Phenotypic and Genotypic Characterization of Coagulase Negative Staphylococci (CoNS) Other than Staphylococcus epidermidis Isolated from Ocular Infections

Abstract

Purpose.: The clinical significance of many species belonging to coagulase-negative staphylococci (CoNS) other than Staphylococcus epidermidis has been increasingly recognized. The present study attempted the characterization of non–S. epidermidis CoNS isolates of ocular origin.

Methods.: The isolates were characterized phenotypically by analytical profile index (API) and genotypically by fluorescent amplified fragment length polymorphism (FAFLP). In addition, the presence of genes that are likely to enhance virulence such as biofilm related genes (icaA and icaB) and methicillin resistance (mecA) were detected.

Results.: API identified seven different species with an identification score varying from 44% to 99%. S. haemolyticus and S. xylosus species identified by API showed good API scores when compared with all other species identified. FAFLP generated 12 clusters from all the isolates and appeared to be more discriminatory. API identification results corresponded well with the FAFLP results only in six clusters. Nearly 40% of isolates showed the presence of mecA and icaAB genes.

Conclusions.: Identification results of these two methods corresponded only in 48.4% of the isolates suggesting that the battery of tests using API were not sufficient enough to identify all the species of CoNS. Therefore, it is sensible to rely on two or more identification methods particularly when API species identification has a lower score. FAFLP genotyping appears to be a reliable and rapid alternative identification method for CoNS that can be used in conjunction with any other phenotypic method.

The clinical significance of many species belonging to coagulase-negative staphylococci (CoNS) other than Staphylococcus epidermidis has been increasingly recognized in recent years. ^1,2 They are also indigenous microflora of the skin and mucous membrane of humans. Isolates of CoNS other than S. epidermidis are primarily associated with nosocomial infections such as endocarditis, prosthesis and intravascular catheter-related infections, and a variety of postoperative infections. ¹ Non–S. epidermidis CoNS are also implicated as a common cause for eyelid and corneal infections. ^{3 –5} Apart from being a common component of the normal ocular flora, these bacteria have also been reported to cause chronic blepharitis, conjunctivitis keratitis, and endophthalmitis. ^3,4,6  

Though non–S. epidermidis CoNS are implicated in variety of infections, ocular infections appeared to be most common. ^4,7 Despite their growing importance, there is still limited information available regarding their molecular epidemiology. Accurate species level identification of CoNS is very much required for the precise measurement of impact, sources, and for knowing their transmission dynamics. Phenotypic and genotypic characterization of these isolates is essential to understand the population dynamics of different species of this group. Strain characterization is imperative because many isolates belonging to CoNS also reside as commensals in the extraocular tissues such as lids and conjunctiva. ^1,8 Many techniques have been used to characterize strains of CoNS responsible for blood stream infections (BSI) and other device-related infections to determine the clonality, and also to conclusively distinguish clinically significant organisms from that of commensals or contaminants. ⁹ Such kind of comparative studies are lacking on CoNS isolates of ocular origin. However, Pinna et al. ⁴ used phenotypic tests as a stand-alone technique for the characterization of ocular CoNS isolates and to date there are no genotyping studies to compare with phenotypic methods. The purpose of the present study was to characterize non–S. epidermidis CoNS isolates of ocular origin phenotypically by analytical profile index (API) and genotypically by fluorescent amplified fragment length polymorphism (FAFLP). 

We attempted different typing approaches such as biotyping through analytical profiling index (API) and genotyping by FAFLP for reliable identification and molecular characterization of non–S. epidermidis CoNS isolates. FAFLP is a PCR-based fingerprinting technology with high resolution and sensitivity that can detect polymorphism at the whole genome level. ^10,11 This technique not only scans for polymorphism in actual restriction sites but also among the nucleotides bordering these sites. As a result, it documents nucleotide sequence variation, insertions, and deletions across the genomes. ¹⁰ This method appears to be a comprehensive approach for understanding microbial genome diversity and evolution. FAFLP genotyping of non S. epidermidis CoNS isolates derived from different ocular disease entities and normal conjunctiva may help in identification of major differences between disease-causing and normal commensals. Furthermore, we also attempted to identify genes that are likely to enhance virulence among these isolates such as genes relevant in biofilm formation (icaA and icaB), invasive, and antimicrobial resistance properties (mecA). ^12,13  

