Genotyping of Epidemic Methicillin-Resistant Staphylococcus aureus Phage Type 15 Isolates by Fluorescent Amplified-Fragment Length Polymorphism Analysis

Abstract

Fluorescent amplified-fragment length polymorphism (FAFLP) analysis was investigated for its ability to identify and subtype isolates of an epidemic methicillin-resistant phage type of Staphylococcus aureus, EMRSA-15. These isolates were also characterized by PCR-restriction fragment length polymorphism (PCR-RFLP) of the coagulase gene and pulsed-field gel electrophoresis (PFGE). For FAFLP, DNA was double digested with restriction enzymes ApaI plus TaqI or EcoRI plus MseI. Site-specific adaptors were ligated to one or the other set of restriction fragments, and PCR amplification was carried out with adaptor-specific primers. Amplified fragments separated on an ABI 377 automated sequencer and analyzed with Genescan version 2.1 software generated FAFLP profiles for all the isolates. The presence or absence of fragments was scored, similarity coefficients were calculated, and UPGMA (unweighted pair group method using arithmatic averages) cluster analysis was performed. Either enzyme-primer combination readily differentiated EMRSA-15 from other methicillin-resistant S. aureus (MRSA) isolates and also revealed heterogeneity within the phage type. The discriminatory power of FAFLP was high. By combining both enzyme-primer data sets, 24 isolates were divided into 11 profiles. PCR-RFLP did not discriminate among these EMRSA-15 isolates. PFGE could discriminate well between isolates but was not as reproducible as FAFLP. All S. aureus and MRSA isolates in this study were typeable by FAFLP, which was easy to perform, robust, and reproducible, with evident potential to subtype MRSA for purposes of hospital infection control.

Methicillin-resistant Staphylococcus aureus (MRSA) is a nosocomial pathogen of worldwide importance (7). Within the United Kingdom, 16 phage types have been epidemic (EMRSA), of which EMRSA-3, -15, and -16 now predominate (4). In 1997 and 1998, an increased incidence of EMRSA-15 was reported. In all, over 1,200 incidents (three or more patients in a single hospital in 1 month infected with the same strain) were reported, with a wider geographical distribution than in previous years (1).

Reliable and rapid typing of EMRSA is needed to implement effective infection control measures. Typing systems should ideally show good discriminatory power and be reproducible, capable of typing all isolates, and easy to use (11). Phage typing has been used to type isolates of S. aureus for over 45 years (21). In this phenotypic technique, strains are classified according to susceptibilities to a set of internationally agreed-upon phages. This simple technique has a high throughput, but some isolates are phage nontypeable or may produce ambiguous results (2).

Various molecular methods have been described for typing isolates of MRSA. They include ribotyping (14), random amplification of polymorphic DNA by PCR (19), insertion sequence profiling (17), PCR-restriction fragment length polymorphism (PCR-RFLP) (8), and, notably, pulsed-field gel electrophoresis (PFGE) (18). PFGE, however, does not reliably produce stable banding patterns for MRSA; variation is seen in interlaboratory studies of defined strain collections (5, 20). Such variation seems to be specific to S. aureus and may be accounted for in part by the presence of variable numbers of lysogenic phage in genomes (10).

The technique of fluorescent amplified-fragment length polymorphism (FAFLP) analysis requires double digestion of the bacterial genome with restriction endonucleases, followed by ligation of adaptor sequences to the ends of restriction fragments. Subsets of fragments can be amplified by stringent PCR, using fluorescently-labelled primers complementary to the adaptor sequences. These can be extended into the restriction fragments by one or two bases to increase their selectivity. Amplified fragments are separated by electrophoresis in a polyacrylamide sequencing gel and visualized by the laser detection system of an ABI automated sequencer. Since different combinations of restriction enzymes can be used, FAFLP has the potential to type any bacterial species. For example, it has been used to investigate an outbreak of invasive disease caused by group A streptococcus (6), where it had discriminatory power superior to that of PFGE. The aim of the present study was to establish whether FAFLP would reproducibly discriminate between isolates of the clinically important epidemic S. aureus phage type EMRSA-15.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

Twenty-four isolates of EMRSA-15 obtained from the Laboratory of Hospital Infection had been assigned to this phage type on the basis of phage lytic patterns, resistance to methicillin and penicillin, and variable resistance to erythromycin and ciprofloxacin. Isolates had been submitted for typing between 1990 and 1997 by different hospital infection control laboratories with a diverse geographical distribution. Isolates listed as from the same region (Table 1) were from the same hospital but from different patients. In one hospital, transmission was known to have occurred between two patients (isolates 499 and 501). In all other instances, the isolates were not epidemiologically related. Three non-EMRSA-15 isolates and the Oxford S. aureus strain NCTC 6731 (isolates 16, 30, 299, and T) served as controls.

