Ocular Pathogen or Commensal: A PCR-Based Study of Surface Bacterial Flora in Normal and Dry Eyes

Abstract

purpose. To compare the bacterial population of the ocular surface of normal and dry eye subjects using conventional culture and 16S rDNA PCR.

methods. Ninety-one subjects were classified as normal (n = 57) or dry eye (n = 34) by using tear break-up time, McMonnies survey, goblet cell density, and meibomian gland assessment. Conventional bacterial culture and broad-range 16S rDNA PCR, cloning, and DNA sequencing were used for bacterial identification. Repeated sampling was performed in a subset of subjects over a 3-month period. The association between goblet cell loss and bacterial counts in a subgroup of subjects was assessed.

results. Most of the bacteria identified by culture were coagulase negative staphylococci, whereas molecular methods demonstrated a considerable number of additional bacteria. Atypical ocular surface bacteria including Rhodococcus erythropolis, Klebsiella oxytoca, and Erwinia sp., were identified in cases of overt inflammation and, surprisingly, on the normal ocular surface. The same bacteria remained on the ocular surface after repeated sampling. Increased bacterial flora was associated with reduced goblet cell density.

conclusions. Molecular analysis revealed a diverse ocular surface bacterial population. In addition to the normal flora, various potentially pathogenic bacteria were identified. The detection of known pathogens in both normal and dry eyes, with minimal signs of infection, presents a diagnostic dilemma. It remains unknown whether their presence is associated with inflammation and reduced goblet cell density or whether they adversely affect the ocular surface predisposing it to abnormal microbial colonization. In the absence of overt clinical infection, it is unknown whether such results should prompt intervention with therapy.

The ocular surface of healthy individuals inherently supports a small population of bacteria, typically coagulase-negative staphylococci (CNS) which are believed to exist as commensals on the mucosa and lid margins. ^{1 2} Under ideal conditions, there is little or no opportunistic bacterial colonization of the conjunctiva or cornea, because of the washing effect of the tears ^{3 4} in conjunction with the action of antibacterial proteins and enzymes within the tear film. ^{5 6 7 8 9}  

Dry eye, due to tear deficiency or excessive tear evaporation, ¹⁰ is often associated with ocular surface conditions such as anterior blepharitis, ^{2 11 12 13 14} meibomian gland disease, ^{2 15 16} keratitis, ^{17 18} and ocular rosacea, ¹⁹ where alterations in the concentration and type of bacteria present have been reported, independent of the presence of conjunctivitis. 

Such disorders, among others, have been associated with several Gram-positive and -negative bacteria, including CNS, ^{5 12 14} Staphylococcus aureus, ^{11 20} Streptococcus sp., Bacillus subtilis, ^{16 21} Rhodococcus sp., ^{22 23} Pseudomonas aeruginosa, ^{17 24} Haemophilus influenzae, Haemophilus aegyptius, ^{16 21} and Klebsiella sp. ^{25 26 27}  

The production of lipases and toxins by many of these colonizing bacteria may induce ocular surface cellular damage and destabilization of the lipid layer of the tear film contributing to tear film instability, inflammation, and symptoms of significant ocular irritation. ^{28 29} Similar symptoms commonly occur in dry eye, without evidence of purulent exudative infection. 

Oral antibiotic therapy has been associated with improved dry eye symptoms, which may be related to a reduction in bacterial counts or bacterial enzymes. ¹³ Therefore, it is reasonable to propose that there may be an important relationship between ocular surface bacteria, tear film function, and ocular surface inflammation. 

To date, investigation of ocular surface microbial flora has relied principally on the application of conventional culture techniques. ^{5 12 16 30 31} However, it is recognized that such methods may frequently be unreliable, due to specific growth requirements of some bacteria along with the often limited size of available samples. ^{32 33} Previous studies have demonstrated the effectiveness of molecular methods in overcoming such obstacles for the detection and identification of single and mixed populations of bacteria in a given sample and as such, are being increasingly used to investigate bacteria of significant clinical importance. ^{19 32 33 34} Whereas Schabereiter-Gurtner et al. ³² applied conventional 16S rDNA PCR in investigating the presence of unculturable ocular bacteria, the study cohort consisted mainly of cases of purulent conjunctivitis, in which large bacterial loads would be expected. 

