16S rDNA-Based Identification of Bacteria from Conjunctival Swabs by PCR and DGGE Fingerprinting

Abstract

purpose. Establishment of a new molecular biology technique for the identification of multiple bacteria from the ocular environment, which can be applied supplementarily to cultivation in cases of severe bacterial infections.

methods. From 60 human conjunctivae (29 with purulent and 31 with nonpurulent conjunctivitis), swabs were taken and DNA was extracted. Fragments of 200 bp, spanning the V3 region of the eubacterial 16S rDNA, were amplified by polymerase chain reaction (PCR) and separated by denaturing gradient gel electrophoresis (DGGE). For phylogenetic identification, DGGE bands were excised and directly sequenced, or 16S rDNA clone libraries were constructed and clones were screened by DGGE. Sequences were compared with sequences of known bacteria listed in the EMBL database. Furthermore, the results were compared with results obtained from conventional cultivation.

results. 16S rDNA could be amplified from 25 of 29 investigated swabs taken from purulent conjunctivitis eyes and from 2 of 31 investigated swabs taken from nonpurulent conjunctivitis eyes. Sixteen samples showed monomicrobial and 11 samples showed polymicrobial infections. The following genera (n is number of samples) were detected: Staphylococcus (n = 8), Corynebacterium (n = 7), Propionibacterium (n = 7), Streptococcus (n = 6), Bacillus (n = 2), Acinetobacter (n = 3), Pseudomonas (n = 3), Proteus (n = 1), and Brevundimonas (n = 1). Four sequences could not be identified to the genus level. They had highest sequence similarities both to sequences of Pantoea and Enterobacter (n = 1), Kingella and Neisseria (n = 1), Serratia and Aranicola (n = 1), and Leuconostoc and Weissella (n = 2), respectively. Culture was only positive for coagulase-negative staphylococci (n = 9), Corynebacteria (n = 3), Staphylococcus aureus (n = 1), Streptococcus sp. (n = 1), Proteus sp. (n = 1), Klebsiella oxytoca (n = 1), and Pseudomonas aeruginosa (n = 1). In total, 45% of the 60 analyzed conjunctival swabs were PCR positive, whereas only 22% were culture positive. No sample positive by culture gave negative results by PCR.

conclusions. 16S rDNA sequence analyses and DGGE fingerprinting are appropriate methods for the detection and identification of monomicrobial as well as polymicrobial ocular infections of bacteria that might not be detected by conventional cultivation.

Bacterial ocular infections are common. Although many cases show a benign course, some can be associated with sight-threatening ocular complications. Identification of the causative pathogens in these cases is mandatory but often difficult because some bacteria have special growth requirements. Furthermore, sample size from ocular tissues is usually small, leading to unreliable cultivation results. Initiation of proper therapy can then be delayed with possible devastating visual consequences. 

Molecular approaches to the identification of bacteria show promising results. The amplification of 16S rDNA of any bacterial species is possible without prior cultivation when broad-range PCR primers targeted to highly conserved regions are applied. The comparison of amplified and sequenced 16S rDNA sequences with sequences of known bacteria in 16S rDNA databases facilitates a subsequent phylogenetic identification. In ophthalmology, the 16S rDNA-based identification of pathogens is still at its beginning and, except in a few studies, is rarely applied. Hykin et al. ¹ and Therese et al. ² used eubacterial primers and Propionibacterium-specific primers to detect bacterial DNA in vitreous samples of patients who had endophthalmitis. Lohmann et al. ³ and Knox et al. ⁴ detected and identified bacteria in corneal scrapings and in vitreous samples of patients who had keratitis and endophthalmitis by amplification and subsequent direct sequencing of 16S rDNA. These studies allowed the simple detection of eubacterial DNA or the identification of monomicrobial infections, whereas pathogens of polymicrobial infections could not be identified by direct sequencing. 

Nevertheless, bacterial infections of the eye are sometimes polymicrobial. In the studies of Ormerod et al. ⁵ and Kunimoto et al., ⁶ it was shown that ocular infections such as endophthalmitis were polymicrobial in up to 32%. Concerning polymicrobial communities, the direct sequencing of mixed 16S rDNA fragments fails, and sequence information can only be obtained through 16S rDNA clone libraries. ^{7 8} To avoid the sequencing of clones containing identical sequences, clone libraries can be screened by restriction fragment length polymorphism analysis (RFLP) or by denaturing gradient gel electrophoresis (DGGE). By applying RFLP, 16S rDNA amplicons are digested with a set of different restriction endonucleases, and DNA fragments are separated in agarose gels, leading to different RFLP profiles of individual 16S rDNA sequences. ^{9 10}  

DGGE facilitates profiling of monomicrobial as well as polymicrobial communities in polyacrylamide gels because of the sequence-specific separation of 16S rDNA amplicons of same length. ¹¹ During gel electrophoresis, short 16S rDNA amplicons migrate toward increasing denaturing concentrations, leading to a partial melting of the DNA helix and to a decrease and subsequent ending of electrophoretic migration. As a consequence, a band pattern is produced in which each band theoretically represents a bacterial taxon. 

