Staphylococcal Persistence Due to Biofilm Formation in Synovial Fluid Containing Prophylactic Cefazolin Staphylococcal Persistence Due to Biofilm Formation in Synovial Fluid Containing Prophylactic Cefazolin Sana S. Dastgheyb,a,c Sommer Hammoud,b Constantinos Ketonis,a,b Andrew Yongkun Liu,a Keith Fitzgerald,a Javad Parvizi,b James Purtill,b Michael Ciccotti,b Irving M. Shapiro,a Michael Otto,c Noreen J. Hickoka Department of Orthopaedic Surgery, Sidney Kimmel Medical College, Thomas Jefferson University, Philadelphia, Pennsylvania, USAa; The Rothman Institute, Philadelphia, Pennsylvania, USAb; National Institute of Allergy and Infectious Diseases, The National Institutes of Health, Bethesda, Maryland, USAc Antibiotic prophylaxis is standard for patients undergoing surgical procedures, yet despite the wide use of antibiotics, break- through infections still occur. In the setting of total joint arthroplasty, such infections can be devastating. Recent findings have shown that synovial fluid causes marked staphylococcal aggregation, which can confer antibiotic insensitivity. We therefore asked in this study whether clinical samples of synovial fluid that contain preoperative prophylactic antibiotics can successfully eradicate a bacterial challenge by pertinent bacterial species. This study demonstrates that preoperative prophylaxis with cefazo- lin results in high antibiotic levels. Furthermore, we show that even with antibiotic concentrations that far exceed the expected bactericidal levels, Staphylococcus aureus bacteria added to the synovial fluid samples are not eradicated and are able to colonize model implant surfaces, i.e., titanium pins. Based on these studies, we suggest that current prophylactic antibiotic choices, de- spite high penetration into the synovial fluid, may need to be reexamined. Infection remains a major concern with orthopedic procedures,although infection rates are generally �1% (1). Because of the high prophylactic antibiotic levels within synovial fluid, the use of minimally invasive techniques, and short surgical times, infection rates remain low. However, with increasing numbers of total knee arthroplasties, infections are projected to exceed 60,000 to 70,000 cases in the United States by 2020 (2, 3). When hardware, such as a prosthesis, is present, infection is recalcitrant to antibiotic treat- ment. Once established, infection can cause prolonged disability, multiple operations, and increased health care costs (3, 4). Ac- cording to the WHO, patients with a surgical site infection have twice the mortality rate, are twice as likely to spend time in an intensive care unit, and are five times more likely to be readmitted to the hospital than uninfected patients (5). To avoid the establishment of infection, antibiotic prophylaxis has become the standard of care (6, 7). For orthopedic procedures, cefazolin (CFZ), which is bactericidal for staphylococci, strepto- cocci, and Escherichia coli (8, 9), is commonly used in the absence of a penicillin allergy (10). Alternative prophylactic antibiotics include fusidic acid, cloxacillin (11), cefixime (12), vancomycin, and gentamicin (13). Interestingly, synovial fluid has been sug- gested to possess antibacterial properties (14–16) that are attrib- uted to hyaluronic acid within the synovial fluid (17), the induc- tion of bactericidal/permeability-increasing protein (15), or unknown compounds, including antimicrobial peptides, within the synovial fluid (14). We recently reported that this ostensi- ble decrease in bacterial numbers, which is perceived to be an “antibacterial” effect, is actually due to the clumping of bacte- ria within synovial fluid, which masks the true microbial load levels (18). The recognition of this behavior also helps to elucidate the difficulties associated with detecting infections. Using standard procedures, the infection rates for revision knee arthroplasties are around 10%, with a 7 to 12% false-negative rate, even in cases in which gross infections are present (19). When more rigorous tech- niques are used to detect infection, �50% of the retrieved im- plants are found to have bacterial contaminants, suggesting that antibiotic prophylaxis has limited success (20). In this study, we sought to determine if the prophylactic anti- biotics, commonly CFZ, present in synovial fluid are able to pre- vent bacterial contamination. Based on our recent studies show- ing that bacterial clumping in synovial fluid dominates its phenotype, we hypothesized that even these high levels of the bac- tericidal antibiotic CFZ would not be able to eradicate a Staphylo- coccus aureus challenge. We used synovial fluid samples from pa- tients who underwent preoperative prophylaxis and established the relative concentrations of active CFZ. We then asked if these concentrations affect S. aureus viability or colonization of a metal surface when bathed in synovial fluid containing pre- operative CFZ. These findings have potential implications for the selection of the most effective means to attain antibacterial prophylaxis, as well as for the methodology to detect joint in- fections. MATERIALS AND METHODS Materials. Samples of synovial fluid were collected both in the