Mild-stretch mechanical ventilation upregulates toll-like receptor 2 and sensitizes the lung to bacterial lipopeptide

Introduction Mechanical ventilation (MV) could prime the lung toward an inflammatory response if exposed to another insult such as bacterial invasion. The underlying mechanisms are not so far clear. Toll-like receptors (TLRs) allow the host to recognize selectively bacterial pathogens and in turn to trigger an immune response. We therefore hypothesized that MV modulates TLR2 expression and in turn modifies responsiveness to agonists such as bacterial lipopeptide (BLP). Method Both in vitro and in vivo experiments were conducted. First, TLR2 expression and protein were measured in the A549 pulmonary epithelial cell line submitted to 8-hour cyclic stretch (20% elongation; 20/minute rate). After a 24-hour period of cyclic stretch, the inflammatory response of the A549 cells to the synthetic BLP, Pam3CSK4, was tested after 8 hours of exposure. In a second set of experiments, healthy anesthetized and paralyzed rabbits were submitted to 8-hour MV (tidal volume = 12 ml/kg, zero end-expiratory pressure; FIO2 = 50%; respiratory rate = 20/minute) before being sacrificed for TLR2 lung expression assessment. The lung inflammatory response to BLP was then tested in animals submitted to 24-hour MV before being sacrificed 8 hours after the tracheal instillation of Pam3CSK4. Results Cyclic stretch of human pulmonary epithelial cell lines increased both TLR2 mRNA and protein expression. Cells submitted to cyclic stretch also increased IL-6 and IL-8 secretion in response to Pam3CSK4, a classical TLR2 ligand. A mild-stretch MV protocol induced a 60-fold increase of TLR2 mRNA expression in lung tissue when compared with spontaneously breathing controls. Moreover, the combination of MV and airway exposure to Pam3CSK4 acted synergistically in causing lung inflammation and injury. Conclusions Mild-stretch MV increases lung expression of TLR2 and sensitizes the lung to bacterial TLR2 ligands. This may account for the propensity of mechanically ventilated patients to develop acute lung injury in the context of airway bacterial colonization/infection.


Introduction
Bacterial superinfection of the lung is a frequent complication in critically ill patients requiring mechanical ventilation (MV) [1]. An incidence of ventilator-associated pneumonia of up to 30% has been reported in patients requiring prolonged MV. Among the bacteria causing ventilator-associated pneumonia, Staphylococcus aureus is a leading source of infection and mortality [2]. Both in vitro and in vivo experimental studies have demonstrated that MV could activate lung cells and induce a proinflammatory response. Although low-stretch MV does not lead to apparent tissue damage or inflammation, it could prime airway cells to respond massively to a second proinflammatory insult such as lipopolysaccharide (LPS) [3][4][5][6]. The subsequent release of large amounts of cytokines is probably responsible for substantial lung injury, particularly through the recruitment of neutrophils mediated by IL-8 secretion [7].
Little is known, however, about the role of Gram-positive bacterial products in this context. In several animal studies, MV has been shown to impair lung bacterial clearance, to increase lung injury and to promote pulmonary-to-systemic microbial translocation [8][9][10][11]. However, the link between these different observations remains to be established.
A prompt immune lung response is necessary to ensure microbial clearance. Early pathogen recognition via the so-called pathogen recognition receptors is considered a first and key step in this process. Among the pathogen recognition receptors, toll-like receptors (TLRs) play an indisputable and major role in the recognition of pathogen-associated microbial patterns. The ligation of TLRs by pathogen-associated molecular patterns induces the signaling of the subsequent inflammatory response. Bacterial lipopeptides (BLPs) are cell wall components of both Gram-positive and Gram-negative bacteria and have been recognized to activate TLR2 in a heterodimeric association with TLR1 or TLR6 [12,13].
TLR2 is mainly expressed by myeloid cells, but epithelial cells from various tissues can also express the receptor [14]. In the airway, both bronchial and alveolar epithelial cells participate in the lung immune response through their ability to produce inflammatory mediators and chemokines [15]. Whereas TLR4 is essential for cell responses to Gram-negative LPS, TLR2 is implicated in innate immune responses to Gram-positive bacteria [12,13]. For example, Tlr2 -/mice are highly susceptible to infections caused by Gram-positive bacteria such as S. aureus [16]. TLR2 expression also seems sensitive to mechanical stress, at least in the endothelium, where it has been shown to be upregulated by various types of mechanical strains [17]. Whereas the concept of lung priming was relatively well described for LPS models depending on the TLR4 pathway [3][4][5][6]18], little is known on the possible implication of other TLRs and related bacterial ligands in MV-dependent lung injury.
To test the hypothesis that MV modulates TLR2 expression, we performed a set of in vitro and in vivo experiments to determine whether cyclic stretch of lung cells and MV modified the expression of TLR2 and to unravel the effect of MV in lung inflammatory response and injury after exposure to a classical TLR2 bacterial ligand.