Methods

Patients, Bacterial Strains, and Phenotypic Characterization

Twenty-two patients with different ocular infections were included in the present study. The patients were enrolled and the study was conducted according to the guidelines set forth in the Declaration of Helsinki. Contemporary isolates from corneal scrapings and vitreous from patients with keratitis and endophthalmitis respectively and isolates obtained from conjunctiva of asymptomatic normal persons, were included in the study. A total of 33 isolates studied included isolates from patients with bacterial keratitis (n = 13), endophthalmitis (n = 9) and asymptomatic individuals (n = 11). The last group contained isolates that were isolated from the conjunctiva of students and staff of L.V. Prasad Eye Institute, Hyderabad. In addition, a single ATCC strain of S. epidermidis (RP62A) was included in the analysis as an out group. All the isolates grown as a sole isolate with a significant amount of growth were selected. The CoNS isolates were identified presumptively by Gram staining, catalase test, and oxidative-fermentative test for glucose. Gram-positive, coagulase-negative, and catalase-positive isolates with fermentation of glucose were collected and stored in BHI/TSB containing 10% glycerol (vol/vol) at −70°C. All isolates were revived on nutrient agar and tested for species identification phenotypically by the analytical profiling index using an API identification system (Staph ID 32; bioMerieux, Marcy I'Etoile, France). The API identification system is a standardized identification system that uses miniaturized biochemical tests and a specially adapted database (V4.0) that relies on 7-digit profile number. The numerical profile was determined by separating tests into groups of three on the result sheet and a value of one, two, or four is indicated for each. By adding together the values corresponding to positive reactions within each group, a 7-digit profile number is obtained that is later used by the database to calculate identification score. Antibiotic susceptibility was determined by Kirby-Bauer disc diffusion method wherein the following clinically relevant antibiotics were tested: amikacin, cefazolin, ceftazidime, ciprofloxacin, chloramphenicol, gentamicin, methicillin or oxacillin, ofloxacin, and vancomycin. Results of antibiotic susceptibility were interpreted according to the Clinical and Laboratory Standards Institute (CLSI) criteria. ¹⁴  

Genotypic Characterization

DNA Extraction.

DNA extraction was performed according to the protocol described earlier with slight modification. ¹⁵ In brief, the isolates were grown at 37°C in brain heart infusion broth and harvested during the logarithmic phase. The cells were washed with Tris EDTA (TE). Twenty microliters of lysostaphin (1 mg/mL), 20 μL of lysozyme (25 mg/mL) and 5 μL of RNAase (10 mg/mL) were added and tubes were kept for incubation at 37°C for 1 hour. Remaining steps involving protein precipitation and DNA extraction were the same as described in the earlier study. ¹⁵ The extracted DNA was dissolved in 0.1 mL of TE and stored at 4°C until required. 

PCR for icaAB and mecA Genes.

Primers were designed for simultaneous amplification of fragment encompassing icaA and icaB genes of certain species CoNS, with the help of previously published sequences. ¹² The presence of ica locus bearing icaA and icaB genes was shown in S. lugdunensis, S. capitis, and S. epidermidis among the CoNS. The ica primers were designed only to amplify certain regions of both the icaA and icaB genes of the ica locus. All non–S. epidermidis CoNS isolates were checked for the presence of mecA using PCR corresponding to the unique penicillin-binding protein (PBP2a or PBP2′) and the primer sequences and PCR reactions were derived from a previous study. ¹³ Presence of icaAB genes were detected by the amplification of the 546 bp fragment. Primers for icaAB were as follows: Forward primer 5′-TTATCAATGCCGCAGTTGTC-3′ and reverse primer 5′-GTTTAACGCGAGTGCGCTAT-3′. Presence of mecA was detected by the amplification of the 967 bp fragment of the mecA gene and the primer sequences for this PCR were as follows: Forward primer 5′-CATTTTGAGTTCTGCACTACC-3′, and a reverse primer 5′-GCAATACAATCGCACATACATTAATAG-3′. 

FAFLP Typing.

