PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
J Orthop Res. Author manuscript; available in PMC Jun 24, 2009.
Published in final edited form as:
PMCID: PMC2701346
NIHMSID: NIHMS94440
A Quantitative Mouse Model of Implant-Associated Osteomyelitis and the Kinetics of Microbial Growth, Osteolysis and Humoral Immunity
Dan Li,1* Kirill Gromov,1,2* Kjeld Søballe,2 J. Edward Puzas,1 Regis J. O’Keefe,1 Hani Awad,1 Hicham Drissi,1 and Edward M. Schwarz1,3
1The Center for Musculoskeletal Research, University of Rochester, Rochester, New York
2The Department of Orthopedics, Aarhus University Hospital, Aarhus, Denmark
3To whom correspondence should be addressed: Dr. Edward M. Schwarz, The Center for Musculoskeletal Research, University of Rochester Medical Center, 601 Elmwood Avenue, Box 665, Rochester, NY 14642, Phone 585-275-3063, FAX 585-756-4727, E-mail: Edward_Schwarz/at/URMC.Rochester.edu
*These authors contributed equally to this work.
Although osteomyelitis (OM) remains a serious problem in orthopaedics, progress has been limited by the absence of an in vivo model that can quantify the bacterial load, metabolic activity of the bacteria over time, immunity and osteolysis. To overcome these obstacles, we developed a murine model of implant-associated OM in which a stainless steel pin is coated with Staphylococcus aureus and implanted transcortically through the tibial metaphysis. X-ray and micro-CT demonstrated concomitant osteolysis and reactive bone formation, which was evident by day 7. Histology confirmed all the hallmarks of implant-associated OM, namely: osteolysis, sequestrum formation and involucrum of Gram-positive bacteria inside a biofilm within necrotic bone. Serology revealed that mice mount a protective humoral response that commences with an IgM response after one weak, and converts to a specific IgG2b response against specific S. aureus proteins by day 11 post-infection. Real-time quantitative PCR (RTQ-PCR) for the S. aureus specific nuc gene determined that the peak bacterial load occurs 11 days post-infection. This coincidence of decreasing bacterial load with the generation of specific antibodies is suggestive of protective humoral immunity. Longitudinal in vivo bioluminescent imaging (BLI) of luxA-E transformed S. aureus (Xen29) combined with nuc RTQ-PCR demonstrated the exponential growth phase of the bacteria immediately following infection that peaks on day 4, and is followed by the biofilm growth phase at a significantly lower metabolic rate (p<0.05). Collectively, these studies demonstrate the first quantitative model of implant-associated OM that defines the kinetics of microbial growth, osteolysis and humoral immunity following infection.
Keywords: Osteomyelitis, Staphylococcus aureus, Osteolysis, Humoral Immunity, Bioluminescent Imaging
Osteomyelitis (OM) is a common infectious disease characterized by progressive inflammation and bone destruction 13. This condition may be caused by a variety of microorganism, however, Staphylococcus aureus is responsible for >80% of these infections 1. Although the total number of osteomyelitis cases is high in that approximately 112,000 orthopaedic device-related infections occur per year in the US at an approximate hospital cost of $15,000–70,000 per incident, the infection rates for joint prosthesis and fracture-fixation devices have been only 0.3–11% and 5–15% of cases respectively over the last decade 2,3, which resulted in a low interest in rigorous prospective clinical studies. Another challenging issue hampering the progress in the OM research field is the absence of a quantitative animal model. It is well known that in vitro cultures rapidly select for growth of organisms that do not elaborate an extracellular capsule. As such, biofilm biology, which plays a critical role in resistance of chronic OM to antibiotic therapy by serving as a dominant protective barrier from the action of antibiotics, can only be studied with in vivo models 4. To date, much of our knowledge of the pathogenesis of OM comes from animal models 5, which have existed in chicken 6, rat 7,8, guinea pig 9, rabbit 10, dog 11, sheep 12, goat 13 and most recently mouse 14. While these models have been used to confirm the importance of bacterial adhesins identified from in vitro assays 1517, none of them contain quantitative endpoints that can determine bacterial load or growth in vivo. Thus, they are of limited value for assessing drug effects, bacterial mutants and the role of host factors during the establishment of chronic OM.
