A Caenorhabditis elegans host model correlates with invasive disease caused by Staphylococcus aureus recovered during an outbreak in neonatal intensive care

1Department of Pathology and Laboratory Medicine; 2Centre for Antimicrobial Resistance, Alberta Health Services/Calgary Laboratory Services/ University of Calgary, Calgary, Alberta; 3Sunnybrook Health Sciences Centre, Toronto, Ontario; 4Department of Microbiology, Immunology and Infectious Diseases; 5Department of Medicine; 6The Calvin, Phoebe and Joan Snyder Institute for Chronic Diseases, University of Calgary, Calgary, Alberta Correspondence: Dr Kunyan Zhang, Department of Pathology and Laboratory Medicine, University of Calgary, 3330 Hospital Drive Northwest, Calgary, Alberta T2N 4N1. Telephone 403-210-8484, fax 403-210-7661, e-mail kzhang@ucalgary.ca Staphylococcus aureus is an important human pathogen and has been a leading cause of nosocomial infection for many decades. It has been shown that S aureus is commonly identified in neonatal intensive care units (NICUs), with up to 80% of neonates colonized with the organism by day 10 in the NICU (1). Neonates in the NICU are more susceptible to S aureus infection due to their immature immune system and high frequency of use of invasive medical devices, particularly intravascular catheters. Therefore, S aureus, including methicillin-resistant strains (methicillin-resistant S aureus [MRSA]) have been frequently identified as causes of outbreaks of infection in NICUs (2). We previously used the nematode, Caenorhabditis elegans, as a host model to determine the virulence of clinical MRSA isolates and demonstrated that the virulence of MRSA to C elegans correlated well with the isolation of MRSA from clinically relevant invasive anatomical sites (3). In the present study, we analyzed clinical strains that were recovered during a methicillin-susceptible S aureus (MSSA) originAl Article


An outbreak and identification of isolates
Sunnybrook Health Sciences Centre is an 1100-bed, universityaffiliated, tertiary-care, teaching hospital in Toronto, Ontario.It is a regional high-risk maternity referral centre with 29 level 3 NICU beds and 12 level 2 beds.MSSA strains were isolated from neonates in the NICU during the outbreak (December 2006 to June 2007) and during the preoutbreak period (January to November 2006).Invasive infection was diagnosed in infants with a compatible clinical syndrome and in whom S aureus was recovered from blood cultures or from some other normally sterile site.Noninvasive infection was believed to be present in neonates with signs and symptoms of infection, who had isolation of S aureus from a nonsterile body site (eg, endotracheal aspirate, eye, skin/soft tissue abscess).S aureus colonization was determined to be present if the organism was isolated from a nonsterile site in a neonate without signs or symptoms of infection, or if it was obtained during culture surveillance as part of the outbreak investigation.Strain M92 is a nematode avirulent control strain and USA300-2406 is a virulent positive-control strain, both of which originate from Calgary, Alberta (3).

C elegans virulence assay
A representative sample of isolates was investigated for strain virulence using the C elegans host virulence model, performed as previously described (3).Briefly, Bristol N2 C elegans nematodes were maintained at room temperature (RT) on nematode growth medium (NGM) plates.For survival assays, 30 nematodes were transferred from NGM plates to a tryptic soy agar (TSA) plate grown with one testing MSSA strain and monitored for survival every 24 h.Plates with heatkilled bacteria were prepared as previously described (3).The experiments were repeated at least three times.Using GraphPad Prism (GraphPad Software, USA), nematode survival rate data were analyzed by the Kaplan-Meier method, and comparison of significant survival difference used the log-rank test.For worm proliferation experiments, all worms were kept on their original plates without nematode transfer, and the accumulated numbers of live larva and adult nematodes were estimated by counting the live worms outside of the bacterial lawn.The C elegans model was assayed by trained technologists 'blinded' to the clinical or epidemiological data, and to the pulsed-field gel electrophoresis (PFGE) typing results.

Quantification of bacterial in vitro growth and in vivo burden
In vitro bacterial growth curves were constructed by reading optical density at 600 nm (OD600).Briefly, overnight brain heart infusion (BHI) bacterial cultures were diluted 1:1000 in fresh BHI broth, and 200 µL of the culture was loaded onto a 96-well plate.Each well was covered with mineral oil.Plates were incubated at 37°C and the OD600 values were measured every 20 min using a Wallac Victor2 multilabel counter (Perkin Elmer, USA).For quantification of in vivo bacterial burden of C elegans, the nematodes were fed representative MSSA isolates, and bacteria were quantified by serial dilution and plating on days 1, 2 and 4. Briefly, eight worms from each tested strain at each time point were washed with phosphate buffered saline (PBS) six times, by sequentially soaking and transferring worms in six 50 µL PBS pools on a TSA plate.The washed worms were then transferred into 100 µL PBS in a 1.5 mL microfuge tube and homogenized.Serial dilutions of the homogenized solutions were then plated to count the live bacteria.

