Low-tidal-volume Mechanical Ventilation Induces a Toll-like Receptor 4–dependent Inflammatory Response in Healthy Mice

All experiments were approved by the Regional Animal Ethics Committee in Nijmegen, The Netherlands, and performed under the guidelines of the Dutch Council for Animal Care and the National Institutes of Health. The current study protocol was designed after a pilot study was performed.

All experiments were conducted in adult male C57BL6 mice (n = 64): TLR4 KO (C57BL6 background) mice (10–12 weeks) (n = 14; body weight 30 [± 3] g) and age-matched WT mice (n = 40; body weight 25 [± 3] g). In a separate set of experiments, TLR2 KO (C57BL6 background) mice (22–24 weeks) (n = 5; body weight 30 [± 2] g) and age-matched WT mice (n = 5; body weight 33 [± 2] g) were used. All KO mice were extensively backcrossed (at least 10 times) and were a gift from Professor Shizuo Akira, M.D., Ph.D. (Osaka University, Osaka, Japan). WT mice were purchased from Charles River (Sulzfeld, Germany).

Mice were anesthetized with an intraperitoneal injection of a combination of ketamine, medetomidine, and atropine as described previously.24Mice in the unventilated groups were killed immediately after induction of anesthesia. Mechanically ventilated (MiniVent®; Hugo Sachs Elektronik-Harvard Apparatus, March-Hugstetten, Germany) animals were orally intubated, ventilated for 4 h, and killed immediately thereafter. The following settings were used during controlled MV: tidal volume, 8 ml/kg body weight; frequency, 150/min; positive end-expiratory pressure, 4 cm H²O; and fraction of inspired oxygen, 0.4. Rectal temperature was monitored continuously and maintained between 36.0°C and 37.5°C using a heating pad.

The first set of experiments was performed to investigate the presence of endogenous ligands for TLR4 in bronchoalveolar lavage (BAL) fluid. BAL was performed after 4 h of MV (group V-WT, n = 8) or directly after induction of anesthesia (group C-WT, n = 8).

The second set of experiments was designed to investigate changes in pulmonary TLR4 and TLR2 expression in WT mice and to investigate changes in cytokine profile in WT and TLR4 KO mice after MV. Blood and lungs were harvested after 4 h of MV in WT (group V^TLR4WT, n = 8) and TLR4 KO mice (group V^TLR4KO, n = 8) or immediately after induction of anesthesia in WT (group C^TLR4WT, n = 8) and TLR4 KO mice (group C^TLR4KO, n = 6).

In the third set of experiments, the role of TLR2 in MV-induced inflammation was studied. Blood and lungs were harvested after 4 h of MV in WT (group V^TLR2WT, n = 5) and TLR2 KO mice (group V^TLR2KO, n = 5).

Lipopolysaccharide was measured in air, tubing, ventilator, and BAL fluid from unventilated animals (n = 4) by Limulus Amebocyte Lysate testing (Cambrex Bio Science, Walkersville, MD; detection limit, 0.06 U/ml) to rule out contamination with lipopolysaccharide in our experimental setting.

Blood was collected by exsanguination, centrifuged at 14,000 rpm (13,000g ) (Eppendorf 5415 C; Nethler-Hinz GmbH, Hamburg, Germany) for 2 min, and plasma was stored at −80°C. Immediately after exsanguination, heart and lungs were carefully removed en block  via  midline sternotomy. The right upper and lower lobes were snap frozen in liquid nitrogen and stored at −80°C. The right middle lobe was fixed for light microscopy. The left lung was homogenized for measurement of cytokine concentrations.