In this study, we developed a novel murine model of implant-associated osteomyelitis in which a stainless steel pin is coated with S. aureus and implanted transcortically through the tibial metaphysis. The resulting infection closely resembles clinical OM. This generates a reproducible abscess without hematogenous spreading or death in >90% of mice. By adapting real-time quantitative PCR (RTQ-PCR) to detect the S. aureus specific nuc gene that is used to assess contaminated food 18, and a bioluminescent strain (Xen29) that can be quantified using longitudinal in vivo bioluminescent imaging (BLI) 19, we have elucidated the pattern of pathogenic growth during the establishment of implant-associated OM. We also performed serology and micro-CT analyses to assess the host response to infection. Our findings demonstrate the establishment of a quantitative small animal surrogate that can be used to evaluate the efficacy of novel interventions for OM.
S. aureus strains and pathogenic challenge
The UAMS-1 strain of S. aureus (ATCC 49230) was obtained from the American Type Culture Collection (Manassas, VA). The bioluminescent strain of S. aureus, Xen29 (derived from ATCC 12600), was obtained from Xenogen Inc. (Cranbury, NJ). Pathogenic challenge was initiated via a contaminated 0.25 mm diameter insect pin (Fine Science Tools, Foster City, CA) that was generated as follows. The pins were autoclaved and stored in 70% ethanol. After air-drying, the pins were incubated in 1.5 ml of an overnight luria broth culture of S. aureus for 20 min. Then, the pins were air dried for 5 minutes before trans-tibial implantation. The inoculating dose of bacteria was determined to be 9.5 +/− 3.7 × 105 CFU of UAMS-1 and 4.2 +/− 0.5 × 105 CFU of Xen29 per pin, by vigorously vortexing the pin in PBS to resuspend the bacteria and growing dilutions on agar plates.
Animal surgery
All animal studies were performed under University Committee for Animal Resources approved protocols. C57BL/6 female 6–8 week-old mice were anesthetized with Ketamine (100 mg/kg) and Xylazine (10 mg/kg), their left tibiae were shaved and the skin was cleansed with 70% ethanol. Infection was initiated by placing the pin transcortically through the tibia via medial to lateral implantation. The pin was bent at both ends for stability and cut adjacent to the skin on both ends, which allowed it to be covered by the skin and to eliminate additional environmental exposure. Once the mice recovered from the anesthesia, they were returned to standard isolator cages without additional treatment until sacrifice, except for the gentamycin treatment group that received 0.1 mg/kg/day i.p.
In vivo radiology and bioluminescent imaging
Longitudinal osteolysis was assessed radiographically using a Faxitron Cabinet x-ray system (Faxitron, Wheeling, IL, USA) as we have previously described 20. Bioluminescent imaging (BLI) of mice infected with Xen29 was performed on day 0, 4, 7, 11, 14, 18 post-infection using a Xenogen IVIS camera system (Xenogen Corporation, Alameda, Calif.). Five minute high sensitivity ventral images were taken at each time point. The BLI was quantified with the LivingImage software package 2.50.1 (Xenogen Corporation, Alameda, Calif.) by analyzing a fixed 1.5 cm diameter region of interest (ROI) centered on the pin. The photon signal was calculated as p/sec/cm2/sr.
