IAI Accepts, published online ahead of print on 14 March 2011 Infect. Immun. doi:10.1128/IAI.00046-11 Copyright © 2011, American Society for Microbiology and/or the Listed Authors/Institutions. All Rights Reserved. 1 Role of the accessory gene regulator agr in community-associated MRSA pathogenesis 2 3 Gordon Y. C. Cheung1, Rong Wang1#, Burhan A. Khan1, Daniel E. Sturdevant2, and Michael Otto1* 4 5 1 Laboratory of Human Bacterial Pathogenesis, National Institute of Allergy and Infectious Diseases, The D o w 6 National Institutes of Health, Bethesda, MD 20892, USA n lo 7 2 Research Technologies Branch, Genomics Unit, Rocky Mountain Laboratories, National Institute of a d e d 8 Allergy and Infectious Diseases, The National Institutes of Health, Hamilton, MT 59840, USA f r o m 9 h t 10 * corresponding author tp : / / 11 Contact information: ia i. a s 12 Laboratory of Human Bacterial Pathogenesis m . o 13 NIAID, NIH rg / o 14 Bldg. 33, 1W10 n J a 15 9000 Rockville Pike n u a 16 Bethesda, MD 20892 ry 2 3 17 USA , 2 0 18 Phone 301 443 5209; Fax 301 480 3633; Email: [email protected] 1 9 b 19 y g u 20 # present address: Meat Safety and Quality Research Unit, U.S. Meat Animal Research Center, P.O. Box e s t 21 166, Spur 18D, Clay Center, NE 68933 22 23 Running title: agr in CA-MRSA 24 1 1 ABSTRACT 2 3 The molecular basis underlying the pathogenic success of community-associated methicillin-resistant 4 Staphylococcus aureus (CA-MRSA) is not completely understood, but differential gene expression has 5 been suggested to account at least in part for the high virulence of CA-MRSA strains. Here, we show that D o 6 the agr gene regulatory system has a crucial role in the development of skin infections in the most w n lo 7 prevalent CA-MRSA strain USA300. Importantly, our data indicate that this is due to discrepancies a d e 8 between the agr regulon of CA-MRSA and those of hospital-associated MRSA and laboratory strains. In d f r o 9 particular, agr regulation in strain USA300 led to exceptionally strong expression of toxins and m h t 10 exoenzymes, up-regulation of fibrinogen-binding proteins, increased capacity to bind fibrinogen, and t p : / / 11 increased expression of methicillin resistance genes. Our findings demonstrate that agr functionality is ia i. a s 12 critical for CA-MRSA disease and indicate that an adaptation of the agr regulon contributed to the m . o 13 evolution of highly pathogenic CA-MRSA. r g / o 14 n J a 15 Keywords: Staphylococcus aureus; MRSA; community-acquired infections; quorum sensing; gene n u a 16 regulation r y 2 17 3 , 2 0 1 9 b y g u e s t 2 1 INTRODUCTION 2 3 Pandemic community-associated methicillin-resistant Staphylococcus aureus (CA-MRSA) strains 4 represent a serious problem for the public health system (25). Previously, MRSA infections were 5 reported exclusively in the hospital setting and in immune-compromised or otherwise predisposed D o 6 patients. In contrast, the recently emerging CA-MRSA strains combine antibiotic resistance with high w n lo 7 virulence and transmissibility, enabling them to spread and cause severe infections in otherwise healthy a d e 8 people (13, 33). The most serious CA-MRSA pandemic is seen in the U.S, where almost all cases are d f r o 9 caused by pulsed-field type USA300 (30), a group of closely related CA-MRSA strains (24) that are now m h t 10 also spreading in other parts of the world (37, 39, 48). t p : / / 11 The virulence potential of S. aureus is mainly determined by its ability to evade human innate host ia i. a s 12 defenses, among which the interaction with phagocytes such as neutrophils plays a preeminent role m . o 13 (12). S. aureus has developed many mechanisms to subvert destruction by human neutrophils, such as r g / o 14 protective capsule or enzymes that eliminate substances produced by neutrophils to kill the bacteria n J a 15 (18). In addition, S. aureus may secrete toxins that lyse human neutrophils and other immune cells, such n u a 16 as alpha-toxin, phenol-soluble modulins (PSMs), and members of the leukocidin family, which include r y 2 17 Panton-Valentine leukocidin (PVL) (12, 18). Furthermore, surface proteins that, among other tasks, 3 , 2 0 18 facilitate adhesion to host tissue are believed to contribute significantly to the establishment of S. 1 9 b 19 aureus infections (11). y g u 20 Virtually all S. aureus toxins are under control of the pivotal virulence regulator agr (31). This system e s t 21 triggers pronounced changes in gene expression at a certain level of cell density by a process called 22 quorum sensing. In addition to toxins, agr is known to up-regulate a wide variety of virulence 23 determinants such as exoenzymes (proteases, lipases, nucleases) and down-regulate expression of 24 surface binding proteins. This adaptation is believed to ascertain production of