ABSTRACT
We have developed a rapid, continuous method for real-time monitoring of biofilms, both in vitro and in a mouse infection model, through noninvasive imaging of bioluminescent bacteria colonized on Teflon catheters. Two important biofilm-forming bacterial pathogens, Staphylococcus aureus and Pseudomonas aeruginosa, were made bioluminescent by insertion of a complete lux operon. These bacteria produced significant bioluminescent signals for both in vitro studies and the development of an in vivo model, allowing effective real-time assessment of the physiological state of the biofilms. In vitro viable counts and light output were parallel and highly correlated (S. aureus r = 0.98; P. aeruginosa r = 0.99) and could be maintained for 10 days or longer, provided that growth medium was replenished every 12 h. In the murine model, subcutaneous implantation of the catheters (precolonized or postimplant infected) was well tolerated. An infecting dose of 10 3 to 10 5 CFU/catheter for S. aureus and P. aeruginosa resulted in a reproducible, localized infection surrounding the catheter that persisted until the termination of the experiment on day 20. Recovery of the bacteria from the catheters of infected animals showed that the bioluminescent signal corresponded to the CFU and that the lux constructs were highly stable even after many days in vivo. Since the metabolic activity of viable cells could be detected directly on the support matrix, nondestructively, and noninvasively, this method is especially appealing for the study of chronic biofilm infections and drug efficacy studies in vivo.
Microbial adhesion and biofilm formation on medical implants is a common occurrence and represents a serious medical problem. Since biofilm microorganisms are difficult to eradicate with antibiotic therapy, chronic, recurrent infections often develop. With the increased use of prosthetic biomedical implants, chronic nosocomial infections have become prevalent in recent years (9, 41). Bacterial colonization of indwelling devices can be a prelude to both systemic infection and malfunction of the device.
A variety of techniques, such as direct microscopic enumeration, total viable count, metabolically active dyes, radiochemistry, and fluorescence, have been used to investigate microbial biofilms (1, 4, 8, 14, 17, 18, 23, 29, 33, 38). While some of these methods are useful for in vitro studies, they have not proved ideal for the investigation of biofilms in experimental infection models. The difficulty in analyzing biofilms in vivo lies in the lack of tools that allow noninvasive longitudinal study design. Assays developed to date, both direct and indirect, are time-consuming and laborious and involve the extraction of bacteria from support surfaces. To better understand and control biofilms on medical devices, rapid, direct, nondestructive, real-time quantitative monitoring methods that are adaptable to the clinical situation are needed. These assays may be used to develop new preventive and therapeutic methods to combat biofilm related infections.
To this end, bioluminescent reporters offer a method of labeling pathogens that is innocuous and allows the sensitive detection of only live, metabolically active cells by optical biophotonic imaging. Since bacterial luciferase requires reduced flavin mononucleotide for the generation of bioluminescence within the cell, it is a good indicator of the metabolic state of cells (22). Many biological applications of bioluminescent and fluorescent reporter gene systems have been developed during the last few years (1, 3, 10, 14, 19-23, 26-28, 33, 35-38). Monitoring a disease process in a living animal by using bioluminescent-tagged pathogens was first demonstrated in 1995 by Contag et al. (5). This noninvasive, rapid, real-time monitoring approach has been applied to a myriad of animal infection models and has proven to have significant advantages over conventional methods for studying disease and treatment in animals (6, 7, 11, 12, 31).
Monitoring bioluminescence as a measure of metabolic activity provides a rapid, quantitative in situ measure of biofilm development and physiological activities of bacteria within biofilms. In this study, we describe the detection of bioluminescent strains of Staphylococcus aureus and Pseudomonas aeruginosa, two of the most predominant biofilm-forming pathogens, using real-time monitoring to nondestructively, image catheter-based infections in an experimental animal model.
(Parts of the present study were presented at the 102nd General Meeting of the American Society for Microbiology, abstr. J6, 2002.)
MATERIALS AND METHODS
Bacterial strains.The bacterial strains used in this study were P. aeruginosa ATCC 19660 and S. aureus ATCC 12600. These strains were engineered for bioluminescence and designated P. aeruginosa Xen 5 and S. aureus Xen 29. If not otherwise indicated, trypticase soy broth (TSB) supplemented with 0.25% glucose (TSBG) (Difco, Detroit, Mich.) was used as the growth medium to culture the bacterial pathogens.
