INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The advent of DNA amplification by PCR has had a great impact on the speed and accuracy with which one can identify a bacterial species or strain. Instead of relying on time-consuming and subjective phenotypic tests, it is now possible to rapidly amplify specific regions of bacterial genomes by PCR and compare them at the sequence level (30, 34). This has the advantage of being independent of the state of the organism (viable or nonviable) and has resulted in the reclassification of some organisms (24). In addition to the reproducibility of PCR, it is extremely sensitive, requiring only small numbers of organisms for analysis. This sensitivity has been exploited as the basis for a number of tests, including the detection of pathogens (4, 15, 16, 21, 24) and the determination of mechanisms of resistance to specific therapeutic agents (8, 33, 35). The reported sensitivity of the technique varies, but the detection by PCR of single organisms or the DNA equivalent to a single organisms has been reported (3). Nested PCRs are particularly useful in situations where a high level of sensitivity is required, as is the case with ocular infections. Use of nested PCRs in a clinical setting has been hampered by the frequent incidence of false-positive results, but techniques have been developed that eliminate this problem (6, 9, 11).

Endophthalmitis is a term referring to severe intraocular inflammation centered around the vitreous cavity and/or anterior chamber of the eye and may be of infectious origin (caused by bacteria or fungi). The challenges presented by this condition to the clinician are considerable, as the severity of the clinical signs varies greatly according to the time to presentation, the inoculum size, and the species of the infecting organism(s) (18, 28). Also, low-grade infections can be difficult to distinguish from purely inflammatory ocular disease. Ideally, all cases of infectious endophthalmitis would be culture proven, but the number of culture-proven cases with typical signs of infectious endophthalmitis varies greatly from center to center (2, 18, 28). It is important to establish a diagnosis and identify the infecting organism, not only because this decides the further management of the patient but also because it justifies the treatment given. Confirmation of the diagnosis is made more difficult by the small volumes of the ocular samples available for analysis (aqueous, 100 to 150 µl; vitreous, 200 to 400 µl). The numbers of organisms required to establish an infection can also be small and may be as low as 14 (31), and often only a few colonies are cultured by routine microbiological methods (usually 40 to 50 CFU). A delay of 24 to 48 h is usual for routine microbiological processing of the specimens, although it may take up to 12 days in the case of fastidious organisms (32). In the absence of a definitive identification of the causal organism, the clinician must commence therapy on an empirical basis, using broad-spectrum antimicrobial agents, because a delay in treatment is often associated with a worse clinical outcome (12).

Clinical cases which are culture negative and respond to antibiotic therapy are considered infectious despite the lack of definitive culture identification. Cultures prove to be negative for a variety of reasons, such as small sample size, sequestration of bacteria on solid surfaces (e.g. intraocular lens, lens remnants, and capsule) leading to low numbers in the liquid sample, the use of antibiotics prior to sampling, and the fastidious nature of some of the organisms which cause intraocular infection (6, 22, 28). The use of molecular techniques has, therefore, been investigated in order to improve the diagnostic rate and reduce the time to diagnosis. This paper describes an integrated protocol describing the direct detection in ocular fluids of pathogens with suspected infective pathology. A nested PCR approach was developed in which primers based on the conserved bacterial 16S rRNA gene sequences were used in the first round of amplification, while a second round of amplification was able to differentiate between gram-positive and -negative pathogens.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Patient sampling. Intraocular (aqueous and vitreous) sampling was undertaken under sterile conditions. Aqueous sampling was undertaken under topical anesthesia, using a 27-gauge (0.33-mm-diameter) needle, and 100 to 200 µl was aspirated. Vitreous sampling was undertaken as a biopsy through the pars plana. After subconjunctival injection of anesthetic, a vitreous tap was performed using a 23-gauge needle which was inserted through the pars plana 3 mm behind the limbus in aphakic eyes and 4 mm behind the limbus in phakic eyes. A total of 200 to 400 µl of vitreous was aspirated.

Microbiological assessment. One drop of vitreous was smeared on a slide for Gram and periodic acid-Schiff staining, and the remainder was immediately plated on blood and Sabouraud agar before transport to the microbiology laboratory. Plates were incubated under aerobic conditions at 37°C. The cultures were transferred to a 30°C incubator if no growth was apparent after 48 h. In experiments where live bacteria were spiked into PCRs, bacteria were streaked out on blood agar (Biomeriux, Basingstoke, United Kingdom) and isolated colonies were inoculated into 3 ml of brain heart infusion (Biomeriux). A serial 10-fold dilution of overnight cultures was prepared in maximum recovery diluent (Oxoid, Basingstoke, United Kingdom), and aliquots were plated in duplicate for enumeration. Aliquots (5 µl) of bacterial suspensions were used for PCR.