Cells and reagents
The human epithelial type II-like A549 cell line was purchased from the American Type Culture Collection (Rockville, MD, USA) and cultured in a 50/50 mixture of DMEM and F-12 nutrient mixture supplemented with 10% FCS, 2 mM L-glutamine, 10 mM HEPES, 50 U/ml penicillin, and 2 μg/ml gentamicin. The NF-B inhibitor Sn50 and the p38 mitogen-activated protein kinase (MAPK) specific inhibitor SB203580 were purchased from Sigma (St Louis, MO, USA). The TLR2specific agonist Pam 3 CSK 4 lipopeptide was purchased from InvivoGen (San Diego, CA, USA).
Cell culture and cyclic stretch A549 epithelial cell lines were seeded onto Bioflex ® plates (Dunn Labortechnik, Asbach, Germany) at a density of 2 × 10 5 cells/well, and were cultured for 48 hours until 70 to 80% confluency was achieved. Mechanical stretch was then performed for different periods of time in fresh medium using the cell stretching apparatus FX-3000 Flexercell strain unit (Flexcell International, Hillsborough, N.C., U.S.A.) in a 37°C, 5% carbon dioxide incubator. The stretching rate was 20 cycles/minute with a square signal, a 1:1 stretch:relaxation ratio, and a 20% maximal equibiaxial elongation, as previously described [7]. In some experiments, Pam 3 CSK 4 was added to the medium (final concentration 1 μg/ml) for an additional 8 hours. At the end of the experiments, conditioned supernatants were immediately collected and kept frozen at -80°C. IL-6 and IL-8 were quantified by ELISA (Endogen, Woburn, MA, USA) in thawed conditioned supernatants. Limits of detection were 45 pg/ml for IL-6 and 65 pg/ml for IL-8.
In rabbits, after sacrifice and exsanguination, the rib cage was opened and lung pieces were cut and RNA was then extracted using the RNA GenElute kit (Sigma). Quantitative PCR was performed using IQ Sybrgreen Supermix (Biorad). Melting curves were performed to ensure the presence of a single amplicon. The following primers were used: rTlr2, The results are expressed as the fold induction using the ΔCt method since the static condition (in vitro) or the spontaneously breathing (SB) animals (in vivo) were always considered the baseline condition (that is, the '1' reference value)