For the BAL procedure, 250 μl sterile saline (0.9%) at 37°C was injected and gently aspirated through the endotracheal tube. Lavage fluid was snap frozen in liquid nitrogen and stored at −80°C. Human embryonal kidney (HEK) 293–TLR4 cell lines were purchased from InvivoGen (San Diego, CA) and cultured according to the manufacturer's guidelines. For stimulation, 5.10⁴cells/well were used in flat-bottom 96-wells plates. HEK293-TLR4 cells were exposed to 50 μl BAL fluid and added to 200 μl culture medium for 24 h. Subsequently, interleukin (IL) 8 was measured in culture supernatants using Luminex bead array technology. To verify the presence of endogenous TLR4 ligands, the BAL fluid was incubated with or without 10 μg/ml Bartonella quintana  lipopolysaccharide as a specific TLR4 antagonist.25,26To rule out lipopolysaccharide contamination, we performed the HEK293-TLR4 assay in the presence and absence of polymyxin B (2 μg/ml) with BAL fluid from ventilated WT mice (n = 4).

For polymerase chain reaction (PCR) analysis of messenger RNA (mRNA), the right upper and lower lobes were homogenized with a micro-dismembrator II (Braun, Melsungen, Germany). Total RNA was extracted in 1 ml TRIzol reagent. Subsequently, 200 μl chloroform and 500 μl 2-propanol (Merck, Darmstadt, Germany) were used to separate the RNA from DNA and proteins. After a wash step with 75% ethanol (Merck), the dry RNA was dissolved in 30 μl water.

To obtain double-strain complementary DNA (cDNA), DNase-treated total RNA 1 μg with oligo dT primers (0.01 μg/ml) was reverse transcribed in a reverse transcriptase PCR with a total volume of 20 μl. Subsequently, quantitative PCR was performed using an ABI/PRISM 7000 Sequence Detection System (Applied Biosystems, Foster City, CA). PCRs of glyceraldehyde-3-phosphate dehydrogenase, TLR2 and TLR4 were performed with Sybr Green PCR Master Mix (Applied Biosystems), 5 μl 1/20 diluted cDNA, and primers in a final concentration of 300 nm in a total volume of 25 μl. The primers were developed using Primer Express® software (Applied Biosystems). Quantification of the PCR signals of each sample was performed by comparing the cycle threshold values, in duplicate, of the gene of interest with the cycle threshold values of the glyceraldehyde-3-phosphate dehydrogenase housekeeping gene. TLR2 and TLR4 mRNA expression was expressed as relative expression to the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase. All primers were validated according to the protocol, and standard curves were all within the tolerable range.

Tumor necrosis factor (TNF) α, IL-6, IL-10, and keratinocyte-derived chemokine (KC) in the homogenized left lung and in plasma were analyzed by enzyme-linked immunosorbent assay (ELISA) (for TNF-α, IL-6, and IL-10: CytoSet, BioSource, Camarillo, CA; for KC: ELISA Kit, R & D Systems, Minneapolis, MN). IL-1α and IL-1β were only assessed in lung homogenates, because of insufficient plasma, using specific radioimmunoassays, as described previously.27Lower detection limits were as follows: IL-1α and IL-1β, 40 pg/ml; TNF-α, 32 pg/ml; IL-6, 160 pg/ml; IL-10, 16 pg/ml; and KC, 160 pg/ml.

The right middle lung lobe was fixed in 4% buffered formalin solution overnight at room temperature, dehydrated, and embedded in paraplast (Amstelstad, Amsterdam, The Netherlands). Sections of 4-μm thickness were used. The enzyme activity of leukocytes was visualized by enzyme histochemistry using chloracetatesterase staining (Leder staining). Leukocytes were counted manually (20 fields per mouse), and after automated correction for air/tissue ratio, the number of leukocytes/μm²was calculated.

Data are expressed as mean (SD) when distributed normally (body weight, endogenous ligands, relative mRNA expression, leukocyte counts) and expressed as median (range) otherwise (cytokine concentrations). Statistical analysis was performed with SAS (SAS Institute Inc., Cary, NC) statistical procedures. Because cytokine concentrations are not normally distributed, Kruskal-Wallis procedures were used, with post hoc  comparisons of subgroups (Duncan). Data of a particular cytokine concentration variable were ranked, followed by analysis of variance in the General Linear Models procedure using the MEANS procedure with the Duncan option and Bonferroni correction for multiple comparisons. For analysis of endogenous ligands, relative mRNA expression, and leukocyte counts, analysis of variance was used on nonranked data with post hoc  comparison of group means. The level of significance was set at P < 0.05.