Thirty-four isolates along with the out group strain RP62A were characterized by FAFLP as described previously. ¹⁶ Using the enzyme combination of EcoRI- MseI we obtained a fingerprint of approximately 44 fragments distributed within the size range of 50 bp to 500 bp. Primer combinations used were EcoRI+0 and MseI+C. FAFLP experiments and analysis were done as per the manufacturer's instructions (AFLP Microbial Fingerprinting kit; Applied Biosystems). To investigate the stability and reproducibility of FAFLP, different DNA preparations from the same strain were subjected to the same reaction conditions at different time intervals and analyzed accordingly. In brief, FAFLP was performed for different DNA preparations from the same strain grown for multiple generations (up to five generations) and then extracted DNA of different generations of the same strain was subjected to FAFLP for more than one time at different time intervals. Under all these experimental conditions, the characteristic amplified fragment profile was reproducibly detected, and none of the fragment sizes varied in any instance by > 2 bp in FAFLP experiments. 

Genomic DNA digestions were performed using restriction enzymes EcoRI and MseI. In each case, DNA was double digested with the endonucleases and the resulting restriction fragments were ligated with the double-stranded adapters, consisting of a core sequence and an enzyme-specific sequence. The restriction ligation reaction was carried out simultaneously in a single step. Preselective PCR was carried out using an unlabeled primer set, one for the EcoRI adapter and another for the MseI adapter, whereas selective PCR was performed by using another primer pair, wherein one primer had the MseI adapter complementary sequence and an additional nucleotide C and the other primer had EcoRI adapter complementary sequence which was labeled with FAM [6-carboxyfluorescein]. The EcoRI and MseI adapters and PCR primers were obtained from a commercial source (AFLP Microbial fingerprinting kit, Applied Biosystems). 

Genescan Detection of FAFLP Fragments.

Selective PCR products, along with formamide loading dye and the internal lane standard 6-carboxy-x-rhodamine (GS 500 Rox), were loaded onto a DNA sequencer (ABI Prism 3100, Applied Biosystems). Fragment separation was continued for 2.5 hours through a performance-optimized polymer 4 (Applied Biosystems). Fragments were detected and compiled (ABI Data Collection Software, version 2.0; Perkin-Elmer, Applied Biosystems). Electropherograms and fragment analysis was performed (GeneScan software version 3.7; Applied Biosystems). Fragments were sized and peak height thresholds were set at 50. Any peak height having lesser value was excluded from the analysis. FAFLP electropherograms were analyzed using two different computer programs (GeneScan 3.7 and Genotyper 3.7; Perkin-Elmer, Applied Biosystems) to detect differentially amplified genomic fragments. FAFLP profile data were converted to binary matrix and used for the construction of dendrograms. Visual inspection of FAFLP profiles was keenly performed for detection of polymorphic and common species-specific signature bands using one software program (GeneScan 3.7; Applied Biosystems). All the isolates showed distinctive FAFLP profiles at the species and strain levels. Among electropherograms of all the isolates obtained from the single computer program (Genescan 3.7; Applied Biosystems), the present study was able to detect polymorphic and monomorphic markers. Distribution of fragments among the different species varied with different numbers of monomorphic fragments (ranging from 26 to 42) in the study. 

In the present study, genetic distance between the isolates was derived from the dendrogram constructed using the unweighted pair group method with arithmetic mean (UPGMA). To define criteria for genetic relatedness, FAFLP was performed for replicates of DNA derived before and after five subcultures of the standard ATCC strains belonging to the three major bacterial species. On observation, all three species shared a minimum of 98% intergel similarity for a specific number of fragments. So, individual isolates belonging to these three species that shared >98% similarity are likely to be identical clones. Isolates that showed >94% up to 98% similarities are considered very closely related and likely to belong to the same species. Isolates that showed <93% similarities (genetic distance of > 7%) are considered to be different species. The criterion derived from the above pilot experiment was applied to all species characterized in the study. 

Results

All 22 isolates from ocular infections were presumptively identified as one of the species of CoNS group. Analytical profiling index using the API identification system identified seven different species in this collection of isolates (Table 1). API identification score varied from 44% to 99%. Majority of S. haemolyticus and S. xylosus species showed good API scores of > 86%. API scores of S. capitis isolates varied from 49% to 97%, whereas all the S. hominis showed API score > 50% except one isolate that showed least score (43%). Only one isolate was identified as S. saprophyticus which showed maximum API score of 99.6%. Two isolates were identified as S. lugdunensis with an API score of 49% and 54%. Majority of commensal isolates belonged to either S. haemolyticus or S. capitis. 