Phenotypic and genotypic characterization of isolates
All S aureus isolates were typed by PFGE according to the Canadian standardized protocol (4).The strains were tested for the presence of Panton-Valentine leukocidin (PVL) genes, and 34 other common S aureus virulence genes, by polymerase chain reaction assay (3).The isolates were characterized by multilocus sequence typing (MLST), staphylococcal protein A (spa) typing and accessory gene regulator (agr) typing as previously described (5).

The outbreak of MSSA infections in an NICU
During a seven-month period (December 2006 to June 2007), 21 infants in the NICU were found to be infected with MSSA.Prevalence (colonization) screens of the nose, umbilicus and groin identified 28 additional infants found to be colonized with MSSA.Following a chart review, 10 additional MSSA infections were found to have occurred during an 11-month period (January to November 2006) before this outbreak (Figure 1A).
PFGE was performed on all but two outbreak isolates (n=47), and all 10 preoutbreak isolates.Two distinct outbreak clones were identified by PFGE.Twelve patients were shown to be infected with a strain designated MSSA-C.This strain was associated with more severe infection and disease, accounting for three of four bloodstream infections that occurred during the outbreak period (Figure 1B).Another strain designated MSSA-G, which appeared to be associated with less invasive disease, was identified in six infants, two of whom were merely colonized (Figure 1B).

Staphylococcus aureus (MSSA) outbreak in the neonatal intensive care unit (NICU) of Sunnybrook Health Sciences Centre, Toronto (Ontario). A Number of S aureus infection and colonization cases during the periods of the preoutbreak (January to November 2006) and outbreak (December 2006 to June 2007). B Invasive infection and noninvasive infection versus colonization from two outbreak-associated strains, MSSA-C and MSSA-G, and other colonization strains unrelated to the outbreak
As expected, several other strains of MSSA were also recovered from neonates (n=21), most of which represented asymptomatic colonization (n=18) (Figure 1B).The majority of these strains had distinct PFGE profiles and were determined to be unrelated to the outbreak.
Nasal swab cultures were performed on NICU staff members early on in the outbreak investigation.Seven employees were infected with the MSSA-C strain but all were asymptomatic.No employee was positive for the MSSA-G strain.With the knowledge that transmission was in fact occurring between health care workers and patients, additional infection control interventions were implemented.As a result, no further transmission occurred within the unit and the outbreak was declared over in June 2007.
Four representative isolates of MSSA-C, three of MSSA-G and three infant-colonized isolates unrelated to the outbreak as determined by PFGE were found, coded as MSSA1 to MSSA10.MSSA1, MSSA2 and MSSA3 were the isolates responsible for three bacteremias in newborns, and MSSA4 caused a skin infection; all of the infections required treatment.MSSA5 was isolated from an umbilical site and MSSA6 and MSSA7 were both isolated from umbilical and eye sites.MSSA8 to MSSA10 were nonoutbreak isolates that were found to colonize on the endotracheal tube, nose or umbilical site.All isolates were submitted for the virulence assays in the C elegans host model and for further molecular characterization.

Double-blinded virulence tests in the C elegans host model
The virulence of the MSSA isolates was determined using the C elegans host model.Isolates MSSA1 to MSSA4 demonstrated nematocidal activity after 24 h.For MSSA1 to MSSA4, killing was noted, respectively, in 45.8%, 72.5%, 73.3% and 68.1% of worms by day 3 and 70.2%, 97.5%, 100% and 94.9% by day 6 (Figure 2A).There were no significant differences between MSSA2 to MSSA4 isolates over the nine-day experiment period (all P≥0.13) with nematocidal activity similar to the virulent positive control strain USA300-2406, while MSSA1 showed relatively less virulence.In contrast, strains MSSA5 to MSSA7 showed similarly intermediate nematocidal activity, with mean killing rates of 26.7%, 23.4% and 26.1% by day 3, and 56.5%, 59.1% and 59.4% by day 6, respectively.There were still 37.0%, 39.4% and 40.6% of worms surviving on day 9, respectively (Figure 2A).However, MSSA8 to MSSA10 showed avirulence or low nematocidal activity.MSSA8 was similar to the negative control colonization strain M92 and did not kill nematodes.Isolates MSSA9 and MSSA10 killed 15.0% and 4.0% by day 3 and 33.8% and 9.4% by day 9,