In the present study, a method is proposed that combines 16S rDNA genotyping with DGGE fingerprinting. Figure 1 shows a scheme of the applied experimental procedure. The microbial communities of conjunctival swabs were investigated by amplifying, cloning, and sequencing of 16S rDNA. Furthermore, monomicrobial and polymicrobial communities were profiled by DGGE. The introduced molecular technique was applied to the detection of bacteria in suppurative conjunctivitis, to obtain an optimized protocol for ocular samples. Especially in ophthalmology, where sample material is very limited and standard diagnostic tests based on cultivation are often negative, 16S rDNA typing in combination with DGGE fingerprinting has the potential to become a promising identification method. 

Methods

Sampling and DNA Extraction

Samples were taken in duplicate from 60 human eyes, in accordance with the tenets of the declaration of Helsinki. Twenty-nine eyes had purulent ocular diseases that were supposed to be most probably due to bacterial infections (i.e., from chalazions and presumed bacterial conjunctivitis and keratitis). Thirty-one control eyes presented nonpurulent conjunctivitis (i.e., from dry eye syndromes, uveitis, and posttraumatic and keratoconjunctivitis epidemica; Table 1 ). Conjunctival swabs were taken with sterile cotton swabs and immediately frozen at −20°C in 2-ml sterile tubes until they were analyzed. From most patients two DNA extractions from duplicate swabs were carried out independently on 2 different days, to avoid false-positive or -negative PCR results. 

For DNA extraction, cotton swabs were mixed with 300 μl DNA extraction buffer I (150 mM Na₂EDTA, 225 mM NaCl; pH 8.5) and 45 μl lysozyme (50 mg/ml). After incubation at 37°C for 30 minutes, 9 μl 25% SDS and 9 μl proteinase K (20 mg/ml) were added. After incubation for an additional 60 minutes at 37°C with agitation, 150 μl of prewarmed (90°C) DNA extraction buffer II[ 100 mM Na₂EDTA, 400 mM Tris-HCl, 400 mM Na₂phosphate buffer (pH 8.0), 5.55 M NaCl, 4% CTAB (hexadecyltrimethyl ammonium bromide); pH 8.0] and 27 μl 25% SDS were added. After incubation at 65°C for 60 minutes the samples were subjected to three cycles of freezing (−80°C) and thawing (65°C). One hundred forty microliters was subsequently used for purification of DNA with the QIAamp Viral RNA Mini Kit (Qiagen, Hilden, Germany) following the protocol of the manufacturer. This kit removes several PCR inhibitors and contains carrier RNA, which facilitates the elution of small amounts of RNA as well as DNA. DNA was eluted with 60 μl ddH₂O and tested for PCR-amplifiable DNA with 16S rDNA-specific primers. 

Amplification of 16S rDNA

Three microliters of the extracted DNA was amplified with primers 27f (5′-AGA GTT TGA TCC TGG CTC AG-3′) ¹² and 907r (5′-CCC CGT CAA TTC ATT TGA GTT T-3′), ¹³ generating a PCR product corresponding to nucleotide positions 8 to 926 of the Escherichia coli 16S rDNA sequence. All reactions were carried out in 25 μl volumes, containing 12.5 pmol of each primer, 200 μM of each deoxyribonucleoside triphosphate, 2.5 μl of 10× PCR buffer (100 mM Tris-HCl, 15 mM MgCl₂, 500 mM KCl; pH 8.3), and 0.5 U of Taq DNA polymerase (Roche Diagnostics, Mannheim, Germany), increased to 25 μl with sterile water. PCR was performed in a Robocycler (Stratagene, La Jolla, CA) with the following thermocycling program: 5 minutes denaturation at 95°C, followed by 30 cycles of 1 minute denaturation at 95°C, 1 minute annealing at 55°C, 1 minute extension at 72°C, and a final extension step of 5 minutes at 72°C. Ten microliters of PCR products was visualized by electrophoresis in 2% (wt/vol) agarose gels and with ethidium bromide (0.5 μg/ml) staining. 

To avoid contamination, all solutions were prepared with sterile water (Sigma-Aldrich, Vienna, Austria), autoclaved twice, and treated with hard UV for 90 minutes in 1-ml aliquots. Furthermore, all steps were performed with aerosol-resistant tips in a CleanSpot PCR/UV Work Station (Coy Laboratory Products, Grass Lake, MI). The preparation of the master mix, the addition of template, and the gel electrophoresis of PCR products were carried out in three separate rooms. For each master mix, two negative controls were carried out through the whole procedure, in which water instead of sample material was used to exclude the possibility of false-positive PCR results through cross-contamination. 