TABLE 1.

MRSA isolate data

Isolate	Yr of isolation/region	Phage pattern	Designated phage type	coa typea	PFGEb	FAFLPc
2	1990/Essex	75 wkd	EMRSA-15	2	P1	A1
3	1991/Birmingham	75 wk	EMRSA-15	2	P1e	A1
17	1995/Blackpool	75 wk	EMRSA-15	2	P5	A1
18	1995/Wigan	75 wk	EMRSA-15	2	P1	A1
29	1995/Glamorgan	75 wk	EMRSA-15	2	P1	A1
496	1996/Keighley	75 wk	EMRSA-15	2	P7	A11
497	1996/Keighley	75 wk	EMRSA-15	2	P7	A10
499	1996/Ormskirk	75 wk	EMRSA-15	2	P1	A1
501	1996/Ormskirk	75 wk	EMRSA-15	2	P11	A1
625	1996/Carlisle	75 wk	EMRSA-15	2	P1	A1
A	1997/Newcastle	42E/81/83C	EMRSA-15	2	P6e	A9
J	1997/Newcastle	75 wk	EMRSA-15	2	P2	A5
U	1997/Newcastle	42E/75/81/90/83C	EMRSA-15	2	P8	A4
K	1997/London	75 wk	EMRSA-15	2	P1	A7
L	1997/Leicester	75 wk	EMRSA-15	2	P10	A8
M	1997/Leicester	83C	EMRSA-15	2	P3	A1
Q	1997/Leeds	75 wk	EMRSA-15	2	P1	A2
R	1997/Leeds	75	EMRSA-15f	2	P1	A1
S	1997/Leeds	75 wk	EMRSA-15	2	P1	A1
28	1995/Merseyside	42E/81/83C	EMRSA-15	2	P4	A1
G	1997/Merseyside	42E/75/81/90	EMRSA-15	2	P2	A1
V	1997/Stafford	75 wk	EMRSA-15	2	P1	A3
P	1997/Stafford	42E/75/81/83C	EMRSA-15	2	P8	A6
F	1997/Middlesex	90/83C	EMRSA-15	2	P3	A8
30	1995/Edinburgh	75	Not designatedf	5	P14	A14
16	1992/Keighley	29/95/75/77/84/96/932	Not designated	6	P12	A15
299	1996/Bristol	29/42E/75/83A/81/	Not designated	1	P13	A13
Tg	Oxford	NDh	Not designated	4	P9	A12

^acoa PCR-RFLP profiles obtained following AluI or CfoI digestion of a 660-bp amplicon from the coagulase gene. 

^bPFGE profile obtained after SmaI PFGE macrorestriction. 

^cFAFLP profile obtained by combining data from the primer pair ApaI+0 and TaqI+G and the primer pair EcoRI+0 and MseI+C (see Fig. 3c). 

^dwk, only reacted at 100 times routine test dilution. 

^eVariable PFGE profiles. 

^fAssigned to EMRSA-15 on the basis of PFGE and antibiogram. 

^gNCTC 6571. 

^hND, not done. 

Bacteria were grown overnight on blood agar plates at 37°C in an aerobic atmosphere. Stock cultures were maintained on Preserver Beads (Technical Service Consultants, Heywood, Lancashire, United Kingdom) at −70°C. Genomic DNA was isolated from plate cultures with lysostaphin-sodium chloride-cetyltrimethylammonium bromide, as described previously (8). The concentration of DNA was estimated by UV spectrophotometry at A₂₆₀ (with a Beckman DU 640) by standard methods (15).

Phage typing.

