Exploratory Study of the Prevalence and Clinical Significance of Tobramycin-Mediated Biofilm Induction in Pseudomonas aeruginosa Isolates from Cystic Fibrosis Patients

Abstract

Exposure to aminoglycoside antibiotics induces biofilm formation by a laboratory strain of the cystic fibrosis (CF) pathogen Pseudomonas aeruginosa. Here, we detected this effect among about half of the clinical isolates from CF patients in a cross-sectional collection, suggesting that biofilm induction may represent a common mechanism of inducible aminoglycoside resistance in CF infections. This induction always occurred at the same tobramycin concentration regardless of MIC, suggesting that the mechanisms of killing and induction may be separable.

People with the genetic disease cystic fibrosis (CF) suffer from chronic airway infections with the Gram-negative bacterium Pseudomonas aeruginosa. These infections are associated with progressive decline in lung function and early death. Evidence suggests that P. aeruginosa infects airways of CF patients as antibiotic-resistant biofilms, which may help to explain why antibiotic treatments are, at best, only marginally effective in treating chronic airway infections in CF patients.

Currently, the antibiotic used most commonly to treat chronic P. aeruginosa airway infections in CF patients is tobramycin. Most children with CF infected with P. aeruginosa are treated both chronically and episodically with this drug. We showed previously that exposure to subinhibitory concentrations of tobramycin and other aminoglycosides induces P. aeruginosa to form higher levels of biofilm, leading to additional resistance to killing (3). Furthermore, by studying P. aeruginosa laboratory strain PAO1, we found that this “defensive” response was mediated by the cyclic diguanylate (CDG) signaling pathway. In PAO1, a CDG phosphodiesterase, which we called the aminoglycoside response regulator (Arr), was required for this response (3). However, Arr is not encoded by all P. aeruginosa clinical isolates from CF patients (5). Furthermore, while we found previously that most of the small number of P. aeruginosa clinical isolates from CF patients that we studied exhibited biofilm induction by tobramycin, the prevalence of biofilm induction in a larger population of clinical isolates from CF patients and the role of Arr in this phenotype are unknown. Similarly, we did not previously explore the relationship between biofilm induction and tobramycin susceptibility (e.g., whether tobramycin-resistant and -susceptible isolates exhibit biofilm induction at the same tobramycin concentrations). Both of these issues must be clarified in considering the design of a clinically useful cotherapeutic agent that would subvert the biofilm induction response. Therefore, we sought to examine the prevalence and clinical mechanism of tobramycin-mediated biofilm induction among a cross-sectional collection of P. aeruginosa clinical isolates from CF patients.

(Portions of this work have been presented in abstract form [1].)

We assembled a cross-sectional collection of 81 nonredundant clinical isolates of P. aeruginosa from 74 children with CF, aged ≤15 years at the time of collection, between 1998 and 2002. Isolates were selected randomly from the CF Foundation-funded Antimicrobial Toolkit collection at Seattle Children's Hospital (SCH), as described previously (4). Rigorous antibiotic treatment data for these patients were not routinely available. The study was approved by the SCH Institutional Review Board.

Biofilm induction was determined by growth of isolates in 96-well polystyrene plates in the presence of 0, 0.1, 0.2, 0.3, and 0.4 μg/ml tobramycin, as described previously (3, 7). With 6 replicates per condition, we found the standard deviation of biofilm induction to be approximately 0.1-fold for the entire collection; therefore, we defined a statistically significant change in biofilm formation to be a 1.2-fold alteration upon drug addition compared to the level with no drug (i.e., a change of ≥2 standard deviations) for this collection. Using this criterion, we found biofilm induction to occur in 40 isolates of the 81 present (49%) and always at 0.2 to 0.3 μg/ml tobramycin (results are shown in Table 1).

TABLE 1.

Summary of results

Characteristic	No. of isolates (%) with characteristic:
	Present	Absent
Induction	40/81 (49)	41/81 (51)
Tobramycin MIC ≥2-fold different from that for PAO1	22/40 (55)	18/40 (45)
High baseline biofilm formation	26/81 (32)	55/81 (68)
Inducibility among isolates with high baseline biofilm formation	11/26 (42)	15/26 (58)
Low baseline biofilm formation	17/81 (21)	64/81 (79)
Inducibility among isolates with low baseline biofilm formation	11/17 (65)	6/17 (35)
Identification of arr by PCR	31/81 (38)	50/81 (62)
Inducibility among isolates positive for arr	17/31 (55)	14/31 (45)
Inducibility among isolates negative for arr	23/50 (46)	27/50 (54)

