Prevalence and Sequence Variation of Panton-Valentine Leukocidin in Methicillin-Resistant and Methicillin-Susceptible Staphylococcus aureus Strains in the United States

Abstract

Panton-Valentine leukocidin (PVL), encoded by the lukSF-PV genes, is a putative virulence factor and marker for community-associated methicillin-resistant Staphylococcus aureus. Here we report the prevalence of PVL among a representative sample of 1,055 S. aureus infection isolates from the United States and describe the sequence variation of the lukSF-PV genes. We performed multilocus sequence typing (MLST) on all isolates and sequenced fragments of the lukSF-PV genes from a sample of 86 isolates. We assigned isolates to a PVL R or H sequence type based on a polymorphism that results in an amino acid change from arginine (R) to histidine (H). Overall, we found that 36% of S. aureus isolates were positive for lukSF-PV. Among the 86 we typed, we identified 72 R variants and 14 H variants. Among the 47 methicillin-resistance S. aureus (MRSA) isolates, 43 harbored the R variant, and among the 39 methicillin-susceptible S. aureus (MSSA) isolates, 29 harbored the R variant. Almost all (97%) of the R variants were found in MLST clonal complex 8 (CC8), while the H variant was broadly distributed among 6 CCs. Within CC8, all 38 MRSA (USA300) and all 28 MSSA isolates harbored the R variant. Of the 20 isolates from blood and the lower respiratory tract, 19 (95%) harbored the R variant. While the R variant had been linked primarily to USA300 MRSA, we found that all CC8 MSSA isolates also contained the R variant, suggesting that some strains of USA300 may have lost methicillin resistance as an adaptation in the community.

INTRODUCTION

Panton-Valentine leukocidin (PVL) is a pore-forming toxin encoded by the lukSF-PV genes (25). PVL is commonly observed in strains of community-associated methicillin-resistant Staphylococcus aureus (CA-MRSA) and is predominant among skin infection isolates (6). PVL has been associated with necrotic infections in humans and forms lytic pores that destroy human neutrophils, monocytes, and macrophages in vitro (5, 6, 14, 15). Nevertheless, the role of PVL in the pathogenesis of S. aureus infections remains controversial, with discordant results from studies using animal models (17, 23, 32). The contribution of PVL to the fitness of S. aureus, including transmission, also remains unclear. While PVL-positive S. aureus was once rare, the frequency of PVL in the United States has increased along with the spread of community-associated MRSA clones (4). Subsequently, there have been reports of PVL-positive clones emerging in the hospital (16) and increasing among invasive disease (5). The predominant PVL-positive clone in the United States is known as USA300. USA300 is a MRSA clone within multilocus sequence typing (MLST) clonal complex 8 (CC8) (29) that harbors the PVL genes. While the USA300 clone is named after a pulsed-field type, subsequent trends in molecular typing have allowed researchers to identify this clone using alternative methods (3, 26). Furthermore, the USA300 pulsed-field type has been shown to contain at least two distinct clones (18). Because the term USA300 has become quite useful in describing the major clone of CA-MRSA in the United States, we apply the term to all CC8, PVL-positive MRSA strains. This nomenclature is appropriate based on the universally observed correlation of CC8, PVL-positive MRSA with “classic” USA300 (3, 18).

As described in a prior report, we studied sequence variation of the lukSF-PV genes in S. aureus isolates collected from international clinical trials in uncomplicated skin infections (22). We identified two major sequence variants of lukSF-PV, the R and H variants, based on a substitution at nucleotide 527 that results in a histidine-to-arginine amino acid change. The R variant was strongly associated with isolates from the United States and particularly the USA300 MRSA clone. Conversely, the H variant was associated with isolates outside the United States and both methicillin-susceptible S. aureus (MSSA) and MRSA. That study was limited to uncomplicated skin and soft-tissue infection (SSTI) isolates 2004 and 2005, and very few PVL-positive MSSA isolates from the United States were found as part of those studies.

We based the present molecular study on isolates collected in the United States between 2004 and 2008 as part of a nationally representative collection of S. aureus clinical culture isolates. Here we report the prevalence of PVL in the United States and expand on our previous work by assessing PVL haplotype variation in MRSA and MSSA isolates and S. aureus isolates from blood, the lower respiratory tract (LRT), and SSTIs.