Generation of bioluminescent S. aureus. S. aureus 12600 was transformed with a modified Photorhabdus luminescence lux operon (12) by using the gram-positive lux transposon plasmid pAUL-ATn 4001 luxABCDE Km r (11), which was introduced into the cells by electroporation as previously described (12, 34). Transformants were grown overnight in TSB containing erythromycin (5 μg/ml) and then plated onto TSB solid medium containing kanamycin (200 μg/ml) to select for clones in which the Tn 4001luxABCDE Km r cassette had transposed and inserted downstream of a promoter. Highly bioluminescent colonies were selected using an IVIS charge-coupled device camera (Xenogen Corp., Alameda, Calif.). One clone, designated S. aureus Xen 29, was selected and further characterized.
Southern blot and inverse PCR analysis of S. aureus ATCC 12600 Tn 4001 luxABCDE Km r.To determine the number of integrations of Tn 4001 luxABCDE Km r into the chromosome of S. aureus Xen 29, Southern blot analysis was conducted using the restriction endonucleases HindIII, SphI, and ClaI, with a PCR-amplified luxA gene fragment as a probe (33). To determine the site of integration, the sequence of the genomic DNA lying upstream of Tn 4001 luxABCDE Km r was amplified by inverse PCR (25), using ClaI-digested genomic DNA and diverging oligonucleotide primers within the lux operon, IR2 (5′ CGT TTC ATT ACC TCT GTT TGA 3′;
Generation of bioluminescent P. aeruginosa. P. aeruginosa ATCC 19660 was made bioluminescent by randomly introducing a Photorhabdus luminescence luxCDABE cassette into its chromosome by conjugating it with Escherichia coli S17-1 λ pir pUT mini-Tn 5 luxCDABE Tc r and allowing transposition of the lux operon to occur (40). To allow bioluminescent P. aeruginosa to be more readily distinguished from the bioluminescent E. coli donor strain, P. aeruginosa was initially made carbenicillin resistant by transforming it with plasmid p4027 (kindly provided by A. M. Kropinski, Queen's University, Kingston, Ontario, Canada). The E. coli-P. aeruginosa mixture was incubated overnight at 37°C, and then 100-μl volumes were spread onto Luria-Bertari plates containing 100 μg of carbenicillin per ml and 20 μg of tetracycline per ml. After overnight incubation at 37°C, the plates were screened for bioluminescent colonies and a highly bioluminescent clone was identified and designated P. aeruginosa Xen 5. Chromosomal DNA of bioluminescent P. aeruginosa Xen 5 was digested with ClaI or AatII and then independently self-ligated. The ligated fragments served as templates for inverse PCR amplification. Primers UTCF1 (5′GTG CAA TCC ATT AAT TTT GGT G 3′) and UTCR (5′ CAT ACG TAT CCT CCA AGC C 3′) were used to amplify the region upstream of the transposon insertion site by using Pfu DNA polymerase (Stratagene, La Jolla, Calif.). The resulting PCR fragments were purified using a PCR purification kit and sequenced with UTCF1 and UTCR as primers. The sequencing results were BLAST searched against the National Center for Biotechnology Information (NCBI) database.
Inverse PCR analysis of P. aeruginosa Xen 5 Tn 5 luxCDABE Tc r.Chromosomal DNA of bioluminescent P. aeruginosa Xen 5 was digested with ClaI or AatII and independently self-ligated. The ligated fragments were then used as templates for inverse PCR amplification of the lux fusion junction, which was amplified using primers UTCF1 and UTCR. The resulting PCR fragments were purified using a PCR purification kit and sequenced with UTCF1 and UTCR as primers. The sequencing results were BLAST searched against the NCBI database.
In vitro bacterial biofilm. (i) Microtiter plate assay.Early adherence to polystyrene surfaces has been suggested to be an indication of the capacity of a bacterial strain to form a biofilm. Thus, both parental strains and their bioluminescent derivatives were tested in a microtiter assay as described previously (4, 16). Briefly, bacterial strains were cultivated overnight in TSBG. The cultures were diluted in fresh TSBG to reach a standardized cell suspension (10 6 CFU/ml), and 100 μl of this suspension was used to inoculate sterile 96-well polystyrene microtiter plates (Nalge Nunc International Corp., Naperville, Ill.). After cultivation for 24 h at 37°C, the wells were gently washed twice with 200 μl of sterile phosphate-buffered saline (PBS). The plates were air dried, and the remaining surface-absorbed cells of the individual wells were stained with 100 μl of 0.1% safranin for 30 s. Absorbance was measured with a micro-ELISA autoreader (Molecular Devices, Sunnyvale, Calif.) at 490 nm. Sterile TSBG lacking cells served as a control, and the value obtained with this well was subtracted from experimental readings. Each assay was performed in quadruplicate.