Phage susceptibilities determined by the method of Blair and Williams (3) at 100 times routine test dilution with the current set of international typing phages are shown in Table 1. The isolates were ascribed to EMRSA-15 on the basis of three lytic patterns: the classic pattern of weak susceptibility to phage 75 (75 wk) (13), a 42E variant pattern, and an 83C variant pattern.

PCR-RFLP of the coagulase (coa) gene.

The coa gene was amplified from all isolates with a RoboCycler gradient 96 platform (Stratagene Ltd., Cambridge, United Kingdom). In a final volume of 50 μl, each reaction mixture contained 75 pmol of both primers 1513 and 2168 (8), 200 μM deoxynucleoside triphosphates, 1× PCR buffer, 3 mM MgCl₂, 1.25 U of Taq DNA polymerase (all from Life Technologies, Paisley, United Kingdom), and 1× bovine serum albumin (New England Biolabs, Hertfordshire, United Kingdom [NEB]). After an initial denaturation step of 94°C for 2 min, the cycling profile was 94°C (30 s), 57°C (30 s), and 72°C (1 min) for 35 cycles. The last cycle had a 5-min extension step at 72°C. Restriction endonuclease analysis of PCR products was performed with AluI and CfoI (Boehringer Mannheim, Lewes, United Kingdom) as described previously (8).

PFGE.

SmaI macrorestriction of genomic DNA was carried out as described for isolates of Streptococcus pyogenes (16). PFGE gel photographs were scanned into the Taxotron software package (Institut Pasteur, Paris, France), and fragments were sized with reference to internal lane markers, using its programs RestrictoScan and RestrictoTyper.

FAFLP.

The enzymes EcoRI and MseI were used to digest approximately 500 ng of genomic DNA from each isolate, and fragments were ligated to double-stranded adaptors as previously described (6). PCR was performed in a 20-μl volume containing 1.5 μl of ligated DNA, 15 μl of Amplification Core mix (Perkin-Elmer Applied Biosystems, Warrington, Cheshire, United Kingdom), 5 pmol of MseI adaptor-specific primer (Perkin-Elmer Applied Biosystems), and 1 pmol of 5-carboxyfluoroscein-labelled EcoRI adaptor-specific primer (EcoRI+0) (Perkin-Elmer Applied Biosystems). The MseI primer contained the extra selective base C (MseI+C).

The enzymes ApaI and TaqI were used to digest DNA as follows: approximately 500 ng of DNA was incubated with 4 U of ApaI (NEB), 1× buffer 4 (NEB), 1× bovine serum albumin (NEB), and 0.5 mg of DNase-free RNase A ml⁻¹ in a final volume of 20 μl at 25°C for 1 h. Five units of TaqI (NEB) was subsequently added to each reaction mixture, and the mixtures were incubated for a further 1 h at 65°C.

A 20-μl solution containing 4 pmol of ApaI adaptors (MWG-Biotech UK Ltd., Milton Keynes, United Kingdom), 40 pmol of TaqI adaptors (MWG-Biotech), 1× T4 ligase buffer (NEB), and 40 U of T4 DNA ligase (NEB) was added to 20 μl of double-digested DNA. The mixture was incubated at 12°C for 17 h, heated at 65°C for 10 min to inactivate the ligase, and stored at −20°C. The sequence of the ApaI adaptor was 3′ CATCTGACGCATGT, 5′ TCGTAGACTGCGTACAGGCC. The sequence of the TaqI adaptor was 3′ TACTCAGGACTGGC, 5′GACGATGAGTCCTGAC.

PCR was performed in a 25-μl volume containing 2 μl of ligated sample, 1× PCR buffer, 2 mM deoxynucleoside triphosphates, 1.5 mM MgCl₂, 0.625 U of Taq DNA polymerase (all from Life Technologies), 5 pmol of TaqI adapter-specific primer (Genosys Biotechnologies, Cambridge, United Kingdom), and 1 pmol of 5-carboxyfluorescein labelled ApaI adaptor-specific primer (ApaI+0) (Genosys). The TaqI primer contained the extra selective base G (TaqI+G). The sequences of the primers are 5′ CGATGAGTCCTGACCGA 3′ (TaqI) and 5′ GACTGCGTACAGGCCC 3′ (ApaI).