MATERIALS AND METHODS

Study design and sampling of S. aureus isolates.

We sampled 1,055 Staphylococcus aureus isolates from Eurofins Medinet, covering 3 time points: 2004, 2006, and 2008. Eurofins Medinet offers global surveillance services and houses one of the world's largest repositories of current, characterized, and clinically relevant bacterial pathogens. At the time of sampling in 2009, Eurofins' S. aureus collection included 14,151 clinical isolates from the selected years of study. We randomly sampled the isolates without replacement to generate our study population. Single patient clinical isolates came from all nine U.S. Census regions; all ages, 0 years to 90 years; inpatients and outpatients; and a variety of specimen sources (Table 1).

Table 1.

Prevalence of Panton-Valentine leukocidin (PVL) in S. aureus isolates by methicillin susceptibility and demographics^a

Group or characteristic	No. of isolates sampled	No. (%) of PVL+ isolates	No. (%) of isolates in groupc
			MRSA		MSSA
			Total	PVL+	Total	PVL+
All isolates	1,055	377 (35.7)	699 (66.3)	336 (48.1)	356 (33.7)	41 (11.5)
Patient sex
Male	597	213 (35.7)	391 (65.5)	187 (47.8)	206 (34.5)	26 (12.6)
Patient age group
0–17 yrs	161	94 (58.4)	109 (67.7)	86 (78.9)	52 (32.3)	8 (15.4)
18–64 yrs	581	249 (42.9)	376 (64.7)	221 (58.8)	205 (35.3)	28 (13.7)
≥65 yrs	313	34 (10.9)	214 (68.4)	29 (13.6)	99 (31.6)	5 (5.1)
Isolate source
SSTI	597	322 (53.9)	411 (68.8)	285 (69.3)	186 (31.2)	37 (19.9)
Blood	130	13 (10.0)	77 (59.2)	12 (15.6)	53 (40.8)	1 (1.9)
LRT	167	12 (7.2)	108 (64.7)	12 (11.1)	59 (35.3)	0 (0)
Other	159	29 (18.2)	102 (64.2)	26 (25.5)	57 (35.8)	3 (5.3)
Region
West	200	95 (47.5)	124 (62.0)	80 (64.5)	76 (38.0)	15 (19.7)
Northeast	262	55 (21.0)	165 (63.0)	47 (28.5)	97 (37.0)	8 (8.2)
Midwest	294	95 (32.3)	211 (71.8)	90 (42.7)	83 (28.2)	5 (6.0)
South	170	81 (47.6)	109 (64.1)	74 (67.9)	60 (35.3)	7 (11.7)
South Atlantic	129	51 (39.5)	89 (69.0)	45 (50.6)	40 (31.0)	6 (15.0)
Settingb
Outpatient	617	276 (44.7)	391 (63.4)	245 (62.7)	226 (36.6)	31 (13.7)
Inpatient	422	93 (22.0)	296 (70.1)	84 (28.4)	126 (29.9)	9 (7.1)

^aTo describe the prevalence of PVL in the United States, we obtained a collection of 1,055 S. aureus isolates from across the nation. Shown here are the weighted frequencies of PVL by demographics and clinical characteristics of those isolates representing the U.S. S. aureus population.

^bSetting information was missing for 16 isolates.

^cTotal percentages are prevalences among all isolates; PVL⁺ percentages are prevalences within MRSA or MSSA.

Isolate characterization.

Eurofins provided oxacillin susceptibility data. Isolates resistant or susceptible to oxacillin are described as MRSA or MSSA, respectively. We performed multilocus sequence typing (MLST) on all 1,055 isolates as previously described by Enright et al. (12), with some modifications of PCR conditions. We carried out PCR of the lukSF-PV genes to detect the presence or absence of Panton-Valentine leukocidin for all 1,055 isolates as described previously (22). We describe any isolate from MLST CC8 that harbors the PVL genes and is MRSA as USA300.

Sampling of PVL-positive isolates for sequencing of lukSF-PV.