Micro-CT analyses
After sacrifice, the pin was removed and the disarticulated tibiae were analyzed by high-resolution (10.5 μm) micro-computed tomography (μCT) (VivaCT 40; Scanco Medical AG, Basserdorf, Switzerland) to render 3D images of the diaphysis as we have previously described 21. Images from each tibia were binarized individually at identical thresholds to allow for unbiased identification of the cortical pinholes and the extent of the osteolytic lesion in the cortices. The three-dimensional binary image was then rendered from the threshold slices, and sagital sections through the tibia were digitally obtained so that the rendered images could be rotated to visualize the pinholes and osteolytic lesions on the lateral and medial cortical surfaces of the specimens. Since the infected pin was introduced through the medial side, and we observed more consistent and reproducible osteolysis on the medial side, we chose to only quantify the area of osteolysis on the medial side. The maximal osteolytic area was computed for each specimen using a semi-automated filter in Adobe Photoshop® CS (Adobe Systems, Inc., San Jose, CA) which thresholds the pinhole area, sums the number of pixels in the thresholded region, and then computes the area in mm2 after calibrating against 1 mm scale bar (9216 pixels/mm2). This area measurement was then used to quantitatively describe the maximum size of the osteolytic cortical lesions for the group as mean ± SEM (Figure 1).
Figure 1
Figure 1
Radiographic progression of trans-tibial implant-associated osteomyelitis
Histologic evaluation of OM
After μCT, the tibial samples were processed for decalcified histology and stained with Orange G/alcian blue (H&E), Gram-stained, or tested for tartrate-resistant acid phosphatase (TRAP) activity as we have described previously 22.
DNA purification and nuc/β-actin real time quantitative PCR (RTQ-PCR)
Infected tibiae were disarticulated and separated from the surrounding soft tissues. Then the tibia was cut into small pieces with scissors, and demineralized individually with 0.5 M EDTA (pH 7.5) in 1.5 ml polypropylene tubes. The tubes were agitated on a shaker at 4°C for 24 h, centrifuged at 5500 rpm for 20 minutes and the precipitate was resuspended in fresh EDTA. This process was repeated 3 to 5 times until the bone fragments were completely decalcified. The samples were then washed three times with ddH2O and the pellet was resuspended in 360 ml buffer ATL (QIAGEN, Valencia, CA) with 40 μl proteinase K (QIAGEN, Valencia, CA), and incubated at 55°C until the bone tissues were fully digested and the solution was clear. Then 400 μl buffer AL (QIAGEN, Valencia, CA) with 5 μl proteinase K was added, and the samples were placed on ice and sonicated five times, at level 15, for 10 seconds with 1-minute resting intervals between the sonications (Microson Ultrasonic Cell Disruptor, Misonix Inc., Farmingdale, NY). After sonication, DNA was purified using DNeasy Tissue kit (QIAGEN, Valencia, CA) according to the manufacturer's protocol. The final sample of DNA was eluted in 100 μl of ddH2O, and stored at 20°C.
RTQ-PCR for the S. aureus-specific nuc gene was performed to quantify the bacterial load with primers 5’-GCGATTGATGGTGATACGGTT -3’ and 5’- AGCCAAGCCTTGACGAACTAA -3’ that amplify a 269-bp product, as previously described 18. The reactions were carried out in a final volume of 20 μl consisting of 0.3 μM primers, 1× Sybr Green PCR Super Mix (BioRad, Hercules, CA), and 2 μl of the purified tibia DNA template. The samples were assayed in triplicate in a Rotor-Gene RG 3000 (Corbett Research, Sydney, AU). In order to control for the integrity of the DNA template between samples we performed RTQ-PCR for the mouse β-actin gene that detects a 124-bp product using primers 5'-AGA TGT GAA TCA GCAAGC AG-3' and 5'-GCG CAA GTT AGG TTT TGT CA-3'. In order to calculate the nuc gene copies in a tibia sample, we first generated a standard curve with S. aureus genomic DNA purified directly from an overnight culture. The mean of the 3 Ct values from each tibia sample were then plotted against this curve to extrapolate the number of nuc genes. This number was then normalized to β-actin and the data are presented as normalized nuc gene copies per sample.