Bacterial isolates used in this study. Following isolation by culture, bacteria were identified using the API biochemical identification system (API Analytab products, Division of Sherwood Medical, New York, N.Y.). A total of 40 strains of 14 bacterial species were tested (see Table 2). All strains were standard NCTC strains (Public Health Laboratory Service, National Collection of Type Culture, Colindale, London, United Kingdom). Individual strains were stored on beads at 70°C (Mast Diagnostics, Bootle, Merseyside, United Kingdom) and subsequently cultured on standard media according to the manufacturers' instructions.

Primer design. The Gram stain-specific primers were designed by creating consensus sequences of a range of common ocular pathogens according to their Gram stain classification and comparing them. The sequences of all primers used in this study are given in Table 1. The gram-positive primer was located between bases 712 and 729 with respect to the sense strand of the Escherichia coli rRNA gene sequence and differed from the gram-negative consensus along its length at 5 of its 18 nucleotides, with a 3-nucleotide mismatch at the 3' end. Similarly, the gram-negative-organism-specific primer differed from the gram-positive-organism-specific consensus at 8 of its 15 nucleotides, with a 4-nucleotide mismatch at the 3' end but was located on the antisense strand. The primers were designed such that differently sized products would be generated, to facilitate an unambiguous assignment of Gram stain classification. The gram-negative-organism-specific PCR resulted in a single 985-bp amplification product, and the multiplex PCR resulted in two PCR products: a pan-bacterial 1025-bp amplicon and a 355-bp product which was specific to gram-positive bacteria.

Nested PCR. Bacterial DNA was extracted using glass beads and alcohol precipitation, as previously described (8). Taq (AmpliTaq LD; Perkin Elmer, Warrington, Cheshire, United Kingdom) for the first round of PCR was pretreated according to the method of Carroll et al. (8). Briefly, prior to PCR amplification the water, buffer, magnesium chloride, and Taq components were mixed and incubated for 30 min at 37°C with 1.0 U of Sau3A1 (Boehringer Mannheim, Lewes, East Sussex, United Kingdom) per U of Taq polymerase. The restriction enzyme was subsequently inactivated by incubation at 95°C for 2 min, following which the deoxynucleoside triphosphates (dNTPs), primers, and template DNA were added and PCR amplification was commenced. Taq for the second round of amplification was used without pretreatment. PCRs were carried out in the proprietary buffers and for the first round of amplification contained a 60 µM concentration of each deoxynucleoside triphosphate (Pharmacia, Little Chalfont, Buckinghamshire, United Kingdom), 3.0 mM Mg²⁺, 2.5 pmol of each of the primers 16SF and 16SR, and 1 U of Taq DNA polymerase in a total volume of 25 µl. The initial denaturation was carried out for 5 min at 94°C, and cycling was performed as follows: 94°C for 10 s, 54.2°C for 10 s, and 72°C for 15 s for 30 cycles (Genius Thermal Cycler; TECHNE, Cambridge, United Kingdom). A second round of amplification used 1 µl of product from the first round, and a Mg²⁺ concentration of 2.5 mM. PCRs specific for gram-negative organisms utilized 5 pmol each of primers NF and N6R. A multiplex PCR which simultaneously detected all species of bacteria and all gram-positive bacteria used 5 pmol each of P2F and NR and 1 pmol of NF. Denaturation was carried out for 5 min at 94°C and cycling was performed at 94°C for 7 s, 60°C for 7 s, and 72°C for 10 s for 30 cycles. Multiple reagent controls from the first round were always subjected to a second round of amplification to control for contamination.

PCR of vitreous and aqueous samples. Samples of vitreous and aqueous humors were received either in sterile tubes which had been sealed in the operating theater or in the syringes used to obtain the sample, after the requirements of the routine diagnostic microbiological service had been fulfilled. Aliquots (5 µl) of vitreous and aqueous humors, either neat or diluted 1/10 with sterile water were used in each PCR after they had been heated to 95°C for 2 min to extract the DNA. Samples were subjected to PCR in duplicate. Positive controls containing 10 ng each of E. coli and Staphylococcus aureus DNA were run for each PCR in neat and diluted vitreous and aqueous humors to check for inhibition of the PCRs by the vitreous. Multiple reagent controls were subjected to two rounds of PCR to control for contamination of reagents.