A stratified sample of the 377 isolates positive for the lukSF-PV genes was selected as follows. We sequenced all lukSF-PV-positive isolates collected from blood and LRT (n = 25) and were successfully able to define the R/H variant type of 20. We also sequenced all lukSF-PV-positive MSSA isolates (n = 41) and were able to successfully define the R/H variant type of 39. Due to the large number of CC8 MRSA (USA300) SSTI isolates (n = 315) and previous work indicating that USA300 SSTI isolates harbor R variant PVL (22, 24, 34), we selected a subset of USA300 SSTI (n = 20) isolates for confirmatory typing. Since all 20 USA300 SSTI isolates harbored the R variant (see Results), no additional PVL sequence typing was performed on USA300 SSTI isolates. We also sequenced all other lukSF-PV-positive isolates that were not included in the above categories (CC1 MRSA, n = 5; CC30 MRSA, n = 3). In summary, we determined R/H types for a total of 86 isolates, 47 of which were MRSA and 39 of which were MSSA.

PVL R/H sequence typing and genetic variation across lukSF-PV.

A 650-bp internal fragment of the lukS-PV gene encoding the R and H variant region of PVL was amplified and sequenced for the R/H typing of the 86 isolates described above using previously described primers (2, 22). To explore the full extent of genetic variation of the PVL genes, a 1,769-bp fragment covering nearly the entire length of lukS-PV and lukF-PV was also amplified and Sanger sequenced for 41 of the 86 isolates. All sequences were aligned with the NCBI reference sequence USA300_TCH1516 (NC_010079) to serve as a comparator. Isolates were assigned to a major PVL variant type based on a nonsynonymous nucleotide substitution at position 527, where the change from a guanine to an adenine results in either an arginine (R variant) or a histidine (H variant) at amino acid position 176 (22). Isolates were assigned to a minor PVL variant type based on additional substitutions in the locus.

RESULTS

lukSF-PV-positive isolates in the United States from 2004 to 2008.

PCR testing revealed that 36% of isolates from this study of 1,055 isolates harbored the lukSF-PV genes (Table 1). The overall frequency of lukSF-PV increased from 16% in 2004 to 44% in 2006, followed by a slight decline in 2008 (40%). The prevalence of lukSF-PV-positive MRSA was 24% in 2004, increasing to 53% in 2006 and up to 54% in 2008. Among MSSA isolates, 9% were lukSF-PV positive in 2004, with an increase to 14% in 2006, followed by a slight decline in 2008 (12%). Among lukSF-PV-positive isolates, MRSA represented the vast majority, accounting for 89% of isolates, while 11% were MSSA. The majority of lukSF-PV-positive isolates were from MLST clonal complex 8 (94%; n = 355), while the remainder were from eight different clonal complexes (CC1, 5, 30, 45, 59, 88, 121, and 377). The USA300 clone (CC8; lukSF-PV-positive MRSA) accounted for 86% of all lukSF-PV-positive isolates observed. SSTIs accounted for 315 USA300 isolates.

Characterization of lukSF-PV sequence variation.

Out of the 86 lukSF-PV-positive isolates sampled to be R/H typed, we identified 72 R variants and 14 H variants. Among the 47 MRSA isolates, 43 isolates harbored the R variant and 4 harbored the H variant. The majority of MRSA isolates were USA300 R variants (n = 38), 5 were CC1 R variants, 1 was a CC1 H variant, and 3 were H variant CC30 isolates. All of the USA300 isolates we typed harbored the R variant. Of the 39 MSSA isolates, 28 were CC8, and all of these harbored the R variant, identical to the USA300 isolates. The one CC377 isolate also harbored this identical R haplotype. The remaining 10 MSSA isolates harbored H variants dispersed among six clonal complexes. The variants were not exclusive to methicillin resistance status, since 28% of the H variants were MSSA and 40% of the R variants were MSSA. In contrast, variants tracked almost exclusively with clonal complex, since all CC8 isolates (n = 66), both MRSA (n = 38) and MSSA (n = 28), were R variants, and CC1 is the only clonal complex in which we observed both R and H variants.

To explore the full extent of genetic variation of the PVL genes, a fragment covering more than 90% of the full sequence of lukSF-PV was also amplified and sequenced for 41 (35 R variants and 6 H variants) of the 86 isolates (Table 2). The genes for PVL were highly conserved; we observed variation at only five nucleotide positions, three of which result in nonsynonymous changes. Two of the three nonsynonymous changes, nucleotide 527 (the R/H variant position) and nucleotide 1729, were described previously. We observed one isolate with an additional novel nonsynonymous change at position 1497, where an insertion of a single adenine results in a frameshift mutation and a premature stop 14 bp downstream.