DGGE Analysis

For the genetic fingerprinting of bacterial 16S rDNA from individual eye swabs, nested PCR reactions were carried out. 16S rDNA fragments corresponding to nucleotide positions 341 to 534 in the E. coli sequence were amplified with the forward primer 341fGC, to which at its 5′ end a 40-base GC-clamp was added (341f: 5′-CCT ACG GGA GGC AGC AG-3′; GC-clamp: 5′-CGC CCG CCG CGC GCG GCG GGC GGG GCG GGG GCA CGG GGG G-3′) ¹¹ and the reverse primer 518r (5′-ATT ACC GCG GCT GCT GG-3′) ¹¹ in 100 μl volumes with 4 μl of PCR product of the first amplification as template DNA. Cycling conditions were as described above. The presence of PCR products was confirmed by analyzing 10 μl of product by electrophoresis in 2% (wt/vol) agarose gels and staining with ethidium bromide before DGGE analysis. Ninety microliters of PCR products was precipitated with 96% EtOH, resuspended in 15 μl ddH₂O, and separated by DGGE. Gel electrophoresis was performed as described elsewhere in a linear denaturant gradient from 25% to 60% in a D GENE-System (Bio-Rad, Munich, Germany). ¹¹ After completion of electrophoresis, gels were stained in an ethidium bromide solution and documented with a UVP documentation system. 

16S rDNA Sequencing

Sequence identification was carried out only with samples showing a PCR product after the first 30 amplification cycles. From some samples sequence information was obtained by excising and direct sequencing of reamplified 200-bp DGGE bands as described elsewhere. ¹⁴ For the direct sequencing of excised DGGE bands, the extracted DNA was reamplified with primer 341f, including an additional T3 sequence at its 5′ end (T3: 5′-AAT TAA CCC TCA CTA AAG-3′) and primer 518r. To identify microorganisms on the basis of longer 16S rDNA fragments, from most samples 900-bp 16S rDNA clone libraries were constructed by cloning 5 μl of PCR product amplified with primers 27f and 907r. Cloning was performed with the pGEM-T Vector System (Promega, Mannheim, Germany), following the protocol of the manufacturer. The ligation products were subsequently transformed into E. coli XLI-Blue, which allows blue-white screening. ¹⁵ To screen for positive clones, clone inserts were amplified with the vector-specific primers SP6 (5′-ATT TAG GTG ACA CTA TAG AAT AC-3′) and T7 (5′-TAA TAC GAC TCA CTA TAG GG-3′). Screening for different clones was carried out by comparing inserts reamplified with primers 341fGC and 518r in DGGE. In a second screening the different clones were compared with the DGGE fingerprint of the corresponding eye swab. Inserts of clones producing PCR products that matched identical positions in the DGGE fingerprint of the eye swab were sequenced. 

For sequencing of clone inserts, fragments were amplified with primers SP6 and T7. One hundred microliters of PCR products was purified with a QIAquick PCR Purification Kit (Qiagen) and sequenced with a LI-COR DNA Sequencer Long Read 4200. ¹⁶ Sequencing reactions were carried out by cycle sequencing with the SequiTherm system (EPICENTRE, Madison, WI) with 2 pmol fluorescently labeled primers and 5 U SequiTherm thermostable DNA polymerase. Clone inserts were partially sequenced (200 or 500 bp) with primers 341f or 518r. Reamplified DGGE bands were sequenced with primer T3. 

Phylogenetic Analyses

The obtained sequences were compared with sequences of known bacteria listed in the EMBL nucleotide sequence database. The FASTA search option for the EMBL database was used to search for close evolutionary relatives. ¹⁷  

Nucleotide Sequence Accession Numbers

The sequences obtained in this study have been assigned in the EMBL database under the accession numbers AJ405008–AJ405055. 

Cultivation

All swabs were cultured by routine in a standard diagnostic labor. Bacterial culture was performed with the aim to detect common causes of conjunctivitis such as Haemophilus influenzae, Staphylococcus aureus, Streptococcus pneumoniae as well as causes of infections in immunocompromised patients (members of the family Enterobacteriaceae and Pseudomonas aeruginosa). Anaerobic cultures and cultures for Neisseria gonorrhoeae and enrichment cultures, which are only performed if unusual organisms are suspected, were not carried out. 

Briefly, swabs were placed into a transport medium (Transswab; Medical Wire & Equipment Co. Ltd., Corsham Wilts, United Kingdom), and after arrival in the laboratory were plated onto columbia agar + 5% sheep blood, chocolate agar + isovitalex + bacitracin and McConkey agar (Becton Dickinson Microbiology Systems, Sparks, MD). The media were incubated in 5% to 7% CO₂ at 35°C for up to 48 hours. Bacteria were identified using standard microbiologic procedures. 