Serology
In order to determine total immunoglobulin isotype (Ig) levels, blood samples were collected from the animals on days 0, 4, 7, 11 and 14 using retro-orbital bleeding, and an ELISA on the sera was performed using the Mouse Typer Sub-Isotyping Kit (BioRad, Hercules, CA) as we have previously described 23. Specific antibodies against S. aureus proteins were detected by western blotting. The protein was obtained from a 100 ml culture of bacteria in log phase, in which the protein extract was prepared using the Complete Bacterial Proteome Extraction Kit (Calbiochem, San Diego, CA) according to the manufacture’s instructions. 20 μg of total S. aureus protein per well was boiled in Laemelli loading buffer and separated in NuPAGE™ 10% Bis-Tris SDS Gels (Invitrogen, Carlsbad, CA) by electrophoresis. The proteins were transferred to a PVDF membrane (Millipore, Billerica, MA) and stained with Ponceau Red (Sigma, St. Louis, MO) to control for protein loading and transfer efficiency. The membrane was then cut into single lanes and blocked with PBS, 0.1% Tween 20 (PBST) and 5% non-fat dry milk for 1 hr at room temperature. Afterwards, each lane was incubated with a unique serum (10 μl serum in 5 ml of PBST) as the primary antibody for 1 hr at room temperature. The strips were then washed 3 times in 0.1% PBST, 15 minutes each at room temperature. The strips were then pooled and incubated with 1.5 μl HRP-conjugated goat anti-mouse IgG antibody (BioRad, Hercules, CA) in 30 ml blocking buffer for 1 hr at RT. The strips were then washed 3 times in PBST, 15 minutes each at room temperature. Finally, the strips were reassembled with the molecular weight marker strip and imaged with ECL+ (Amersham) chemiluminescence autoradiograph.
A murine transtibial model of implant-associated osteomyelitis
In our initial attempt to develop a quantitative model of implant-associated OM we chose an intramedullary implant approach, since this represents the serious infections in patients with total joint replacements. However, repeated attempts failed to provide evidence that a reproducible abscess could be generated (data not shown). This led us to conclude that it is challenging to consistently achieve a quantitative model of intramedullary implant associated-OM, although others have recently succeeded 24.
Next, we investigated a transtibial implant approach that mimics OM of external fixation pins. In our initial experiments, an insect pin coated with Xen29 was surgically implanted into the left tibia of mice. Longitudinal x-rays demonstrated osteolysis adjacent to the pin within 7 days (Figure 1A). Moreover, this mode of infection led to a highly reproducible localized abscess in >90% of the mice, and never resulted in detectable hematogenous spreading, sepsis or death. In order to quantify the osteolysis, we performed a time-course study in which the infected tibiae were analyzed by μCT (Figure 1B and C). These results are consistent with sequestrum formation in which osteoclastic bone resorption of cortical bone around the infected implant occurs with concomitant reactive periosteal bone formation.
The presence of OM in the mice was confirmed by histological analyses on tibiae that received infected or sterile pins. Figure 2 demonstrates that the tibial transcortical pin model contains all of the salient features of chronic OM including: sequestrum and involucrum formation, osteoclastic resorption of the cortical bone and Gram stained extracellular bacteria and biofilm that reside in the necrotic bone surrounding the implant. None of the negative controls, including heat killed S. aureus and non-pathogenic E. coli (data not shown), demonstrated these features.
Figure 2
Figure 2
Histological evaluation of the trans-tibial implant-associated model of osteomyelitis
Kinetics of infection and the host immunity during the establishment of OM
Since it is impossible to effectively extract live bacteria from infected bone to quantify the in vivo bacterial load due to the calcified matrix and biofilm, we adapted an RTQ-PCR method to determine the number of nuc gene per tibia as a surrogate outcome measure. Using PCR primers specific for murine β-actin and S. aureus nuc, we first validated the sensitivity and specificity of the assay (detection of <10 copies per sample with a single peak; Supplemental Figure A). Then, we used this assay to complete a time-course study on infected tibiae samples to assess the in vivo bacterial load during the establishment of OM. Figure 3A and B shows that the bacterial load peaked on day 11 post-surgery and dropped sharply thereafter. The sharp drop suggests that the host has generated an effective immune response that clears the bacteria.