Bronchoalveolar lavage fluid from unventilated WT mice resulted in increased production of IL-8 by HEK293-TLR4 cells compared with the medium. This could only be partly reduced by coincubation with a highly specific TLR4 antagonist, indicating that these cells generate IL-8 in a TLR4-independent manner as well (see Discussion, Role for Endogenous Ligands). BAL fluid from ventilated WT mice induced a significantly higher IL-8 response compared with BAL from unventilated WT mice, and coincubation with a highly specific TLR4 antagonist did significantly reduce the IL-8 response (fig. 1). From the experiments performed in the presence of TLR4 antagonist, it can be derived that the TLR4-dependent response was approximately fivefold higher in ventilated WT mice compared with unventilated WT mice.

Mechanical ventilation increased the relative mRNA expression of TLR4 and TLR2 in lung tissue of ventilated WT mice as compared with unventilated WT mice (fig. 2).

Mechanical ventilation in WT mice increased levels of IL-1α, IL-1β, and KC in lung homogenates (fig. 3). However, in TLR4 KO mice, MV did not increase IL-1β in lung homogenates, and the MV-induced increase in KC was less pronounced. Interestingly, TLR4 KO did not reduce the response of IL-1α to MV. Neither MV nor TLR4 KO affected TNF-α, IL-6, or IL-10 concentrations in lung homogenates.

In WT mice, MV was associated with significantly increased levels of IL-6, KC, and TNF-α in plasma (fig. 4). In TLR4 KO mice, MV did not increase plasma levels of IL-6 and TNF-α. MV did result in increased levels of KC, although the response was less pronounced than in WT mice (fig. 4).

Mechanical ventilation in TLR2 KO mice did not result in statistically different cytokine levels in lung tissue homogenates and plasma compared with their age-matched WT ventilated animals (figs. 5 and 6).

In WT mice, MV increased the number of pulmonary leukocytes (table 1). In TLR4 KO mice, MV did not affect the number of leukocytes in the lung. Unventilated TLR4 KO mice expressed a higher number of pulmonary leukocytes when compared with unventilated WT mice.

The number of pulmonary leukocytes in ventilated TLR2 KO mice was not different from those in the age-matched ventilated WT mice (table 1).

No lipopolysaccharide contamination was detected in our experimental setting and BAL fluid from unventilated mice.

This study is the first to reveal several key findings regarding the role of TLR4 signaling in the development of MV-induced inflammation in healthy mice. Within the time frame studied, we found that low-tidal-volume MV resulted in increased expression of endogenous TLR4 ligands in BAL and enhanced mRNA levels for TLR4 in lung homogenates. In addition, short-term MV resulted in the increase of inflammatory cytokines in the lung and in a systemic inflammatory response in plasma. MV-induced inflammation seemed at least partially TLR4 dependent.

The current study confirms previous findings regarding changes in cytokine profile, induced by MV, in lungs and plasma of healthy mice.24Little has been published about the effects of low-tidal-volume MV on cytokines in lungs and plasma of healthy animals. In accord with our results are an increase of IL-1β mRNA expression in lung tissue,28an increase of TNF-α concentration in plasma,29and the lack of response of IL-6 and TNF-α in lung homogenates30,31after low-tidal-volume MV. The finding of an increased response of IL-6 in plasma but not in lung homogenate can be a time-dependent effect, as shown in a previous study.24IL-6 levels peaked in lung homogenates after 2 h of MV and decreased thereafter. However, in plasma, the highest value was reached after 4 h. In contrast to our results, Dhanireddy et al.  30found no increase of KC in lung homogenate and plasma after low-tidal-volume MV, and Takenaka et al.  32did not find an increase of IL-6 or TNF-α in plasma after low-tidal-volume MV. These different results might be partly explained by differences in ventilator settings and experimental setup in these studies.

