Pseudomonas aeruginosa is a Gram negative bacteria producing a wide array of virulence factors frequently responsible for chronic airway infections in cystic fibrosis (CF) or chronic obstructive pneumonia disease (COPD) patients, as well as acute nosocomial airway infections in intensive care units 123.

In acute P. aeruginosa pneumonia, the functional consequences, and particularly lung fluid movements, have been studied extensively. Lung fluid balance is the result of fluid movements following active ion transport by functional alveolar cells, and permeability of the alveolar capillary barrier. In P. aeruginosa-induced acute lung injury (ALI), distal airspace fluid clearance (DAFC) is typically increased at 24 hours through a TNF-α pathway 4. Studies have also shown that the capacity of maintaining alveolar active fluid transport is correlated with patient outcome in ALI 56. Lung liquid clearance (LLC) is another functional marker reflecting the capacity of the lung to evacuate fluid instilled in the alveoli outside the lung, LLC involves DAFC, epithelial and endothelial permeabilities 7. We previously showed that, even though DAFC is upregulated, LLC is decreased at both 4 and 24 hours in ALI 7 reflecting a major endothelial injury overwhelming the alveolar response.

In chronic infection, these functional consequences on lung fluid balance are less clear. In the 70's, Cash developed an experimental model of chronic pneumonia by intra tracheal injection of P. aeruginosa embedded in agar beads 8. Most of the work performed with this model has focused on immunological, inflammatory, or nutritional aspects 9101112. To the best of our knowledge, no previous work has tried to evaluate alveolar permeability and lung fluid transport in P. aeruginosa chronic lung infection. In order to elucidate these functional aspects we studied lung fluid transport in an experimental model of chronic P. aeruginosa lung infection in the rat. After the validation of the experimental model, we studied alveolar function: alveolar-capillary barrier permeability, lung liquid clearance, distal airspace fluid clearance and its pharmacologic stimulation.

In our study we validated an experimental model allowing us to explore alveolar function in chronic P. aeruginosa lung infection through measurements of lung liquid movements. In this model of chronic P. aeruginosa lung infection, after observing an initial increase of both alveolar permeability and lung fluid movements, we characterized an impairment of DAFC where, even though DAFC returned to baseline, it remained unresponsive to pharmacological stimuli.

In the first part of our work, we validated, on several parameters, the chronic infection model previously described by Cash et al 8. After reaching a peak on the second day of the infection and decreasing from the 5^th to the 15^th day, lung bacterial load persisted for 3 weeks. These results, as well as the analysis of the BAL and the histological features, are consistent with the literature 8111718.

Since, in this model, P. aeruginosa is associated with agar beads, we performed, as control groups, instillation of sterile agar beads. Sterile agar bead instilled rats did not show any evidence of weight loss and although they did present an increase in BAL cellularity, there were no PMN's except a slight increase on the second day which failed to reach a statistical significance (data not shown). This result is consistent with the literature, Nacucchio et al showed that agar beads alone could not reproduce the same level of injury than P. aeruginosa in agar beads 19.

From this first part of our work, we concluded that the model of chronic infection with P. aeruginosa is adequate, based on clinical, bacteriological, cytological and histological data.

Although a clinical study has reported increased lung permeability in COPD patients infected by P. aeruginosa 20, few studies have focused on the consequences of chronic lung infection on alveolar function and particularly fluid movements. In our study, lung fluid movements were maximal on the 2^nd day post infection. We observed an increase of alveolar-capillary barrier permeability, DAFC and overall lung liquid clearance. A normal lung wet to dry weight ratio was a consequence of this adequate alveolar response. This contrasts sharply with the data we obtained in an acute lung injury model where LLC dramatically decreased and W/D weight ratio increased at 4 and 24 hours after Pseudomonas instillation 7.

In our chronic model, following the increase in both permeability and lung liquid clearance, we observed an improvement in permeability with a return to baseline of these 2 parameters on the 5^th day.

The St group presented a moderate increase in permeability (Asp/plasma ratio: 0.31 ± 0.04), it has previously been reported that agar beads could alone be responsible for a moderate increase in permeability 19. However, taking into account the association of the other parameters validating the model (clinical, bacteriological, cytological and histological), this effect does not challenge the model.