For both enzyme-primer combinations, touchdown PCR cycling conditions were employed as described previously (6). FAFLP products were stored at −20°C prior to gel electrophoresis.

Gel analysis.

FAFLP fragments were separated in 5% denaturing (sequencing) polyacrylamide gels on an ABI Prism 377 automated DNA sequencer. Gels 0.2 mm thick were prepared from LongRanger Singel (Flowgen, Lichfield, Staffordshire, United Kingdom) according to the manufacturer’s instructions. FAFLP products (1 μl) were added to 1.75 μl of loading dye (a mixture containing 1.25 μl of formamide, 0.25 μl of loading solution [dextran blue in 50 mM EDTA], and 0.25 μl of the internal size marker [Perkin-Elmer]). To size ApaI/TaqI fragments accurately, GeneScan-2500 markers labelled with the red fluorescent dye 6-carboxy-χ-rhodamine (ROX) were used as internal size standards, while for EcoRI/MseI fragments, GeneScan-500 markers (ROX labelled) were used. The sample mixture was heated to 95°C for 2 min, cooled on ice, and immediately loaded into the gel. Running conditions were 2.5 kV at 51°C for 10 h for ApaI/TaqI fragments or 1.68 kV at 51°C for 10 h for EcoRI/MseI fragments. The well-to-read distance was 48 cm in both cases.

Fragment analysis.

Fragments were sized with GeneScan version 2.1 software. Peak height thresholds were set at 50; any peak heights less than this value were not included in the analysis. Electropherograms of all fragment profiles were visually inspected for polymorphisms, with the presence and absence of fragments scored in a binary matrix, and were recorded as a text (tab-delimited) file in Excel version 5.0 (Microsoft). Dice coefficients of similarity were calculated with in-house software. Cluster analysis was performed by UPGMA (NEIGHBOR program of PHYLIP) and then displayed with TREEVIEW (12).

RESULTS

Characterization by existing molecular methods.

Twenty-four EMRSA-15 isolates had identical PCR-RFLP patterns following CfoI or AluI digestion of a PCR amplicon corresponding to a fragment of approximately 660 bp from the coagulase gene. This was RFLP pattern 2 (8). The four non-EMRSA-15 isolates had distinct coa gene RFLP patterns corresponding to patterns 1 (isolate 299), 4 (isolate T), 5 (isolate 30), and 6 (isolate 16).

Following SmaI macrorestriction of genomic DNA, fragments ranging in size from 75 to 700 kbp were produced from all isolates by PFGE. Eleven EMRSA-15 isolates had identical macrorestriction profiles containing eight fragments (Fig. 1). This profile was designated P1 (Table 1). Eleven EMRSA-15 isolates differed from this predominant profile by one to three fragments (profiles P2 to P8). Two EMRSA-15 isolates and the four non-EMRSA-15 isolates had four or more differences (profiles P9 to P14). When PFGE was repeated, two isolates (isolate 3 and isolate A) exhibited one- to two-band differences compared to their initial PFGE profiles (Fig. 1).

FIG. 1.

RestrictoTyper software-derived schematic of PFGE fragments following SmaI macrorestriction of MRSA genomic DNA. The open boxes in lanes 11 and 27 represent fragments present only on repetition of PFGE with a second extract of DNA; the asterisk denotes a fragment absent on repetition. The sizes of fragments are indicated in kbp. Lane 1, isolate 30 (profile P14); lane 2, 299 (profile P13); lane 3, 16 (profile P12); lane 4, 501 (profile P1); lane 5, L (profile P10); lane 6, T (profile P9); lane 7, U (profile P8); lane 8, P (profile P8); lane 9, 496 (profile P7); lane 10, 497 (profile P7); lane 11, A (profile P6); lane 12, 17 (profile P5); lane 13, 28 (profile P4); lane 14, F (profile P3); lane 15, M (profile P3); lane 16, G (profile P2); lane 17, J (profile P5); lane 18, R (profile P1); lane 19, V (profile P1); lane 20, S (profile P1); lane 21, K (profile P1); lane 22, 625 (profile P1); lane 23, 499 (profile P1); lane 24, 29 (profile P1); lane 25, 18 (profile P1); lane 26, Q (profile P1); lane 27, 3 (profile P1); lane 28, 2 (profile P1).