Table 2.

Nucleotide variation observed among 41 full-length lukSF-PV sequence fragments^a

CC and no. of isolates sequenced	Methicillin resistance profile	lukSF-PV sequence variant	Nucleotide at position:
			527b	663	1396	1497c	1729b
Reference			G	T	A		G
CC1
4	MRSA	R2	G	T	A		A
1	MRSA	H1	A	G	G		A
CC5
1	MSSA	H2	A	G	A		A
CC8
17	MRSA	R1	G	T	A		G
13	MSSA	R1	G	T	A		G
1	MSSA	R1	G	T	A	A	G
CC30
1	MRSA	H2	A	G	A		A
1	MSSA	H1	A	G	G		A
CC121
2	MSSA	H2	A	G	A		A

^aWe sequenced an internal fragment of lukSF-PV comprising more than 90% of the full PVL gene sequence from 41 of the 86 isolates we R/H typed. Shown here is a summary of the lukSF-PV sequence variation among those isolates. The right side of the table lists the five nucleotide positions that were variable. The nucleotides at those positions from reference sequence of USA300_TCH1516 are shown in the top row. Underneath are the corresponding nucleotides observed in the isolates in this study. Nucleotides differing from sequence of the reference strain are in bold font.

^bNonsynonymous change.

^cNovel insertion.

We identified five unique sequences, including two R variant haplotypes which differ at nucleotide 1729, where the change from an adenine to a guanine results in a glutamic acid (R1 variant) or a lysine (R2 variant). The major R variant (R1) accounted for all CC8 isolates, which includes all USA300 (n = 17) and all CC8 MSSA (n = 14) isolates. All of these sequences were identical, with the exception of one sequence from a CC8 MSSA isolate that had the novel insertion at nucleotide 1497. The minor R variant (R2) accounted for the remaining 5 R variant sequences, and all were from CC1 isolates. This haplotype was identical to the haplotype previously associated with the MW2 strain (also from CC1) of CA-MRSA (22), and because of its frequency, we have discriminated the R variants, R1 and R2.

We observed two H variant subtypes (H1 and H2) that differed at nucleotide positions 663 and 1396. We found two H1 variants in CC1 and CC30. H2 variants were also found in CC30 (n = 1), CC5 (n = 1), and CC121 (n = 2). We did not find any sequence variation between isolates of the same multilocus sequence type. The majority of clonal complexes harbored only one haplotype, with the exception of CC30, where MSSA harbored the H1 variant and MRSA harbored the H2 variant.

PVL sequence variants in isolates associated with SSTIs, blood, and LRT.

Among 59 SSTI isolates, we found 46 R variants and 13 H variants. Among 20 blood (n = 11) and LRT (n = 9) isolates, 19 harbored the R variant and 1 harbored an H variant. Among the 19 isolates from blood and LRT that harbored the R variant, all were from CC8 and 18 were MRSA (USA300). The remaining isolate was an H variant MRSA blood isolate from CC1. Isolates from other sources (n = 7) were all R variants.

DISCUSSION

While PVL-positive S. aureus has received significant attention in recent years, to our knowledge this is the first reported estimate of the overall prevalence of lukSF-PV among S. aureus isolates from a nationally representative sample in the United States. The relatively high overall prevalence (36%) is consistent with widespread reports of the prevalence of the PVL-positive USA300 clone of CA-MRSA in the US (27, 28, 29). Furthermore, the frequency of PVL is highest in isolates from children and SSTIs, which represent known risk factors for CA-MRSA. While PVL-positive S. aureus was first identified in the mid-1990s and is now recognized as a significant pathogen in the United States (6), we have identified the years 2004 to 2006 as an important time frame. During that time, the frequency of lukSF-PV increased nearly threefold.