The current study demonstrates that TLR4 plays a role in development of the inflammatory response after MV in healthy lungs. MV in TLR4 KO did not increase IL-1β in lung homogenates, and increases of KC in lung homogenates were less pronounced compared with WT mice. KC and IL-1β are found to play an early and central role in lung injury induced by hemorrhage.14,33IL-1β also is involved in lung injury induced by liver injury.34Recently, IL-1α rather than IL-1β was identified as a key mediator in sterile inflammation in response to injured cells.35Our results demonstrate that both IL-1α and IL-1β are involved in the inflammatory response after MV, but only the increase in IL-1β was found to be TLR4 dependent. KC is a chemoattractant but also has a direct cytotoxic effect.36Jiang et al.  17found KC to be produced by pulmonary epithelial cells in a TLR-dependent manner; however, this was in direct response to bleomycin. More recently, functional TLR4 expression was found to be critical in the KC increase after hemorrhage.14Our results show that KC production in response to MV in healthy lungs seems to be at least partly TLR4 dependent.

The clinical relevance of the cytokine up-regulation by MV is the resulting proinflammatory state. This makes the host more vulnerable to a possible “second hit” (e.g. , major surgery).37,Vice versa , MV itself can be the second hit where an already comprised host exists (e.g. , MV in the critically ill patient).38–40Recently, this “two-hit” hypothesis was linked to TLR4 reactivity.41Inhibition of TLR4 may be an effective strategy to prevent or reduce MV-induced pulmonary and systemic inflammation.

The activation and attraction of leukocytes is an important feature in VILI.42Leukocytes are thought to be a principal culprit in sterile inflammation causing tissue damage.35KC is a major attractant for leukocytes.43In a previous study,24we have shown that 4 h of MV in healthy mice preserves alveolar integrity but induces a pulmonary leukocyte influx after the increase of KC in the lung. In the current study, this pulmonary leukocyte influx is confirmed; however, in TLR4 KO mice, MV did not affect the number of pulmonary leukocytes. Recently, Frink et al.  14published a hemorrhage model in which the pulmonary leukocyte influx after hemorrhage was also found to be TLR4 dependent.

In the current study, the number of pulmonary leukocytes was significantly higher in unventilated TLR4 KO mice compared with unventilated WT mice. TLR4 KO mice also showed increased lung homogenate levels of KC and IL-1β compared with WT mice at baseline. Frink et al.  14did not find differences in leukocyte content before hemorrhage; however, in that study, TLR4 mutant mice were used, which have a different background (C3H/HeJ). Because in these mice the defect is not complete, partial responses of TLR4 could be preserved. In the current study, TLR4 KO mice were used. The absence of TLR4 increases their sensitivity to infections.44However, before the experiments, these mice did not exhibit clinical signs of discomfort or chronic infection (i.e. , abnormal behavior, weight loss). The higher number of pulmonary leukocytes in unventilated TLR4 KO mice is most likely the result of the increased lung homogenate levels of KC.

Recently, a number of studies suggested that endogenous ligands exists that can bind to TLR4 and induce an inflammatory response.18–23,45By using HEK293-TLR4 reporter cells, which produce IL-8 in the presence of TLR4 agonists, we sought for endogenous TLR4 ligands in BAL fluid sampled from unventilated and ventilated WT mice. The IL-8 response in BAL fluid from ventilated mice was remarkably enhanced compared with unventilated mice, and the addition of a highly specific TLR4 antagonist decreased the induction of IL-8, indicating the presence of endogenous TLR4 ligands in BAL fluid from ventilated mice. HEK293-TLR4 cells incubated with BAL fluid from unventilated mice also induced an IL-8 response. This interesting observation could suggest that endogenous ligands for TLR4 are present in the normal (unventilated) lung and released in the air spaces. This might be explained by direct lung injury, induced by performing the BAL procedure as lung lavage is potentially injurious, especially when large volumes are used.46We used small volumes for our BAL procedure, but nevertheless cannot exclude that this induces some degree of lung injury and subsequent release of endogenous ligands. The fact that the IL-8 response could only partly be prevented by coincubation with a TLR4 antagonist suggests that other non–TLR4-mediated factors play a role. We found that the HEK293-TLR4 cells can also respond to recombinant mouse TNF-αin vitro  (personal communication, L.A.J. and M.G.N., August 2007). Because small concentrations of TNF-α can be present in BAL fluid of healthy mice,47this most likely caused the IL-8 response in the unventilated mice. Most importantly, the IL-8 response was much higher in BAL fluid from ventilated WT mice compared with unventilated WT mice, and coincubation with a TLR4 antagonist prevented this response. This indicates increased concentrations of endogenous TLR4 ligands after MV.