The purpose of this study was to investigate the normal ocular surface bacterial flora and assess how it varies in dry eye. Bacterial identification in conjunctival swab and impression cytology (IC) samples was assessed by conventional culture techniques and 16S rDNA–based PCR with cloning and DNA sequencing. 

Methods

Study Participants and Sample Collection

The nature of the study and procedures involved were fully explained to all participants, and informed consent was obtained before recruitment. The use of volunteers complied with the tenets of the Declaration of Helsinki, and institutional ethics committee approval was obtained. A total of 91 subjects (34 dry eye and 57 normal healthy subjects) were recruited to the study through advertising posters, personal communication, and clinical referrals of preoperative patients with presenting signs and symptoms of dry eye. Exclusion criteria included pregnancy, use of oral or topical antibiotics at the time of the study and 3 months before testing, use of prescribed eye medication, active ocular infection, or conjunctivitis. The study cohort comprised a population from both rural and urban areas, and no subject had spent any significant period in a hospital setting. 

Diagnosis of Dry Eye

All dry eye diagnosis was performed by determination of tear break-up time (TBUT) detected by fluorescein, ³⁵ McMonnies dry eye questionnaire, ³⁶ reduced goblet cell density, biomicroscopic examination of the ocular surface and meibomian glands, ³⁷ and presenting dry eye signs and symptoms. The sequence of testing was as follows: external eye examination; biomicroscopic examination of meibomian glands, lids, lid margins, conjunctiva, and tear film; conjunctival IC specimen; and TBUT with fluorescein. The pattern of testing remained constant for all subjects; one ophthalmologist and one investigator performed all examinations and measurements throughout the study. The demographics of the study subjects are presented in Table 1 . 

Assessment of Dry Eye Symptoms

An interview with an ophthalmologist was conducted to ascertain the presence, type, and frequency of symptoms in each subject. A detailed medication and general health history was documented specifically to determine whether any subject was taking medication or had any clinical condition that might have influenced tear production or stability. 

The study cohort did not include any subject with Sjögren’s syndrome, arthritis, or diabetes, and no subjects were taking any medications, which may have influenced the presence of dry eye. All subjects completed a McMonnies dry-eye–based questionnaire ³⁶ with a score of ≥14, ^{38 39 40 41 42 43} indicating the presence of dry eye as previously recommended by Albietz and Bruce, ³⁸ Gullion and Maissa, ³⁹ and Johnson and Murphy. ⁴⁰  

Meibomian Gland and Ocular Surface Grading

A biomicroscopic examination was performed at a slit lamp to assess the presence of meibomian gland disease, lid erythema and swelling, conjunctiva erythema, and edema and tear film debris in both eyes. Grading was categorized according to the criteria laid down by the Allergen Restasis study group (personal communication, 2005; Allergen, Irvine, CA) and by Foulks and Bron. ³⁷ Grade 0 indicated a normal ocular surface with no meibomian gland blockage; grade 1 indicated plugging (blockage) of one or two glands; grade 3 indicated that one to three glands were blocked; and grade 4 was allocated to subjects with plugging of three or more meibomian glands. Grades 3 or above were regarded as positive for the presence of meibomian gland dysfunction (MGD) and ocular surface abnormality. 