clinic dur- ing knee aspirations and in the operating room during total knee arthro- plasties, with the permission of the Thomas Jefferson University institu- tional review board. Cefazolin was obtained from AAP Pharmaceuticals. Kirby-Bauer disks were obtained from BBL. Wheat germ agglutinin (WGA) staining for polysaccharide intercellular adhesin (PIA) (also Received 17 October 2014 Returned for modification 14 December 2014 Accepted 21 January 2015 Accepted manuscript posted online 26 January 2015 Citation Dastgheyb SS, Hammoud S, Ketonis C, Liu AY, Fitzgerald K, Parvizi J, Purtill J, Ciccotti M, Shapiro IM, Otto M, Hickok NJ. 2015. Staphylococcal persistence due to biofilm formation in synovial fluid containing prophylactic cefazolin. Antimicrob Agents Chemother 59:2122–2128. doi:10.1128/AAC.04579-14. Address correspondence to Noreen J. Hickok, noreen.hickok@jefferson.edu. Copyright © 2015, American Society for Microbiology. All Rights Reserved. doi:10.1128/AAC.04579-14 2122 aac.asm.org April 2015 Volume 59 Number 4Antimicrobial Agents and Chemotherapy o n A p ril 5 , 2 0 2 1 a t C A R N E G IE M E L L O N U N IV L IB R h ttp ://a a c.a sm .o rg / D o w n lo a d e d fro m http://dx.doi.org/10.1128/AAC.04579-14 http://dx.doi.org/10.1128/AAC.04579-14 http://aac.asm.org http://aac.asm.org/ called poly-N-acetylglucosamine [PNAG]) was performed using Alexa Fluor 488-labeled WGA (Life Technologies). One-millimeter titanium alloy (Ti6Al4V) wires were obtained from Goodfellow Metals. Bacterial strains. The following strains of S. aureus were used for the study: methicillin-susceptible S. aureus (MSSA) ATCC 25923, methicil- lin-resistant S. aureus (MRSA) clinical isolate TJU2, MRSA LAC (USA300), MSSA Xen 36 (Perkin-Elmer), and MSSA AH1710, a strain that constitutively expresses green fluorescent protein (GFP) that was used to visualize staphylococcal surface colonization (a kind gift from Alexander Horswill, University of Iowa [21]). In addition, Staphylococcus epidermidis strain ATCC 35984 was used in the disk diffusion assays. Qualitative synovial fluid microscopy. Directly after the samples were collected, 1 ml of each sample was removed and imaged by light microscopy at 20� magnification in order to determine the overall cellu- larity and/or crystallinity of the sample. Disk diffusion assays. Each sample of synovial fluid was tested for the presence of antibiotics by an analysis of the zone of inhibition (ZOI) of bacterial growth around a Kirby-Bauer disk, in which antibiotic concen- tration is proportional to the distance from the disk. In brief, 6-mm Kirby- Bauer disks were impregnated with 20-�l samples of synovial fluid, dried overnight, and placed on tryptic soy broth (TSB) agar plates streaked with lawns of S. aureus (200 �l, at a concentration of 107 CFU/ml). Monitoring aggregate formation. An initial inoculum of 104 CFU/ml of MRSA USA300 was incubated in 5 ml of TSB or naive synovial fluid (no preoperative antibiotics) at 180 rpm and 37°C for 8 h. The resultant ag- gregates of USA300 were imaged using digital photography and scanning electron microscopy. In some cases, 108 CFU/ml of S. aureus ATCC 25923 was incubated in synovial fluid containing preoperative CFZ at 180 rpm and 37°C for 3 h and, after aggregate formation, with 10 �g/ml ethidium bromide for 20 min at 37°C; these were imaged using digital photography on a UV light box. Scanning electron microscopy. Large aggregates of bacteria in the synovial fluid were removed using forceps and fixed for 1 h in 4% para- formaldehyde (PFA). The fixed aggregates were serially dehydrated for 10 min each in 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, and 100% ethanol (% ethanol in double-distilled water [ddH2O]). The samples were then affixed onto a coverslip (Nunc), sputter-coated with palladium, and visualized using a Hitachi TM-1000 scanning electron microscopy (SEM) (Ibaraki, Japan). Wheat germ agglutinin staining. Wheat germ agglutinin staining for PIA was performed using Alexa Fluor 488-labeled WGA (Life Technolo- gies) at a concentration of 0.1 mg/ml (Jefferson and Cerca [22]) for 30 min at 37°C. Direct counting of S. aureus survival. The 200-�l samples of synovial fluid (SF) (samples SF4.1 to SF4.5) were challenged with 106 CFU/ml of a clinically isolated strain of S. aureus (TJU2), incubated under static con- ditions for 24 h, and serially diluted (with phosphate-buffered saline [PBS]). The entire 200-�l volume of each sample was plated on TSB agar to ensure that all surviving bacteria were detected. The samples of synovial fluid (200 �l each) were challenged for 24 h at 37°C under static conditions with high inocula (between 107 and 109 CFU/ml) of S. aureus ATCC 25923, a laboratory-adapted strain of S. au- reus, which is highly susceptible to CFZ in TSB, even at a high initial inoculum (18, 23). The samples were serially diluted in PBS, and the entirety (200 �l) of each sample was plated on TSB agar. Monitoring S. aureus survival using Xen 36. A 108-CFU/ml inocu- lum of the luciferase-expressing strain Xen 36 was incubated with 200 �l of fresh samples of synovial fluid from patients given preoperative CFZ (samples SF4.1 to SF4.5) or with 200 �l of samples of TSB