Results

Amplification of 16S rDNA and DGGE Fingerprinting

Generally, PCR was done twice with template DNA obtained from each of the duplicate extractions per sample. Samples producing reproducible PCR products four times were regarded as PCR positive. In cases of negative PCR results in all amplifications, the samples were regarded as PCR negative. Five samples showed very faint bands in agarose gels in only one of the duplicate DNA extractions. Therefore, they were amplified twice more under the same conditions. Because these very faint bands were not reproducible, these samples were also regarded as PCR negative. In total, 27 of 60 investigated eye swabs showed reproducible PCR products after 30 amplification cycles (Table 1) . Two of the PCR-positive samples (A4 and A9) belonged to the eyes with nonpurulent conjunctivitis and were initially not supposed to show bacterial colonization (Table 2) . Four of the PCR-negative samples were initially supposed to be PCR positive according to the symptoms (purulent conjunctivitis). Thirty-three of the 60 investigated samples were regarded as PCR negative. 

DGGE analysis and sequence identification were carried out only with the 27 PCR-positive samples. DGGE fingerprints derived from the conjunctival swabs taken in duplicate per patient showed the same patterns (data not shown). Table 1 gives an overview of the conjunctival swabs from which 16S rDNA could be amplified successfully and furthermore an overview of the PCR-negative conjunctival swabs. 

Sequence Analysis of Individual DGGE Bands

Ten sequences were obtained by the direct sequencing of excised DGGE bands. Thirty-eight sequences were obtained by sequencing of cloned 16S rDNA inserts. During the screening no clones could be detected that were not present in the original DGGE band patterns. From samples A6 and A9 no clones or excised DGGE bands were sequenced (Table 2) . From the excised and reamplified DGGE bands sequence information between 106 and 173 bp was obtained, some of which had a high number of ambiguous bases, because of the sometimes bad quality of DNA. From clone inserts, between 199 and 550 bp were sequenced. Table 2 shows the results of comparative sequence analyses obtained from the EMBL database. Sequence homologies to sequences of known bacteria in the EMBL database ranged between 90% and 100%. Most sequences had similarity values between 98% and 99.8%. In total, 16 samples showed monomicrobial and 11 samples showed polymicrobial infections. Of the 11 samples showing polymicrobial infections, 7 samples had infections with two different genera, 3 samples with three different genera, and 1 sample with four different genera. Table 3 gives an overview of how often the identified genera were detected in the 27 PCR-positive samples. The most frequently detected genera were Staphylococcus and Corynebacterium, followed by Propionibacterium and Streptococcus. The remaining sequences were affiliated with the genera Bacillus, Acinetobacter, Brevundimonas, Pseudomonas, and Proteus. For most sequences affiliated with Streptococcus, Staphylococcus, and Corynebacterium, the sequence similarities obtained from the EMBL database were the same for different species within the same genus, which prevented an identification to the species level. For six sequences no identification to the genus level could be obtained because of similar sequence similarity values to the different genera Leuconostoc and Weissella (A22-K1 and A22-K2), Pantoea and Enterobacter (A20-K1 and A20-K2), Kingella and Neisseria (A19-K1), and to Serratia and Aranicola (A19-K4), respectively. 

Comparison of DGGE Fingerprints

To compare the DGGE positions of all sequences, the sequenced clones and DGGE bands were reamplified and analyzed parallel in DGGE. Figure 2 shows the DGGE patterns of the reamplified clones and excised DGGE bands of all 27 PCR-positive samples. DGGE fingerprints of 12 samples contained one band, fingerprints of 9 samples contained two bands, fingerprints of 5 samples contained three bands, and fingerprints of 2 samples contained four bands. Table 3 gives an overview of how many different positions in DGGE could be found among sequences affiliated with the same genus. Five different positions in DGGE were found among 9 Staphylococcus-affiliated sequences. Three different positions in DGGE were found among 6 Streptococcus-affiliated sequences. Nine different positions in DGGE were found among 10 Corynebacterium-affiliated sequences. Two different positions in DGGE were found among 3 Lactobacillaceae-affiliated sequences. Only different positions in DGGE were found among 4 Acinetobacter-, 4 Pseudomonas-, 2 Bacillus-, and 2 Enterobacter/Pantoea-affiliated sequences. All sequences affiliated with Propionibacterium had the same position in DGGE. 

Comparison of Cultivation and PCR

No sample positive by culture gave negative results by PCR, whereas 55% of the PCR-positive results were culture negative. Culture was only positive for coagulase-negative staphylococci (n = 9), Corynebacterium spp. (n = 3), S. aureus (n = 1), Streptococcus sp. (n = 1), Proteus sp. (n = 1), Klebsiella oxytoca (n = 1), and P. aeruginosa (n = 1; Table 3 ). In seven cases (samples A3, A4, A5, A7, A16, A18, and A25) coagulase-negative staphylococci and in one case (sample A4) K. oxytoca were detected only by cultivation but not by PCR in apparently polymicrobial infections (Tables 2 and 3) . 

Discussion

In the present study, the conjunctival flora of 60 eyes was investigated by molecular means, including PCR, DGGE, and 16S rDNA sequence analyses. Working with conjunctival swabs, only very little bacterial material is available. The amplification of 16S rDNA is therefore a useful method for the investigation of such samples. However, the extreme sensitivity of nucleic acid amplification techniques enhances the possibility of detection of clinically irrelevant or contaminating target sequences. If used improperly, the amplification of 16S rDNA might give misleading results, especially with clinical samples that contain relatively few pathogenic microorganisms. We have observed that the exceeding of 35 amplification-cycles led to amplification of contaminating DNA in some of the reagent-only controls, despite precautions such as the use of aerosol-resistant tips and the spatial separation of DNA extraction, PCR, and the gel electrophoresis of PCR products. As amplification was therefore restricted to 30 cycles, the normal conjunctival flora could not be detected under the given laboratory circumstances, because of the low number of bacteria present on the normal conjunctiva. 

In total, eubacterial DNA could be amplified from 27 of 60 investigated eye swabs. PCR-positive results were obtained from 25 of 29 specimens taken from purulent conjunctivitis and 2 samples taken from eyes with dry eye syndrome. Four samples derived from eyes with purulent conjunctivitis were unexpectedly PCR negative. Reasons for these negative PCR results were in two cases fungal infections and in two cases herpes simplex virus infection. Negative PCR results from control eyes had been expected. In cases of intraocular disease vitreous fluid and aqueous humor have to be investigated, because conjunctival swabs do not contain the pathogenic organism. Control eyes with contact lens wear history, dry eye syndrome, allergy, and trauma had been suspected to have no bacterial infection. 

Comparison of sequences with sequences listed in the EMBL database revealed that most of them had sequence similarities of 98% to 100% to sequences of known genera. As the investigated sequences were only parts (between 121 and 536 bp) of the approximately 1600-bp-long 16S rDNA, a clear phylogenetic affiliation was obtained only to the genus and not to the species level. Generally, a reliable phylogenetic identification based on partial 16S rDNA analysis is often only possible to the genus level. ^{18 19} For some sequences a clear identification to the genus level was not possible. Two clones had identical similarities to sequences of the genera Enterobacter and Pantoea, both members of the Enterobacteriaceae and proposed to be reclassified together in the genus Pantoea. ²⁰ Three clones had similarities identical to sequences of the genera Leuconostoc and Weissella, both members of the Lactobacillaceae and proposed to be reclassified together in the genus Weissella. ²¹ One clone had sequence similarities identical to sequences of the genera Serratia and Aranicola, both members of the Enterobacteriaceae. Another clone had a 90% similarity value to sequences of the genera Kingella and Neisseria, both members of the Neisseriaceae. The low sequence similarity value to known bacteria reveals that this sequence represents a new genus related to the Neisseriaceae. The problem of inadequate phylogenetic identification could be improved by cloning and sequencing of the entire 16S rDNA. Furthermore, because the number of 16S rDNA sequences in public databases is increasing day by day, a better identification of so far unknown bacteria may be possible in the future. 

The obtained results are mostly in accordance with results of other studies based on cultivation. Most bacteria frequently observed in this study are expected pathogenic organisms causing infections in human eyes and lids. Different species of Bacillus, Proteus, Pseudomonas, Serratia, and especially of Corynebacterium, Staphylococcus, and Streptococcus have been found to be part of the normal conjunctival flora as well as to play roles as pathogens in different ocular diseases. ^{22 23 24 25} Propionibacterium acnes has been detected to be part of the normal anaerobic conjunctival flora as well as to be one of the causative agents of late-onset endophthalmitis ^{23 26 27} and corneal ulceration. ²⁸ Concerning the other detected bacteria, little is known from the literature on a possible pathogenic character in human eyes. However, in neonates and infants, these bacteria have been associated with more or less severe non-eye diseases. ²⁹ To our knowledge, Acinetobacter spp. and Enterobacter agglomerans/Pantoea ananas are not very commonly detected in the human eye. E. agglomerans and Acinetobacter lwoffii are rarely isolated from eyes that have endophthalmitis. ^{25 30} Pantoea may cause fever, shaking chills, sepsis, and osteomyelitis. Pantoea was previously observed in six conjunctival swabs of patients who had conjunctivitis, unfortunately without knowledge of the clinical course, ³¹ and was involved in endophthalmitis after foreign body penetration. ³² The detected organism with the highest sequence similarities to the genera Leuconostoc and Weissella has not yet been found in eyes. Members of the genus Leuconostoc are facultatively anaerobic, catalase-negative, Gram-positive cocci and exhibit an intrinsic resistance to vancomycin. Neonates may be colonized during delivery by Leuconostoc inhabiting the maternal genital tract. Leuconostoc was also encountered in cerebrospinal fluid, peritoneal dialysate fluid and wounds but is supposed to have very little virulence for healthy humans. ³³ Kingella kingae is a small Gram-negative rod and may be involved in suppurative arthritis, osteomyelitis, spondylodiskitis, endocarditis, transient bacteremia, meningitis, pulmonary infections, dactylitis, and subglottic and epiglottic infections. K. kingae, also known to cause eyelid abscesses and endophthalmitis, was observed in corneal ulceration. ^{34 35 36} Detection by conventional culture is difficult and cultures should be examined once per week for a total of 3 weeks. However, only 5% of Kingella infections will be detected by this procedure. Brevundimonas is a pseudomonad, rarely encountered in human infection. Brevundimonas vesicularis may be a virulent organism involved in central nervous system infections and bacteremia, including nosocomial infections. ³⁷  

