The presence of bacterial biofilm, tolerance to antibiotics and dysfunctional activity of phagocytic cells are all related to difficulties in eradicating foreign-body infections. We aimed to quantify the presence of intracellular Staphylococcus aureus and to study the extent to which the intracellular activity of antibiotics might determine their efficacy against an experimental rat tissue-cage model of foreign-body infection.
Using this model, animals were treated for 7 days with 100 mg/kg/day levofloxacin or 200 mg/kg/12 h cloxacillin, or were left untreated. Antibiotic efficacy was evaluated by means of bacterial counts from tissue-cage fluid (TCF); these counts were derived separately in total, intracellular and extracellular bacteria. The presence of intracellular bacteria was checked by electron microscopy. Population analysis was performed with surviving bacteria recovered at the end of levofloxacin therapy.
Among a total number of bacteria (mean log cfu/mL ± SD) from TCF of 6.86 ± 0.6, we identified 6.38 ± 0.8 intracellular bacteria and 5.57 ± 0.5 extracellular bacteria. Levofloxacin was more efficient than cloxacillin (P < 0.05) against both intracellular and extracellular bacteria. The killing activity of levofloxacin against the intracellular population was higher than against the extracellular bacteria (P = 0.1). The frequency of levofloxacin-resistant mutants among surviving bacteria at the end of levofloxacin therapy was similar to that for the wild-type strain.
Intracellular bacteria accounted for the largest proportion of the total inoculum in this model of foreign-body infection. The intracellular activity of an antibiotic seems to be an additional relevant factor in the antibiotic response to these infections.
Difficulties in eradicating foreign-body infections have been related to several issues such as bacterial biofilm, tolerance to antibiotics and functional abnormalities in the activity of phagocytic cells that are in contact with the foreign body.1–5
The intracellular location of bacteria may be considered an additional mechanism of virulence allowing the microorganism to achieve greater protection against antibiotics.6,7 However, the potential role of these intracellular microorganisms in limiting antibiotic efficacy in foreign-body infections is not well known.
Staphylococcus aureus is one of the most frequent microorganisms responsible for device infections, and it is also able to penetrate and survive within phagocytic cells.1,8,9 The existence of intracellular S. aureus in the setting of foreign-body infection has recently been related to disease relapse and the presence of more resistant strains.10,11
The rat model of tissue-cage infection reasonably mimics chronic prosthetic infection and has provided relevant information about foreign-body infections over the last 15 years.12–15
The aims of the present study were to screen for the presence of and to quantify intracellular S. aureus in this rat model of chronic foreign-body infection. We also studied the extent to which the intracellular activity of antibiotics may determine their final therapeutic efficacy, using cloxacillin and levofloxacin as antistaphylococcal agents.
Materials and methods
Microorganism and antibiotic susceptibilities
We used a methicillin-susceptible S. aureus ATCC 29213 strain.
The MICs and MBCs (mg/L) were determined according to standard CLSI recommendations.16 They were 0.5 and 1, respectively, of both levofloxacin and cloxacillin.
The study was approved by the Ethical Committee for Animal Experiments at the University of Barcelona. We used the model reported by Lucet et al.,12 which has provided enough (contrasted) evidence of the bacterial biofilm formation associated with foreign-body infections; all the methodology was described in detail in our previous reports.14,15 In brief, two Teflon tissue cages (with two coverslips each) were subcutaneously implanted in rats and 3 weeks later tissue-cage fluid (TCF) was infected with S. aureus. After 3 weeks, TCF was obtained to perform bacterial counts (day 1) and therapy was intraperitoneally administered for 7 days. Twenty-four hours after the end of treatment (day 8), TCF was again recovered to quantify bacterial counts (see below). Figure S1 [available as Supplementary data at JAC Online (http://jac.oxfordjournals.org/)] shows the macroscopic granulation adhered to a tissue cage after 3 weeks of infection.
Therapeutic groups were 100 mg/kg/day levofloxacin, 200 mg/kg/12 h cloxacillin and controls. The dosage of each antibiotic was selected according to previous pharmacokinetic studies following the methodology previously reported in detail,14 and on the basis of previous data about the relationships in drug concentrations between extravascular fluids and serum in our model.17 The main pharmacokinetic parameters and their relationships to MIC were reported in our previous paper.14
Bacterial counts of total, intracellular and extracellular bacteria
The TCF recovered from infected animals was divided into two samples, which were processed in parallel to quantify bacterial counts.