Figure 2) Virulence and in vitro and in vivo bacterial growth features of methicillin-susceptible S aureus (MSSA) strains associated with and unrelated to the outbreak. A Kaplan-Meier survival plots of nematodes fed the invasive outbreak strain MSSA-C (MSSA1 to MSSA4), noninvasive outbreak strain MSSA-G (MSSA5 to MSSA7) and nonoutbreak colonization strains (MSSA8 to MSSA10), in comparison with nematode virulent (USA300-2406) and avirulent (M92) control strains. B In vitro growth curves of strains MSSA1to MSSA10 in brain heart infusion (BHI) at 37°C for 24 h. C Caenorhabditis elegans in vivo bacterial burden of representative strains (MSSA8, MSSA5 and MSSA4) from the avirulence, intermediate and high virulence groups, respectively, on days 1, 2 and 4. CFU Colony forming units
Can J Infect Dis Med Microbiol Vol 23 No 3 Autumn 2012 respectively.In addition, differential effects of different MSSA strains on the larval stage worms were observed.Both strains MSSA-C (MSSA1 to MSSA4) and MSSA-G (MSSA5 to MSSA7) efficiently inhibited larval proliferation, with less than 100 larva observed on the tested plates on day 9.However, strains MSSA8 to MSSA10 produced a large number of second-and third-generation larvae, accumulating more than 1000 worms on day 9 (data not shown).The experiments suggested that MSSA1 to MSSA4 were highly virulent, MSSA5 to MSSA7 were intermediately virulent and that MSSA8 to MSSA10 were avirulent or of low virulence in the C elegans model.Interestingly, only when these results were subsequently compared with clinical data was it determined that MSSA1 to 4 were the outbreak isolates associated with invasive infection (MSSA-C), MSSA5 to MSSA7 were associated with the noninvasive outbreak isolates (MSSA-G) and MSSA8 to MSSA10 were associated with the outbreak-unrelated colonization strains.
The nematocidal activities of the highly and intermediately virulent strains (MSSA1 to MSSA4 and MSSA5 to MSSA7) were abolished when the worms were fed with heat-killed bacteria (data not shown), suggesting that killing requires the presence of live bacteria.To determine whether the low or non-nematocidal activities of the MSSA strains were due to deficient bacterial growth, bacterial in vitro growth curves were constructed and the in vivo burden of representative strains were calculated.As shown in Figure 2B, all the strains except MSSA8, regardless of high or low virulence, showed similar growth curves.The avirulent strain MSSA8 exhibited a slower growth pattern than the other strains.Further comparisons of the bacterial in vivo burden on days 1, 2 and 4 did not denote any significant difference among the representative strains (MSSA8, MSSA5 and MSSA4) from the avirulence, intermediate and high virulence groups (P>0.34 [Mann Whitney test]) (Figure 2C).

Molecular (genotypic) and phenotypic characterization of MSSA isolates
All 10 isolates were further characterized by MLST, spa, and agr typing in addition to PFGE (Figure 3).All MSSA-C isolates (MSSA1 to MSSA4) were clustered together by PFGE and carried the same MLST type (ST109), spa type (t209) and agr type (II), whereas all MSSA-G strains (MSSA5 to MSSA7) shared identical PFGE patterns and genetic profiles (MLST type ST5, spa type t071 and agr type II).The other three colonization strains each had a unique genetic profile.Strain MSSA8 was MLST type ST942, spa type t1445 and agr type III.Strain MSSA9 was MLST type ST45, spa type t1156 and agr type I, and strain MSSA10 was MLST type ST8, spa type t334 and agr type I (Figure 3).
PVL and other common virulence gene tests revealed that only MSSA8 strains carried PVL and tst genes while the other nine strains were negative for both of these virulence genes (Figure 3).Different groups of MSSA strains possess distinct virulence gene profiles, although different isolates within the same group (MSSA-C and MSSA-G) shared an identical virulence gene pattern (Figure 3).Both MSSA-C (MSSA1 to MSSA4) and MSSA-G (MSSA5 to MSSA7) group isolates had similar virulence gene profiles except for chp, sdrC, hysA and sak, which were present in the former group of MSSA strains but not in the latter.However, these genes were also present in MSSA8 to MSSA10, while all but chp were present in MSSA10.It was also noted that all isolates except MSSA8 carried hla (Figure 3).