with or without 20 �g/ml CFZ for 24 h at 37°C in a standing incubator and then imaged for chemiluminescence (ImageQuant LAS 4000). In other experiments, initial inocula of 103, 104, and 105 CFU/ml of Xen 36 were incubated for 12 h with naive synovial fluid or with different samples of synovial fluid from patients given preoperative CFZ (samples SF4.6 to SF4.8); luciferase expression was then measured by a luminom- eter. S. aureus adherence to titanium pins. A total of 107 CFU/ml of S. aureus AH1710 was added to 1 ml of synovial fluid samples containing preoperative CFZ and incubated 37°C for 24 h. With naive synovial fluid (SF), TSB, or PBS, the preincubation step was omitted. Three Ti alloy (Ti6Al4V) pins were then added and incubated at 37°C for 24 h. The retrieved pins were imaged by confocal laser scanning microscopy. The pins were also incubated with 107 CFU of S. aureus AH1710 in 100 �l of PBS plus 10% TSB or 100 �l of synovial fluid containing preopera- tive CFZ for 48 h. After imaging, the pin incubated in synovial fluid con- taining preoperative CFZ was washed 3� with PBS and incubated in 1 ml of TSB at 37°C for 24 h. After 3 PBS washes, the adherent bacteria were visualized by confocal laser scanning microscopy (CLSM) (Olympus Fluo- View 300). To determine the number of adherent bacteria, three 2-mm Ti6Al4V pins were incubated under static conditions in 200 �l of synovial fluid containing preoperative CFZ and 107 CFU/ml of strain ATCC 25923 for 18 h at 37°C. The pins were removed with sterile forceps, gently washed 3� in PBS, sonicated for 10 min in 0.3% Tween 20-PBS, diluted, plated onto 3M Petrifilm plates, and counted. Statistical analysis. Because of the size of the retrieved synovial fluid samples, some samples could only be tested in one set of 2 to 3 replicates. In those cases, multiple independent synovial fluid samples were always examined but not always averaged. Some evaluations were repeated at least in triplicate in �3 independent experiments. In those cases, statisti- cal analyses were performed, and differences between the values and groups were tested using a paired t test. Significance was set at a P value of �0.05. RESULTS Preoperative cefazolin is present in synovial fluid. We collected synovial fluid samples from patients who were undergoing total knee arthroplasties (with preoperative CFZ) or who were under- going knee aspirations in the office (no antibiotics [naïve]). The fluid had a golden color and tended to be either clear (Fig. 1A) or somewhat cloudy with crystals apparent. Under microscopy, the samples ranged from highly cellular (Fig. 1B) to few visible cells (Fig. 1C). Using a disk diffusion assay in which the clear zone (zone of inhibition [ZOI]) is proportional to the CFZ concentra- tion, we tested the antibiotic content of the synovial fluid obtained from subjects who had been given 2 g of CFZ 1 h prior to surgical procedures (synovial fluid with preoperative CFZ) against differ- ent strains of staphylococci. As expected, a disk containing naive synovial fluid did not inhibit S. aureus ATCC 25923 growth (Fig. 1Di), whereas synovial fluid with 2 g of preoperative CFZ caused a ZOI due to CFZ diffusion from the disk (Fig. 1Dii). The ability of this synovial fluid with preoperative CFZ to cause a ZOI was also observed with S. aureus isolated from a patient with septic arthritis (Fig. 1Diii); with S. aureus AH1710, a genetically engineered iso- late that expresses green fluorescent protein from a plasmid under chloramphenicol selection (Fig. 1Div); with S. epidermidis ATCC 35984 (Fig. 1Dv); and with a clinical MRSA isolate (from Jefferson Hospital; Fig. 1Dvi). The zones of inhibition (ZOI) were measured digitally and compared to the ZOI measured from a control disk incubated with that strain (Fig. 1E). Based on these comparisons, we deduced that synovial fluid with preoperative CFZ had �200 �g/ml CFZ (200� the MIC for S. aureus ATCC 25923 [23]). S. aureus forms aggregates in synovial fluid. We compared S. aureus growth in TSB to that in synovial fluid (Fig. 2A). The bac- teria in TSB grew to a cloudy suspension, whereas in naive syno- vial fluid (nSF), the sample appeared largely clear, with a large Bacterial Persistence in Cefazolin in Synovial Fluid April 2015 Volume 59 Number 4 aac.asm.org 2123Antimicrobial Agents and Chemotherapy o n A p ril 5 , 2 0 2 1 a t C A R N E G IE M E L L O N U N IV L IB R h ttp ://a a c.a sm .o rg / D o w n lo a d e d fro m http://aac.asm.org http://aac.asm.org/ clump floating in the fluid (Fig. 2Ai, red box) and a cloudy pre- cipitate on the bottom of the tube. When the clump was examined more closely (Fig. 2Aii), it appeared to be a large slimy aggregate of bacteria; similar aggregates were observed with three different sy- novial fluid samples with 2 g of preoperative CFZ (samples SF6.1, SF6.2, and SF6.3; Fig. 2Aiii), emphasizing the prevalence of aggre- gate formation. The very dense matrix of the bacterial aggregate in naive synovial fluid was visualized by scanning electron micros- copy (Fig. 2Aiv). Using confocal laser scanning microscopy, a three-dimensional reconstruction was compiled of aggregated S. aureus in synovial fluid with 2 g of preoperative CFZ (sample SF4.1) and showed a dense reticulated structure reminiscent of a biofilm (Fig. 2Av). We thus asked if aggregates in synovial fluid containing preoperative CFZ express a polysaccharide matrix characteristic of biofilms. When stained for polysaccharide inter- cellular adhesin (PIA) (also called poly-N-acetylglucosamine [PNAG]) with wheat germ agglutinin (WGA), S. aureus in syno- vial fluid with 2 g of preoperative CFZ (samples SF6.1, SF6.2, and SF6.3) clearly aggregated, and the aggregates showed green fluo- rescence, indicating the production of biofilm polysaccharide in the aggregated bacteria (Fig. 2Bi). In three different trials with TSB, S. aureus did not aggregate, and staining for PIA was punc- tate (Fig. 2Bii). Preoperative cefazolin in synovial fluid does not eradicate a bacterial challenge. Based on our previous studies and our find- ings in Fig. 2, S. aureus readily clumped in synovial fluid contain- ing preoperative CFZ. We thus asked to what extent this CFZ- containing synovial fluid was able to eradicate a bacterial challenge. When a clinical isolate of S. aureus (from a patient with septic arthritis) was incubated with CFZ-containing synovial fluid, 4 out of 5 different synovial fluid samples (SF4.1 to SF4.5) showed no significant decrease in bacterial viability after a 24-h challenge. Only 1 sample, SF4.1, showed a marked decrease (�3 log) from the initial inoculum (Fig. 3A). Using the same five sam- ples of synovial fluid, a visual analysis of S. aureus Xen 36 chemi- luminescence revealed that luciferase activity appeared to be com- pletely abolished in TSB containing CFZ. Conversely, luciferase activity persisted in samples SF4.1 to SF4.5, as seen in Fig. 3A, with noticeably lower activity in SF4.1 and robust activities in SF4.3 to SF4.5. By a direct measurement of luminescence, naive synovial fluid samples challenged with 102, 103, and 104 CFU/ml of S. aureus Xen FIG 1 Synovial fluid containing preoperative antibiotics. (A) Representative image of a synovial fluid sample obtained during a total knee arthroplasty that contains preoperative CFZ. (B) Typical microscopic image of synovial fluid displaying high cellularity. (C) Typical microscopic image of synovial fluid displaying low cellularity. (D) Zones of inhibition (ZOI) obtained with synovial fluid containing preoperative antibiotics (2 g of CFZ). In all, at least eight different synovial fluid samples were tested on each strain in triplicate. The images shown are representative of the ZOI obtained across all samples. Shown within the panel are ZOI associated with synovial fluid from a knee aspiration that does not contain antibiotics (naive SF) (i), synovial fluid containing preoperative antibiotics in S. aureus ATCC 25923 (ii), synovial fluid containing preoperative antibiotics in an S. aureus clinical isolate (iii), synovial fluid containing preoperative antibiotics in S. aureus AH1710 (iv), synovial fluid containing preoperative antibiotics in S. epidermidis ATCC 35984 (v), and synovial fluid containing preoperative antibiotics in an S. aureus MRSA isolate (Thomas Jefferson University Hospital [TJU2]) (vi). (E) ZOIs measured digitally with ImageJ using the synovial fluid samples and strains shown in panel D. Each circle represents the average ZOI from at least three disks/strain. The red diamonds represent the ZOI obtained with 200 �g/ml CFZ/disk. Dastgheyb et al. 2124 aac.asm.org April 2015 Volume 59 Number 4Antimicrobial Agents and Chemotherapy o n A p ril 5 , 2 0 2 1 a t C A R N E G IE M E L L O N U N IV L IB R h ttp ://a a c.a sm .o rg / D o w n lo a d e d fro m http://aac.asm.org http://aac.asm.org/ 36 for 12 h showed a stepwise increase in luminescence. When the samples with preoperative CFZ were challenged with Xen 36, SF4.6 and SF4.7 showed levels of luminescence that were similar to those of the samples with 103 to 104 CFU in naive synovial fluid, independent of the size of the starting inoculum. Sample SF4.8, which had been dosed with 1 g of preoperative CFZ, showed a stepwise increase in luminescence with increasing inocula (Fig. 3C). In contrast, in TSB (Fig. 3D), luciferase activity was high with all three inocula; after 12 h, TSB containing 20 �g/ml CFZ resulted in markedly decreased luminescence, and these levels were �10� lower than those measured in CFZ-containing synovial fluid. Of note, the CFZ concentrations in SF were �10-fold greater than those in TSB. S. aureus in synovial fluid containing antibiotics readily ad- heres to metal surfaces. We reasoned that if viable bacteria were indeed present in synovial fluid containing preoperative CFZ, they should readily adhere to titanium alloy (Ti) pins, thus modeling biofilm initiation in the presence of an implant. After 24 h of incubation, �10,000 CFU/ml of bacteria were recovered from Ti pins that had been inoculated with 107 CFU/ml of S. aureus in the presence of TSB (Fig. 4A). Importantly, no bacterial colonization was detected on Ti pins in which S. aureus was incubated in TSB�CFZ before the addition of the pin. In contrast, between 200 and 2,500 CFU of adherent S. aureus (average � standard devia- tion, 1,156 � 832 CFU) were recovered from Ti pins incubated in S. aureus-inoculated