Assessment of Conjunctival Goblet Cells

Conjunctival epithelial and goblet cells were harvested with 5 × 8-mm strips of 0.22-μm cellulose acetate filter paper (Biopore; Millipore Ltd., Watford, UK). Subsequent to periodic-acid Schiff (PAS) staining, goblet cell density (GCD) was graded as previously described by Anshu Munshi et al. ⁴⁴ An estimation of the nucleocytoplasm ratio was noted, and cytologic grading was performed according to criteria laid down by Saini et al. ⁴⁵ Grade 1 indicated the presence of >30 goblet cells/4 high-power fields (HPFs), with small, round epithelial cells having a nucleus-to-cytoplasm ratio of 1:2. Grade 2 indicated a large cell sheet consisting of larger polygonal epithelial cells, with a decreased nucleocytoplasmic ratio of 1:3 and the presence of 15 to 30 goblet cells/4 HPF. Grade 3 was assigned to samples demonstrating 5 to 15 goblet cells/4 HPF with epithelial cells displaying a further decrease in nucleus-to-cytoplasm ratio. Grade 4 was given if less than five goblet cells/four HPFs were present and in cases in which large epithelial cells with pyknotic nuclei were visible. For the purposes of this study, grades 3 and 4 were classed as indicative of a reduced GCD. 

Tear Film Beak-Up Time

Fluorescein sodium was instilled in the inferior conjunctiva of both eyes by using sterile paper strips (Florets; Chauvin Pharmaceuticals Ltd., Romford, UK) moistened with a drop of sterile saline. The subject was asked to blink several times and TBUT was determined by measuring the time lapse in seconds between the instillation of the fluorescein and the appearance of the first dry spots on the cornea, visualized with the cobalt blue filter of the slit lamp. The mean of three successive readings was taken as the overall TBUT, and a cutoff of ≤7 seconds was an indication of tear film instability. 

Subjects were considered to have dry eye if a positive McMonnies dry eye symptom survey score of ≥14, in addition to a positive score in one or more other test (TBUT, meibomian glands, or GCD), was noted in at least one eye. 

Sampling of Ocular Surface Bacteria

Sample collection was performed in a clean ophthalmic consulting room, and the examining ophthalmologist wore sterile gloves to minimize contamination of test samples by foreign bacteria that may have been present in the surrounding environment. In addition, negative control swab and IC paper samples were taken at the time and place of subject testing, to confirm the lack of environmental contamination. Each control was subsequently subjected to the same bacterial analysis as the test samples. 

Ocular specimens for bacterial analysis were collected concurrently from healthy control subjects and patients with dry eye by using sterile cotton swabs (Bibby Sterilin Ltd., Stone, UK) and sterile IC filters (Millipore UK, Ltd.). Culture swab samples were taken from the posterior lid margin and lower conjunctival sac before being placed directly into a sterile swab holder containing Stuart’s transport medium. 

After instillation of topical anesthetic, swabs for PCR analysis were cut into sterile DNase- and RNase-free nucleic acid extraction tubes by using sterile scissors. Duplicate swabs were randomly processed from individual subjects as control samples during the DNA extraction process. 

Conjunctival cytology impressions were obtained with 0.22 μm sterile cellulose acetate filters (∼5 × 8 mm; Millipore UK, Ltd.) at the slit lamp by gently pressing the filters onto the bulbar conjunctiva with forceps, and placing the filters into a sterile, DNase- and RNase-free nucleic acid–extraction tube. 

Swab samples for molecular analysis were collected from one eye, and IC specimens for molecular analysis were taken from the other eye. Repeat sampling of conjunctival swabs over a 3-month period was performed for a cohort of five subjects with dry eye and four normal subjects. 

Microbial Culture

Each conjunctival culture sample was plated onto one chocolate agar and blood agar, supplemented with horse blood (Oxoid Ltd., Basingstoke, UK) within 2 hours of sampling and incubated at 35°C in 5% (vol/vol) CO₂ and at 37°C in aerobic and anaerobic conditions, respectively. Positive bacterial cultures were identified by Gram stain and a Staphylococcus identification system (API; bioMerieux, Marcy l’Etoile, France), in which samples were deemed to be culture negative when no growth was observed after 48 hours. 

DNA Extraction

Bacterial genomic DNA was extracted directly from swab and IC samples within 24 hours of sampling (FastDNA Spin Kits for Soil; BIO101; Anachem Ltd., Bedfordshire, UK), according to the manufacturer’s instructions. Positive extraction control samples of Staphylococcus epidermidis isolated from culture swabs and negative extraction control samples in nuclease-free water (NFW) were included in all experiments. 