(ii) Catheter-associated biofilm.Biofilms of S. aureus and P. aeruginosa were developed on a 14-gauge Teflon intravenous catheter (Abbocath-T; Burns Vet Supply, Vancouver, Wash.). Briefly, the catheter was cut into 1-cm segments and each piece was sterilized with 70% ethanol and air dried. Bacterial biofilms were developed on the catheter by placing individual segments into tubes containing 1.0 ml of a cell suspension (10 4 CFU/ml) in TSBG in the exponential phase of growth. After incubation for 2 to 3 h at 37°C, colonized catheters were recovered aseptically and rinsed once with TSBG to remove unbound bacteria. The catheters were then incubated further in fresh TSBG at 37°C for various time intervals. At specific time points, two or three catheters were removed for quantitative analysis of biofilm development. During the incubation time, the medium was decanted and replaced every 12 h with fresh TSBG. Control catheters were prepared as above but without the bacterial inoculum. Monitoring both bioluminescence and viable counts allowed assessment of the kinetics of biofilm formation on catheter material.
Experimental model of infection and monitoring of bioluminescent biofilms.The experimental foreign-body infection in murine model developed by Rupp et al. (32) was established with slight modifications. All experimental procedures were carried out according to the following protocol approved by the Institutional Animal Care and Use Committee. BALB/c female mice (Charles River, Wilmington, Mass.) weighing 18 to 22 g were anesthetized with Ketamine (Burns Vet Supply) at 100 mg/kg and Xylazine (Burns Vet Supply) at 5 mg/kg, their flanks were shaved, and the skin was cleansed with Betadyne and alcohol. An 8- to 10-mm skin incision was made and dissected to create a subcutaneous tunnel, into which 1 cm segment of intravenous catheter was implanted at a distance of at least 2 cm from the incision. The incision was then covered with intact skin and closed with surgical staples, and the skin was disinfected. One catheter segment was inserted on each side of each animal. Infection was induced by either implanting a precolonized catheter carrying a defined inoculum or by infecting the lumen of the sterile catheter with defined quantities of bacterial suspension in PBS postimplantation. The postimplant inoculum was introduced by injection into the catheter lumen via a 31-gauge needle, approximately 1 h after the implantation procedure. In each experiment, an extra group of animals was inoculated with vehicle (PBS), to serve as a negative control. Mice were imaged for a maximum of 5 min, at various time points following inoculation, using an IVIS camera system (Xenogen Corp.). Total photon emission from defined regions of interest within the images of each mouse was quantified using the LivingImage software package (Xenogen Corp.). The photon signals from the catheter were quantified from the ventral image of each mouse. After the final imaging time point, the mice were humanely killed and the infected catheters were surgically removed for enumeration of bacteria by both bioluminescence imaging and conventional plate count method. Bacteria recovered at the end of the experimental period were compared with the inoculated strain for bioluminescence.
Extraction and quantification of bacteria from the catheter biofilm assay. (i) In vitro.Two to three catheters were removed from the incubating chamber at the appropriate time according to each experimental objective. The catheters were rinsed in fresh TSB and imaged to quantify the bioluminescence signal. They were then transferred to a separate tube containing 1 ml of TSB. The tubes were placed in an ultrasonic bath at 38.5 to 40.5 kHz (VWR, San Francisco, Calif.), sonicated for 5 to 10 min, and votexed for 1 min to remove the biofilm bacteria from the support surface. To assess for complete removal of the biofilm bacteria, the catheters were imaged at different time intervals and the loss of bioluminescence signal was used to define the complete biofilm removal protocol. The suspension of bacteria that was removed from the catheter were diluted, plated on Trypticase soy agar, and incubated at 37°C for colony counting. Correlation between CFU counts and bioluminescent signal (relative light units, RLU) was determined by plotting RLU versus CFU.