Electrophoresis and imaging. Following PCR amplification, products were resolved on a 1% agarose-Tris-acetate-EDTA gel and visualized using ethidium bromide under UV illumination, and results were recorded using the UVP Ltd. (Cambridge, United Kingdom) gel documentation system.

Sequencing of PCR products. PCR products were electrophoresed on 1% agarose gels, and the bands were excised, extracted using the Qiaquick Gel extraction kit (Qiagen, Crawley, West Sussex, United Kingdom), and sequenced using the fluorescent dye terminator sequencing system (ABI). Sequences were submitted for BLAST searching for similarity to other sequences (1). Consensus sequences were generated and compared using DNASTAR (Madison, Wis.) software.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The sensitivity and specificity of the Gram stain-specific primer pairs was evaluated on a range of common pathogens and is detailed in Table 2. The sensitivity of the gram-negative-organism-specific primers was 10 fg of DNA per reaction in all of the species tested. In contrast, there was a wide variation in the sensitivity of the gram-positive-organism-specific primer pair, from 100 fg to 1 pg, reflecting the broad genotypic and phenotypic diversity of this group. A multiplex PCR was developed using the primers NF-NR and P2F. The sensitivity of this PCR was identical to that observed for individual PCRs. Examples of these PCRs carried out with serial dilutions are shown in Fig. 1. None of these primer sets amplified human lymphocyte DNA or genomic DNA from Candida albicans or Aspergillus fumigatus under the conditions tested. In the multiplex PCR the 1,025-bp amplicon is the product of the primer pair NF-NR, which are both universal bacterial primers, while the 355-bp amplicon is specific for gram-positive bacteria. The product of the primer pair NF-N6R that is specific for gram-negative bacteria is 985 bp. Evaluation of the potential of these primers to amplify DNA from whole bacteria was undertaken by spiking various numbers of bacteria into water, 5 µl of which was used in the PCRs. The limits of detection (number of organisms) of the primer pairs for E. coli, Pseudomonas aeruginosa, S. aureus, and Streptococcus pyogenes were 5, 24, 4, and 4, respectively, and the primers were capable of detecting between 8 × 10² and 4.8 × 10³ organisms per ml. The multiplex and gram-negative-organism-specific PCRs were applied to four samples of intraocular fluid with suspected infective pathology and two samples from an eye with intraocular inflammation as a control. A comparison was made between the results obtained by Gram staining, culture, and PCR, and a summary is given in Table 3. In all culture-positive samples the results of PCR and culture were 100% concordant. Also, the results of subsequent DNA sequencing matched the identity of the bacterium as isolated by culture. Although in this study PCR was applied to these samples retrospectively, a definitive result could have been reported 3.5 h after receipt of the sample.

View larger version (21K): [in this window] [in a new window]   FIG. 1.   Sensitivities of the primer sets were evaluated using dilutions of DNA in water. (A) Multiplex PCR in which the 1,025-bp amplicon is the product of the NF-NR primers (universal bacterial primers) and the 355-bp amplicon is the product of the P2F-NR primers (specific for gram-positive bacteria). The template DNA was S. aureus NCTC 8532. Lane 1, 10 ng of DNA; lane 2, 1 ng of DNA; lane 3, 100 pg of DNA; lane 4, 10 pg of DNA; lane 5, 1 pg of DNA; lane 6, 100 fg of DNA; lane 7, 10 fg of DNA; lane 8, DNA ladder; lane 9, negative control. (B) Gram-negative-organism-specific PCR in which the 985-bp amplicon is the product of the primers NF-N6R. The template DNA was E. coli NCTC 10418. Lane 1, 10 ng of DNA; lane 2, 1 ng of DNA; lane 3, 100 pg of DNA; lane 4, 10 pg of DNA; lane 5, 1 pg of DNA; lane 6, 100 fg of DNA; lane 7, 10 fg of DNA; lane 8, DNA ladder; lane 9, negative control.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The detection by PCR of bacterial DNA from body sites which are normally sterile, has been used to improve the rate of microbiological diagnosis for cerebrospinal fluid, synovial fluid, and vitreous (15, 19, 26). This study has confirmed the usefulness of molecular techniques in establishing the presence of infection and has further developed them by determining the Gram stain status of the infecting bacterium. These techniques were also able to confirm the presence of bacteria in a patient with culture-negative endophthalmitis, who demonstrated a clinical history and signs highly suggestive of an infective etiology and who responded well to antibiotic therapy, thereby providing further evidence of the infective etiology of the condition. Samples from a patient with a case of chronic intraocular inflammation served as controls and remained PCR negative. Gram-positive organisms are isolated from intraocular samples in 58 to 96% of cases, e.g., Staphylococcus spp. (coagulase-negative staphylococci and S. aureus), Streptococcus spp., Bacillus cereus, and Propionibacterium acnes (13, 14, 17, 18, 23). Gram-negative organisms account for a smaller percentage of culture positive cases, comprising 4 to 29% in different studies (5, 13, 14, 17, 18, 20, 23). Gram-negative organisms typically isolated from ocular infections include E. coli, Proteus mirabilis, Serratia marcescens, Klebsiella pneumoniae, Haemophilus influenzae, and P. aeruginosa. Initial treatment of patients presenting with presumed bacterial endophthalmitis is aided by the Gram staining of samples and is guided by the results of this rapid test. Compared to infection with gram-positive bacteria, infections with gram-negative bacteria are associated with a greater inflammatory response and poorer visual prognosis: a reflection of the toxins produced and the greater virulence of these organisms (20). The Gram stain status of the infecting bacterium is, therefore, important because it allows targeted antimicrobial therapy in the later stages of management and has implications for prognosis and final visual outcome. In the clinical setting, however, this test is usually negative (no organisms seen) and, therefore, is only undertaken when sufficient sample is available for the full array of culture media to be inoculated. PCR techniques, on the other hand, only require very small amounts of clinical sample (5 µl) and are not only rapid but sensitive and efficient in allowing a diagnosis of infection to be made. A prospective study with larger numbers of clinical samples would be useful and is now required.