Flow cytometry
A549 cells stretched or kept in static conditions for 24 hours were removed from culture plates using pH 7.4 PBS containing 0.1% ethylenediamine tetraacetic acid and were used immediately for fluorescence-activated cell sorting (FACS). For each analysis, 10 6 cells were washed three times with 1 ml PBS containing 0.4% BSA (Sigma) at 4°C. Cells were incubated in PBS/BSA containing 5 μg/ml mouse anti-huTLR2 mAb (T2.5 clone) antibody or an isotype control antibody (mouse IgG1; Biosource, Camarillo, CA, USA), at 4°C for 30 minutes. Cells were then washed three times with PBS and resuspended PBS containing 10 μg/ml goat anti-mouse IgG phycoerythrin-labeled antibody, at 4°C in the dark for 30 minutes. For intracellular staining, cells were pre-fixed with 1% formaldehyde and permeabilized with IC Perm Buffer (Biosource) supplemented with 4% BSA. Staining was then performed as described above. FACS was carried out using FacScan and analyzed with the CellQuest software (Becton Dickinson, San Jose, CA, USA). Results were expressed as the mean fluorescence intensity ratio, defined as the mean fluorescence intensity of cells stained with an antigen-specific antibody divided by the mean fluorescence intensity of cells incubated with the isotype control antibody.

Animal experiments
Male New Zealand White rabbits (body weight 2.7 to 3.0 kg) were obtained from the Elevage scientifique des Dombes (Romans, France). They were placed in individual cages and were fed ad libitum with water and nutriments, according to current recommendations mentioned in the Guide for the Care and Use of Laboratory Animals (National Institutes of Health No. 92-23, revised 1985). The Dijon Faculty of Medicine Ethical Committee approved the experimental protocol.
In a first set of experiments, animals were randomly allocated to the SB group (n = 4) or the MV group (n = 4) ( Figure 1). Animals from the former group were immediately sacrificed by a lethal dose of penthotal, whereas the latter were orally intubated as previously described [8,9]. The animal was then connected to a pressure-controlled ventilator (Babylog ® ; Drager, Lübeck, Germany). MV was performed in the supine position with continuous infusion of ketamine and pancuronium bromide. The peak inspiratory pressure was set at 12 cmH 2 O with zero end-expiratory pressure, a respiratory rate of 30/minute and an inspired fraction of oxygen of 0.5. The tidal volume was measured at the onset of MV by placing a pneumotachometer between the endotracheal tube and the ventilator circuit (12.6 ± 0.8 ml/kg). This corresponds to mild-stretch MV. Animals from the MV group were submitted to MV during an 8-hour period before being sacrificed. In both groups, animals were exsanguinated by venous puncture and lungs were harvested for RNA extraction. Additional samples were obtained for microscopic examination. A tissue sample of 1 cm 3 was fixed in formalin and embedded in paraffin. Four-micrometer sections were obtained and colored with hematoxylin and eosin.
In a second set of experiments, rabbits were orally intubated as described above and subjected to MV for a 24-hour period before being given intrabronchially either 500 μg Pam 3 CSK 4 diluted in 2 ml isotonic saline or vehicle alone, through a silicon catheter introduced into the tracheal tube and pushed until one lower lobe bronchus was reached. The animals were then kept under MV during an 8-hour period before being sacrificed. An arterial catheter was inserted in most of these animals for blood sampling and blood pressure monitoring. Controls were SB animals that received 500 μg Pam 3 CSK 4 and were allowed to go back to their cage before being sacrificed 8 hours later.

Statistical analysis
Data are presented as the mean (standard deviation). The Mann-Whitney U test or the Kruskal-Wallis test was used to compare continuous variables whether the number of groups was two or more. P ≤0.05 was considered significant.

Results
In vitro studies Cyclic stretch induces the expression of Toll-like receptors in lung epithelial cells A sixfold increase in epithelial cell content in TLR2 mRNA was detected by quantitative PCR in stretched A549 cells when compared with cells kept in static condition ( Figure 2a). TLR4 mRNA was increased to the same extent in stretched cells but MD-2 and CD14 mRNA levels remained unchanged.