CFZ-containing synovial fluid. When the pins were examined for presence of GFP-expressing bacteria, a Ti pin that was not challenged with bacteria showed no fluorescence (Fig. 4Bi) and served as a negative control. Abundant fluorescence was apparent when a pin was incubated with S. au- reus in PBS plus 10% TSB, a minimal growth medium (Fig. 4Bii). In synovial fluid containing preoperative CFZ, S. aureus coloniza- tion was not initially detectable (Fig. 4Biii) but became apparent when the colonized pin was removed from the synovial fluid and incubated in TSB for an additional 24 h. This additional incuba- tion supported sufficient bacterial growth to be detected visually (Fig. 4Biv). DISCUSSION Antibiotic prophylaxis during orthopedic procedures is used to minimize the establishment of infection. In this paper, we show that synovial fluid containing CFZ due to a 2-g preoperative pro- phylactic protocol has attenuated efficacy against S. aureus in sy- novial fluid. Furthermore, we show that Ti pins in synovial fluid that contains CFZ due to preoperative prophylaxis can be colo- nized by added S. aureus at clinically significant numbers. In the presence of metal implants, it is now known that both in vitro (24, 25) and in vivo, as few as 100 CFU of S. aureus are sufficient to establish infection (26). Accompanying this diminished level of efficacy of antibiotics in synovial fluid, our in vitro findings show that in general, up to 90% of bacteria survive, suggesting that preoperative prophylaxis may be ineffective in the synovial fluid environment. Taken together, these in vitro studies suggest that the current standard for prophylaxis may be inadequate to combat S. aureus contamination. Antibiotic penetration is thought to determine the effective- ness of prophylaxis (13, 27–29). Because CFZ is bactericidal, ob- tains high levels within the synovial fluid within 30 to 60 min, and has a broad spectrum of coverage, it is recommended for preop- erative prophylaxis in procedures that breach the joint capsule (33). Based on our data, postprophylactic synovial fluid concen- trations are �100 to 200 �g/ml, which is �400� the MIC for S. aureus (9). These concentrations were determined through the testing of 23 synovial fluid samples containing preoperative CFZ. Despite these CFZ concentrations, a large number of S. aureus bacteria persist and are seen to form clumps in the antibiotic- containing synovial fluid. These clumps are biofilm-like in that they are stained by WGA, which stains the polysaccharide coating of S. aureus biofilms containing PIA. Our recent work shows that the formation of these clumps depends on a proteinaceous com- ponent that is critical for clump formation (18). This aggregation causes an accompanying decrease in virulence and a biofilm-like phenotype (18; S. S. Dastgheyb, M. Otto, and N. J. Hickok, un- published data). In addition to the protein content, the viscosity (150 � 50 MPa [30]) of synovial fluid favors the formation of biofilms, whether anchored or floating. Therefore, while plank- tonic S. aureus in TSB might be effectively eradicated by 200 �g/ml CFZ, S. aureus readily survived in this same concentration of an- tibiotic in synovial fluid. This recalcitrance to the effects of CFZ was found in several strains, including a clinical strain of S. aureus, and with inocula ranging from 103 to 107 CFU/ml. It is worth FIG 2 Bacterial growth and clumping in synovial fluid. (A) i, 5-h culture of S. aureus incubated in TSB or naive SF (nSF) (no antibiotics). These clumps have already been described by Dastgheyb et al. (18). Note the S. aureus aggregate that is boxed in red. ii, magnification at 10� of the aggregated S. aureus ATCC 25923 that was shown within the red box. iii, aggregated bacteria after incuba- tion in SF6.1, SF6.2, or SF 6.3, each containing 2 g of CFZ from preoperative prophylaxis, visualized by ethidium bromide staining, and photographed. Scale bar 1.5 mm. iv, scanning electron micrograph of an aggregate formed in naive synovial fluid. Scale bar 50 �m. v, three-dimensional reconstruc- tion of confocal laser scanning micrographs of an S. aureus ATCC 25923 ag- gregate formed in synovial fluid containing 2 g of preoperative CFZ. Similar images were obtained with other synovial fluid samples containing preopera- tive CFZ. (B) WGA staining for PIA/PNAG in S. aureus ATCC 25923 shown incubated in three synovial fluid samples (i) (left to right: SF3.1, SF3.2, and SF3.3) or three separate cultures of TSB without antibiotics (ii). Bacterial Persistence in Cefazolin in Synovial Fluid April 2015 Volume 59 Number 4 aac.asm.org 2125Antimicrobial Agents and Chemotherapy o n A p ril 5 , 2 0 2 1 a t C A R N E G IE M E L L O N U N IV L IB R h ttp ://a a c.a sm .o rg / D o w n lo a d e d fro m http://aac.asm.org http://aac.asm.org/ noting that throughout the study, the higher doses of bacteria have been used to more readily visualize the aggregation phenomenon that we are describing. However, as seen in the data in Fig. 3, the lower levels of bacterial contamination show similar behavior at higher levels. We would argue, therefore, that the high inocula used throughout will model the more clinically relevant lower levels of bacteria. Bacterial survival in CFZ-containing synovial fluid is illus- trated by the adherence of S. aureus to a titanium alloy pin. While a titanium alloy was used, the most common alloy used in knee hardware is cobalt-chrome-molybdenum. Data from others (31, 32) and our unpublished data (S. Dastgheyb, A. E. Villaruz, K. Y. Le, V. Y. Tan, A. C. Duong, S. S. Chatterjee, G. Y. C. Cheung, H.