The PCR protocol described in this paper incorporates a number of safeguards, such that a result can be reported with certainty. The pretreatment of the Taq DNA polymerase ensures that false positives due to intrinsic contamination of the enzyme are avoided. The use of both neat and 1/10 dilutions of the intraocular fluid for analysis, as well as for positive controls, ensures that a negative PCR result is not due to inhibition by components of the aqueous and vitreous. Inhibition of PCR by ocular fluids has been reported by Wiedbrauk et al. (36) and was observed in this study (Fig. 2). The effects of routine dilution on all samples were not tested but were found to be required in the analysis of 44% of samples in our subsequent studies (29; N. Okhravi, P. Adamson, A. Dunlop, H. M. A. Towler, M. M. Matheson, and S. Lightman, unpublished data). As the inhibition of the reaction was overcome by diluting the clinical sample in all cases, further studies to elucidate the nature of these inhibitory factors were not undertaken. Aqueous samples were found to require dilution more frequently than vitreous samples; therefore, one can assume the inhibitory factor(s) is present to a greater degree in the former (29; Okhravi et al., unpublished data).

View larger version (27K): [in this window] [in a new window]   FIG. 2.   Results of PCR from a clinical case of culture-positive bacterial endophthalmitis secondary to gram-positive bacteria. The vitreous sample was subjected to PCR as described in the text and electrophoresed on a 1% agarose-TAE gel. Patient sample PCR results appear in duplicate (lanes 3 to 6). Lane 1, multiplex PCR of gram-negative DNA in water (positive control); lane 2, multiplex PCR of gram-positive DNA in neat vitreous (positive control); lanes 3 and 4, multiplex PCR of patient sample (vitreous, diluted 1/10); lanes 5 and 6, multiplex PCR of patient sample (neat vitreous); lane 7, gram-negative-organism-specific PCR of patient sample (vitreous, diluted 1/10); lane 8, Gram negative PCR of gram-negative DNA in water (positive control); lane 9, multiplex PCR of gram-negative DNA in vitreous (positive control); lane 10, multiplex PCR of gram-negative DNA in water (positive control); lane 11, gram-negative-organism-specific PCR of gram-negative DNA in vitreous; lane 12, gram-negative-organism-specific PCR of gram-negative DNA in water; lane 13, DNA ladder; lane 14, negative control (water only, no vitreous); lane 15, negative control (vitreous and water).

As the sensitivity of the primers varied with each bacterial species it was not possible, due to the limited supply of ocular sample, to test the sensitivity of each reaction with ocular fluids as well as water. However, other studies in our laboratory have demonstrated that the sensitivity in water was the same as that in vitreous as long as two rounds of PCR were used (29). Although the PCR protocol developed in this study was developed specifically for ocular samples, it has the potential to be used in other clinical settings where only small volumes of clinical samples are available and a high degree of sensitivity is required.