16S rDNA Amplification

Broad-range 16S rDNA PCR was used for the detection of any bacterial DNA present in individual samples, without any prior cultivation. Amplification of highly conserved regions of the 16S rDNA gene was facilitated using 0.1 μM (each) universal 16S rDNA broad range primers (Table 2) ; 200 μM dATP, dGTP, dCTP, and dTTP; 10 mM Tris-HCl; 2.5 mM MgCl₂; 1.25 units Taq DNA polymerase (Amplitaq; Applied Biosystems [ABI], Warrington, UK) and 4 μL of DNA template were made up to a final volume of 50 μL in NFW. PCR was performed in a thermocycler (model 2400; Perkin-Elmer, Wellesley, MA): 96°C for 5 minutes followed by 40 cycles of 96°C for 1 minute, 55°C for 1 minute, and 72°C for 1 minute, followed by a final extension at 72°C for 10 minutes. A positive (S. epidermidis genomic DNA) and negative control (NFW) were included in each PCR run. Broad-range 16S rDNA primers and amplicon sizes used to determine the presence of any bacteria are shown in Table 2 . Randomly selected PCR products were assessed by electrophoresis on 2% (wt/vol) agarose gels independently on different days, to assess the possibility of false-positive or -negative PCR results. To minimize contamination, all extractions and PCR reactions were prepared in a category 2 PCR hood, ⁵⁰ by using sterile filter tips and instrumentation pretreated with UVA light. 

DNA Sequence Analysis

Sequencing of the amplified 16S rDNA was subsequently performed for identification of any individual bacteria. Positive PCR amplicons were purified (QIAquick PCR Purification Kit; Qiagen Ltd., Crawley, UK) and subsequently prepared for automated sequencing (Big Dye Terminator ver. 3.1 cycle sequencing kit; ABI), as described by the manufacturer. PCR products were cycle sequenced in both directions and analyzed on a sequencer (model 3100 Genetic Analyser; ABI). The resultant sequence chromatograms were assessed (Chromas software; Technelysium Pty. Ltd., Tewantin, Queensland, Australia), and DNA sequences were interpreted by comparing initial sequences with those stored in the GenBank Data System using the basic local alignment software tool (BLAST; National Center for Biotechnology Information, Bethesda, MD). ⁵¹ DNA sequences displaying >97% shared identity with known sequences in the database were assigned to that species. 

Cloning of 16S rDNA Amplicons

Cloning was performed for the identification of individual bacteria in samples containing a mixed population. Sequencing traces of the 16S rDNA gene displaying mixed populations were cloned into a vector (pGEM-T with the pGEM-T Easy Vector System; Promega, Southampton, UK). Identification of individual cloned inserts was achieved through sequencing of 16S rDNA gene fragments using vector-specific pUC/M13 primers forward, (5′-GGC GGC CGC GGG AAT TCG ATT-3′) and reverse (5′-GCC GCG AAT TCA CTA GTG ATT-3′) (Promega) and the same cycling conditions as described earlier. 

Statistical Analysis

Statistical significance for comparison of different groups was performed with the Fisher exact test. The relationship between swab and IC samples analyzed by PCR was investigated by using the McNemar variation of the χ² test. The difference between GCD and mean bacterial count was analyzed by ANOVA and the Tukey post hoc test (Minitab, State College, PA). 

Results

Microbiology

Conventional culture was used to assess the presence and identification of bacteria in the conjunctival sac (Table 3) . The bacteria were quantified by counting the number of colony forming units (CFU) visible. Because of unforeseen complications in the transportation of culture samples from 11 subjects, the results could not be accurately determined and are therefore not included in the results. 

In total, 83% (n = 67) of all conjunctival swab samples subjected to routine microbial analysis (n = 80) produced a positive result. In normal subjects, 75% (n = 37) were positive for culture, and 97% (n = 30) were positive in subjects with dry eye (DE). By culture, CNS species (S. epidermidis, S. lentus, S. xylosus, and S. sauri) were the most frequently isolated bacteria, the most predominant being S. epidermidis, found in 100% of all positive samples. S. lentus was identified in a small number of swabs taken from both normal and DE subjects. S. xylosus was cultured only in a sample from a normal subject while S. sauri was present only in a subject with DE. No bacterial species, other than CNS, were identified from any normal conjunctival swab samples. Bacillus was the only bacteria identified, cultured from a single DE swab sample. 