In this study, the presence of polymicrobial infections of coagulase-negative staphylococci and K. oxytoca with Corynebacterium spp., Streptococcus spp., Pseudomonas spp. or Propionibacterium spp. raised the problem of detecting the coagulase-negative staphylococci and K. oxytoca by PCR. The presence of coagulase-negative staphylococci might have been overestimated by selective cultivation as a result of cultivation-dependent population shifts. Because of the preferential amplification of the more abundant template DNA of those bacteria that could not be detected by cultivation, rare bacteria might have failed to be amplified in a sufficient amount. Concerning polymicrobial infections, on the one hand the template DNA of rare bacteria can be outcompeted in the amplification process by template DNAs of bacteria that are present in greater numbers ^{11 38 39} ; on the other hand selective cultivation can lead to an overestimation of a certain organisms. Both cases lead to discrepancies between PCR and culture. 

The combination of cloning and genetic fingerprinting by DGGE allows the identification of polymicrobial infections. Although the excising and direct sequencing of DGGE bands seems to be a more rapid method, cloning leads to longer and a higher quality of sequence information, which facilitates a more reliable phylogenetic identification. The construction of DGGE markers, containing 16S rDNA fragments of bacteria relevant for ocular diseases, would facilitate and accelerate the interpretation of DGGE fingerprints. In this study, it is noticeable that especially among the genera Corynebacterium, Acinetobacter, and Pseudomonas sequences could be distinguished by DGGE. 

Although being an elegant method for the investigation of especially polymicrobial infections, DGGE has been used so far only in a few studies for bacterial identification in clinical specimens. ^{40 41 42} The broad-range nature of the method allows the detection of rare, unexpected, or fastidious pathogens. Especially in ophthalmology, where only little sample material is available from the outset and cultivation results are often negative, the sensitive broad-range amplification of 16S rDNA in combination with DGGE could become a promising detection and identification method. In the present study, no sample positive by culture gave negative results by PCR, whereas 55% of the PCR-positive results were culture negative. This indicates that PCR is a more reliable and significantly more sensitive method for the detection of bacterial infections than cultivation. Similar observations were made by Lohmann et al., ⁴³ whose PCR results in vitreous samples were positive in 92%, whereas the cultivation results were positive in only 24%. Culturing failed to detect Propionibacterium acnes and Actinomyces israelii, which are known to be fastidious and slowly growing bacteria. Therese et al. ² investigated vitreous samples, in which 44% of culture-negative samples showed positive PCR results. Okhravi et al. ⁴⁴ investigated 37 aqueous and vitreous samples with suspected infection, of which 100% were PCR positive and 54% were culture negative. Ley et al. ⁴¹ detected bacterial DNA of Pseudomonas spp., Acinetobacter spp., Escherichia spp., Moraxella spp., Staphylococcus spp., and Bacillus spp. in 20 blood samples that were culture negative. In addition, several other studies in medical microbiology revealed that the PCR technique is more sensitive in the detection of bacteria than cultivation. ^{3 7 8 45}  

The introduced molecular technique was applied to the detection of bacteria in suppurative conjunctivitis, to establish the technique for ocular samples. Results of the present pilot study reveal that 16S rDNA genotyping in combination with DGGE is more sensitive than conventional cultivation. Although the introduced method is not completely free of all biases, it can be used supplementarily or as an alternative to cultivation, especially in infections caused by bacteria with unusual growth requirements, for patients who have been unsuccessfully treated with antibiotics or who suffer from sight-threatening or chronical bacterial infections that cannot be cultured. 

 Submitted for publication July 19, 2000; revised November 15, 2000; accepted November 22, 2000.

 Commercial relationships policy: N.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked“ advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Figure 1 shows a scheme of the applied experimental procedure, beginning with DNA extraction from eye swabs, followed by amplification of 16S rDNA, construction of clone libraries, genetic fingerprinting by DGGE, and sequencing.

Table 1.

 

Overview of the Diagnosis and of PCR and Cultivation Results of 60 Investigated Conjunctival Swabs

Diagnosis	No. of Samples	Positive PCR Results	Positive Cultivation Results
Purulent conjunctivitis, chalazion	29	25	12
Dry eye syndrome	2	2	1
Uveitis	10	—	—
Keratoconjunctivitis epidemica	3	—	—
Contact lens wear	3	—	—
Dry eye syndrome	7	—	—
Trauma, allergy or previous keratoplasty	6	—	—

Table 2.

