s, 53°C for 10 s, 72°C for 70 s, and finally one cycle of 72°C for 4 min and cool down to 4°C. The PCR product was purified with a Centri-Spin column (Princeton Separations, Adelphia, NJ, USA) following the instructions. The purified CY5-labeled PCR products were then dried by Speed-Vac centrifugation (Savant, Farmingdale, NY, USA).

2.4Genomic DNA isolation for PCR

One gram (wet weight) of fresh fecal sample was homogenized in 9 ml of 0.85% NaCl, then centrifuged at 200×g for 10 min to collect the upper phases. This centrifugation step was repeated three times, then the upper phases were centrifuged at 7000×g for 10 min to collect the pellets. The pellets were mixed with 350 μl of solution A (Easy-DNA kit, Invitrogen, Carlsbad, CA, USA) and transferred to a 1.5-ml tube, then heated at 65°C for 10 min. RNase A (3 μl×2 mg ml−1) was added to the solution and heated at 65°C for 10 min. The mixture was added with solution B (150 μl from Easy-DNA kit) and mixed well, then added with 500 μl of chloroform and mixed again. The tube was centrifuged for 10–20 min at 16 000×g. The upper phase was transferred to a new tube and 1 ml ethanol was added and mixed. After cooling the tube at −20°C for 20 min, the tube was centrifuged at 16 000×g for 10 min. The supernatant was discarded and the pellet was washed once with 1 ml 70% EtOH, then air-dried. The pellet was resuspended with 50 μl of TE buffer. The DNA concentration was determined by agarose gel electrophoresis with DNA standards. The DNA was diluted with 1% Triton X-100 to 1:20 or 1:200 (1–10 ng μl−1 DNA) then used for PCR amplification.

Bacterial DNA for each of the intestinal microflora species tested was isolated by the same method, however, the first step was to collect the cells from 5 ml pure culture of each bacterial species by centrifugation at 7000×g for 10 min without pre-washing.


The microarray slides need prehybridization and blocking before hybridization (1–24 h). The 5×Block solution provided by MWG was diluted with autoclaved ddH2O to 1×Block solution (4×SSC, 0.5% sodium dodecyl sulfate (SDS) and 1% bovine serum albumin (BSA)). The slides were incubated in this solution for 45 min at 42°C, then washed carefully by dipping the slides into water six times at room temperature. The slides were dried by centrifugation for 1 min at 3000×g in an IEC clinical centrifuge with IEC CAT 801 Rotor (International Equipment Company, Needham Heights, MA, USA).

The dried, purified, CY5-labeled PCR products were dissolved in 10 μl of MWG hybridization buffer (50% formamide, 6×SSC, 0.5% SDS, 50 mM Na-phosphate, pH 8.0, 5×Denhard's). The tube with the products was heated for 3 min in a boiled water bath, then immediately placed into ice water for 2 min. The solution was collected by brief centrifugation and applied onto the oligo area on the microarray slides by using a micropipetman. A small glass coverslip that was pre-washed, autoclaved, and dried was used to cover the hybridization solution on the array area. The slides were placed into a hybridization chamber (Corning Inc., Corning, NY, USA) and then immersed in a water bath for hybridization overnight at 42°C.

After hybridization, the coverslip was removed by washing the slides 5 min with 2×SSC, 0.1% SDS. All of the washing steps were conducted at room temperature. The slides were then washed 5 min with 0.5×SSC, 0.1% SDS, and then washed 5 min with 0.5×SSC only. The slides were dried by centrifugation as described above, and kept in a slide box in the dark at room temperature until the results were read by a GSI lumonics ScanArray 5000 (Packard Biosciences, Perkin-Elmer, USA).


3.1Preliminary experiments

The effects of length of the oligonucleotide on the hybridization efficiency were determined. We evaluated 20-, 30-, and 40-mer oligonucleotides in our assay. We found that a 40-mer oligonucleotide gave a stronger signal than a 30-mer oligonucleotide, and the 20-mer had almost no signal (data not shown).

We have also tested if the 16S rDNAs of predominant bacterial species in the human gastrointestinal tract tested could be amplified by the primers Amp-F and Amp-R. Fig. 1 shows that 1.5-kb PCR products (16S rDNAs) were generated by the primers from all of the 20 bacterial species tested. A lower annealing temperature (53°C) was used to ensure amplification because there were some base mismatches at both ends of the 16S rDNA sequence for different bacterial species. Due to the lower annealing temperature, some non-specific PCR products were produced (Fig. 1). However, these minor products should not interfere with the hybridization results because the major PCR products (full size 16S rDNA, 1.5 kb) can be hybridized with the oligonucleotide probe.

3.2Hybridization results for the samples prepared from pure bacterial cultures and human feces

PCR amplified 16S rDNAs from the 20 bacteria species were hybridized with oligonucleotide probes that were attached on glass slides. The aim of this study was to determine if the 16S rDNA probes for each bacterial species were designed properly based on the GenBank DNA sequence database. Fig. 2A illustrates the results of the oligonucleotide-microarray method for the 20 bacterial species. Since the oligonucleotide probes were printed by the company without regular arrangement on the slides, we indicated the locations of each probe in Fig. 2B for ease in interpretation. Most probes gave positive results. However, the three probes for each bacterial species gave variable results with different fluorescent intensity. The more CY5-labeled target (PCR products) bound to the probe on the array, the higher intensity of the signal will be produced as indicated by the colors, white (saturated), then red, yellow, green, and blue. We designed three probes for each species to ensure the sensitivity and specificity.

Figure 2.

Results from microarray method for 20 bacteria. A: Hybridization results. Signal: blue (+), green (++), yellow-white-red (+++), no signal (−). B: The locations of all 60 oligo probes.