Cyclic stretch increases intracellular TLR2 in lung epithelial cells
Cells were harvested after a 24-hour cyclic stretch for FACS analysis. No TLR2 was detected in A549 cells in a static condition in nonpermeabilized cells as well as in permeabilized cells. After 24 hours of cell stretching, TLR2 became detectable in permeabilized epithelial cells and to a lesser extent in nonpermeabilized cells (Figure  2b, c). TLR4 and MD-2 remained undetectable in intact or permeabilized cells regardless of the mechanical stress. Finally, cyclic stretch did not influence CD14 expression. The promonocytic human THP1 cells expressing TLR2, TLR4, CD14 and MD-2 were used as positive controls for the mAbs used in the experiments (data not shown).
Cyclic stretch increases the inflammatory response of lung epithelial cells to bacterial lipopeptide A549 cells submitted to a 24-hour stretch released higher levels of IL-6 and IL-8 in response to an additional 8-hour stimulation by the synthetic BLP Pam 3 CSK 4 than did cells not submitted to cyclic stretch (Figure 3a). A synergistic effect between cyclic stretch and Pam 3 CSK 4 was also observed at the IL-6 and IL-8 mRNA level (Figure 3b).
In vivo studies Mechanical ventilation in healthy rabbits causes a marked increase of lung TLR2 mRNA MV induced a marked increase in lung TLR2 mRNA, much greater than those observed for TLR4, MD-2, and CD14 ( Figure 5).
In these ventilated rabbits, no macroscopic lung abnormalities were noted. Microscopic studies, however, showed features of mild lung injury, including septal thickening and inflammatory cell infiltration together with a small increase in IL-8 mRNA (Figures 6 and 7) features absent in lungs from SB animals.

Mechanical ventilation increases lung inflammation induced by bacterial lipopeptide
Despite MV and lung insult by Pam 3 CSK 4 , hemodynamic stability and adequate blood oxygenation were achieved in our model (Table 1)   The endobronchial instillation of the BLP Pam 3 CSK 4 induced more lung injury in rabbits previously ventilated for 24 hours compared with those who were SB during that period of time ( Figure 6). These microscopic abnormalities included more inflammatory cell infiltration and more pronounced alveolar septa thickening.
Although Pam 3 CSK 4 alone did not induce much IL-8 mRNA, we observed a clear synergistic effect between MV and BLP on lung IL-8 gene expression (Figure 7).