Conventional culture demonstrated a general increase in the overall mean number of bacteria present in samples taken from DE versus normal subjects (26 vs. 18 CFU). All negative control samples produced a negative result, with no growth observed after 48 hours. 

PCR Amplification of 16S rDNA

DNA extracted from 109 conjunctival swabs and 90 conjunctival IC samples was tested for the presence of bacterial DNA by conventional broad-range 16S rDNA PCR. Table 4provides an overview of results for positive swab and IC samples in normal and DE subjects. Sixty-four samples were regarded as PCR negative, with no bands visible on ethidium bromide–stained agarose gels. No bands were detected in all negative control swab and IC samples. As expected, no samples that were culture positive gave negative results by PCR. 

Based on the Fisher exact test, the number of PCR-positive IC samples between normal and DE patients is not significantly different (P = 0.13). There was also no statistically significant difference in the number of positive PCR swabs between normal and DE subjects, as determined by the Fisher exact test (P = 0.28). 

There was no evidence of a significant lack of correlation between the two proportions (75% and 65%) of positive swab and IC samples, as determined by the McNemar test on 72 matched-paired subjects within our study cohort. 

Comparison of Culture versus PCR and DNA Sequencing

Overall, a considerable difference was noted in the increased diversity of bacterial populations determined by DNA sequencing compared with cultivation (Table 5) . Table 5displays a comparison of the overall bacterial genera identified by routine culture methods with those identified through direct DNA sequencing and sequencing of cloned 16S rDNA inserts in all conjunctival samples. GenBank sequence accession numbers for the sequences in this study are AY692487 and DQ972937 through DQ972950 (http://www.ncbi.nlm.nih.gov/Genbank; provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD.). 

Initial direct sequence chromatographs for 19 normal and 14 DE subjects indicated the presence of mixed bacterial populations. Twenty-five PCR products from 8 DE and 8 normal subjects were selected for cloning, and 70 cloned inserts were subsequently sequenced in both directions. Of the 70 insertsselected, the following bacterial genera were identified: S. epidermidis (n = 3), Staphylococcus sp. (n = 6), uncultured bacterium (n = 23), R. erythropolis (n = 14), Propionibacterium (n = 2), Corynebacterium sp. (n = 2), Erwinia sp. (n = 18), Klebsiella oxytoca (n = 1), and Klebsiella sp. (n = 1). Most of the bacteria identified by DNA sequencing were detected in normal and DE subjects (Table 6) . However, certain bacteria were found in samples from DE subjects only, including Bacillus sp., Propionbacterium acnes, and K. oxytoca. In addition Erwinia sp. was identified only in sequences of DNA from healthy eyes. 

All bacteria initially identified through molecular analysis were subsequently detected after repeat sampling of nine subjects (n = 5 DE, n = 4 N) over a 3-month period, demonstrating that the same bacterial species remained on the ocular surface of these subjects. 

A subgroup of 27 subjects with different goblet cell densities (n = 9 goblet cell grade 2, n = 9 goblet cell grade 3, n = 9 goblet cell grade 4) were selected for a preliminary study to determine whether there was an association between the level of ocular surface bacteria and GCD. Grade 2 was allocated to normal subjects only (Fig. 1A) . The results indicated a trend of increased bacteria quantity, as determined by culture, with a reduction in GCD. Bacterial levels (CFU/swab) were higher overall in subjects with grades 3 or 4 than in those of normal control subjects. A statistically significant difference (P = 0.005) was noted between the mean bacterial counts in the three groups, as determined by one-way ANOVA. The Tukey post hoc test for individual pair-wise comparisons demonstrated a statistically significant difference in bacterial counts only between the control group and grade 4 (Fig. 1B) . 