Figure 3
Figure 3
Kinetics of infection and host immunity during the establishment of OM
Since effective clearance of S. aureus is partially dependent on humoral immunity, we evaluated antisera from infected mice over the course of infections. In order to determine when mice produce specific high affinity antibodies against S. aureus proteins after infections, we performed western blots on whole-cell bacterial extracts using sera from the challenged animals as the primary antibody. The results demonstrated the generation of specific anti-S. aureus IgG antibodies by day 11, which increase thereafter (Figure 3C). An assessment of total Ig levels in the sera of these challenged mice demonstrated an initial increase in IgM levels after one week, which was converted to IgG2b at two weeks (Figure 3D). Taken together, these results indicate that establishment of implant-associated OM is consistent with classical microbial pathogenesis and immunity in which the bacteria enjoy an initial growth period in the naïve host that is terminated by an acquired humoral response. The finding that mice are unable to completely eradicate S. aureus infection after generating a functional humoral response is consistent with the fact that antibodies cannot penetrate biofilm.
Bioluminescent imaging (BLI) quantification of bacterial metabolism during the establishment of OM
While the results of our RTQ-PCR studies demonstrate its ability to quantify bacterial load in vivo, the assay has two fundamental shortcomings that limit its utility. The first is the sacrificial endpoint, which prohibits longitudinal studies. The second is that RTQ-PCR cannot distinguish between metabolically active bacteria and dormant microbes within biofilm. Since this resolution is critical for the investigation of pathogenic mechanisms and evaluation of novel interventions, we examined the utility of BLI in our transtibial implant model. In our time-course studies with Xen29, only background signal was detected in mice that receive a sterile pin (Figure 4) or infected mice treated with parenteral gentamycin (data not shown). In contrast, the BLI of infected, untreated tibiae demonstrated a sharp 4-fold increase from baseline on day 4, which subsequently dropped to background levels by day 11.
Figure 4
Figure 4
Bioluminescent Imaging (BLI) quantification of bacterial growth during the establishment of chronic osteomyelitis
Based on the finding that BLI peaks on day 4 and nuc levels peak on day 11 along with the facts that bioluminescent of S. aureus Xen 29 results from the reaction of enzymes and protein substrates synthesized by the lux operon and that nuc is a single copy gene in the S. aureus chromosome, we hypothesized that BLI is a measure of S. aureus metabolic activity (protein synthesis), while nuc levels are indicative of bacteria number. This theory also predicts that the BLI:nuc ratio is greater early on in infections compared to dormancy in latent infection. To test this, we performed linear regression analyses of BLI versus nuc gene copy number before the presence of biofilm on day 2 and on day 18 when humoral immunity limits the bacteria to biofilm growth. These analyses demonstrated that there is no relationship between BLI and nuc gene copy number on day 2 (Figure 5A), while there is a highly significant correlation between these outcome measures on day 18 (Fig 5B). Moreover, the finding that the BLI/nuc ratio is 15 times greater on day 2 vs. day 18 (Figure 5C) further substantiates BLI as an outcome measure of metabolic activity and nuc levels as a measure of bacterial load.
Figure 5
Figure 5
BLI is a function of bacterial metabolism
Although infection rates following orthopaedic surgery are considered to be low 2,3, implant-associated OM remains a catastrophic outcome that often requires revision surgery and can lead to sepsis and death 1. This problem is compounded by the emergence of multi-drug resistant strains and the absence of effective treatments for patients with methicillin-resistant S. aureus (MRSA) 1,3. Progress in this area has been limited due to the absence of a quantitative animal model that can be utilized to elucidate molecular targets and evaluate novel interventions. From our assessment of the literature, we surmise that a quantitative OM model has not been developed for two reasons. First, the field has been overly concerned with intramedullary implant models, since this represents the more serious clinical condition 813. Unfortunately, we found that this model gave rise to highly variable (temporal and spatial) lesions, making a reproducible, quantitative model very challenging, although others have recently succeeded 24. In contrast, we have found that implantation of an infected transcortical pin always produces lesions adjacent to the pin, and never results in distal OM, hematogenous infection or death (Figure 1). Thus, while it could be argued that the mechanisms of infection and treatment modalities differ between intramedullary and external fixation implants, we find the transcortical pin model to be the best approach to study OM in vivo.
