Metabolic Activity of Staphylococcus epidermidis Is High during Initial and Low during Late Experimental Foreign-Body Infection

Abstract

Foreign-body infection (FBI) is notoriously resistant to eradication by antibiotic treatment. It is hypothesized that reduced bacterial metabolic activity contributes to this resistance. We examined the metabolic activity of Staphylococcus epidermidis in 204 samples recovered during in vitro foreign-body colonization and in 424 samples recovered during in vivo FBI in a rat model. Metabolic activity was measured by determining the amount of 16S rRNA per genome by quantitative PCR. The initial foreign-body-associated growth proved to be a metabolically active process, both in vitro and in vivo. The initial 16S rRNA content was similar to that observed during in vitro exponential-growth phase. However, during late in vivo FBI, a 114-fold (P ≪ 0.0001) decrease in the 16S rRNA content was observed, indicating that there was markedly decreased metabolic activity. This decreased metabolic activity during late FBI can explain at least in part why such infections are so difficult to eradicate with conventional antibiotic treatment.

Coagulase-negative staphylococci (CoNS) are the leading cause of foreign-body infections (FBI), such as infections of intravascular devices, prosthetic valves, endovascular grafts, orthopedic endoprotheses, and cerebrovascular shunts (14). CoNS are opportunistic pathogens that mainly depend on the presence of these foreign bodies to become pathogenic (14). The formation of a biofilm on the foreign body has been shown to be a key factor in the pathogenesis of staphylococcal FBI (2, 20). The clinically most important characteristic of CoNS FBI is their persistent nature. Although antibiotics usually suppress inflammatory symptoms quite easily, eradication of the infection generally requires removal of the device (23). The exact nature of the resistance remains unclear, but decreased metabolic activity of the foreign-body-associated bacteria is postulated to be an important factor. Numerous reports have demonstrated that metabolically less active bacteria are tolerant to the killing effects of most antibiotics (3, 4, 9-11). One paper suggested that there was decreased metabolic activity of Klebsiella pneumoniae after prolonged incubation on polycarbonate filter membranes (28). Data on bacterial metabolic activity during FBI are, however, still lacking. In this study we aimed to examine the metabolic activity in CoNS during in vitro and in vivo FBI.

In heterotrophic bacteria, the cellular concentration of ribosomes has been shown to be proportional to the total protein synthesis rate and thus to the metabolic activity over a wide range of conditions (8, 19). During ribosome synthesis, rRNA synthesis is the rate-limiting step (17, 19). Thus, the amount of rRNA per cell can be used as a measure of metabolic activity (8). This link was initially demonstrated in Escherichia coli and was subsequently confirmed in several bacterial species that belong to distantly related genera, such as Bacillus, Vibrio, Pseudomonas, Thermotoga, and Synechococcus (16, 22). A similar correlation was indirectly documented in Staphylococcus aureus during growth following starvation recovery (6). It therefore seems reasonable to assume that the link between rRNA and protein synthesis is essential to bacterial physiology and to use the amount of rRNA per cell as a measure of metabolic activity in staphylococci (8). A recently developed technique for direct quantification of bacterial transcripts during in vitro or in vivo FBI by means of a quantitative PCR was used to quantify the amount of rRNA per cell in a large number of in vivo and in vitro samples (25, 26). The amount of rRNA per cell was quantified by determining the amount of cDNA per genome (or the cDNA/genomic DNA [gDNA]ratio) (25, 26). By using this technique, it was demonstrated previously that structural biofilm genes, such as ica and aap, as well as mecA, were mainly transcribed during early in vivo FBI but not during late in vivo FBI (25). A similar method has been described by Goerke and coworkers for the study of gene expression in in vitro and in vivo S. aureus infections (12, 13, 29).

(Preliminary reports of this work were presented at the 41st Interscience Conference on Antimicrobial Agents and Chemotherapy, Chicago, Ill., 2001, and at the 12th European Congress of Clinical Microbiology and Infectious Diseases, Milan, Italy, 2002.)

MATERIALS AND METHODS

Bacterial strain.

For all experiments, a previously well-characterized biofilm-producing Staphylococcus epidermidis strain, strain 10b, which was isolated from a patient with a proven catheter-related bloodstream infection, was used (25, 26, 27). This strain has a 100% infection rate in the rat model used (27).

Cloning and quantification of the number of plasmid copies.