(ii) In vivo.After the final imaging time point, the mice were humanely killed and the catheters were gently removed from the subcutaneous tissue by making a skin incision at approximately 2 cm from the implant wound. Harvested catheters were imaged and biofilm bacteria were detached from the catheters as described above, in addition to being enumerated by conventional colony count assay. Catheters from control mice were included in each experiment to assess the adequacy of aseptic and surgical techniques.
Determination of LD 50.The dose of an individual strain that resulted in 50% lethality (LD 50) for mice was determined using the biofilm infection model. Briefly, groups of five mice were each implanted with catheters containing inocula of 10 3, 10 5, 10 6, 10 7, and 10 8 CFU. Each mouse in the group was subjected to implantation of two catheters. The mice were observed for 4 days, and the LD 50 was determined using Reed and Muench proportional-distance calculation method (30). The potential lethal effect of implants or the surgical procedure was examined by implanting sterile catheters. None of the noninfected mice died throughout the observation period.
RESULTS
Characterization of S. aureus Xen 29.Southern blot analysis demonstrated only one copy of Tn 4001 luxABCDE Km r to be inserted in the chromosome of S. aureus Xen 29. The chromosomal integration site of Tn 4001 luxABCDE Km r was determined by BLAST analysis of the sequence obtained from the inverse PCR fragment. It was determined that the lux transposon had inserted after nucleotide 393 of a 455-nucleotide open reading frame, designated "SA2154' (analyzed against the S. aureus N315 genome located in the NCBI database).
Characterization of P. aeruginosa Xen 5.The inverse PCR results indicated that the transposon Tn 5 luxCDABE had inserted into the P. aeruginosa gene PA4974, which encodes a probable secretion protein similar to the outer membrane protein TolC of Vibrio cholerae and E. coli. Upstream of PA4974 was thiC and downstream was PA4975; both of these ORF were in the opposite direction to the PA4974.
Comparisons of transformants and parental strains for biofilm formation.To assess the effect on biofilm formation of insertion of the lux transposon cassette into the bacterial chromosome, both wild-type and bioluminescent derivative strains were tested for their ability to form biofilms in vitro by two different quantitative methods. There was a slight increase in adherence of the lux constructs of both P. aeruginosa and S. aureus to polystyrene microtiter plates compared to the parental strains. Consistent with these results, the lux constructs also showed a slightly greater propensity to form biofilms on catheter material when assessed by the conventional colony count method (Table 1). The enhanced biofilm formation seen in the lux constructs is unlikely to be due to the growth rate, since the generation times between the strains were almost identical (Table 1). This slight increase in biofilm formation among lux-containing strain merits further investigation, although we do not think it significantly affected the reported results.
Comparative characteristics of parental and bioluminescent derivatives.
Bioluminescence and viability during biofilm development in vitro catheter assays.Bioluminescent images of 2-day-old biofilms of P. aeruginosa Xen 5 and S. aureus Xen 29 on 1-cm segment of 14-gauge Teflon intravenous catheters were recorded using an IVIS camera and are shown in Fig. 1. The intensity of the bioluminescent signal in P. aeruginosa Xen 5 appeared to be higher than that in S. aureus Xen 29, which could be due to a number of factors particular to the organism. Also, P. aeruginosa Xen 5 exhibited a higher bioluminescence activity when grown in liquid media. No bioluminescence was observed in the wild-type strains or noncolonized (control) catheter. This finding confirms that bioluminescence is specific to the lux constructs; the signal could be detected even when the cells were embedded in a thick extracellular matrix (Fig. 2).
Monitoring the level of bioluminescence activity of P. aeruginosa and S. aureus in a 2-day-old biofilm on catheter segments. Images were acquired with the IVIS camera and are displayed as pseudocolor images, with variations in color representing light intensity at a given location. Red represents the most intense light emission, while blue correspond to the weakest signal. Note the lack of light signal from the wild-type and control catheter, indicating the specificity of the detection system. The color bar indicates relative signal intensity.
Phase-contrast micrograph of a wet mount, showing 5-day-old biofilms of P. aeruginosa Xen 5 (A) and S. aureus Xen 29 (B). A small lump of catheter biofilm was detached from the catheter surface to show the bacterial aggregates.