Table 1.

 

Phenotypic, FAFLP Genotypic Profiles, Antibiograms, and PCR Results of mecA and ica Operon of Ocular CoNS Isolates Other than S. Epidermidis

Serial Number	Laboratory Number	Disease Diagnosis	API Identification	API Seven-Digit Profile	API Score	FAFLP Cluster	PCR Amplification		Antibiotic Resistogram
							mecA	icaAB
1	L2171/04	Keratitis	S. haemolyticus	2 2 3 2 1 5 1	99	II	+	−	Cf, Cip, G, Of
2	L1011/04	Keratitis	S. haemolyticus	6 6 1 2 1 5 1	92	II	+	+	Cz, Cip, G, Ak, Ca
3	L2481/04	Keratitis	S. haemolyticus	6 6 3 2 1 5 1	97	II	+	+	Cz, Cf, Cip, G, V, Of
4	L1151E	Endophthalmitis	S. haemolyticus	6 6 3 2 1 5 1	86.9	II	+	+	Cz, Cip, G, Of
5	L2170/04	Keratitis	S. haemolyticus	2 2 3 2 1 5 1	99	II	+	+	Cf, Cip, G, Of
6	L1543/04	Keratitis	S. haemolyticus	6 7 3 2 1 5 1	49.6	VI	−	−	S to all
7	L2145/04	Keratitis	S. haemolyticus	2 6 3 2 1 5 1	99	VII	+	+	Ak, Cf, Cip
8	L1202/04	Endophthalmitis	S. haemolyticus	6 3 0 2 1 1 3	64	XI	−	−	S to all
9	C1	Commensal	S. haemolyticus	6 2 3 2 1 5 1	89.7	II	−	+	Cz
10	C2	Commensal	S. haemolyticus	2 4 3 2 1 5 1	86.3	II	−	−	S to all
11	C3	Commensal	S. haemolyticus	2 2 0 2 1 6 1	97	V	−	−	S to all
12	C4	Commensal	S. haemolyticus	6 7 3 2 1 1 1	49	VII	−	−	S to all
13	L1714/04	Endophthalmitis	S. hominis	6 7 1 2 1 1 3	61.9	IV	−	+	S to all
14	165	Endophthalmitis	S. hominis	6 7 1 2 1 5 3	51.7	IV	+	−	Ca, Cip, G
15	1391/04	Keratitis	S. hominis	6 2 0 2 1 1 2	57.1	V	+	−	Cip
16	L998/04	Endophthalmitis	S. hominis	6 3 0 0 1 5 2	65.5	X	−	−	Ca
17	L1247/04	Endophthalmitis	S. hominis	6 7 1 2 1 5 3	53	XII	+	+	S to all
18	C5	Commensal	S. hominis	6 2 0 2 1 1 3	74.1	VII	−	−	S to all
19	C6	Commensal	S. hominis	6 3 1 2 1 5 1	43.5	VIII	+	−	S to all
20	L492/04	Endophthalmitis	S. capitis	6 3 0 2 1 1 1	95.6	III	−	+	S to all
21	L1648/04	Keratitis	S. capitis	6 3 0 6 1 0 3	55.3	VII	−	+	Cip, Of
22	C7	Commensal	S. capitis	6 7 2 2 1 1 3	78	VII	−	+	S to all
23	C8	Commensal	S. capitis	6 3 0 2 1 1 3	64.3	XII	+	+	S to all
24	C9	Commensal	S. capitis	6 2 0 2 1 1 3	49	XII	−	−	S to all
25	C10	Commensal	S. capitis	6 6 0 2 1 1 1	97	IX	−	−	S to all
26	L1562/04	Keratitis	S. warneri	6 7 3 2 1 1 3	59.2	VII	+	−	S to all
27	L1624/04	Endophthalmitis	S. warneri	6 7 1 0 1 1 3	83.4	VIII	+	+	S to all
28	L237/04	Endophthalmitis	S. xylosus	6 7 3 7 5 4 3	99	I	−	−	S to all
29	L1002/04	Keratitis	S. xylosus	6 7 0 2 5 7 2	86.9	IX	−	−	Cip, G, Of
30	L2250/04	Keratitis	S. saprophyticus	6 2 0 2 1 1 1	99.6	VII	−	−	S to all
31	L1561/04	Keratitis	S. lugdunensis	6 7 1 0 1 5 2	54.2	V	+	−	S to all
32	L1773/04	Keratitis	S. lugdunensis	6 7 1 0 1 5 2	54.2	V	−	−	S to all
33	C11	Commensal	S. lugdunensis	6 7 0 2 1 5 2	48.7	XI	−	−	S to all