An important finding from this study is that all CC8 lukSF-PV-positive isolates, including USA300 isolates, share the R variant haplotype and that most of the lukSF-PV-positive MSSA in the United States is from CC8. This finding is consistent with a model in which a fraction of USA300 isolates from CC8 lose the mecA gene over time while retaining PVL and possibly other elements of the USA300 genetic background essential to its fitness. The mecA gene is carried on a mobile element, SCCmec, and there is evidence that some clinical MRSA isolates have lost all or part of the SCCmec cassette (8). Expression of mecA imposes a fitness cost on isolates in environments without antibiotic use (11, 19), such as community settings where USA300 emerged, and there may be evolutionary pressure for some CA-MRSA strains to lose the SCCmec cassette while retaining key virulence or fitness factors. Indeed, there is evidence that MSSA isolates produce higher concentrations of PVL than MRSA strains in soft tissue infections (31). While relatively uncommon, CC8 MSSA with PVL has the potential to expand in communities where CA-MRSA has thrived. For example, there was recently a report of an outbreak of a methicillin-susceptible relative of USA300 among football players in New England (13). We cannot rule out the alternative hypothesis that the lukSF-PV-positive CC8 MSSA strains we observed had never acquired SCCmec in the past, but the more parsimonious explanation is that these strains were formerly MRSA. First, the presence of mecA among CC8 had been observed long before the emergence of the lukSF-PV-positive USA300 clone (7). Second, the lukSF-PV-positive CC8 MSSA strains observed in this study harbor R-variant lukSF-PV sequences that are identical to the lukSF-PV sequences from USA300, indicating common descent rather than independent acquisition of lukSF-PV.

The results presented here are consistent with prior reports which show the H variant to be distributed across more clonal complexes than the R variant (10, 22, 30). Consistent with our previous work on samples from the United States, the majority of H variants we observed were from MSSA isolates, but other studies have shown that H variant PVL-positive MRSA is not uncommon outside the United States (10, 30). The R variant remains primarily associated with CC8, where it has expanded, mainly due to the remarkable success of the USA300 clone. This situation should persist, since the prophage carrying lukSF-PV in USA300 has a mutation that renders it unable to excise (33). However, the fact that the R and H variants are found in multiple clonal complexes demonstrates that spread of PVL via horizontal transfer among lineages is a key feature of its evolution. The prophage carrying the PVL genes may transfer via phage transduction (21).

This study expands on our previous study of lukSF-PV variation in uncomplicated SSTIs by assessing variation in blood and LRT isolates. While PVL is primarily associated with skin infections, 7% of LRT isolates and 10% of blood isolates in this study harbored the lukSF-PV genes and 95% harbored the R variant. However, we cannot conclude that the R variant is uniquely adapted to cause blood or lower respiratory tract infections, because the frequency of R variants among SSTIs was also very high in the United States. Tong et al. (30) reported that R and H variants were present at similar frequencies among sepsis isolates, suggesting that both isoforms are capable of causing invasive disease. While the role of PVL in human pathogenesis has been a matter of some controversy (7, 23, 31), it has been shown that PVL-positive isolates have caused invasive diseases such as bacteremia (27) and necrotizing pneumonia in human patients (9). Observation of PVL H variants in a blood isolate suggests that, whatever role PVL may play in invasive pathogenesis, the capacity is not limited to a single variant. Regardless, if PVL is directly or indirectly responsible for pathogenesis, it is undeniably associated with the successful genetic makeup of USA300 and its repertoire of virulence factors (28, 29). Both the H variant and the R variant have been shown to be leukotoxic against human polymorphonuclear neutrophils (1). This leukotoxicity could be an important component of host immune evasion during invasive disease (20).

In conclusion, 36% of Staphylococcus aureus isolates in this collection of infection isolates from the United States were positive for the lukSF-PV genes. These genes are highly conserved, with two major variants found in the United States. However, we identified a novel haplotype of the R variant which warrants further surveillance. The R variant was the predominant type among both lukSF-PV-positive MRSA and MSSA isolates in the United States. The H variant was mainly distributed among the clonally diverse MSSA. We have expanded our previous work by showing that the R and H variants are present in blood and lower respiratory tract disease isolates as well as SSTIs in the United States. While the R variant accounted for 19 of 20 blood and lower respiratory tract isolates, it was not the exclusive variant, since we also observed the H variant in one blood isolate. Most lukSF-PV-positive MSSA isolates in the United States were from CC8 and harbored the R variant, suggesting that some strains of USA300 may be losing methicillin resistance as an adaptation in the community.