To determine if this microarray method could detect these predominant bacteria in the human gastrointestinal tract, we tested human fecal samples from three healthy adult volunteers (Fig. 3, Table 2). As expected, the healthy human individuals had similar microfloras: The most predominant intestinal microfloras present in the fecal samples were Bacteroides thetaiotaomicron (BT), Bacteroides vulgatus (BV), Bacteroides distasonis (BD), Clostridium clostridiiforme (CC), Clostridium leptum (CL), Fusobacterium prausntzii (FPR), Peptostreptococcus productus (PSP), Ruminococcus obeum (ROB), Ruminococcus bromii (RBR), Ruminococcus callidus (RCA), Ruminococcus albus (RAL), Bifidobacterium longum (BIL), Bifidobacterium adolescentis (BIA), Eubacterium biforme (EBI), Eubacterium aerofaciens (EAE), and Escherichia coli (ECO). However, some unique differences for the bacterial species were found in the fecal samples from each individual. For example, human 1 had a strong response for E. biforme (EBI), but the other two individuals did not have this species (Fig. 3). Human 3 has a signal for Bacteroides fragilis (BF), and no signal for B. adolescentis (BIA), E. aerofaciens (EAE), and E. coli (ECO) (Fig. 3C).

Figure 3.

Results of the microarray method for fecal samples from three healthy human subjects. Signal: blue (+), green (++), yellow-white-red (+++), no signal (−). A: Human 1. B: Human 2. C: Human 3.

Table 2.  Results read from microarray method in Fig. 3 for three human fecal samples
  1. Signal: blue (+), green (++), yellow-white-red (+++), no signal (−).

SpeciesProbe no.Human 1Human 2Human 3

The results from this study indicate that predominant microflora can be successfully identified using the oligonucleotide-microarray method.

3.3Application of oligo-microarray method in the examination of fecal sample from a person with long-term diarrhea

A person with long-term diarrhea gave fecal samples for diagnostic evaluation before and after treatment with two antibiotics: spiramycin (0.1 g×2 each time) and amoxicillin (0.25 g×2 each time), four times a day, for 10 days. After antibiotic treatment, the person orally ingested some live anaerobic bacteria for a 10-day period. The composition of the fecal flora was monitored before any treatments (Day −20 for control), after 10 days antibiotic treatment but before using the anaerobic bacteria (Day 0), and after using the anaerobic bacteria 10 days (Day 10). Fecal samples for each test period were analyzed by the microarray method, and the results are shown in Table 3. We found that the person with long-term diarrhea lost many of the normal intestinal bacterial species including BV, BD, RBR, RCA, BIA, EBI, and EAE, but had very high concentration of B. fragilis (BF). After antibiotic treatment, only E. coli and BF were detected in the fecal sample. After using the anaerobic bacteria for 10 days, the population of BF decreased and many normal human intestinal microfloras recovered and symptoms of diarrhea went away. PCR analysis of the person's fecal samples gave similar results (data not shown).

Table 3.  Results read from microarray method for a person with long-term diarrhea
  1. Signal: blue (+), green (++), yellow-white-red (+++), no signal (−). The samples tested were from the same person in different time: Day −20, before using antibiotics; Day 0, after using the two antibiotics for 10 days and just before using the anaerobic bacteria; Day 10, after using the anaerobic bacteria for 10 days.

SpeciesProbe no.Day −20Day 0Day 10


The data from this study showed that a microarray technique using species-specific oligonucleotide probes for the detection of predominant bacterial species in the human gastrointestinal tract is a useful method to monitor the populations of anaerobic bacteria in human fecal samples. We observed some variability in the predominant flora from each of the human subjects. However, our results are in agreement with other studies reported on the numerically important predominant anaerobic bacteria in fecal samples. For example, B. vulgatus, Fusobacterium prausnitzii, B. thetaiotaomicron, Ruminacoccus spp., P. productus and Eubacterium spp. were often reported among the 10 most predominant bacteria in fecal flora [1–3,8]. The results obtained by microarray method in this study indicated that these bacteria could be detected with very strong signals from human fecal samples.

PCR methods for identifying these bacterial species [21–23] were also conducted for comparison to the microarray method (data not shown). The results showed that both methods detected the numerically predominant strains and were in agreement with reported predominant anaerobic bacteria.

From Fig. 2 we can see some negative results for the positive control samples made from reference bacterial species. We checked the hairpins and dimers for the oligos and found that the hybridization results are not dependent on these two factors. For example, oligo 2 has highest hairpin energy (−9.4 kcal), but it gave very strong positive results. The three oligonucleotide probes for each bacterial species gave different fluorescent intensities in the microarray slide, which could be due to the oligo sequence (GC vs. AT) or the target location, i.e. the target secondary structure may interfere with the hybridization. In addition, there are sequence conflicts in the GenBank database for the same bacterial species, suggesting that there may be errors in several of the probes. For example, we found that Lactobacillus acidophilus ATCC332 that we used as a reference strain is now re-named as Lactobacillus johnsonii, suggesting that this strain has a different 16S rRNA sequence to other L. acidophilus strains. In our future investigations, we will design new probes to replace these negative probes and design additional probes for detection of other intestinal bacterial species present in human fecal samples.

In conclusion, an oligonucleotide-microarray method was developed for the detection of 20 predominant human intestinal bacterial species from human fecal samples. The results obtained from this study indicate that the oligo-microarray method is a reliable method for the detection of the intestinal bacteria in the fecal samples and the technique is quite comparable with other reported methods.


We thank Drs. Bruce Erickson and Dr. Seong-Jae Kim for critical review and Ms. Pat Fleischer for clerical assistance.