specific sets of virulence 3 1 determinants at the time of infection when they are most needed: binding proteins at the beginning, 2 when cell density is low and adhesion to host tissue is crucial; toxins and degradative exoenzymes when 3 the infection is established, nutrients need to be acquired from host tissues, and the concomitant 4 activation of the host’s immune system requires production of immune evasion factors. The timely 5 activation of agr in vivo and its importance for virulence of S. aureus have been demonstrated (8, 22, D o 6 49), even though these studies often only used laboratory strains. The role of agr in clinical strains of S. w n lo 7 aureus is less well understood. Particularly, no reports about the impact of agr on the virulence of CA- a d e 8 MRSA have been published. d f r o 9 Previous work indicated a crucial role of differential gene expression in the evolution of CA-MRSA m h t 10 virulence and suggested that agr control has a major function in establishing the exceptional virulence t p : / / 11 potential of the most predominant CA-MRSA strain USA300 (27, 29, 33, 45). Thus, we here investigated ia i. a s 12 the role of agr in the CA-MRSA strain USA300 in detail, using an animal skin infection model and m . o 13 genome-wide analysis of agr-dependent gene expression in comparison with HA-MRSA and laboratory r g / o 14 strains. We demonstrate a strong impact of agr on CA-MRSA (USA300) virulence in experimental skin n J a 15 infection and provide evidence indicating that an adaptation of agr-dependent gene regulation n u a 16 contributed to the evolution of virulence in the CA-MRSA strain USA300. r y 2 17 3 , 2 0 1 9 b y g u e s t 4 1 MATERIALS AND METHODS 2 3 Bacterial strains and growth conditions 4 S. aureus strains used in this study were LAC (CA-MRSA, pulsed-field type USA300) (6), MW2 (CA-MRSA, 5 pulsed-field type USA400) (1), Sanger 252 (HA-MRSA, pulsed-field type USA200) (21), and RN6390 D o 6 (laboratory strain, derived from strain 8325) (32). RN6390 and USA300 both belong to sequence type 8. w n lo 7 Deletion mutants in the agr regulatory system were produced by transduction of the transposon a d e 8 insertion from strain RN6911 in this study (strain 252) or previously (4). The agr system is entirely d f r o 9 deleted in these strains, except for a 3’ part of RNAIII, which is not transcribed owing to absence of the m h t 10 corresponding promoter. All mutants were verified by analytical PCR, DNA sequencing of the flanking t p : / / 11 regions, and real-time (RT)-PCR for absence of the RNAIII transcript. Main cultures for microarray ia i. a s 12 analysis and quantitative (q)RT-PCR were inoculated from pre-cultures grown overnight to an OD600 of m . o 13 0.1. r g / o 14 n J a 15 RNA isolation, transcriptional profiling, and quantitative RT-PCR n u a 16 RNA isolation from cultures grown to the indicated time points, removal of remaining DNA, cleanup, r y 2 17 cDNA synthesis and labeling were performed as previously described (28). Biotinylated S. aureus cDNA 3 , 2 0 18 was hybridized to custom Affymetrix GeneChips (RMLChip 7) with 100% coverage of chromosomal 1 9 b 19 genes from USA300 and scanned according to standard GeneChip protocols (Affymetrix). Each y g u 20 experiment was replicated 3 times. Affymetrix GeneChip Operating Software (GCOS v1.4, e s t 21 http://www.affymetrix.com) was used to perform the preliminary analysis of the custom GeneChips at 22 the probe-set level. Subsequent data analysis was performed as described (28). The complete set of 23 microarray data was deposited in NCBIs Gene Expression Omnibus (GEO, 24 http://www.ncbi.nlm.nih.gov/geo/) and is accessible through GEO Series accession number GSE18793. 5 1 Quantitative RT-PCR was performed as described (26), with at least 3 independent samples. All 2 oligonucleotides were synthesized by Sigma (Table 1). 3 4 Fibrinogen adhesion assay 5 To measure adhesion to fibrinogen, microtiter plates were coated with 100 µl/well of a 10 µg/ml D o 6 fibrinogen suspension at 4°C overnight. Afterwards, an equal volume of a 5 mg/ml BSA solution was w n lo 7 added and the plates were incubated at 37°C for 2 h. Plates were washed 4 times with PBS, after which a d e d 8 100 µl of bacterial solutions normalized to an OD600nm of 1 were added. Plates were incubated at 37°C f r o 9 for 1 h. Adhered bacteria were counted using microscopy. m h t 10 tp : / / 11 Mouse subcutaneous infection model ia i. a s 12 The subcutaneous abscess model was performed as described (45). Briefly, Crl: SKH1-hrBR hairless mice m . o 13 were between 4 and 6 weeks of age at the time of use. Mice were inoculated with S. aureus from mid- r g / o 14 exponential growth phase (3 h) with the indicated CFU in 50 µl as described (44). Animals were n J a 15 examined for skin lesions at 24-h intervals for a total