To assess the feasibility of using bioluminescence as a quantitative indicator of biofilm bacteria, studies were performed to compare bioluminescence to the number of viable cells on catheter material (Fig. 3). During log phase growth, the luminescence output was proportional to the bacterial biomass, as determined by the number of culturable and bioluminescent cells. Bioluminescence was found to closely correlate with viable-cell count, yielding correlation coefficients of 0.99 and 0.98 for P. aeruginosa Xen 5 and S. aureus Xen 29, respectively (Fig. 4). However, once the cultures reached stationary phase, the bioluminescence decreased as expected and was no longer correlated with the number of cells, indicating a decrease in growth and metabolic activity of the population at stationary phase (data not shown). This is probably due to oxygen and/or substrate limitations since the reduced expression of bioluminescence within the biofilm could be restored by replacement of the culture medium with fresh medium every 12 h. This procedure not only replenished the growth medium but also removed planktonic cells, leaving only the sessile bacterial population embedded in the biofilm. Alternatively, this decrease in bioluminescence might be due to the accumulation of inhibitory molecules or to a combination of these factors. For P. aeruginosa and S. aureus, the bioluminescent signal and CFU reached a peak after 2 days of incubation, after which the numbers remained relatively constant, presumably due to the bacterial population reaching the maximum size that was able to colonize the catheter surface. The mean RLU of 6 × 10 6/catheter for P. aeruginosa Xen 5 and 8 × 10 5/catheter for S. aureus Xen 29 approximately corresponded to 1.8 × 10 7 CFU/catheter and 6 × 10 6 CFU/catheter for P. aeruginosa Xen 5 and S. aureus Xen 29, respectively.
Growth and bioluminescence curves of P. aeruginosa Xen 5 and S. aureus Xen 29 grown on catheter surfaces. Viable counts are reported as CFU per catheter, and bioluminescence is represented as RLU measured using the IVIS camera. Each data point is the mean and standard error for three or four catheters. Bioluminescence was determined at each time point immediately prior to determination of the viable-cell count.
Scatter plots of viable cells and bioluminescence data to demonstrate the relationship between viable counts and bioluminescence for P. aeruginosa Xen 5 (A) and S. aureus Xen 29 (B).
LD 50 studies in the biofilm in vivo model.The LD 50s for the parental and lux derivatives in the biofilm model after 48 h are shown in Table 1. In terms of the LD 50 and survival time, there was no major difference between the wild-type and engineered strains, suggesting that the integration of the lux cassette into the bacterial chromosome had no effect on bacterial pathogenicity or survival in mice.
Experimental mouse infection model.Figure 5 illustrates the real-time biophotonic images of representative mice, infected with bioluminescent P. aeruginosa Xen 5 or S. aureus Xen 29, over a 20-day catheter-based infection. Both strains produced a significant bioluminescent signal in mice, allowing the progression of infection to be monitored noninvasively in the experimental model. The total photon emission from the infected sites was quantified using LivingImage software, and cumulative results are shown in Fig. 6. Following implantation of precolonized catheters, the bioluminescence measurements increased exponentially over 24 h. One day after implantation, the bioluminescent signal reached approximately 10 5 RLU/catheter, and it remained moderately stable for all S. aureus and P. aeruginosa doses tested until the termination of the experiment on day 20 (Fig. 6). The kinetics of biofilm development as a measure of bioluminescence in postimplantation infection was also similar, except that the bioluminescent signal reached a peak after 2 days as opposed to 1 day postimplantation in the precolonized catheter model. The number of CFU recovered from catheters following the 20-day imaging time point had increased up to ∼10 6 to 10 7 CFU/catheter over all inoculating doses of 10 3 to 10 5 CFU/catheter (Table 2). The strong bioluminescent signal detected from implanted catheters suggests that the biofilm population remained metabolically active throughout the experiment, and the increase in biomass, as measured by bioluminescence and CFUs, over the starting inoculum confirmed that there was local proliferation (in vivo) and colonization of biofilm bacteria on the catheter matrix during the course of infection (Table 2). Successful infection is defined as the recovery of viable pathogens from the catheter in numbers greater than the initial infective dose at the time of sacrifice. The inoculum studies showed that 100% of the animals implanted with precolonized catheters with all three doses (10 3, 10 4, or 10 5) of S. aureus or P. aeruginosa developed catheter-associated infection by the day of sacrifice. An infection rate of 100% was reached with the highest inoculum of either S. aureus or P. aeruginosa in the groups where infection was started 1 h after implanting the catheter (Table 2). In the groups of animals that were inoculated with 10 3 or 10 4 CFU of P. aeruginosa or S. aureus per catheter, the infection rate was 50 and 75% respectively, suggesting that the animals inoculated postimplantation with lower doses of bacteria were more capable of eliminating the infection and thus preventing the establishment of a biofilm. The