 Ak, amikacin; C, Commensal; Ca, ceftazidime; Cf, cefataxime; Cip, ciprofloxacin; Cz, cefazolin; G, gentamicin; Of, ofloxacin; V, vancomycin.

Antibiotic susceptibility tests reveal that nine of 22 isolates were resistant to three or more antibiotics and all 11 commensals were sensitive to most of the antimicrobial agents tested. The mecA gene was detected in 42% (14) of isolates and icaAB gene was positive in 39.3% (13 out of 33). Among the endophthalmitis isolates icaAB was detected in 55% (5/9) and mecA was present in four isolates (44%). Keratitis isolates showed 78% positivity for mecA and 46% (6/13) of keratitis isolates were icaAB-positive. Thirty-six percent (4/11) of isolates from asymptomatic subjects showed amplification of icaAB. Only two out of 11 commensal isolates from asymptomatic subjects showed amplification of mecA. 

A total of 34 isolates (one reference and 33 clinical isolates) were examined by FAFLP. 

The single primer combination EcoRI+0 and MseI+C generated a total of 31 to 54 differently sized fragments experimentally ranging in size from 50 to 500 bp for all the isolates. Under all these experimental conditions, the characteristic amplified fragment profile was reproducibly generated reassuring their consistency over time, and none of the fragment sizes varied in any instance by more than two bp. Dendrograms constructed by UPGMA using binary data generated through FAFLP profiles generated 12 clusters from all the isolates (Fig. 1). The dendrogram depicted the genetic differences and relatedness among the 34 FAFLP profiles. The degree of polymorphism found among non–S. epidermidis CoNS isolates in our study is at the level of 17%. The 12 main clusters were designated using roman numerals (I–XII) (Fig. 1). Maximum number of isolates fell in two main clusters, namely II (n = 7) and VII (n = 7). Cluster II had only the S. haemolyticus species in it and all the isolates were from keratitis except one isolate (L1151/05) which was from endophthalmitis. Pathogenic reference strain RP62A formed a separate clade and it showed a difference of > 3% with the nearest cluster that is cluster I that had only one isolate which was identified as S. xylosus by API. Cluster VII comprised isolates belonging to five different species (this speciation was according to API). Cluster IV had isolates belonging to only S. hominis species. Identification results for API corresponded with FAFLP results only in six clusters (I, II, III, IV, VI, and X). Isolates obtained from asymptomatic individuals were seen scattered in a majority of FAFLP clusters. 

Figure 1.

 

Dendrograms constructed by UPGMA showing 12 clusters from all of the 34 isolates, depicting genetic relatedness among the non–S. epidermidis CoNS isolates. The CoNS isolates causing endophthalmitis and keratitis are indicated by the suffixing “E” and “K” following the strain identification number respectively. Commensal isolates are prefixed as C followed by their strain identification number.

Discussion

During the past decade, there have been changing trends in the practice and progress of medicine that have resulted in the emergence of CoNS as a major cause of nosocomial infections. ¹⁷ Previous studies on CoNS have shown that apart from S. epidermidis, a large number of isolates from ocular infections were S. haemolyticus and S. hominis. ^18,19 Our study also showed similar preponderance of S. haemolyticus wherein 36% of isolates belonged to this species. In another study, apart from S. epidermidis, other CoNS species like S. warneri, S. capitis, S. hominis, S. xylosus, S. simulans, S. equorum, and S. lugdunensis were isolated from ocular infections. ⁴ The present study also witnessed ocular infections with these species except S. simulans and S. equorum. We believe that this increase in ocular infections by non–S. epidermidis CoNS species may be either real or a result of improvement in diagnosis and identification of species of Staphylococcus. 