-S. Joo, N. J. Hickok, and M. Otto, submitted for publication) sup- port the adherence of S. aureus to both materials. By direct count- ing, approximately 103 of the 104 bacteria inoculated into the sy- FIG 3 Bacterial survival in antibiotic-containing synovial fluid. (A) Survival of S. aureus incubated in synovial fluid from patients given 2 g of preoperative (pre-op) CFZ (samples SF4.1 to SF4.5). Red dashed line represents initial inoculum. (B) Xen-36 chemiluminescence after 24 h incubation in TSB (0 or 20 �g/ml CFZ), or 5 samples of synovial fluid containing 2 g of preoperative antibiotics (samples SF4.1 to SF4.5). (C) Intensity of Xen 36 luminescence after 12 h of incubation of 100, 1,000, or 10,000 CFU in 100-�l samples of naive synovial fluid or SF4.6, SF4.7, and SF4.8 from patients given 2, 2, and 1 g of preoperative CFZ, respectively. (D) Intensity of Xen 36 luminescence after 12 h of incubation with 100, 1,000, or 10,000 CFU in 100-�l samples of TSB containing 0 or 20 �g/ml CFZ. The numbers of repeats with the number of independent experiments in parentheses are indicated for each panel. FIG 4 Bacterial survival and adherence in the presence of antibiotic-containing synovial fluid. (A) CFU recovered from Ti alloy pins incubated in TSB or in 8 synovial fluid samples containing 2 g of preoperative CFZ containing 107 CFU/ml of S. aureus for 24 h. All bacterial recoveries were performed independently three times in three independent experiments. (B) All experiments used three independent pins/experiment, and each experiment was repeated three times. i, control Ti alloy pin that has not been exposed to S. aureus. ii, adherence of S. aureus (green fluorescence) to Ti alloy pin in PBS plus 10% TSB. iii, adherence of S. aureus to Ti alloy pin in synovial fluid containing preoperative antibiotics. iv, pin in panel iii after removal from synovial fluid and 24 h of incubation in TSB. The green fluorescence is due to bacterial growth during this time frame. Scale bar 500 �m. Dastgheyb et al. 2126 aac.asm.org April 2015 Volume 59 Number 4Antimicrobial Agents and Chemotherapy o n A p ril 5 , 2 0 2 1 a t C A R N E G IE M E L L O N U N IV L IB R h ttp ://a a c.a sm .o rg / D o w n lo a d e d fro m http://aac.asm.org http://aac.asm.org/ novial fluid adhered to the pin. The recently published work of Dastgheyb et al. (18) established that despite early differences in proliferation rates, the rates of S. aureus proliferation in TSB and synovial fluid at 24 h are approximately equal. Thus, this differ- ence in colonization is unlikely to be due to a simple difference in proliferation. Independent of surface adhesion, the nonadherent bacteria in synovial fluid exhibit a biofilm-like phenotype and demonstrate a biofilm-like recalcitrance to antibiotic treatment. In synovial fluid, it is clear from these studies and our previous work that this aggregation alters antibiotic sensitivity and may have an impact on the numbers of bacteria that colonize an im- plant. Importantly, while performing these experiments, we relied on multiple synovial fluid samples to validate our results (1), as volumes were limited (2), and the use of multiple samples strengthened our observations. Interestingly, synovial fluid rap- idly reforms after joint replacement so that the synovial environ- ment is reestablished with a new more susceptible surface for bac- terial colonization (i.e., the implant). It is important to realize that when infection has been established, an important part of the treatment is extensive lavage and removal of any fibrin clots; based on the data in this paper, this removal becomes critical, as this clot is an important structure in the maintenance of indolent bacteria. In an arthroplasty setting, preoperative antibiotics are accom- panied by the removal of synovial fluid during the arthrotomy, followed by blood clot formation, and an additional one to two doses of antibiotic are given postoperatively. These additional procedures and antibiotic treatments may be sufficient to sterilize the knee environment, but this scenario needs to be explicitly tested in light of our findings. Importantly, the results we report may be especially pertinent to clinical situations in which synovial fluid is aspirated without performance of joint replacements. A single preoperative dose of antibiotic is typically recommended, and the data here suggest that this is inadequate. Further explora- tion of clinical synovial fluid samples is warranted. ACKNOWLEDGMENTS We thank Alexander Horswill for generously supplying the fluorescent S. aureus strain AH1710. This study was supported by NIH grants HD06153 and DE019901 to N.J.H., as well as the Intramural Research Program of the National Insti- tute of Allergy and Infectious Diseases (NIAID), the National Institutes of Health (NIH) (grant ZIA AI000904-13 to M.O.), and a T32 NIH training grant (T32-AR-052273 to I.M.S.). We declare no conflicts of interest. REFERENCES 1. Pulido L, Ghanem E, Joshi A, Purtill JJ, Parvizi J. 2008. Periprosthetic joint infection: the incidence, timing, and predisposing factors. Clin Orthop Relat Res 466:1710 –1715. http://dx.doi.org/10.1007/s11999 -008-0209-4. 