lower incidence of infection seen in this group is most probably explained by the actual dose inoculated in this group being slightly lower than that in the group infected with precolonized catheters (Table 2). Local phagocytic activity may also have eradicated the infection before the pathogens were able to colonize the catheter. This normal host defense may be especially effective in clearing the lower doses of infecting pathogens. Despite these differences, subcutaneous implantation of the biofilm-containing catheter (pre- or postinfected) seemed to be well tolerated by the mice. An inoculum of ∼10 3 to 10 5 CFU of S. aureus and P. aeruginosa per catheter resulted in a reproducible, localized persistent infection surrounding the catheter until the termination of the experiment on day 20 (Fig. 5 and 6). Doses above 10 6 CFU/catheter for P. aeruginosa and 10 8 CFU/catheter for S. aureus resulted in 100% mortality, whereas an inoculum of ≤10 5/catheter produced a chronic infection. No bacteria could be cultured from implants that did not have a bioluminescent signal, suggesting that monitoring of infection via bioluminescence is a reasonable measure of bacterial load. All the S. aureus and P. aeruginosa organisms recovered from catheters with 20-day infections were shown to be bioluminescent, demonstrating the stability of lux constructs even after many days in vivo.
Real-time monitoring of P. aeruginosa Xen 5 (A) and S. aureus Xen 29 (B) biofilms in a mouse model. Precolonized catheters were implanted at subcutaneous sites with doses ranging from 10 3 to 10 5 CFU, and growth of the biofilm was monitored by detecting photon emission over a 20-day time course using an IVIS camera. Similar results were obtained when the catheter bed was inoculated with doses ranging from 10 3 to 10 5 CFU after subcutaneous implantation of sterile catheters.
Growth and bioluminescence curves of P. aeruginosa Xen 5 (A) and S. aureus Xen 29 (B) biofilms in mice infected with precolonized catheters carrying various inocula. Each data point is the mean and standard error for two or three mice. Each mouse was subjected to implantation of two catheters. The viable counts in each catheter were determined immediately after removal of the catheters from the implanted sites and are shown in the upper quadrants of the plot. Identical results were obtained when similar doses of bacteria (10 3, 10 4, and 10 5 CFU/catheter) were injected into the implant site after implantation, except that the RLU value peaked 2 days after infection.
P. aeruginosa Xen 5 and S. aureus Xen 29 catheter infection in mice
DISCUSSION
In this study, we have demonstrated the use of biophotonic imaging to improve the monitoring of catheter-associated biofilm infection in mice. Bioluminescent P. aeruginosa and S. aureus strains were generated using a complete bacterial lux operon, and these strains were used to study the infection process in real time in living animals. Unlike experiments with the insect luciferase (luc), external addition of any substrate for bioluminescence is not a prerequisite in our system. Furthermore, due to its linkage to cellular metabolism, luciferase activity in vivo indicates the physiological state of the intact organism. Using an IVIS camera, we were able to visualize the bioluminescent, catheter-associated biofilm bacteria directly through the skin of live animals.
The direct assessment of an in vivo biofilm in real time without exogenous sampling has several unique features. The technique provides both spatial and sequential information about the progression of infection. Unlike using physical biochemical indicators, using luciferase as the reporter ensures that the signal observed is from viable metabolically active cells, as they exist in the biofilms. Because of its nondestructive and noninvasive nature, the imaging procedure can be performed repeatedly, allowing each animal to be used as its own control over time, overcoming the problem of animal-to-animal variations. The biostatistics is improved through collection of multiple data points from the same animal. Thus, the overall number of animals required is reduced. Furthermore, the ability to monitor the pathogen burden quantitatively without exogenous sampling considerably reduces the time and cost.
A number of animal models have been developed for studying aspects such as different biomaterials, pathogenesis, and treatment of foreign-body infections (2, 13, 24, 32, 39, 41). These models have provided vital information about the progress of infection and antibiotic pharmacodynamics. However, they generally require the sacrifice of animals at each sampling point and rely on traditional colony counting procedures, with lengthy incubation periods, for assessment of microbial numbers. Such procedures also require the handling of individual samples, during which variability arises from the difficulties in disrupting cell aggregates. These aggregates are not released from the biofilm in the form of a homogeneous suspension and thus cannot be easily recovered and quantified. Moreover, the ex vivo methods by which these events are typically monitored require the removal of tissue and consequently the loss of contextual influences of the living animal. Analysis of biofilms in vivo is greatly enhanced by tools that allow noninvasive quantitative study. The noninvasive approach described here also allows for experimental protocols that are significantly more rapid and accurate than conventional techniques.