Surveillance studies of bacterial resistance are imperative for a better utilization of antimicrobials in any clinical setting as well as to decide on the empiric antimicrobial therapy. Antibiotic susceptibility of CoNS is unpredictable and multidrug resistance including methicillin resistance is more common in CoNS due to indiscriminate use of antibiotics. ^4,20 Hence it is prudent to make antibiotic susceptibility testing mandatory in all cases of clinically significant ocular infections caused by CoNS. Forty-one percent of isolates from the disease-causing group of isolates were resistant to three or more antibiotics as well as high methicillin resistance (78%) was noted among the isolates obtained from patients suffering from keratitis in our study. In contrast, methicillin resistance was comparatively lower among the isolates obtained from healthy volunteers. Presence of mecA gene that mediates methicillin resistance has been earlier proposed as a marker for discriminating between disease-causing and contaminating strains. ^12,13 Our results substantiate earlier findings of prevalence of mecA gene in disease-causing strains. 

Previous studies have shown that presence of ica operon among clinically significant CoNS is vital for their virulence. ^12,21 However, contradictory views have also suggested that ica is not a prerequisite for establishing CoNS infection. ²² A recent study showed no differences in the distribution of the ica between the clinically significant isolates and non-postoperative bacterial keratitis isolates. Nevertheless, the ability to produce biofilm was found to be present significantly more in disease-causing keratitis isolates than among non-postoperative bacterial keratitis isolates. ^7,23 It is also presumed that disease-causing icaAB-positive strains were selected because of their biofilm-forming capacity at the time of invasive therapy from a population of essentially icaAB and biofilm-negative commensal strains that are thought to be the prototype of an avirulent CoNS subpopulation. ²⁴ This model is based on many observations of higher prevalence of icaADBC operon in invasive strains than in commensal strains. ^12,13 Concurringly, we also documented more cases of icaADBC detection (55%) in invasive non–S. epidermidis CoNS strains causing endophthalmitis than in commensal or keratitis strains. Therefore, it appears that CoNS infections are typically endogenous in character and the populations with invasive characters are recruited and selected from the pool of commensal strains present in the patients, which is responsible for the disease. 

Identification results for API and microbial genotyping corresponded only in 48.4% of the isolates. API identified isolates of S. haemolyticus accurately with an identification score greater than 90. Majority of S. haemolyticus isolates were correctly identified by both API and FAFLP and the results were corresponding well. However, API failed to identify other species of CoNS with a good score. API identification of S. warneri, S. hominis, S. captis and S. lugdunensis always accompanied lower API identification scores ranging from 40 to 75. Identification results of API not corresponding with the FAFLP results while identifying certain species, implies that API has misidentified many isolates. Such noncorrespondence of API with FAFLP suggests that the battery of biochemical tests of API were not able to identify or not sufficient to identify all the species of CoNS. Hence, it is better not to consider, or one can say it is not prudent, to rely on API alone particularly when a species is identified by API with a lower score. Promisingly, FAFLP was able to discriminate more isolates by generating a total of 12 clusters when compared with 7 groups identified by API. Therefore, it is possible that with aid of the referral CoNS FAFLP database, one can accurately identify all the species of CoNS. Moreover, FAFLP did not detect major differences between disease-causing isolates and normal commensals implying that these two are not genetically distinct and disease might be sourced endogenously. In addition, disease-wise clustering was not evident using this method with an exception of cluster II. Furthermore, FAFLP genotyping in this study showed a clear heterogeneity among the isolates of single species and revealed that there were no predominant clones in the study population. Microbial genotyping DNA typing appears to be most reliable as a rapid genotypic method, particularly well suited to the epidemiologic study of CoNS isolates. 

Footnotes

 Supported by Grant No. F. No. 36-190/2008 (SR) from the University Grants Commission (UGC), Government of India (KP).

Footnotes

 Disclosure: A.R Makki, None; S. Sharma, None; A. Duggirala, None; K. Prashanth, None; P. Garg, None; T. Das, None