of 14 days. Skin lesion dimensions were measured n u a 16 daily with a caliper. Length (L) and width (W) values were applied to calculate the area of lesions using ry 2 3 17 the formula of L x W. All animals were euthanized after completion of the entire procedure. Animal , 2 0 18 studies were approved by the Animal Care and Use Committee (IUCAC number ASP LHBP 1E), National 1 9 b 19 Institute of Allergy and Infectious Diseases. y g u 20 e s t 21 6 1 RESULTS 2 3 Impact of agr on the outcome of skin infection caused by CA-MRSA 4 To investigate the role of agr during infection, we performed a mouse subcutaneous infection model, 5 which mimics skin and soft tissue infections as the most common manifestation of disease caused by D o 6 CA-MRSA (30) (Fig. 1). Subcutaneous injection of the same number of bacteria (~ 3 x 107) resulted in w n lo 7 strong abscess formation using strain LAC (USA300), while abscesses were significantly smaller with a d e 8 strains LAC agr, 252, and 252 agr (Fig. 1A). Furthermore, animals infected with strain LAC developed d f r o 9 necrosis of the epidermis and dermis with epidermal ulceration (Fig. 1B,C). In contrast, strains LAC agr, m h t 10 252, and 252 agr caused chronic, subcutaneous abscesses lacking the dermal and epidermal necrosis t p : / / 11 noted with the LAC strain. Mice infected with the laboratory strain RN6390 (at 106 CFU) developed ia i. a s 12 abscesses that were dermonecrotic and larger than those in mice infected with RN6911 (agr), which m . o 13 never showed necrosis. Finally, animals injected with PBS control or heat-killed cells of any of the used r g / o 14 strains did not develop abscesses, supporting the idea that bacterial production of toxins rather than the n J a 15 host’s response to pro-inflammatory components of the staphylococcal surface plays a major role for n u a 16 the development of CA-MRSA induced skin disease. These results demonstrate that the agr system has a r y 2 17 crucial impact on the development of skin infections by the CA-MRSA strain USA300. 3 , 2 0 18 1 9 b 19 Agr activity y g u 20 To investigate the mechanistic basis of the impact that the agr system has on the virulence of strain e s t 21 USA300, we first determined agr activities during in-vitro growth (32, 35). While maximal agr activity in 22 strain 252 was reached 2 h later compared to strains USA300 and RN6390, most likely owing to slower 23 growth and delayed onset of the post-exponential growth phase in that strain, we found that maximal 24 activities of agr in strains LAC and 252 were not significantly different (p=0.55, Student’s t-test) (Fig. 2). 7 1 This indicates that the significantly lower virulence potential of strain 252 and the observed differential 2 impact of agr on virulence in these two strains are not due primarily to differences in agr activities. 3 4 Agr-dependent gene regulation in CA-MRSA: Overview, toxins and exoenzymes 5 It has been reported that the agr regulon in clinical strains may differ from the scheme of agr regulation D o 6 that had been established using laboratory strains (2, 35). This prompted us to evaluate the hypothesis w n lo 7 that the agr regulons rather than agr activities may account for the differential impact agr has on the a d e 8 virulence potential in CA- versus HA-MRSA. To that end, we analyzed the agr regulons of strains LAC, d f r o 9 252, and RN6390 by genome-wide transcriptional profiling. m h t 10 Analysis of the total up- and down-regulated genes did not reveal pronounced differences: strain t p : / / 11 252 showed 360 differentially regulated genes, strain LAC 262, and strain RN6390 228. Forty-three ia i. a s 12 genes were differentially regulated in all 3 strains. Genes coding for proteases, lipases, phenol-soluble m . o 13 modulins (PSMs), and other toxins were strongly up-regulated by agr in all 3 strains (Table 2, r g / o 14 Supplemental Table S1). The most strongly regulated toxin genes in strain LAC were those coding for the n J a 15 Panton-Valentine leukocidin (PVL) and PSMs. While the microarray only contained samples for the psmβ n u a 16 genes, owing to the small size of the other psm genes, qRT-PCR demonstrated exceptionally strong agr r y 2 3 17 control and exceptionally high expression in strain LAC also for the psmα genes (Fig. 3A). Furthermore, , 2 0 18 the strong impact of agr on PSM expression was verified on the protein level for all PSMs (Fig. 3B). In 1 9 b 19 accordance with previous results (45), expression of PSMs was overall lower in the 252 strain compared y g u 20 to strain LAC. Production of the PSMα peptides, the major contributors among PSMs to CA-MRSA skin e s t 21 infection and bacteremia (45), was considerably higher in strain LAC than in strains 252 and RN6390, in 22 both absolute and relative terms. Moreover, up-regulation by agr of secreted