FAFLP: general considerations.

The selective primers used for each FAFLP were chosen empirically. In preliminary experiments, the primers ApaI+0 and TaqI+G and the primers EcoRI+0 and MseI+C produced profiles that could be easily scored and that revealed differences between isolates. Other nucleotide substitutions in the TaqI or MseI selective primers produced profiles with large numbers of small fragments. These were not sufficiently resolved to score easily and revealed fewer differences between isolates.

DNA was extracted from all isolates and subjected to FAFLP with both enzyme-primer combinations. Isolates were later recovered from stocks frozen at −70°C, the DNA was reextracted, and the samples were again subjected to FAFLP. The fragment profiles from different DNA extracts of the same isolate were shown to be reproducible. It was found that profiles resulting from degraded DNA extracts (visualized by agarose gel electrophoresis) contained certain extra-low-intensity fragments. This phenomenon was not seen in profiles generated from DNA preparations with undegraded DNA of the same isolate. Profiles exhibiting such fragments were discarded.

FAFLP with ApaI+0 and TaqI+G.

The range and size of fragments produced by FAFLP with the enzyme-primer combination ApaI+0 and TaqI+G was approximately 80 fragments per isolate, ranging from 55 to 820 bp. Nineteen EMRSA-15 isolates had identical fragment profiles (Fig. 2a). Five others had profiles very similar to this predominant profile, containing one extra fragment or lacking one or two different fragments. The extra fragment present in isolates 496 and 497 was 557 bp; in isolate V it was 192 bp, and in isolate K it was 578 bp. The following fragments, characteristic of the predominant profile, were not found in certain isolates: 556 bp (isolates 496 and 497), 607 bp (isolates 497 and J), 721 bp (isolate 496), and 733 bp (isolates 497 and J). The four non-EMRSA-15 isolates each exhibited over 60 fragment differences with respect to the predominant EMRSA-15 profile.

FIG. 2.

GeneScan version 2.1 software-derived electropherograms of the predominant EMRSA-15 FAFLP profile for ApaI+0 and TaqI+G (a) and for EcoRI+0 and MseI+C (b). The solid arrowheads and peaks indicate fragments that are present in the predominant profile but absent in other EMRSA-15 isolates (sizes are indicated in base pairs). The open arrowheads indicate the absence from the predominant profile of a polymorphic fragment that is present in another EMRSA-15 isolate.

FAFLP with EcoRI+0 and MseI+C.

Approximately 70 fragments per isolate, ranging from 55 to 320 bp, were produced by EcoRI+0 and MseI+C. To resolve the large number of fragments smaller than 200 bp (Fig. 2b), gels (48-cm well-to-read distance) were run at a lower voltage (1.68 kV). Fifteen EMRSA-15 isolates had identical profiles. Nine others had profiles which differed from this predominant profile by one to five fragments. The extra fragment present in isolate Q was 152 bp; in isolate K there were fragments of 228 and 246 bp, and in isolate 496 there were fragments of 240 and 302 bp. The following fragments, characteristic of the predominant profile, were not found in certain isolates: 100 bp (isolates 496, 497, A, and P), 108 bp (isolates A, F, and L), 111 bp (isolate P), 121 bp (isolates A, F, and L), 204 bp (isolates 496, 497, A, F, and L), and 296 bp (isolates 496, 497, A, P, and U). The four non-EMRSA-15 isolates each exhibited over 20 fragment differences with respect to the predominant EMRSA-15 profile.

Cluster analysis.

Three dendrograms were constructed on the basis of the ApaI+0-TaqI+G and EcoRI+0-MseI+C FAFLP data (Fig. 3a and b). All show clear separation between the EMRSA-15 isolate and the four unrelated isolates. Nine isolates with the classical EMRSA-15 phage pattern (75 wk) and three with variant patterns showed the predominant FAFLP profile A1 (see below). However, seven isolates with the classical phage pattern had unique FAFLP profiles.

FIG. 3.

UPGMA dendrograms derived from FAFLP data. (a) ApaI+0 and TaqI+G; (b) EcoRI+0 and MseI+C; (c) combined data from both primer pairs. The d values were calculated by using Simpson’s index of diversity (9). The horizontal scale bars represent 5% divergence.