We observed that during in vitro growth, the bioluminescence of P. aeruginosa and S. aureus was affected by the growth phase. There was a decline in bioluminescence as the biofilm became older, most probably due to a decreased cellular metabolic activity. We were able to maintain a linear relationship between bioluminescence levels and CFU by supplying fresh media frequently to the biofilm bacteria, giving an indication of decreased substrate pools for the luciferase reaction. Interestingly, this was not an issue in our experimental infections, implying that an adequate supply of factors needed for the light-emitting enzymatic reaction were available in vivo. Extraction of bacteria from infected mice confirmed a good correlation between bioluminescence data and number of viable bacteria in the catheter. Such a method of monitoring in vivo bioluminescent organisms in living animals has been described for several other pathogens in acute infection and was demonstrated to correspond well to bacterial CFU data (5-7, 11, 12, 31). However, it is important to also measure a nonluminescent end point, since the emitted light can decrease without affecting the viability of the cells (15). In the present study, we demonstrated a good correlation between photon count imaging and viable counts in vivo (r = 0.98). We also confirmed at least a 2-log-unit increase in both light output and CFU for S. aureus and P. aeruginosa strains above the initial inoculating doses. This indicates significant bacterial growth on catheters following implantation. These observations can be especially appealing for the analysis of the efficacy of antimicrobial compounds in vivo, since the effectiveness could be rapidly monitored without the need for exogenous sampling and culturing. More importantly, once a good correlation between light signal and CFU is established, the bioluminescence could be used to monitor the metabolic status and bacterial load at the site of infection.
In the present study, to maximize the clinical relevance, two parameters were simulated: (i) the effective inocula required to produce and maintain a stable foreign-body infection, and (ii) the assessment of timing of the infection (i.e., implantation of established biofilm catheters versus infection after device implantation). For the first parameter, various doses of biofilm bacteria that would make it possible to maintain a stable infection in mice over a longer period was examined. For the second parameter, infection was produced either by implanting a precoated catheter with a defined inoculum or by infecting the catheter lumen with a defined inoculum after implanting a sterile catheter into the subcutaneous space of mice. Implanting infected catheters (with a standardized biofilm formed on catheter) mimics intraoperative contamination with members of the skin flora, e.g., infections that originate from bacterial seeding of the device. Delaying the infection until after device implantation mirrors postoperative colonization of the implanted device. Both approaches are clinically relevant and address the problems associated with device-related infections, as well as allowing the study of prevention and treatment strategies of infections related to medical devices.
A potential limitation of this model is that the implanted foreign body is not mechanically functioning to flush any planktonic bacteria present. However, the model presents several of the key characteristics of foreign-body infection that have been observed clinically. First, the infection became localized to the catheter so that distant spread was limited. Second, the infection became chronic, with only a slow increase in the bacterial counts following establishment of the infection. Third, doses as low as 10 3 CFU were capable of producing a stable chronic biofilm infection by both pathogens, thus demonstrating success in reproducing infection without requiring a large number of bacteria. The low inocula that were able to establish a stable infection probably resemble the case in clinical situations. Another characteristic of our model is that infection can be induced in a reproducible manner with two of the bacterial species most commonly implicated in foreign-body infection.
This study presents a new model for studying chronic infections by gram-positive and gram-negative bacterial biofilms. The method described here is rapid, uses fewer animals, and enables in vivo monitoring of the pathogen burden and the metabolic activity of a cell population under complex environmental conditions. Additionally, this model may be useful for the study of the pathogenesis and antibiotic susceptibility of both gram-positive and gram-negative pathogens within biofilms or within the biofilm environment or in the context of a biofilm. Experiments intended to establish the applicability of the model in efficacy assessments of therapeutic agents are under way.
ACKNOWLEDGMENTS
We thank E. Reynolds and C. Dalesio (Graphics and Communications, Xenogen Corp.) for assistance with drawings.
FOOTNOTES
- Received 25 June 2002.
- Returned for modification 10 October 2002.
- Accepted 28 October 2002.
- Copyright © 2003 American Society for Microbiology