 

Overview of the Diagnosis, Cultivation, and Sequence Analysis Results of the 27 PCR-positive Conjunctival Swabs

Sample	Diagnosis	Sequence Name*	Phylogenetic Affiliation	%, †	Cultivation
A1	Purulent conjunctivitis	A1–B1 (T3/121/0)	Staphylococcus spp.	100.0	Coagulase-negative staphylococci
A2	Purulent conjunctivitis and chalazion	A2–K1 (341f/200/0)	Staphylococcus caprae	100.0	Coagulase-negative staphylococci
				A2–K2 (518r/455/0)		Uncultured Corynebacterium	98.0
A3	Chalazion	A3–K1 (341f/503/0)	Corynebacterium accolens/fastidiosum	99.8	Coagulase-negative staphylococci
				A3–K2 (341f/497/0)		Uncultured Corynebacterium sp.	99.6
							Corynebacterium amycolatum/asperum	99.4
A4	Dry eye syndrome	A4–K1 (341f/502/2)	Corynebacterium macginleyi	99.0	Coagulase-negative staphylococci, Klebsiella oxytoca, Corynebacteria
				A4–K2 (341f/506/1)		Corynebacterium sp.	99.8
							Corynebacterium amycolatum/ asperum/variabilis	94.5
A5	Purulent conjunctivitis	A5–K1 (341f/503/0)	Uncultured Corynebacterium sp.	99.6	Coagulase-negative staphylococci
A6	Purulent conjunctivitis	A6–B1, not sequenced	Has the same position in DGGE as Corynebacterium sp.		Corynebacteria
A7	Purulent conjunctivitis	A7–K1 (518r/453/0)	Corynebacterium asperum	99.8	Staphylococcus aureus
A8	Chalazion	A8–B1 (T3/141/0)	Corynebacterium bovis	98.6	Corynebacteria
				A8–B2 (T3/151/0)		Propionibacterium acnes	99.3
A9	Dry eye syndrome	A9–B1, not sequenced	Has the same position in DGGE as Propionibacterium acnes		—
A10	Purulent conjunctivitis and keratitis	A10–B1 (T3/151/7)	Propionibacterium acnes	98.1	—
A11	Purulent conjunctivitis	A11–K1 (341f/550/0)	Acinetobacter johnsonii	99.8	—
				A11–K2 (341f/449/0)		Pseudomonas putida	99.8
				A11–B2 (T3/149/9)		Streptococcus sp.	91.3
A12	Purulent conjunctivitis	A12–B1 (T3/172/0)	Acinetobacter lwoffii	99.4	—
A13	Purulent conjunctivitis	A13–K1 (341f/200/1)	Acinetobacter johnsonii/lwoffii	98.0	—
				A13–K2 (341f/200/0)		Acinetobacter sp.	99.5
							Acinetobacter johnsonii/lwoffii	99.0
				A13–K3 (341f/200/0)		Brevundimonas vesicularis	99.0
A14	Purulent conjunctivitis	A14–B1 (T3/139/0)	Staphylococcus pasteuri	99.3	—
				A14—B3 (T3/106/8)		Propionibacterium acnes	91.5
A15	Purulent conjunctivitis	A15–K1 (341f/200/0)	Staphylococcus capitis	100.0	—
				A15–K2 (341f/200/3)		Propionibacterium acnes	98.5
A16	Purulent conjunctivitis	A16–K1 (518r/481/0)	Streptococcus salivarius	99.2	Coagulase-negative staphylococci
A17	Purulent conjunctivitis	A17–K1 (341f/200/1)	Staphylococcus pasteuri	98.0	—
				A17–K2 (341f/200/1)		Staphylococcus pasteuri	99.0
				A17–K3 (341f/200/0)		Streptococcus bovis/waiu/suis	96.5
A18	Purulent conjunctivitis and keratitis	A18–K1 (341f/498/0)	Propionibacterium acnes	99.4	Coagulase-negative staphylococci
		A18–K2 (341f/537/4)	Streptococcus mitis/oralis	98.1
A19	Purulent conjuntivitis and keratitis	A19–K1 (341f/475/0), ‡	Kingella kingae/Neisseria sp.	90.3	—
				A19–K2 (341f/199/0)		Staphylococcus capititis	99.5
				A19–K3 (341f/200/0)		Streptococcus bovis/waiu/suis	97.5
				A19–K4 (341f/518/0)		Serratia proteamaculans/Aranicola proteolyticus	100.0
A20	Chalazion and purulent conjunctivitis	A20–B1 (T3/150/1)	Staphylococcus capitis/cohnii/ saprophyticus/epidermis	99.3	—
		A20–K1 (341f/536/0), ‡	Enterobacter agglomerans/Pantoea ananas	99.3
		A20–K2 (341f/536/0), ‡	Enterobacter agglomerans	99.4
			Pantoea annans	99.3
		A20–K3 (341f/531/1)	Propionibacterium acnes	99.5
A21	Purulent conjunctivitis	A21–K1 (341f/519/0)	Weissella confusa/kandleri	97.1	Streptococci, Coagulase-negative staphylococci
				A21–K2 (341f/200/2)		Streptococcus mitis/pneumoniae/ oralis/infantis	96.5
		A21–K3 (341f/200/0)	Staphylococcus capitis/cohnii/ saprophyticus/epidermis	100.0
A22	Purulent conjunctivitis	A22–K1 (341f/543/0)	Leuconostoc mesenteroides	94.7	—
		A22–K2 (341f/518/0), ‡	Leuconostoc paramesenteroides	96.7
			Weissella sp.	96.5
A23	Purulent conjunctivitis	A23–K1 (T3/200/0)	Proteus mirabilis	99.5	Proteus sp.
A24	Purulent conjunctivitis	A24–K1 (341f/530/1)	Pseudomonas aeruginosa	99.6	Pseudomonas aeruginosa
A25	Purulent conjunctivitis	A25–K1 (341f/532/1)	Pseudomonas sp.	98.9	Coagulase-negative staphylococci
				A25–K2 (341f/533/1)		Pseudomonas sp.	99.3
			Pseudomas migulae/putida	98.9
A26	Purulent conjunctivitis and keratitis	A26–K1 (341f/146/0)	Bacillus subtilis	97.3	—
A27	Purulent conjunctivitis	A27–B1 (T3/173/1)	Bacillus clarkii	94.3	—