Our results showed an increase of the DAFC at 48 hours post infection. In acute lung injury, the initial alveolar response is usually towards an increase of DAFC which many authors have documented in septic shock 21, or after endotoxin administration 22. In septic shock, this increase was related to the release of endogenous catecholamines. In acute P. aeruginosa pneumonia, increased DAFC can be related to either Pseudomonas exoproducts 15 or to a TNF-α dependent mechanism during the first 24 hours of the infection 4. We tested in our model whether TNF-α or catecholamines could explain our results. TNF-α was not detectable and systemic endogenous epinephrine or nor-epinephrine not different from controls on the 2^nd or the 5^th day. TNF-α is produced during the early phase of pneumonia, and its short half life probably explains the absence of detectable levels at 48 hours. A dynamic evaluation of TNF-α production with serial samples or antibody neutralization experiments would be helpful to precisely study the role of TNF-α. We therefore did not rule out that TNF-α may have triggered an inflammatory response which could be responsible for the increased DAFC. Other potential mechanisms such as Transforming Growth Factor β remain to be investigated 23.

Surprisingly, on the fifth day, DAFC returned to baseline along with the improvement in permeability. Although it is logical to see an improvement in permeability, consistent with a decrease of the bacterial burden and an adequate host response, DAFC was expected to remain increased. The persisting presence of the pathogen in the alveoli and many factors only related to its presence would normally lead to a persistent increase of DAFC 15. We therefore decided to evaluate if a normal increase in DAFC could be elicited on the 5^th day post infection in response to known pharmacological stimuli 2425. In the normal lung, intra-alveolar administration of terbutaline generates a DAFC increase of approximately 30% 26. We observed comparable results in our study in control animals as well as animals which received only sterile beads. In our model, on the 5^th day, terbutaline intratrachéal administration did not change DAFC. However the lack of effect may be due to airway inflammation and an inability to adequately deliver the drug, we therefore decided to use intraperitoneal administration with the same agent. Our results also show the absence of DAFC increase. We then hypothesized that the absence of response to the stimulation might be related to the persistence of the pathogen in the alveoli. To test this hypothesis, we injected the animals with ceftazidime to sterilize the lungs on the 5^th day. Sterilization was achieved but failed to restore DAFC stimulation with terbutaline. To explain this impairment of DAFC, different hypotheses still remain to be investigated concerning these agonist's receptors and their regulation. Other authors have shown in different situations that either an internalization or a decrease of affinity of the receptors 27 could be observed. Another hypothesis could be a lost of sensitization through a decrease of the AMPc dependent signal transmission. It was shown, in vitro, on tracheal cells that a continuous or repeated exposure to isoproterenol could lead to a lost of sensitization 28.

If this unresponsiveness exists in patients, the absence of an adapted DAFC response in chronic lung infection could lead to major damage in the presence of any new lung injury. Although chronic lung infection has not been isolated, per se, as an aggravating factor associated to mortality in COPD patients admitted in an intensive care unit, a pre-existing underlying pathology is associated with a worsening of the prognosis in community and nosocomial pneumonia 2930. DAFC impairment might be part of the answer to this effect of underlying disease.

In conclusion, chronic P. aeruginosa pneumonia is characterized initially at 48 hours by an increased alveolar-capillary barrier permeability and an adapted host response with an increased DAFC and LLC preserving a normal lung wet to dry weight ratio. On the 5^th day, DAFC remains non responsive to pharmacological stimulation even after bacterial elimination. This impairment of DAFC could represent one of the factors responsible for the increased susceptibility of chronically infected patients to other respiratory insults.

SB and FA were responsible for the acquisition of the data. KF and MOH made substantial contributions to the drafting of the manuscript and the analysis of the data. TP performed the radioactive labelling of the albumin (I¹³¹). EK was involved in the revision of the manuscript and the English editing. XL performed all the histological analysis. BG was involved in the acquisition of the data, the design and the conception of the study as well as the drafting of the article. All the authors read and approved the final manuscript.