2. Kurtz SM, Lau E, Schmier J, Ong KL, Zhao K, Parvizi J. 2008. Infection burden for hip and knee arthroplasty in the United States. J Arthroplasty 23:984 –991. http://dx.doi.org/10.1016/j.arth.2007.10.017. 3. Kurtz SM, Lau E, Watson H, Schmier JK, Parvizi J. 2012. Economic burden of periprosthetic joint infection in the United States. J Arthro- plasty 27:61– 65 e61. http://dx.doi.org/10.1016/j.arth.2012.02.022. 4. Borens O, Corvec S, Trampuz A. 2012. Diagnosis of periprosthetic joint infections. Hip Int 22(Suppl 8):S9 –S14. http://dx.doi.org/10.5301/HIP .2012.9565. 5. World Health Organization. 2009. WHO guidelines for safe surgery 2009: safe surgery saves lives. World Health Organization, Geneva, Switzerland. http://whqlibdoc.who.int/publications/2009/9789241598552_eng.pdf. 6. Block JE, Stubbs HA. 2005. Reducing the risk of deep wound infection in primary joint arthroplasty with antibiotic bone cement. Orthopedics 28: 1334 –1345. http://dx.doi.org/10.3928/0147-7447-20051101-13. 7. Darouiche RO. 2003. Antimicrobial approaches for preventing infections associated with surgical implants. Clin Infect Dis 36:1284 –1289. http://dx .doi.org/10.1086/374842. 8. Schurman DJ, Hirshman HP, Kajiyama G, Moser K, Burton DS. 1978. Cefazolin concentrations in bone and synovial fluid. J Bone Joint Surg Am 60:359 –362. 9. McAllister TA, Tait SC, Perman MG, Park AC. 1976. Clinical bacteri- ology of cephazolin. Scott Med J 20:220 –227. 10. de Lalla F. 2001. Antibiotic prophylaxis in orthopedic prosthetic surgery. J Chemother 13(Suppl 2):48 –53. http://dx.doi.org/10.1179/joc.2001.13 .Supplement-2.48. 11. Somekh E, Golan T, Tanay A, Poch F, Dan M. 1999. Concentration and bactericidal activity of fusidic acid and cloxacillin in serum and synovial fluid. J Antimicrob Chemother 43:593–596. http://dx.doi.org/10.1093/jac /43.4.593. 12. Somekh E, Heifetz L, Dan M, Poch F, Hafeli H, Tanai A. 1996. Penetration and bactericidal activity of cefixime in synovial fluid. Antimi- crob Agents Chemother 40:1198 –1200. 13. Ritter MA, Barzilauskas CD, Faris PM, Keating EM. 1989. Vancomycin prophylaxis and elective total joint arthroplasty. Orthopedics 12:1333– 1336. 14. Gruber BF, Miller BS, Onnen J, Welling R, Wojtys EM. 2008. Antibac- terial properties of synovial fluid in the knee. J Knee Surg 21:180 –185. http://dx.doi.org/10.1055/s-0030-1247816. 15. Inman RD, Chiu B. 1996. Comparative microbicidal activity of synovial fluid on arthritogenic organisms. Clin Exp Immunol 104:80 – 85. http://dx .doi.org/10.1046/j.1365-2249.1996.d01-650.x. 16. Pruzanski W, Leers WD, Wardlaw AC. 1974. Bacteriolytic and bacteri- cidal activity of sera and synovial fluids in rheumatoid arthritis and in osteoarthritis. Arthritis Rheum 17:207–218. http://dx.doi.org/10.1002/art .1780170303. 17. Pirnazar P, Wolinsky L, Nachnani S, Haake S, Pilloni A, Bernard GW. 1999. Bacteriostatic effects of hyaluronic acid. J Periodontol 70:370 –374. http://dx.doi.org/10.1902/jop.1999.70.4.370. 18. Dastgheyb S, Parvizi J, Shapiro IM, Hickok NJ, Otto M. Effect of biofilms on recalcitrance of staphylococcal joint infection to antibiotic treatment. J Infect Dis, in press. 19. Ackerman D, Lett P, Galat DD, Jr, Parvizi J, Stuart MJ. 2008. Results of total hip and total knee arthroplasties in patients with synovial chondro- matosis. J Arthroplasty 23:395– 400. http://dx.doi.org/10.1016/j.arth.2007 .06.014. 20. Jacovides CL, Kreft R, Adeli B, Hozack B, Ehrlich GD, Parvizi J. 2012. Successful identification of pathogens by polymerase chain reaction (PCR)-based electron spray ionization time-of-flight mass spectrometry (ESI-TOF-MS) in culture-negative periprosthetic joint infection. J Bone Joint Surg Am 94:2247–2254. http://dx.doi.org/10.2106/JBJS.L.00210. 21. Malone CL, Boles BR, Lauderdale KJ, Thoendel M, Kavanaugh JS, Horswill AR. 2009. Fluorescent reporters for Staphylococcus aureus. J Mi- crobiol Methods 77:251–260. http://dx.doi.org/10.1016/j.mimet.2009.02 .011. 22. Jefferson KK, Cerca N. 2006. Bacterial-bacterial cell interactions in bio- films: detection of polysaccharide intercellular adhesins by blotting and confocal microscopy. Methods Mol Biol 341:119 –126. http://dx.doi.org /10.1385/1-59745-113-4:119. 23. Dhawan VK, Maier MK, Nayyar M, Chuah SK, Thadepalli H. 1984. In vitro activity of cefmenoxime, cefotaxime, latamoxef, cefazolin, nafcillin and vancomycin against 53 endocarditis and bacteremic strains of Staph- ylococcus aureus. Chemotherapy 30:328 –330. http://dx.doi.org/10.1159 /000238288. 24. Antoci V, Jr, King SB, Jose B, Parvizi J, Zeiger AR, Wickstrom E, Freeman TA, Composto RJ, Ducheyne P, Shapiro IM, Hickok NJ, Adams CS. 2007. Vancomycin covalently bonded to titanium alloy pre- vents bacterial colonization. J Orthop Res 25:858 – 866. http://dx.doi.org /10.1002/jor.20348. 