The second roadblock towards the development of a quantitative OM model is the difficulty in extracting individual live bacteria, classically known as colony forming units (CFU), from infected bone. Here we demonstrate two independent methods to overcome this obstacle. The first, nuc RTQ-PCR, is a highly specific and sensitive method that has been successfully used to quantify S. aureus levels in contaminated cheese 18. It should be noted that although we control for DNA integrity via β-actin RTQ-PCR, we have no way of knowing the total nuc yield following our rigorous extraction procedures. Thus, it is likely that the nuc genes per tibia that we present represent a fraction of the total number of bacteria originally present in the sample. Furthermore, it is likely that our ability to extract nuc genes from infected tibiae becomes less efficient when the bacteria are residing in dense biofilm. Thus, the rather low nuc gene levels observed in latent infections (day 18), may be an under representation of the actual bacterial load. This may also explain the remarkable effects of the anti-S. aureus antibodies on nuc gene levels (Figure 3), as it is known that biofilm provide bacteria with an immune-privileged environment. Nevertheless, our RTQ-PCR results provide the first demonstration that: i) in vivo bacterial load levels can be quantified during the establishment of OM, and ii) that the peak bacterial load is coincident with the generation of humoral immunity against the bacteria.
As mentioned above, two critical aspects of OM research that cannot be addressed by our RTQ-PCR approach are longitudinal studies and assessment of microbial metabolic activity. As research on the regulators of in vivo biofilm formation has become a central focus in this field, developing a surrogate outcome measure of pre and post-biofilm growth is of great value. To this end we have investigated the utility of BLI, which has emerged as a research technique with enormous potential 25. Contag and colleagues were the first to utilize bioluminescent strains of pathogens to study infection longitudinally in mice 26. Subsequent studies have proven the value of BLI over conventional methods in studying disease and therapeutic interventions in animals 25. More recently, this group generated bioluminescent S. aureus Xen29 for this purpose 19. In our hands, Xen29 behaves essentially the same at UAMS-1, which is the most widely used strain of S. aureus used in OM research 913. Thus, we adopted this approach to be compatible with μCT quantification of osteolysis and RTQ-PCR in our model of OM.
The first interesting result we obtained with BLI was that it peaked on day 4 (Figure 4), well before the peak of bacterial load and the appearance of adaptive immunity (Figure 3). Since BLI is a function of metabolic activity (protein synthesis), not bacterial load, our interpretation of these results is that there is a brief exponential bacterial growth phase that peaks on day 4, and is followed by the initiation of the biofilm growth phase that occurs at a low metabolic rate. Once this biofilm growth phase is initiated, perhaps due to the exhaustion of nutrients in the necrotic tissue, the bacterial load continues to increase. Microbes are shed from the biofilm until protective immunity is acquired on day 11. Afterwards, bacterial persistence is restricted to the biofilm in necrotic tissue indefinitely. To test this theory we postulated that at a given time during infection, nuc copy number (1 nuc gene per bacterium) only correlates with BLI (mean rate of luxA-E protein synthesis for all of the bacteria) during periods of homogenous growth. In support of this theory we found that no correlation exist during concomitant planktonic (high metabolic rate) and colonized (lower metabolic rate) bacterial growth, which is known to occur 48hr after infection, and when the inter-animal variability between the proportion of planktonic to colonized bacteria is expected to be great (Figure 5A). In contrast, when the bacteria are restricted to biofilm growth (low metabolic rate), due to acquired immunity at day 18, and there is no inter-animal variability between the proportions of planktonic to colonized bacteria, a significant correlation exists between nuc and BLI (Figure 5B). Furthermore, we found that the BLI:nuc ratio early on in infection (day 2) is 15-fold greater vs. chronic infection (day 18). While further studies are needed to formally demonstrate the kinetics of biofilm formation in vivo, our findings clearly distinguish these two phases of bacterial growth.