The index of discriminatory power, D (9), was 0.55 for ApaI+0 and TaqI+G and 0.72 for EcoRI+0 and MseI+C. Six isolates (28, A, F, M, P, and U) that displayed the predominant ApaI+0 and TaqI+G profile were differentiated by EcoRI and MseI+C. Two isolates (J and V), indistinguishable by EcoRI and MseI+C, were differentiated with ApaI+0 and TaqI+G. When the two sets of data were combined, the index of discriminatory power was 0.79, and the predominant profile, A1, contained 13 of 28 isolates (Fig. 3c).

DISCUSSION

Phage typing and PFGE are currently used to type MRSA. However, phage typing has poor discrimination and fails to type all S. aureus isolates, with 15 to 20% found to be nontypeable (2). In this study, for example, phage typing failed to distinguish EMRSA-15 isolate R from the non-EMRSA-15 isolate 30. The reproducibility of PFGE for S. aureus has also been questioned (5, 20).

The high-resolution genotyping technique of FAFLP has been successfully used to investigate an outbreak of invasive disease by the gram-positive pathogen S. pyogenes (6). In the present study, we asked whether FAFLP was applicable to S. aureus and whether it could reproducibly identify and subtype a major epidemic phage type of MRSA. We provided evidence that S. aureus is fully typeable by FAFLP. In this and other studies, we have not found any S. aureus isolates (>100 tested, including different EMRSA phage types and sporadic strains) that fail to generate FAFLP profiles. This compares well with phage typing (2).

FAFLP is easy to perform for S. aureus, but the electrophoresis run (up to 10 h) is longer than that required to separate FAFLP profiles of S. pyogenes (6). This is due to the presence of many fragments of less than 100 bp that need to be resolved in S. aureus profiles. Digestion of genomic DNA and overnight ligation of adaptor sequences is performed on day 1, PCR followed by electrophoresis of amplified fragments in the polyacrylamide sequencing gel is completed on day 2, and the profiles are analyzed on day 3. In our hands, PFGE of S. aureus, which takes 3.5 to 4 days, is not as robust and is more labor-intensive than FAFLP.

The discriminatory power of FAFLP is determined by the choice of restriction enzymes and by the chosen selectivity of the primers. In the present study, we were able to further increase its discriminatory power by combining data from different enzyme-primer combinations. The range of fragments for the primer pair ApaI+0 and TaqI+G (55 to 820 bp) was greater than that for the primer pair EcoRI+0 and MseI+C (55 to 320 bp). Nonetheless, the discriminatory power for ApaI+0 and TaqI+G was lower. The dendrograms derived from either primer pair (Fig. 3a and b) demonstrate that FAFLP distinguishes isolates of EMRSA-15 from non-EMRSA-15 isolates. Moreover, the dendrogram derived from combined data (Fig. 3c) divided the 24-phage-type EMRSA-15 isolates into 11 profiles. The enzyme-primer combinations used in this study are suitable for FAFLP investigations of other S. aureus isolates.

FAFLP for all isolates in this study was reproducible over time: profiles were stable for different DNAs extracted successively from an isolate. For S. aureus this reproducibility was superior to that of PFGE—our results and other studies (5, 20) show that PFGE of S. aureus can be insufficiently reproducible. For example, on repetition of PFGE, isolate A yielded one extra band, while isolate 3 yielded one extra band and one absent band. PFGE has been adopted in preference to phage typing for subtyping S. aureus (2). However, we suggest that it may be inherently less suitable than FAFLP for genotyping S. aureus, since its display of a single rare-cutter digest of the whole genome is easily distorted by loss or gain of lysogenized prophages from the genome (10). By contrast, FAFLP simultaneously samples approximately 80 loci distributed randomly across the S. aureus genome, and FAFLP profiles are apparently not affected in the same way by large mobile DNAs. Thus, in our study, while the discriminatory power of FAFLP approximated that of PFGE for EMRSA-15, PFGE was less reproducible.

In summary, we have demonstrated that FAFLP can identify and subtype MRSA. In this study it readily defined a genotype for the epidemic phage-type EMRSA-15, distinguished it from those of other MRSA, and revealed EMRSA-15 to be a clone complex.