(i) One sample of TCF was sonicated (150 W × 1 min; Afora, Madrid, Spain) and 100 µL of this fluid and its 10-fold dilutions were cultured on Mueller–Hinton agar (MHA) plates to quantify bacterial counts (total bacteria).12,14
(ii) The other sample of TCF was centrifuged (100 g for 5 min) to separate the pellet and the supernatant in order to quantify the intracellular and extracellular bacterial counts, respectively.18 The pellet was resuspended in PBS and exposed to gentamicin (100 mg/L for 2 h; Sigma-Aldrich), an antibiotic that does not penetrate eukaryotic cells. Gentamicin was then removed by washing twice with PBS, before disrupting the cells by sonication. Finally, 100 µL of this fluid and its 10-fold dilutions were cultured as above to quantify intracellular bacteria. The supernatant from TCF was filtered (FS Bioblock, 5 µm) and checked for the absence of leucocytes using Turk solution (acetic acid and methylene blue prepared in-house); this filtered fluid was cultured as indicated above to quantify extracellular bacteria.
To avoid carryover antimicrobial agent interference, the sample was placed on the plate in a single streak down the centre, allowed to absorb into the agar until the plate surface appeared dry and the inoculum was then spread over the plate.
To count the number of leucocytes from TCF, 10 µL of each sample was diluted 1:1 with Turk solution and read with a Neubauer chamber under an optical microscope.
Population analysis of surviving bacteria
For samples from the levofloxacin group we analysed the surviving population following the recommended methodology.19 Briefly, viable bacteria recovered from TCF at the end of levofloxacin therapy were cultured overnight in Trypticase soy broth (TSB), before being adjusted to an inoculum of 1010 cfu/mL. A sample of 100 µL of this fluid and its 10-fold dilutions were cultured (at 37°C for 48 h) on agar plates containing several levofloxacin concentrations (0.5, 0.6, 0.8 and 1 mg/L) and the number of colonies was counted.
All the procedures followed the recommended methodology.20 Pellets of cells from gentamicin-treated TCF were fixed in 2.5% glutaraldehyde in 0.1 M PBS (pH 7.4). Cells were then washed in PBS buffer and dehydrated through a graded ethanol series. Finally the pellets were embedded in resin, thin sectioned and examined with an electron microscope (JEOL 1011).
All bacterial counts are presented as log cfu/mL (mean ± SD). Differences in bacterial counts for treated and untreated animals were analysed for statistical significance using analysis of variance (ANOVA). An unpaired Student's t-test with the Bonferroni correction was used post hoc to determine statistical significance.
The killing rate constant (k) was calculated by applying the equation k = ln Nt – ln N0/t, where N0 is the number of viable bacteria at time 0 and Nt is this number at time t.
For all tests, differences were considered statistically significant when P < 0.05.
Preliminary studies were performed in order to check the accuracy of the methodology for identifying different bacterial populations.
After 3 weeks of infection the total bacterial count in TCF (mean log cfu/mL ± SD) was 6.86 ± 0.6 (n = 10); the mean leucocyte count (polymorphonuclear leucocytes) from TCF at day 1 was 3.6 × 109/L. Following previously described methodology, we centrifuged the TCF to separate the pellet and the supernatant. The pellet (intracellular bacterial population) was exposed to gentamicin for 2 h to exclude the presence of extracellular bacteria; bacterial counts before and after exposure to gentamicin were, respectively, 6.47 ± 0.8 and 6.38 ± 0.8 (Figure 1). Finally, we checked the presence of intracellular bacteria after this procedure using electron microscopy (Figure 2).
The recovered supernatant (extracellular bacterial population) was filtered (FS Bioblock, 5 µm), after which the absence of leucocytes was confirmed using Turk solution. The extracellular bacteria were exposed to gentamicin for 2 h to ensure that they were killed in a different manner to intracellular bacteria; results of bacterial counts before and after exposure to gentamicin were, respectively, 5.57 ± 0.5 and 3.48 ± 0.2 (Figure 1).
We observed that the largest proportion of bacteria had an intracellular location and that treatment with gentamicin did not significantly kill these intracellular bacteria, in contrast to what occurred with extracellular bacteria (killing > 2 log cfu/mL at 2 h). All procedures were confirmed to be harmless for bacteria and we noted that the sum of intracellular and extracellular bacteria did not exactly achieve the total bacteria.
Once all the methodology had been checked and standardized we conducted therapeutic experiments. At day 1 there were no significant differences between the groups in bacterial counts for each compartment. These bacterial counts (mean log cfu/mL ± SD for total, intracellular and extracellular bacteria) were, respectively: 6.72 ± 0.9, 6.3 ± 1 and 5.6 ± 0.9 for controls (n = 12); 7 ± 0.5, 6.5 ± 0.5 and 5.5 ± 0.6 for levofloxacin (n = 17); and 6.9 ± 0.8, 6.3 ± 0.9 and 5.5 ± 0.8 for cloxacillin (n = 18).
At the end of treatment both therapeutic regimens were more efficient than controls (P < 0.05); total bacterial counts were 7.3 ± 0.6 for controls, 5.2 ± 0.8 for levofloxacin and 5.9 ± 0.7 for cloxacillin. We again noted that the largest proportion of bacteria after therapy had an intracellular location and that they remained viable. Levofloxacin was better than cloxacillin (P < 0.05); activity against intracellular and extracellular bacteria was analysed separately and showed significantly greater efficacy for levofloxacin in both compartments (P < 0.05) when compared with cloxacillin. A comparison between groups in terms of intracellular and extracellular bacterial counts is represented in Figure 3.