In summary, the development of a quantitative model of implant-associated OM opens the field to research that was previously unapproachable. New avenues include: the identification of biofilm genes through the characterization of mutant phenotypes, identification of diagnostic and/or protective antigens, and facile evaluation of infection-resistant implants such as the Selfprotective 'smart' devices 27, and novel drug efficacy. Although there are some fundamental limitations of this model with regard to its ability to accurately represent the true clinical problems associated with infected intramedullary implants and prostheses, we find that this transtibial mouse model, which displays all of the salient features of infected external fixation pins, to be a quantitative model suitable for most of these needs.
Supplementary Material
01
Acknowledgments
The authors would like to thank Laura Yanoso for technical assistance with the micro-CT and Krista Scorsone for technical assistance with the histology. This work was supported by research grants from the US Army Medical Research Acquisition Activity (USAMRAA), Orthopaedic Trauma Research Program (OTRP) W81XWH-07-1-0124, and the National Institutes of Health PHS awards AR48681, DE17096, AR52674, AR51469, AR46545, AR54041 and AR53459.
1. Darouiche RO. Treatment of infections associated with surgical implants. N Engl J Med. 2004;350-14:1422–1429. [PubMed]
2. Lew DP, Waldvogel FA. Osteomyelitis. Lancet. 2004;364-9431:369–379. [PubMed]
3. Toms AD, Davidson D, Masri BA, Duncan CP. The management of peri-prosthetic infection in total joint arthroplasty. J Bone Joint Surg Br. 2006;88-2:149–155. [PubMed]
4. Costerton JW, Stewart PS, Greenberg EP. Bacterial biofilms: a common cause of persistent infections. Science. 1999;284-5418:1318–1322. [PubMed]
5. Norden CW. Lessons learned from animal models of osteomyelitis. Rev Infect Dis. 1988;10-1:103–110. [PubMed]
6. Daum RS, Davis WH, Farris KB, Campeau RJ, Mulvihill DM, Shane SM. A model of Staphylococcus aureus bacteremia, septic arthritis, and osteomyelitis in chickens. J Orthop Res. 1990;8-6:804–813. [PubMed]
7. Rissing JP, Buxton TB, Weinstein RS, Shockley RK. Model of experimental chronic osteomyelitis in rats. Infect Immun. 1985;47-3:581–586. [PMC free article] [PubMed]
8. Lucke M, Schmidmaier G, Sadoni S, Wildemann B, Schiller R, Stemberger A, Haas NP, Raschke M. A new model of implant-related osteomyelitis in rats. J Biomed Mater Res B Appl Biomater. 2003;67-1:593–602. [PubMed]
9. Passl R, Muller C, Zielinski CC, Eibl MM. A model of experimental post-traumatic osteomyelitis in guinea pigs. J Trauma. 1984;24-4:323–326. [PubMed]
10. Worlock P, Slack R, Harvey L, Mawhinney R. An experimental model of post-traumatic osteomyelitis in rabbits. Br J Exp Pathol. 1988;69-2:235–244. [PubMed]
11. Varshney AC, Singh H, Gupta RS, Singh SP. Experimental model of staphylococcal osteomyelitis in dogs. Indian J Exp Biol. 1989;27-9:816–819. [PubMed]
12. Kaarsemaker S, Walenkamp GH, vd Bogaard AE. New model for chronic osteomyelitis with Staphylococcus aureus in sheep. Clin Orthop. 1997;339:246–252. [PubMed]