The detection of biofilm induction among these clinical isolates may be impacted by the magnitude of biofilm formation in the absence of tobramycin; for example, an isolate with a high baseline level of biofilm formation may not be able to be induced significantly. Conversely, an isolate with low baseline biofilm formation may simply be defective for biofilm growth and may thus be uninducible. To test whether the ability of an isolate to form any biofilm in the absence of tobramycin (baseline biofilm formation) was predictive of inducibility, we compared the baseline for each isolate to that of a control strain, laboratory strain PAO1. Twenty-six isolates displayed a baseline biofilm formation ability at least 1.5-fold higher than that for PAO1; of these isolates, 11 (42%) exhibited induction. Conversely, of the 17 isolates with a decreased baseline (≤1.5-fold that of PAO1), 65% exhibited induction. In summary, there was no clear relationship between biofilm inducibility and baseline biofilm formation, suggesting that magnitude of inducibility is not simply a surrogate marker of the inherent ability to form biofilm.

The identification of Arr as required for full biofilm inducibility by tobramycin in PAO1 (3) suggested the possibility of a specific pathway for a phenotypic response to aminoglycosides, potentially separable from the mechanism of aminoglycoside killing. To further explore this possibility, we examined the relationship between the MIC of tobramycin and the concentration at which biofilm formation was induced among our collection of clinical isolates. Among the 40 isolates displaying induction, 22 (55%) exhibited tobramycin MICs ≥2-fold different from that for PAO1 (of these 22 isolates, 11 exhibited higher MICs and 11 lower); however, as noted above, induction for all of these isolates always occurred in the same tobramycin concentration range. Of the 7 isolates classified as “resistant,” due to a tobramycin MIC of ≥16 μg/ml, 3 exhibited biofilm inducibility. Similarly, while the mucoid phenotype has been associated with a poor clinical response to tobramycin (2), of the 24 isolates we identified as mucoid, 8 (33%) exhibited inducibility, indicating no clear relationship between mucoidy and biofilm inducibility. These data support the concept that the mechanisms of growth inhibition and killing by tobramycin and that of biofilm induction may be separable. As the mechanism of tobramycin resistance among clinical isolates from CF patients is most often described as “impermeability” (6), likely due to either poor penetration or avid extrusion through membrane pumps, these findings further suggest that biofilm induction occurs via events that occur at the cell surface. To examine the specificity of this response, we tested a subset of isolates, chosen to reflect a range of biofilm inducibility by tobramycin, for their responses to two other aminoglycosides used frequently in CF care, amikacin and gentamicin. We found that the same pattern of biofilm induction was conferred by all three aminoglycosides but that, as we found for PAO1 previously (3), tobramycin was the most effective in the strains exhibiting induction, followed by amikacin and then gentamicin. This apparent selectivity further suggests a common mechanism for aminoglycoside-mediated biofilm induction.

While Arr was required for a full tobramycin response in PAO1, the arr gene is not encoded by all clinical isolates. To examine the relationship between Arr and biofilm induction among our collection of isolates, we used a PCR approach to identify the presence or absence of the arr gene. Two different primer sets (CCAGGCATCGGCACGATAG/GCTCGGCGTTGTCTCGGAC and GCGAGCCACTGACATTGACAC/GCGGCGGGACAGTCTCTTTC) were designed to amplify different portions of the arr gene that are specific for that gene, in order to minimize amplification of a paralogous gene (such as a c-di-GMP regulator encoded elsewhere on the chromosome). These primer pairs generated products of different sizes, which aided discrimination. For each experiment, PAO1 and PA14 (the latter of which does not encode arr [5]) were also tested as positive and negative controls, respectively. Of the cross-sectional isolates, 38% tested positive for the presence of the arr gene. Of these, 55% exhibited biofilm induction as defined above, while 45% did not. Of the isolates without the arr gene, 46% showed biofilm induction, while 54% did not. Therefore, there is no clear relationship between the presence of the arr gene in clinical isolates and biofilm inducibility by tobramycin. As there are numerous c-di-GMP regulators encoded by P. aeruginosa (3, 5), some of which are homologous to arr, it may be that, in clinical isolates, other c-di-GMP regulators fulfill the role played by Arr in PAO1.

In conclusion, we found that approximately half of the P. aeruginosa clinical isolates from CF patients exhibited significant biofilm induction by tobramycin. The presence of inducibility was independent of (i) the inherent abilities of those isolates to form biofilm, (ii) the presence or absence of the arr gene, and (iii) the susceptibilities of those isolates to tobramycin. Therefore, biofilm induction may represent a common mechanism of inducible aminoglycoside resistance during chronic P. aeruginosa infections in CF patients. The mechanism of biofilm induction in clinical isolates remains to be fully elucidated but may represent a useful therapeutic target. Future studies are required to identify this mechanism, as well as the relationship between biofilm induction, response to therapy, and prior exposure to antibiotics.