Killing rates (k) against intracellular and extracellular bacteria were, respectively, 0.0296 log cfu/h and 0.0239 log cfu/h for levofloxacin, and 0.0173 log cfu/h and 0.014 log cfu/h for cloxacillin. We noted that the levofloxacin killing rate was higher against intracellular than against extracellular bacteria, although this difference was not statistically significant (P = 0.1).
Population analysis (n = 8) of surviving bacteria at the end of levofloxacin therapy showed no changes in initial MIC. The frequency of mutants growing on agar plates with 0.5 and 0.6 mg/L levofloxacin was, respectively, 6.7 × 10–8 and 3.46 × 10–8 for in vivo samples (n = 8); and 8.05 × 10–8 and 9 × 10–9 for the wild-type strain ATCC 29213. No mutant colonies were grown on plates containing 0.8 or 1 mg/L levofloxacin.
Bacterial biofilms determine the limited efficacy of antibiotics in foreign-body infections.1,3 In the S. aureus tissue-cage infection model, bacterial biofilm is formed by surface-adhering microorganisms (on the tissue-cage and coverslip surfaces) and the evaluation of infection is usually monitored by analysing the inflammatory and infected fluid in the vicinity of the foreign body (TCF), which is probably coming from the continuous turnover established with the adjacent biofilm.2,5,13 Several previous studies with this model have repeatedly emphasized the relevance of bacterial tolerance to antibiotics as the main factor underlying the therapeutic problems of device infections.2,21
Additionally, the presence of dysfunctional leucocytes with viable intracellular bacteria in the TCF has been related to the inability to prevent experimental foreign-body infections and the maintenance of these infections.4,13 Furthermore, the presence of intracellular S. aureus in some phenotypic forms, such as small colony variants, has recently been reported as the cause of relapses in human orthopaedic prosthetic infections.10,11
In the present study we were surprised to find that most of the microorganisms isolated from TCF were identified as intracellular bacteria, which was in agreement with that reported in a previous study but in a higher proportion.22 It is likely, therefore, that our model of a chronic 3-week-old device infection mimics well a localized purulent prosthetic device infection.
When the efficacy of antimicrobial therapy was evaluated, interesting differences between extracellular and intracellular compartments were observed. The greater efficacy of levofloxacin compared with cloxacillin was found in both compartments, but it was greater in the intracellular compartment. Differences observed in the extracellular population were probably related to the higher tolerance to cloxacillin in comparison with that to levofloxacin found in this S. aureus population. The killing rate showed a tendency to be higher against the intracellular population than against the extracellular population; in all likelihood this suggests an additional role for this intracellular location in the limited antibiotic efficacy against foreign-body infections. One limitation when interpreting our results is the probable continuous turnover between intracellular and extracellular populations; however, since the presence of viable intracellular bacteria is highly relevant at both the beginning and end of treatment it seems reasonable to assume that better intracellular killing by the antibiotic represents greater efficacy.
Several studies have reported that all antibiotics have reduced activity against intracellular S. aureus, not only due to the difficulty of the antibiotic reaching the microorganism in its intracellular location, but also because phenotypic changes in this location confer greater antibiotic resistance.6,23 Fluoroquinolones achieve greater intracellular activity in comparison with β-lactams;6,24 in particular, levofloxacin showed one of the best activities against S. aureus and the in vitro intracellular killing of drugs correlated well with their in vivo efficacy in experimental abscesses.25,26 Our results regarding efficacy are consistent with the good intracellular killing activity of levofloxacin, and our analyses of viable bacteria at the end of treatment also support the lack of development of resistance in this location. Since all surviving bacteria recovered were grown on conventional agar plates (MHA) they were not analysed for particular bacterial phenotypic variants such as small colony variants or persisters. In this way, our results are preliminary and further studies could be performed to determine the role of these bacteria in the treatment failure of foreign-body infection.
In conclusion, we have noted that the intracellular bacterial inoculum represents the most important fraction of the total inoculum even at the end of antibiotic treatment. In the context of an experimental model of foreign-body infection, both phenotypic tolerance to antimicrobials and the intracellular location of microorganisms appear to determine the limited efficacy of antibiotics. The intracellular activity of an antibiotic seems to be a relevant factor in the antibiotic response to this kind of infection.
This work was supported by a research grant from the Ministerio de Sanidad y Consumo, Instituto de Salud Carlos III (FIS 04/005) and grants from the Spanish Network for Research in Infectious Diseases (REIPI C03/14; REIPI RD06/0008). O. M. was supported by a grant from the REIPI.
None to declare.
We thank V. Moreno from the Universitat de Barcelona for his assistance with the statistical analysis; and C. Jimenez from the Hospital Universitari de Bellvitge for her technical assistance with the electron microscopy.