13. Salgado CJ, Jamali AA, Mardini S, Buchanan K, Veit B. A model for chronic osteomyelitis using Staphylococcus aureus in goats. Clin Orthop Relat Res. 2005;436:246–250. [PubMed]
14. Marriott I, Gray DL, Tranguch SL, Fowler VG, Jr, Stryjewski M, Scott Levin L, Hudson MC, Bost KL. Osteoblasts express the inflammatory cytokine interleukin-6 in a murine model of Staphylococcus aureus osteomyelitis and infected human bone tissue. Am J Pathol. 2004;164-4:1399–1406. [PubMed]
15. Patti JM, Jonsson H, Guss B, Switalski LM, Wiberg K, Lindberg M, Hook M. Molecular characterization and expression of a gene encoding a Staphylococcus aureus collagen adhesin. J Biol Chem. 1992;267-7:4766–4772. [PubMed]
16. Johansson A, Flock JI, Svensson O. Collagen and fibronectin binding in experimental staphylococcal osteomyelitis. Clin Orthop Relat Res. 2001;382:241–246. [PubMed]
17. Elasri MO, Thomas JR, Skinner RA, Blevins JS, Beenken KE, Nelson CL, Smeltzer MS. Staphylococcus aureus collagen adhesin contributes to the pathogenesis of osteomyelitis. Bone. 2002;30-1:275–280. [PubMed]
18. Hein I, Lehner A, Rieck P, Klein K, Brandl E, Wagner M. Comparison of different approaches to quantify Staphylococcus aureus cells by real-time quantitative PCR and application of this technique for examination of cheese. Appl Environ Microbiol. 2001;67-7:3122–3126. [PMC free article] [PubMed]
19. Francis KP, Joh D, Bellinger-Kawahara C, Hawkinson MJ, Purchio TF, Contag PR. Monitoring bioluminescent Staphylococcus aureus infections in living mice using a novel luxABCDE construct. Infect Immun. 2000;68-6:3594–3600. [PMC free article] [PubMed]
20. Zhang X, Schwarz EM, Young DA, Puzas JE, Rosier RN, O'Keefe RJ. Cyclooxygenase-2 regulates mesenchymal cell differentiation into the osteoblast lineage and is critically involved in bone repair. J Clin Invest. 2002;109-11:1405–1415. [PMC free article] [PubMed]
21. Koefoed M, Ito H, Gromov K, Reynolds DG, Awad HA, Rubery PT, Ulrich-Vinther M, Soballe K, Guldberg RE, Lin AS, O'Keefe RJ, Zhang X, Schwarz EM. Biological Effects of rAAV-caAlk2 Coating on Structural Allograft healing. Mol Ther. 2005;12-2:212–218. [PubMed]
22. Ito H, Koefoed M, Tiyapatanaputi P, Gromov K, Goater JJ, Carmouche J, Zhang X, Rubery PT, Rabinowitz J, Samulski RJ, Nakamura T, Soballe K, O'Keefe RJ, Boyce BF, Schwarz EM. Remodeling of cortical bone allografts mediated by adherent rAAV-RANKL and VEGF gene therapy. Nat Med. 2005;11-3:291–297. [PMC free article] [PubMed]
23. Schwarz EM, Krimpenfort P, Berns A, Verma IM. Immunological defects in mice with a targeted disruption in Bcl-3. Genes Dev. 1997;11-2:187–197. [PubMed]
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. Vancomycin covalently bonded to titanium alloy prevents bacterial colonization. J Orthop Res. 2007 [PubMed]
25. Contag CH, Bachmann MH. Advances in in vivo bioluminescence imaging of gene expression. Annu Rev Biomed Eng. 2002;4:235–260. [PubMed]
26. Contag CH, Contag PR, Mullins JI, Spilman SD, Stevenson DK, Benaron DA. Photonic detection of bacterial pathogens in living hosts. Mol Microbiol. 1995;18-4:593–603. [PubMed]
27. Parvizi J, Antoci V, Jr, Hickok NJ, Shapiro IM. Selfprotective smart orthopedic implants. Expert Rev Med Devices. 2007;4-1:55–64. [PubMed]