Discussion

The microbial flora of the ocular surface has commonly been determined through conventional culture techniques. Such methodology, as used in the present study, detected an overall higher level of bacteria in patients with DE than in normal subjects. All samples deemed positive by culture were identified as CNS, with only a single sample displaying two different bacterial genera. These findings concurred with previous studies identifying CNS to be the predominant isolate cultured from healthy eyes. ^{1 5 31}  

A much greater diversity of bacteria was detected when molecular cloning and DNA sequencing were used in parallel samples. This methodology identified several additional common ocular bacteria (Corynebacterium and Propionibacterium) as well as some not typically associated with the normal ocular surface, previously linked with disease states such as endophthalmitis ⁵² and keratitis. ^{22 53} Such bacteria included R. erythropolis, ⁵² Klebsiella sp., and Erwinia sp. ⁵⁴  

One particular organism, Rhodococcus, was found in both normal subjects and those with disease in keeping with DE symptomatology. Identification of this bacterium in subjects with no apparent ocular symptoms raised the question of whether Rhodococcus is indeed part of the normal ocular flora in certain individuals. However, because it is not commonly associated with the normal ocular flora, it is important that it be suspected as a possible pathogen in a subject with DE. 

Klebsiella, recently associated with contamination of preservative free eye drops ⁵⁵ was identified through sequencing of cloned rDNA fragments in both normal and DE subjects, presenting a diagnostic dilemma for the ophthalmologist regarding whether to instigate treatment in seemingly quiet eyes. 

Repeated sampling after a 3-month period confirmed the presence of Klebsiella in subsequent analysis, indicating that this organism may be consistently present on the conjunctiva of some individuals. This finding suggests that certain bacteria are part of the commensal bacterial flora permanently resident on the lid margins and conjunctival sac of the normal healthy ocular surface of a subset of the population. 

The results of this study highlight that the understanding of bacterial diversity comprising microbial communities in DE is incomplete, if not inaccurate. The use of molecular techniques, which circumvent standard culture methods and often differ from cultivation results, has increased our knowledge of the bacterial diversity of the conjunctiva. Some of the diagnostic difficulties produced through molecular investigation are caused by its extreme sensitivity, allowing the detection of microbes that are present, but may not be implicated in the clinical condition. It is very difficult therefore to propose a therapeutic regimen to treat such organisms in the absence of ocular symptoms. The sensitivity of molecular analysis allows the accurate detection and confirmation of slow-growing, cultivation-resistant bacteria and bacteria with unusual growth requirements, especially in specimens where there is often a limited amount of bacterial sample. Such limitations in culture methodologies could explain in part why, by culture, there were negative results for bacteria other than CNS. 

The detection of known pathogenic bacteria in DE and normal subjects presents a diagnostic dilemma of whether such pathogens can be implicated in causing or exacerbating ocular surface inflammation. Little is known from the literature about the relevance and implications of some of these bacteria in the context of ocular surface microflora and their potential role in conditions such as DE. Their presence alone would perhaps predispose a patient to development of abnormal ocular surface microbial infection or colonization. 

In keeping with previous studies, demonstrating elevated levels of S. epidermidis in patients with blepharitis or keratitis, ^{2 11 56 57} we found an association between the incidence of blepharitis and increased bacterial count in some subjects. The relationship between S. epidermidis, and these conditions may be related to the production of lipolytic enzymes, and the results from this study and previous work suggest a potential association between this group of bacteria and ocular surface conditions. ¹³  

By assessing bacterial load and GCD in a subgroup of 27 subjects, a trend of increasing bacterial count with a decrease in goblet cells was observed. Further studies are ongoing to investigate whether bacterial colonization of the ocular surface alters the number and function of conjunctival goblet cells. This trend has been noted in other bacterial inflammatory conditions, with previous studies demonstrating the depletion of rectal goblet cells after colonization by bacteria after only a few days. ⁵⁸ However, although an increased bacterial population is often associated with the initiation of disease, it is possible that an already compromised environment, such as thinned tear film in DE, facilitates bacterial growth. The existence of a weakened and thinned tear film in DE may allow the infiltration and colonization of the ocular surface by several micro-organisms. 