degradative enzymes such 23 as lipases and proteases was much stronger in strain LAC than in strain 252, while further virulence 24 genes such as lrgA and lrgB involved in bacterial programmed cell death (36) and staphylokinase (sak) 8 1 (3) were down-regulated by agr in strain 252 but not in strain LAC. While the hla (α-toxin) gene is not 2 functional in strain 252 owing to a nonsense mutation (5), it is still interesting to note that hla was up- 3 regulated by agr in strain LAC while down-regulated in strain 252. Data achieved using qRT-PCR and 4 Western blot confirmed a strong impact of agr on hla expression in strain LAC (Fig. 4). Remarkably, the 5 relative abundances of hla transcript and α-toxin in the 6 investigated strains were similar to the D o w 6 differences in virulence that we observed in the skin infection model, in keeping with the notion that α- n lo 7 toxin is a key virulence determinant in CA-MRSA skin infections (23) and serving as an explanation for a d e d 8 the high virulence observed for strain RN6390. f r o 9 m h t 10 Methicillin resistance tp : / / 11 So far no reports describing an impact of agr on methicillin resistance exist, except for our previous ia i. a s 12 observation of agr-dependent expression of mecA in the CA-MRSA strain MW2 (USA400) (35). Here, we m . o 13 found that genes involved in the resistance to methicillin (mecA, mecR1) were strongly up-regulated by rg / o 14 agr in the CA- but much less in the HA-MRSA strain, which was confirmed by qRT-PCR of the mecA gene n J a 15 (Fig. 4). In accordance with the idea that agr impacts methicillin resistance in strain LAC, oxacillin had a n u a 16 stronger growth impact on the LAC agr than the LAC wild-type strain at low concentrations, while no ry 2 3 17 impact of agr on methicillin resistance was detectable for strain 252 (Fig. 5A). These results show that , 2 0 18 agr has a gene regulatory effect on the expression of methicillin resistance genes and the methicillin 1 9 b 19 resistance phenotype in CA-MRSA. y g u 20 e s t 21 Surface binding proteins and fibrinogen binding capacity 22 According to microarray and qRT-PCR analyses, expression of surface binding protein genes in strain LAC 23 differed significantly from the other strains and only partially followed the general dogma of agr 24 regulation, which predicts a negative effect of agr on the expression of those genes. First, transcription 9 1 of sdrC, sdrE, fnbA, fnbB, and empbp was either only moderately down-regulated by agr in strain LAC, or 2 not affected (Table 2). Only with the protein A gene spa did we observe the predicted strong negative 3 agr effect (Fig. 4). Notably, strong expression and positive rather than negative regulation by agr of the 4 fibrinogen-binding protein genes clfA and efb in strain LAC were in clear contrast to the low expression 5 of surface binding protein genes in the other wild-type strains and the notion about commonly negative D o 6 regulation of this class of genes by agr. Fibrinogen binding assays confirmed that the observed w n lo 7 differences in clfA and efb expression translated to a high capacity of strain LAC to bind fibrinogen and a a d e 8 strong effect of agr on that capacity, which was much lower or absent in the other strains (Fig. 5B). d f r o 9 Thus, expression and agr-dependent regulation of surface protein genes in strain LAC are profoundly m h t 10 different from HA-MRSA and laboratory strains, allowing for a simultaneous expression of surface t p : / / 11 binding proteins, in particular those involved in binding of fibrinogen (ClfA, Efb), and universally agr-up- ia i. a s 12 regulated virulence determinants such as toxins and exoenzymes. m . o 13 r g / o 14 Metabolism and arginine deiminase n J a 15 The arc operon responsible for arginine deiminase activity has been discussed as a potential contributor n u a 16 to pH homeostasis and immune evasion, impacting colonization and virulence properties (17, 41). In r y 2 17 addition to the core genome-encoded arc operon (arc2), the arc operon present on the arginine 3 , 2 0 18 catabolic mobile element (ACME), arc1, which among CA-MRSA is uniquely found in USA300 clones, has 1 9 b 19 drawn particular interest, as it may contribute to the exceptional virulence and transmissibility features y g u 20 of these predominant CA-MRSA strains (14-15, 17). The core-genome and ACME-encoded arc operons e s t 21 were down-regulated by agr in USA300, similar to a single, third arc gene (arcC3), but expression of the 22 arc operons was low overall (Table 2, Fig. 4). Interestingly, the sum of arcA1 and arcA2 transcript levels 23 in strain LAC was not as high as the level of arcA2 transcript in strain 252, which is inconsistent with the 24 idea that arginine deiminase activity is responsible for the higher pathogenic potential of LAC compared 10