The 16S rRNA gene was cloned in the pGEM-T easy vector (Promega, Madison, Wis.) as previously described (24, 26). Pure plasmid DNA was obtained by using a High Pure plasmid isolation kit (Roche Diagnostics GmbH, Mannheim, Germany) and was quantified with a GeneQuant RNA/DNA calculator (Amersham Pharmacia Biotech, Uppsala, Sweden) at a wavelength of 260 nm. The number of gene copies in the plasmid stock solution was 2.11 × 10¹⁰ copies per μl (95% confidence interval, 2.06 × 10¹⁰ to 2.17 × 10¹⁰ copies per μl).

16S rRNA contents of planktonic and sessile bacteria during in vitro foreign-body colonization in a physiologic salt solution.

The in vitro model has been described previously. Briefly, a frozen culture of S. epidermidis 10b was grown to the end of the exponential growth phase, pelleted, and resuspended in 0.9% NaCl. Fragments (length, 7 mm) of a commercial polyurethane intravenous catheter (Arrow International, Reading, Pa.) were added to this bacterial suspension and placed in a water bath at 37°C. After different incubation periods, nucleic acids (RNA and DNA) were extracted instantaneously by a method adapted from the method of Cheung et al. (1) as previously described (25, 26). Briefly, for bacteria in suspension (planktonic bacteria), an aliquot of a bacterial suspension was rapidly cooled on ice. The bacteria were pelleted, and the pellet was suspended in 500 μl of NAES buffer (50 mM sodium acetate [pH 5.1], 10 mM EDTA, 1% sodium dodecyl sulfate) and added to a FastRNA blue tube (Bio 101, Carlsbad, Calif.) with 500 μl of acidified phenol-chloroform (5:1; pH 4.5; Ambion, Austin, Tex.) at room temperature. For the bacteria that adhered to a catheter segment, the colonized catheter fragment was washed with 1 ml of 0.9% NaCl and added directly to a FastRNA tube containing 500 μl of NAES buffer and 500 μl of acidified phenol-chloroform. The FastRNA tubes were shaken for 23 s at 6,000 rpm with a FastPrep instrument (FP 120; Bio 101, Savant, Holbrook, N.Y). After shaking, the tubes were centrifuged, and 90% of the supernatant (450 μl) was precipitated with 520 μl of isopropyl alcohol and 35 μl of 3 M sodium acetate. The pellet was washed with 70% ethanol and resuspended in 150 μl of RNase-free water. Fifty microliters of this sample was diluted 1/10 and used for quantification of gDNA. The remaining 100 μl was purified with an Rneasy mini kit (Qiagen, Hilden, Germany) and treated with RNase-free DNase (Qiagen) on Rneasy columns according to the instructions of the manufacturer. The RNA was finally dissolved in 60 μl of RNase-free water. Reverse transcription of the RNA was performed with Moloney murine leukemia virus reverse transcriptase (Promega) as previously described (25, 26).

Nucleic acid was isolated at time zero (n = 18) (just before the bacteria were suspended in 0.9% NaCl) and at 10 min (n = 48), 35 min (n = 48), 60 min (n = 48), 120 min (n = 40), and 180 min (n = 25) both from the bacteria that were adhering to the catheters (referred to as sessile bacteria) and from planktonic bacteria. Data for most time points were generated in at least three different experiments; the only exceptions were the 120- and 180-min points (two independent experiments).

16S rRNA content of sessile bacteria during in vivo FBI.

We used a previously described model in which first-generation descendants of inbred Fisher rats harbored under germfree conditions since 1965 were used (27). These rats, exposed to normal rat flora from birth, were labeled ex-germfree Fisher rats. In this model, which resembles the other subcutaneous models for FBI, catheter fragments were inoculated with a small amount of S. epidermidis 10b before implantation, which resulted in a 100% infection rate (27). The experimental conditions have been described in detail previously (25, 26). Briefly, 7-mm catheter fragments were incubated for 20 min in a suspension of S. epidermidis in 0.9% NaCl as described above. After 20 min, the suspension with catheter fragments was placed on ice. The back of each rat was shaved over a large area, and the skin was generously disinfected with 0.5% clorhexidine in 70% alcohol and allowed to dry for 2 min. A 10-mm skin incision was made at the base of the tail of each adult ex-germfree Fisher rat, and the subcutis was carefully dissected to create four subcutaneous tunnels. In each rat, three or eight catheter fragments were quickly inserted subcutaneously at least 2 cm from the incision; the distance between two fragments was at least 1 cm. For catheter explantation, animals were sacrificed by using diethyl ether. The skin was disinfected as described above, and the naked catheter segments were gently removed from the subcutaneous tissue. All catheter fragments from one animal were used for one time point. In each experiment, baseline expression levels in sessile bacteria before implantation were determined (time zero; n = 22). A total of 402 polyurethane catheter segments were implanted subcutaneously in the rat model, and these segments were explanted at 10 different times. The times used were 15 min (n = 31), 1 h (n = 34), 2 h (n = 33), 4 h (n = 34), 6 h (n = 33), 12 h (n = 34), 24 h (n = 34), 2 days (n = 39), 7 days (n = 34), and 14 days (n = 96). Data for each in vivo time point were generated in at least five independent experiments by using at least five different rats. Nucleic acid isolation and cDNA synthesis were performed immediately after explantation as described above; the only difference was that the catheters were not washed with 0.9% NaCl before nucleic acid extraction.