The heterogeneity of the ocular flora may be modified in certain individuals, whereby bacteria may be introduced to the ocular surface through extraneous means, ranging from increased rubbing of the eyes to the introduction of contaminated eye drops. Contamination is a factor that should be recognized and taken into account when managing patients with DE. In this study, we did not specifically ask subjects whether they rubbed their eyes to determine any association of increased bacterial density, although we observed a large variation in the bacterial quantity between seemingly normal patients, with increased bacteria populations noted in some normal healthy eyes. Although an increased bacterial population at any given site presents the potential for infection, in this particular investigation a lack of association between bacterial count and levels of inflammation was noted in several of the subjects, indicating that the S. epidermidis population remained nonvirulent, providing no stimulus to inflammation. Bacterial regulatory factors and the individual host immune response may influence the exertion of bacterial virulence and subsequent symptoms. Although in certain individuals a stimulus may not elicit an inflammatory reaction, others may show an acute inflammatory response to the same stimulus, perhaps due to intervariations in innate or acquired immunity among individuals. 

However, an increase in bacterial count may induce detrimental effects on normal cellular function through a quorum-sensing mechanism, when the bacterial population reaches a high concentration. ^{59 60} Such a mechanism whereby bacterial pathogenesis is initiated due to an increase in cell–cell signaling in elevated bacterial populations has been demonstrated to occur in other ocular pathogens, including S. aureus and P. aeruginosa. On reaching a certain concentration, these bacteria have been shown to initiate their virulence and overcome the host immune response. ^{59 60} Further studies are currently under way to explore the relationship between quorum-sensing and bacterial expression of virulence factors in subjects with DE, ocular inflammation, and raised levels of bacteria. 

The use of PCR and sequencing may redefine the normal ocular surface microbial flora. The improved sensitivity of these tests over normal culture techniques, however, can be a double-edged sword producing new information of which the clinical relevance is as yet not fully determined. Until we have fully defined the normal ocular flora by using molecular technologies, it will be difficult to determine which bacteria are normal commensals and which may be implicated in ocular surface disease. 

 Supported by the Department for Employment and Learning in Northern Ireland.

 Submitted for publication May 18, 2007; revised July 25, 2007; accepted September 20, 2007.

 Disclosure: J.E. Graham, None; Jon. E. Moore, None; X. Jiru, None; John.E. Moore, None; E.A. Goodall, None; J.S.G. Dooley, None; V.E.A. Hayes, None; D.A. Dartt, None; C.S. Downes, None; T.C.B. Moore, None

 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.

 

Table 1.

 

Results of Dry Eye Evaluation Tests and Demographics of Study Subjects

Subjects	McMonnies Questionnaire ≥14	TBUT ≤7 sec	Meibomian Glands (Grade 3 or 4)	Goblet Cell Grading (Grade 3 or 4)	Gender	Mean Age (y)	Age Range (y)
Control (n = 57)	5	15	3	10	Male (n = 24)	50 ± 21	22–80
					Female (n = 33)	38 ± 17	20–79
Dry eye (n = 34)	34	29	14	18	Male (n = 13)	52 ± 15	22–78
					Female (n = 21)	46 ± 14	22–76

 Age is expressed as the mean ± SD.

Table 2.

 

Universal Primer Sequences Used to Amplify Various Regions of the 16S rDNA Gene, Amplicon Size, and Reference

Primer Name	Sequence	Position on 16S rDNA Gene	Amplicon Size (bp)	Reference
P11P (f)	5′-GAG GAA GGT GGG GAT GAC GT-3′	1174–1193*	216	46
P13P (r)	5′-AGG CCC GGG AAC GTA TTC AC-3′	1389–1370*
PSL (f)	5′-AGG ATT AGA TAC CCT GGT AGT CCA-3′	706–729, †	570	47
XB4 (r)	5′-GTG TGT ACA AGC CCC GGG AAC-3′	1324–304, †
PSL (f)	5′-AGG ATT AGA TAC CCT GGT AGT-3′	706–729*	313	48
PSR (r)	5′-ACT TAA CCC AAC ATC TCA CGA CAC-3′	1019–995*
27 (f)	5′-AGA GTT TGA TCM TGG CTC AG-3′	27–519	492	49
519 (r)	5′-GWA TTA CCG CGG CKG CTG-3′

 f, forward primer; r, reverse primer.

