Respiratory tract colonization of intubated patients by P. aeruginosa is initiated by the colonization of the intubation device by bacteria either originating from the digestive flora of the patient (endogenous acquisition) or transmitted via handling of the device by health care workers (exogenous acquisition) . This colonization requires efficient bacterial adherence to the inert surface and the formation of a biofilm . P. aeruginosa then reach the respiratory tract either by leakage around the cuff of the intubation device, by shearing of secretions containing embedded bacteria from the device during the inspiratory air flow, or by contamination and direct inoculation of the respiratory tract by suction tubes introduced through the intubation device by health care workers [23–25].
Both in vitro studies and animal models have suggested that QS plays a major role in the virulence of P. aeruginosa . However in situ and in patient data concerning the potential role of QS during colonization and infection by P. aeruginosa in humans remain scarce. QS-activity has been linked to the detection of P. aeruginosa AIs in the lungs of cystic fibrosis patients [13–16], as well as in lung transplanted patients . So far studies have only shown that most of the isolates originating from colonized intubated patients are QS-proficient [18, 22].
In the present study we show that the two P. aeruginosa autoinducers, 3-oxo-C12-HSL and C4-HSL, can be detected in situ in biofilms covering intubation devices retrieved from patients colonized by P. aeruginosa. Moreover all P. aeruginosa isolates collected either from these biofilms or from tracheal aspirates produced these AIs in vitro. Whereas 3-oxo-C12-HSL has been previously shown to play a role in the differentiation of P. aeruginosa biofilms [7, 26], C4-HSL seems to be important during the maturation stage of biofilm development , for the total amount of biofilm formed , and for the maintenance of biofilm architecture . In our study, isolates recovered from the biofilm of intubation devices produced higher levels of C4-HSL than 3-oxo-C12-HSL in vitro. Previous observations have suggested that C4-HSL is produced in higher quantities than 3-oxo-C12-HSL in P. aeruginosa biofilms in the lungs of CF patients [13, 14]. Singh et al. have even suggested that this AI ratio might serve as a biomarker for the biofilm mode of growth, as the mucoid isolates from the sputum of CF patients inverted their AI ratio when sub-cultured in liquid medium . In contrast, in our study a higher level of C4-HSL as compared to 3-oxo-C12-HSL was observed when isolates were grown in liquid cultures, suggesting that this ratio is not useful as a biomarker for biofilm growth for these isolates. None of the isolates collected from our patients had a mucoid phenotype. This discrepancy could therefore be explained by inherent differences in biofilm development patterns between mucoid and non-mucoid strains. So far there is no evidence that non-mucoid P. aeruginosa colonizing the lung of intubated patients grows in biofilms. Such a hypothesis deserves further investigations.
Autoinducers have been previously recovered from biofilms formed on urinary tract catheters  and our results show for the first time that QS-signaling molecules are also produced in biofilms covering intubation devices. Whether the local concentrations of AIs on the intubation devices that we measured are physiologically relevant and are sufficient to activate the expression of QS-dependent genes remains speculative. However microarray analyses have shown that QS is a continuum and that the expression of some genes is already activated even at low AI concentrations . It is therefore likely that even low concentrations of AIs on the intubation devices are physiologically relevant.
Intriguingly, the in vitro production of both AIs varied with the site of collection. Isolates from tracheal aspirates produced higher levels of C4-HSL but lower levels of 3-oxo-C12-HSL than their genotypically identical counter-parts isolated from intubation devices, even after several in vitro passages. This observation suggests that different microenvironments select P. aeruginosa isolates with various AI production capacities. Moreover the production of elastase, and the capacity to adhere to an inert surface and to produce a biofilm also varied with the site of isolation. Tracheal aspirate isolates produced higher levels of elastase, but were less able to adhere and produce biofilm than their counter-parts recovered from intubation devices. LasB elastase is one of the major virulence factors controlled by the P. aeruginosa QS-circuit . In vitro its production is regulated by the QS-circuit and depends mainly on the production of 3-oxo-C12-HSL [9, 10]. LasB elastase is believed to allow P. aeruginosa to invade surrounding tissues by its broad range enzymatic activity and to facilitate blood stream invasion by degradation of elastin fibers in the lamina propria of blood vessels .
The observed differences in phenotypes between isolates obtained from tracheal aspirates, compared to those from the intubation devices, make sense. Isolates growing in the biofilm on intubation devices might not invest energy in producing elastase, but might be primed to adhere and form biofilms, whereas isolates growing in the lungs might find an advantage in producing elastase. For tracheal aspirate isolates, levels of 3-oxo-C12-HSL production correlated positively with the capacity to produce elastase. In contrast for isolates originating from the intubation device, 3-oxo-C12-HSL levels did not correlate with elastase production, but correlated weakly with the capacity to adhere and to form biofilms on inert surfaces. These observations suggest that the previously described control of LasR on elastase production might not apply to all isolates depending on the site they were collected from. It seems that the microenvironment on the intubation devices selects for phenotypes that produce high levels of 3-oxo-C12-HSL but without concomitant increased elastase production. In contrast the lower lung environment seems to select for isolates that produce less 3-oxo-C12-HSL but with a positive control of this AI on elastase production. Two of our patients (12 and 13) received antimicrobial therapies during their mechanical ventilation. We can therefore not exclude that different concentrations of these drugs at the two sites might have selected different phenotypes. However this seems unlikely as isolates of patient 8 presented the same differences in phenotype in the absence of any exposure to antimicrobial therapies.
We detected no correlations between the levels of C4-HSL production and elastase activity, adhesion or biofilm formation. Strikingly, these phenotypes remained stable for each isolate independently of its origin even after several in vitro passages. This suggests that mutations in specific regulators, that affect the production of AIs and the expression of QS-dependent phenotypes might be selected in particular microenvironments. Indeed lasR and rhlR mutants of P. aeruginosa strains isolated from intubated patients have been previously described . Such mutations influence the bacteria capacity to produce AIs, and are expected to affect QS-dependent phenotypes. We therefore sequenced both lasR and rhR in 6 TA and 6 ID isolates (Table 2). None of them harbored mutations compared to the wild type strain PAO1 (data not shown). This explains why all the isolates analyzed in the present study were QS-proficient. Phenotypic variations not linked to QS but also influencing biofilm formation have been described in P. aeruginosa isolates from CF patients . The nature of the mutational events possibly selected either during biofilm growth on intubation devices, or inside the lung are presently under investigation. The identification of these events would have major implications for our understanding of bacterial selection in specific environments and the dynamics of bacterial populations.
QS-inhibition has been suggested as a new therapeutical approach against P. aeruginosa infections. For instance macrolides interfere with the production of AIs [8, 30] and were shown to retard biofilm formation . Our result might be important if QS-inhibitors that interfere with the production and/or activity of one, or both, of the autoinducers are used in the prevention or treatment of P. aeruginosa VAP. Indeed, our data suggest that inhibition of QS might have different effects on P. aeruginosa isolates depending on their microenvironnement.