Taqman quantitative PCR.

Quantification of both the cDNA and the gDNA was performed with the ABI Prism 7700 sequence detection system (PE Applied Biosystems, Foster City, Calif.) as previously described (24-26). During each run, a standard dilution of the plasmid with a known quantity was included to permit gene quantification with the supplied software according to the instructions of the manufacturer. In each run a negative control (distilled water) was included. The Taqman primers and probes used have been described elsewhere (24).

Statistical analysis.

All statistical analyses were performed with SAS (version 8; SAS Institute, Inc., Cary, N.C.) and with SPSS (version 11.01; SPSS Inc., Chicago, Ill.). Since the in vitro and in vivo cDNA/gDNA ratios were not normally distributed at any time point (Shapiro-Wilk test for normality), all data were log₁₀ transformed in order to fulfill the requirements of normality (18).

For the in vitro data, two hypotheses were tested. A significant change in gene expression levels over time within one group (sessile or planktonic) was tested with a one-way analysis of variance (ANOVA). A significant difference in the evolution over time of the gene expression levels between the sessile group and the planktonic group was tested with a two-way ANOVA. When the one-way ANOVA was significant, two-sided univariate tests with a correction for multiple comparisons were done (Bonferroni test) to locate the significant differences. When the two-way ANOVA was significant, multiple two-sided t tests were done with a correction for multiple testing (α = 0.05/5 = 0.01).

For the in vivo data, a one-way ANOVA was used to test if there was a significant evolution of the expression levels over time. When the one-way ANOVA test was significant, the two-sided Bonferroni multiple-comparison method was used to determine which time points differed at α = 0.05, with a correction for multiple comparisons.

RESULTS

16S rRNA contents of planktonic and sessile bacteria during in vitro foreign-body colonization in a physiologic salt solution.

The results are summarized in Fig. 1. In planktonic bacteria a progressive decrease in cellular 16S rRNA content was observed (P < 0.0001, as determined by one-way ANOVA, with the Bonferroni test locating significant differences between time points that were not immediately successive), whereas in sessile bacteria a temporary increase was followed by a decrease in the rRNA content (P < 0.0001, as determined by one-way ANOVA, with the Bonferroni test locating significant differences between expression levels at 10 and 35 min on the one hand and the last two expression levels on the other hand). This difference in evolution of the rRNA content between sessile and planktonic bacteria growing under the same conditions was significant (P < 0.0001, as determined by two-way ANOVA). The maximal 16S rRNA content in sessile bacteria was 1.181 log₁₀ cDNA/gDNA, which is the same order of magnitude as the maximal 16S rRNA content observed during in vitro exponential growth in a rich medium (1.48 log₁₀ cDNA/gDNA directly after inoculation of the bacteria into brain heart infusion broth) (26). The difference in 16S rRNA content between sessile and planktonic bacteria was 0.97 log₁₀ or 9.4-fold (P = 2.94 × 10⁻⁷, as determined by a t test) at 10 min and 0.96 log₁₀ or 9.1-fold (P = 7.05 × 10⁻⁶, as determined by a t test) at 35 min.

FIG. 1.

Evolution of the 16S rRNA contents of sessile and planktonic bacteria after inoculation into 0.9% NaCl in vitro. The solid line indicates the 16S rRNA contents of sessile bacteria. The dashed line indicates the 16S rRNA contents of planktonic bacteria. The 16S rRNA content is expressed as the log₁₀ cDNA/gDNA ratio on the y axis. The error bars indicate standard errors. The x axis indicates time.

16S rRNA content of sessile bacteria during in vivo FBI.

The evolution of the rRNA content during in vivo FBI is summarized in Fig. 2. Given the rapid changes in rRNA content during early in vivo FBI observed during preliminary experiments, multiple time points were included during the first 24 h. After implantation of the catheter fragments into the rats, the 16S rRNA content increased, and it peaked after 15 min. The maximal rRNA content at this time was 1.1324 log₁₀, which is the same order of magnitude as the maximal rRNA content observed during in vitro exponential growth in a rich medium (26). Following this peak, a 2.056 log₁₀ or 114-fold decrease (95% confidence interval, 1.81 log₁₀ to 2.30 log₁₀; P = 2.04 × 10⁻³³, as determined by a t test) in the 16S rRNA content was observed. This decrease was highly significant (P < 0.0001, as determined by one-way ANOVA). There were significant differences between all distant time points (P < 0.01 for 36 pairs, as determined by a Bonferroni test). The decrease in the 16S rRNA/DNA ratio was most pronounced during the first 24 h of in vivo FBI and leveled off after 48 h.