25. Costerton JW. 2005. Biofilm theory can guide the treatment of device- related orthopaedic infections. Clin Orthop Relat Res 437:7–11. http://dx .doi.org/10.1097/00003086-200508000-00003. 26. Li D, Gromov K, Soballe K, Puzas JE, O’Keefe RJ, Awad H, Drissi H, Schwarz EM. 2008. Quantitative mouse model of implant-associated osteomyelitis and the kinetics of microbial growth, osteolysis, and hu- Bacterial Persistence in Cefazolin in Synovial Fluid April 2015 Volume 59 Number 4 aac.asm.org 2127Antimicrobial Agents and Chemotherapy o n A p ril 5 , 2 0 2 1 a t C A R N E G IE M E L L O N U N IV L IB R h ttp ://a a c.a sm .o rg / D o w n lo a d e d fro m http://dx.doi.org/10.1007/s11999-008-0209-4 http://dx.doi.org/10.1007/s11999-008-0209-4 http://dx.doi.org/10.1016/j.arth.2007.10.017 http://dx.doi.org/10.1016/j.arth.2012.02.022 http://dx.doi.org/10.5301/HIP.2012.9565 http://dx.doi.org/10.5301/HIP.2012.9565 http://whqlibdoc.who.int/publications/2009/9789241598552_eng.pdf http://dx.doi.org/10.3928/0147-7447-20051101-13 http://dx.doi.org/10.1086/374842 http://dx.doi.org/10.1086/374842 http://dx.doi.org/10.1179/joc.2001.13.Supplement-2.48 http://dx.doi.org/10.1179/joc.2001.13.Supplement-2.48 http://dx.doi.org/10.1093/jac/43.4.593 http://dx.doi.org/10.1093/jac/43.4.593 http://dx.doi.org/10.1055/s-0030-1247816 http://dx.doi.org/10.1046/j.1365-2249.1996.d01-650.x http://dx.doi.org/10.1046/j.1365-2249.1996.d01-650.x http://dx.doi.org/10.1002/art.1780170303 http://dx.doi.org/10.1002/art.1780170303 http://dx.doi.org/10.1902/jop.1999.70.4.370 http://dx.doi.org/10.1016/j.arth.2007.06.014 http://dx.doi.org/10.1016/j.arth.2007.06.014 http://dx.doi.org/10.2106/JBJS.L.00210 http://dx.doi.org/10.1016/j.mimet.2009.02.011 http://dx.doi.org/10.1016/j.mimet.2009.02.011 http://dx.doi.org/10.1385/1-59745-113-4:119 http://dx.doi.org/10.1385/1-59745-113-4:119 http://dx.doi.org/10.1159/000238288 http://dx.doi.org/10.1159/000238288 http://dx.doi.org/10.1002/jor.20348 http://dx.doi.org/10.1002/jor.20348 http://dx.doi.org/10.1097/00003086-200508000-00003 http://dx.doi.org/10.1097/00003086-200508000-00003 http://aac.asm.org http://aac.asm.org/ moral immunity. J Orthop Res 26:96 –105. http://dx.doi.org/10.1002 /jor.20452. 27. Mini E, Grassi F, Cherubino P, Nobili S, Periti P. 2001. Preliminary results of a survey of the use of antimicrobial agents as prophylaxis in orthopedic surgery. J Chemotherapy 13(Suppl 2):73–79. http://dx.doi.org /10.1179/joc.2001.13.Supplement-2.73. 28. Norden CW. 1991. Antibiotic prophylaxis in orthopedic surgery. Rev Infect Dis 13(Suppl 10):S842–S846. http://dx.doi.org/10.1093/clinids/13 .Supplement_10.S842. 29. Yamada K, Matsumoto K, Tokimura F, Okazaki H, Tanaka S. 2011. Are bone and serum cefazolin concentrations adequate for antimicrobial pro- phylaxis? Clin Orthop Relat Res 469:3486 –3494. http://dx.doi.org/10 .1007/s11999-011-2111-8. 30. Jebens EH, Monk-Jones ME. 1959. On the viscosity and pH of synovial fluid and the pH of blood. J Bone Joint Surg Br 41-B:388 – 400. 31. Berry CW, Moore TJ, Safar JA, Henry CA, Wagner MJ. 1992. Antibac- terial activity of dental implant metals. Implant Dent 1:59 – 65. http://dx .doi.org/10.1097/00008505-199200110-00006. 32. Gómez-Barrena E, Esteban J, Medel F, Molina-Manso D, Ortiz-Perez A, Cordero-Ampuero J, Puertolas JA. 2012. Bacterial adherence to sepa- rated modular components in joint prosthesis: a clinical study. J Orthop Res 30:1634 –1639. http://dx.doi.org/10.1002/jor.22114. 33. American Academy of Orthopaedic Surgeons. 2004. Recommendations for the use of intravenous antibiotic prophylaxis in primary total joint arthroplasty. AAOS information statement, information statement 1027. American Academy of Orthopaedic Surgeons, Rosemont, IL. Dastgheyb et al. 2128 aac.asm.org April 2015 Volume 59 Number 4Antimicrobial Agents and Chemotherapy o n A p ril 5 , 2 0 2 1 a t C A R N E G IE M E L L O N U N IV L IB R h ttp ://a a c.a sm .o rg / D o w n lo a d e d fro m http://dx.doi.org/10.1002/jor.20452 http://dx.doi.org/10.1002/jor.20452 http://dx.doi.org/10.1179/joc.2001.13.Supplement-2.73 http://dx.doi.org/10.1179/joc.2001.13.Supplement-2.73 http://dx.doi.org/10.1093/clinids/13.Supplement_10.S842 http://dx.doi.org/10.1093/clinids/13.Supplement_10.S842 http://dx.doi.org/10.1007/s11999-011-2111-8 http://dx.doi.org/10.1007/s11999-011-2111-8 http://dx.doi.org/10.1097/00008505-199200110-00006 http://dx.doi.org/10.1097/00008505-199200110-00006 http://dx.doi.org/10.1002/jor.22114 http://aac.asm.org http://aac.asm.org/ Staphylococcal Persistence Due to Biofilm Formation in Synovial Fluid Containing Prophylactic Cefazolin MATERIALS AND METHODS Materials. Bacterial strains. Qualitative synovial fluid microscopy. Disk diffusion assays. Monitoring aggregate formation. Scanning electron microscopy. Wheat germ agglutinin staining. Direct counting of S. aureus survival. Monitoring S. aureus survival using Xen 36. S. aureus adherence to titanium pins. Statistical analysis. RESULTS Preoperative cefazolin is present in synovial fluid. S. aureus forms aggregates in synovial fluid. Preoperative cefazolin in synovial fluid does not eradicate a bacterial challenge. S. aureus in synovial fluid containing antibiotics readily adheres to metal surfaces. DISCUSSION ACKNOWLEDGMENTS REFERENCES