Discussion
Herein, we show that cyclic stretch of lung cells and MV in rabbits upregulate TLR2. MV-dependent TLR2 upregulation was associated with a marked increase in lung cell activation and lung injury after a challenge with Pam 3 CSK 4 , a classical TLR2 agonist. These results are consistent with cyclic stretch increasing lung cell reactivity to TLR2 bacterial agonists. Prestretching type II-like human alveolar A549 cells induced a marked increase in the induction of proinflammatory mediators such as IL-6 and IL-8 in response to BLP; a synergistic effect was observed between MV and TLR2 activation by Pam 3 CSK 4 . We cannot, however, exclude that the increased expression of TLR2 by A549 cells is solely and directly caused by cyclic stretch. Actually, the release of inflammatory mediators by stretched cells might account at least in part for the observation reported herein [19].
We failed to detect TLR2 at the surface of epithelial cells by FACS. Depending on cell types, TLRs can be expressed at the surface or intracellularly. TLR4, for example, is expressed at the surface of myeloid cells but in intracellular compartments in epithelial cells such as A549, BEAS-2B cells, intestinal cells as well as endothelial cells [20,21]. Similarly, TLR2 -which is also expressed at the plasma membrane in myeloid cells -is found intracellularly in corneal epithelial cells [22]. Using a permeabilization protocol, we showed that cyclic stretch induced intracellular TLR2 upregulation after cyclic stretch in A549 cells. TLR2 mRNA was also increased by cyclic stretch in A549 cells. Cyclic stretch has been shown to induce proinflammatory gene transcription in lung epithelial cells via either NF-B or MAPK pathways [7]. We therefore tested the effect of pharmacological inhibitors of the NF-B and the p38 MAPK signalization pathways on cyclic stretchdependent induction of TLR2. The p38 MAPK inhibitor (SB203580) abrogated cyclic stretch-induced TLR2 gene expression, whereas NF-B had no effect. These findings are consistent with the fact that the MAPK pathway is critical in the cyclic stretch-induced cell activation. These data also indicate that TLR2 regulation is not entirely dependent on the NF-B pathway, depending on the stimulus [23]. The involvement of the p38 MAPK pathway in TLR2 regulation by cyclic stretch is suggestive of TLR2 mRNA stabilization (post-transcriptional effect) rather than de novo TLR2 transcription, but this remains to be ascertained using TLR2 transcription assays.
We next tested whether these in vitro findings could be reproduced in an animal model. We showed that mild-stretch MV induced a marked upregulation of lung TLR2. Furthermore, the treatment with BLP of rabbits ventilated for 24 hours induced a marked IL-8 expression into the airway and significant lung injury, as compared with SB rabbits. This is the first report showing the MV-induced upregulation of TLR2, and a synergistic effect between MV and BLP. These results are reminiscent of those found with the LPS/CD14/TLR4 pathway [3][4][5][6]24], and highlight the increased reactivity of lung cells to Gram-positive bacterial products and bacteria induced by MV via the TLR2 pathway [25,26]. Moreover, in contrast with previous findings, we observed that TLR2 and TLR4 could be differentially upregulated by cyclic stretch. Accordingly, the TLR2 upregulation was strikingly more pronounced, especially in the in vivo model. Interestingly, similar findings were obtained in vitro in human endothelial cells submitted to either a laminar or a turbulent flow [17]. Such data suggest that differences exist between TLRs with respect to their expression regulation by abnormal mechanical strains.
Our work adds to a growing body of literature implicating pattern-recognition receptors of the innate immunity in the development of ventilator-induced lung injury. A large tidal volume MV increased TLR4 and TLR2 expression in the lungs of preterm lambs, but the effect of the corresponding agonist was not tested [27]. Others have demonstrated that lung hyperinflation increased TLR4 expression in rats with abdominal sepsis [25]. In contrast, TLR4 knockout mice were protected against ventilator-induced lung injury due to lung hyperinflation [28]. Interestingly, it was shown that even protective MV applied to healthy mouse lungs could modify TLR expression [29].  Physiological parameters recorded in rabbits after a 24-hour mild-stretch mechanical ventilation (T1) and 8 hours following the tracheal instillation of bacterial lipopeptide (T2). HR, heart rate; MAP, mean arterial pressure.
Any clinical translation should be made cautiously, however, since the chosen ventilator settings were not in accordance with the current guidelines. The aim was to cause overdistension within healthy lungs, a goal difficult to achieve with the recommended settings (that is, 6 to 8 ml/ kg and positive end-expiratory pressure). Such lung-protective settings, however, could lead to regional overdistension when applied to injured lungs [30,31]. Another limitation is that we have not studied the relationship between lung stretch magnitude and TLR2 expression since only one degree of cell elongation as well as one tidal volume value were tested. We cannot therefore conclude definitively that TLR2 expression is stretch dependent.

Conclusions
Cyclic stretch mimicking MV in vitro induced TLR2 expression and enhanced TLR2 reactivity to BLP of epithelial human cells, most probably through a p38 MAPK-dependent effect. These findings were in part reproduced in a rabbit model, where mild-stretch MV increased lung injury in response to the instillation of a TLR2-dependent BLP. Translated into the clinical arena, our findings suggest that the increase of the TLR2 expression of the lung epithelia caused by MV may prime the lung, which in turn will overreact to the presence of colonizing or infecting Gram-positive bacteria within the airway and worsen lung injury. The potential impact on lung bacterial clearance warrants further studies.

Key messages
• Mechanical ventilation magnifies lung inflammatory response to a Gram-positive bacterial product, and in turn worsens lung injury.
• The stretch-dependent TLR2 upregulation could account for such findings.
• The p38 MAPK pathway could play a critical role.