*  Sequences numbered with –B were obtained by direct sequencing of excised DGGE bands; sequences numbered with –K were obtained by shotgun cloning of 900-bp 16S rDNA PCR products (sequencing primer/number of sequenced bases/number of ambiguous bases).

†  Percent similarity of partial 16S rRNA coding sequences to sequences of their closest bacterial relatives available in the EMBL nucleotide sequence database.

‡  A phylogenetic identification to the genus level was not possible.

Table 3.

 

Overview of the Abundance of Bacterial Genera and Number of Their Different Positions in DGGE

Identified Genera	Phylogenetic Affiliation	PCR-Positive Swabs*	DGGE, †	Culture-Positive Swabs*
Gram-positive, aerobic
Corynebacterium	Actinomycetales/Corynebacteriaceae	7	9/10	3
Staphylococcus	Bacillus/Staphylococcus group	8	4/9	10
Streptococcus	Bacillus/Clostridium group, Streptococcaceae	6	3/6	1
Bacillus	Bacillus/Clostridium group, Bacillaceae	2	2/2	—
Gram-positive, obligate anaerobic
Propionibacterium	Actinomycetales/Propionibacteriaceae	7	1/7	—
Gram-positive, faculative anaerobic
Leuconostoc/Weissella	Bacillus/Clostridium group, Lactobacillaceae	2	2/3	—
Gram-negative, aerobic
Acinetobacter	γ-Proteobacteria, Moraxellaceae	3	4/4	—
Pseudomonas	γ-Proteobacteria, Pseudomonas group	3	4/4	1
Brevundimonas	α-Proteobacteria, Caulobacter group	1	1/1	—
Kingella/Neisseria	β-Proteobacteria, Neisseriaceae	1	1/1	—
Pantoea/Enterobacter	γ-Proteobacteria, Enterobacteriaceae	1	1/1	—
Proteus	γ-Proteobacteria, Enterobacteriaceae	1	1/1	1
Serratia/Aranicola	γ-Proteobacteria, Enterobacteriaceae	1	1/1	—
Klebsiella oxytoca	γ-Proteobacteria, Enterobacteriaceae	—	—	1

*  Number of eye swabs in which the corresponding genera were detected.

†  Number of different positions in DGGE/number of investigated sequences.

Figure 2.

 

Detail of the ethidium bromide–stained 16S rDNA DGGE fingerprints of reamplified clones and excised DGGE bands derived from 27 eye swabs, from which 16S rDNA fragments could be amplified successfully after 30 amplification cycles. DGGE bands A1–B1, A6–B1, A11–B2, and A19–K4 were almost not visible in the figure and their positions in DGGE were therefore marked with arrows. Lanes and DGGE bands are numbered according to the nomenclature of samples listed in Table 2 . Lanes M represent the standard reference pattern. The following reference species were used: Borrelia sp. (I), Pasteurella canis (II), Listeria monocytogenes (III), Actinobacillus pleuropneumonie (IV), Staphylococcus aureus (V), Bordetella bronchiseptica (VI), Escherichia coli (VII), Pasteurella hemolytica (VIII), Rhodococcus sp. (IX) and Frankia sp. (X).