DISCUSSION
Invertebrates have been used as host models for studying the virulence of pathogens because of their convenience and simplicity.However, such models may not accurately reflect the interactions between humans and microorganisms because invertebrates are distantly related evolutionarily.It is known that some genes involved in the innate immune responses of invertebrates and mammals are highly conserved (6).Sifri et al (7) showed that the host p38 MAP kinase pathway, which is involved in human neutrophil antimicrobial response, is also important for C elegans resistance against S aureus infection.Similarly, certain key virulence factors of S aureus, such as agr, hla and spa, are commonly required for pathogenicity in both nematodes and mammalian hosts (7).We previously showed that the nematocidal activity of the clinical MRSA strains correlated well with clinical and epidemiological data collected in the Calgary Health Region over a six-year period (3).In the current blinded investigation, we observed a high degree of correlation between nematocidal activity (by MSSA strain virulence) and the clinical scenario corresponding to the outbreak setting with respect to invasive and noninvasive infections versus colonization.Together, these results strongly support the use of C elegans as a robust host model to study the virulence of S aureus strains.
The nematocidal activities of high-virulence MSSA strains required the presence of live bacteria, as heat-killed bacteria lost their virulence, which was consistent with previous studies (3,7).However, all MSSA strains, regardless of high or low nematocidal activities, had similar in vitro growth curves, suggesting that the bacterial in vitro growth pattern does not exert an influence on bacterial nematocidal activities.Although the avirulence strain MSSA8 had a lower in vitro growth rate, it showed a similar colonization and replication ability within the nematode gut as the high and intermediate virulence strains MSSA4 and 5.These results suggest that bacterial virulence factors from live bacteria, other than bacterial replication, are involved in nematocidal activities.
Our previous findings suggested that the virulence of MRSA in C elegans may be determined by the specific genetic background of the MRSA strain (3).In the present study, the high (MSSA1 to MSSA4), intermediate (MSSA5 to MSSA7) and low (MSSA8 to MSSA10) nematocidal MSSA strains possessed unique genetic characteristics and distinct virulence gene profiles, which further supports the above conclusions.MSSA1 belonged to the invasive outbreak strain (MSSA-C) and shared identical genetic and virulence gene profiles with the other isolates (MSSA2 to MSSA4) within the strain group (Figure 3).However, MSSA1 demonstrated slightly less nematocidal activity than its counterparts MSSA2 to MSSA4 (the mean killing rate was 70.2% versus 97.5% over a nine-day experimental course, respectively) (Figure 2A), indicating that it may have some degree of unrelatedness.Interestingly, the antibiotic resistance tests revealed that MSSA2 to MSSA4 isolates were resistant to clindamycin and erythromycin, while the MSSA1 strain was susceptible to both of these agents (data not shown).This further supports the hypothesis that the MSAA2 to MSSA4 isolates may have acquired genes mediating resistance to these agents in addition to other virulence gene(s) favouring their propensity for dissemination, invasiveness and evolution.Comparison of the virulence gene profiles of the virulent (MSSA-C) and intermediately virulent (MSSA-G) strains demonstrated that some virulence genes, such as the exotoxin gene chp, adhesive molecular gene sdrC, and exoenzyme genes hysA and sak, were missing from the MSSA-G strains but not the MSSA-C strains, implying that these genes may impart the ability to cause invasive infection.However, some of these genes were present in the avirulent (or low virulent) MSSA8 to MSSA10 strains.In addition, the toxic shock syndrome (tst) gene has been shown to be associated with an increase in the incidence of bloodstream infections (8).PVL genes have also been suggested to be one of the major virulence factors associated with severe infections due to S aureus and the propensity of community-associated MRSA strains to disseminate (9,10).Only one MSSA8 isolate in the current study carried PVL and tst genes, although that isolate was associated with colonization instead of infection and avirulence in the C elegans model.These results are in agreement with our previous finding that multiple virulence factors or certain combinations of virulence factors acting in concert may be responsible for the nematocidal activity (3).Further studies are needed to elucidate the elaborate mechanisms involved in C elegans killing (virulence) by S aureus.
The present study was the first correlational observation validating the C elegans host virulence model in characterizing the virulence of outbreak and nonoutbreak strains of S aureus.The present study may be limited by its relatively small sample size, as well as the possibility of misclassification of invasive versus noninvasive isolates based on clinical criteria.In summary, we have demonstrated that the C elegans host model provides a validated approach to studying S aureus virulence determinants and could lead to a better understanding of bacterial pathogenesis in staphylococcal infections.