FIG. 2.

Evolution of the 16S rRNA content during 2 weeks of in vivo FBI. (Top) Evolution of the 16S rRNA content during the entire period studied. (Bottom) Evolution of the 16S rRNA content during the first 24 h. The 16S rRNA content is expressed as the log₁₀ cDNA/gDNA ratio on the y axis. The error bars indicate standard errors.

DISCUSSION

In the present study we evaluated the metabolic activity in a large number of S. epidermidis samples recovered from in vitro and in vivo FBI by using 16S rRNA transcript analysis. This study was the first study to substantiate (in S. epidermidis) the previous hypotheses that initial foreign-body colonization is an active process (14) and that bacteria have decreased metabolic activity during late FBI (23).

An increase in the 16S rRNA content to levels similar to those observed during the early exponential growth phase in broth was observed during in vitro foreign-body colonization in a poor medium, as well as during initial in vivo foreign-body-associated growth. As previously demonstrated, this increase coincided with increased transcription of the ica genes during initial in vitro foreign-body colonization and with a high level of transcription of several genes involved in the synthesis of structural bacterial or biofilm compounds (ica, aap, mecA, and gmk) during initial in vivo foreign-body-associated growth (25). Thus, it can be hypothesized that the higher metabolic activity during initial FBI is necessary for adaptation to a new growth mode and for construction of a biofilm. The active nature of the initial foreign-body-associated growth may have two clinical implications. It may explain the importance of the timing of prophylactic administration of antibiotics in the prevention of surgical-wound infections, as observed in a large, randomized trial (5). In this trial the risk for developing surgical-wound infections was lowest when antibiotics were administered shortly before the onset of the operation and was 6.7- and 5.8-fold higher when antibiotics were administered more than 2 h before incision and more than 3 h after incision, respectively. Most surgical-wound infections are related to foreign bodies, such as sutures, implanted foreign bodies, and drains (15). When antibiotics are administered just before incision, maximal tissue levels can be expected during the operation. Furthermore, administration of antibiotics just before incision results in peak tissue concentrations when the molecular targets of the antibiotics have their highest levels of expression in the infecting bacteria. Second, the efficacy of minocycline-rifampin-coated catheters for prevention of catheter-related bloodstream infections can also be explained in part by our observations (7, 21). Minocycline acts on the ribosomal complex, and rifampin acts on RNA transcription. The high levels of expression of the targets of these antibiotics in the initial phase of foreign-body colonization may make the bacteria more vulnerable to the local concentrations of these antibiotics and account for the clinical success of minocycline-rifampin-coated catheters.

A 114-fold decrease in the 16S rRNA content was observed during the 2-week course of FBI in the rat model. The number of gDNA copies correlated very well with the number of CFU during the first 12 h of in vivo growth, but differences of 0.83 log₁₀ (3.8-fold) and 0.69 log₁₀ (4.8-fold) were observed between gDNA levels and the number of CFU after 24 h and 1 week, respectively (data not shown). At these late times, the number of CFU may be an underestimate of the number of viable bacteria due to the growth of cell aggregates as single CFU or due to auxotrophic variants. On the other hand, it is conceivable that—especially at 1 and 2 days—the temporary accumulation of DNA from dead bacteria results in overestimation of the extent of the decrease in the 16S rRNA content. Nevertheless, the large difference was highly significant and contrasted with the constant high levels of expression of the atlE gene and the moderate decrease in the expression of the gmk gene as previously described (25). Although the correlation between metabolic activity and ribosomal content may be weaker in very slowly growing bacteria (19), these observations provide strong evidence which supports the hypothesis that the bacteria recovered from chronic foreign-body-associated infections in vivo have reduced metabolic activity compared with the activity under other growth conditions (8, 17, 19). However, these observations do not permit us to determine which metabolic activity in CoNS should be considered normal or a reference metabolic state. As mentioned above, there is a large amount of experimental evidence that indicates that metabolically less active bacteria have reduced susceptibility to many antibiotics (3, 4, 9-11). Consequently, the observed decreased metabolic activity during late FBI can explain at least in part why these infections are so difficult to eradicate with conventional antibiotic treatment.