*   Escherichia coli (GenBank Accession No. J01859).

†   Stenotrophomonas maltophilia (GenBank Accession No. AY169434).

Table 3.

 

Overview of Bacterial Cultivation

Diagnosis	Positive Swabs	Mean CFU/Swab	Identified Genera
Normal (n = 49)	37	18 ± 21	S. epidermidis (n = 37)
			S. lentus (n = 2)
			S. xylous (n = 1)
Dry eye (n = 31)	30	26 ± 22	S. epidermidis (n = 30)
			S. lentus (n = 1)
			S. sauri (n = 1)
			Bacillus sp. (n = 1)

 Data are the number of tested swabs. Standard deviations of mean bacterial counts (CFU/swab) are shown. n, number of samples.

Table 4.

 

Overview of Positive 16S rDNA PCR Results for the Detection of Bacterial DNA in Conjunctival Swab and IC Samples

Diagnosis	PCR Positive Swabs	PCR Positive IC	Total Samples PCR Positive
Normal	40/59	28/51	68
Dry eye	39/50	28/39	67

 Data are the number of swab and impression cytology samples in both groups. A greater number of swab and impression cytology samples were positive by PCR in DE subjects than in the normal group.

Table 5.

 

Comparison of Bacterial Genera Identified by Culture or DNA Sequencing in Conjunctival Samples of All Normal and DE Subjects

Bacteria Identified by Culture	Bacteria Identified by DNA Sequencing
Coagulase negative staphylococci	Coagulase negative Staphylococcus sp.
		Staphylococcus epidermidis
Bacillus sp.	Bacillus sp.
	Rhodococcus erythropolis
	Rhodococcus sp.
	Uncultured bacterium
	Corynebacterium sp.,
	Propionibacterium acnes,
	Klebsiella sp.
	Klebsiella oxytoca
	Erwinia sp.

Table 6.

 

Comparison of Bacterial Genera Identified in 16 Normal and 12 DE Subjects Using DNA Sequencing

Normal Controls (n = 16)	Dry Eye Subjects (n = 12)
Coagulase negative Staphylococcus sp. (n = 1)	Coagulase negative Staphylococcus sp. (n = 4)
Staphylococcus epidermidis (n = 12)	Staphylococcus epidermidis (n = 6)
Rhodococcus erythropolis (n = 5)	Rhodococcus erythropolis (n = 2), Rhodococcus sp. (n = 1)
Uncultured bacterium (n = 10)	Uncultured bacterium (n = 8)
Corynebacterium sp., (n = 3)	Corynebacterium sp. (n = 2)
Klebsiella sp. (n = 1)	Klebsiella sp. (n = 2), Klebsiella pneumoniae(n = 1)
	Klebsiella oxytoca (n = 2)
Propionibacterium (n = 3)	Propionibacterium acnes (n = 2)
Bacillus sp. (n = 1)	Bacillus sp. (n = 2)
Erwinia sp. (n = 1)

  n, number of subjects.

Figure 1.

 

(A) Photomicrographs of representative conjunctival IC specimens stained with periodic acid-Schiff. (A) Grade 2: a normal control cytologic profile with a high number of goblet cells present. (B) Grade 3: a reduced number of goblet cells. (C) Grade 4: distinct squamous metaplasia of the conjunctival epithelium and complete absence of goblet cells. (D) Interval plot of the mean bacterial count (with 95% CI) in a subgroup of 27 subjects with different IC grades (control, n = 9, grade 3, n = 9; and grade 4, n = 9). There was a significant difference in the mean bacterial counts (cfu/swab) between the control group and the grade 4 group (P = 0.005). Magnification: (A, B) ×200; (C) ×400.

The authors thank the microbiology staff of the Royal Group Hospitals (Belfast) for their help and technical support.