In a noninfectious lung injury model, Jiang et al.  17found hyaluronan, which is released from the extracellular matrix, to act as an endogenous ligand and initiate an inflammatory response in a TLR4-dependent manner. Hillman et al.  48showed that heat shock protein 70, serum amyloid A-3, and TLR2 and TLR4 mRNA were increased in BAL fluid after brief, large-tidal-volume ventilation in fetal sheep. Because the identification of specific endogenous ligands for TLR4 in healthy lungs after MV was beyond the scope of the current study, further investigation is needed.

The current data suggest that TLR2 does not play a role in the MV-induced inflammatory response after short-term MV. Although the relative expression of TLR2 in WT mice was up-regulated, the ventilated TLR2 KO mice did not show differences in cytokine levels compared with age-matched ventilated WT mice after 4 h of MV. This is in accord with preliminary findings in TRIF-deficient mice in which the inflammatory response to MV was clearly suppressed (manuscript in preparation). Because TLR2 signaling is MyD88 dependent and TLR4 is MyD88 or TRIF dependent,49it is likely that the TLR4–TRIF pathway is the dominant pathway for the inflammatory response to MV. However, further studies are needed to establish the role of TLR2 in MV-induced inflammation, because we cannot exclude that TLR2 plays a role in the inflammatory response beyond 4 h of MV.

Factors other than MV, possibly affecting TLR4 and 2 signaling, were carefully avoided. Contamination with lipopolysaccharide is suggested to be a confounding factor in many studies.50We therefore excluded lipopolysaccharide contamination during the experiments. The possibility of triggering an inflammatory response by invasive procedures (i.e. , insertion of an intraarterial line) and subsequent bacterial contamination37was eliminated by performing experiments in noninvasively monitored animals. Previously, cardiorespiratory stability in invasively monitored animals has been documented.24The possible immune-modulating effects of anesthetics have been studied extensively.51Ketamine, for example, is known to have an inhibitory effect on lipopolysaccharide-induced cytokine production,52–55possibly by suppressing TLR4 expression.56Recently, it has been demonstrated that ketamine alone (without lipopolysaccharide) also attenuated cytokine production in humans in the direct postoperative period after elective abdominal surgery.57In the current study, all animals received ketamine. Ideally, an additional control group of spontaneously breathing animals during ketamine, medetomidine, and atropine anesthesia is needed. However, this will result in hypoventilation with severe respiratory acidosis and hemodynamic instability.

Interestingly, a phase III trial is being conducted to investigate the effects of TLR4 inhibition in patients with severe sepsis.‡‡The current study supports a role for TLR4 in the inflammatory reaction after MV in healthy lungs. Increasing the understanding of the innate immune response to MV and the contribution of MV to the “multiple-hit” concept may lead to future treatment advances in VILI, in which TLR4 may serve as a therapeutic target.

The authors thank Francien van de Pol, Ing., and Ilona van den Brink, Ing. (Lab Technicians, Department of Anesthesiology, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands), for their expertise in performing the animal experiments; Ineke Verschueren, Ing. (Lab Technician, Nijmegen Centre for Infectious Diseases, Radboud University Nijmegen Medical Centre), for her help with cytokine assays; and Shahla Abdollahi-Roodsaz (Ph.D. Student, Rheumatology Research and Advanced Therapeutics, Department of Rheumatology, Radboud University Nijmegen Medical Centre) for her help with endogenous ligand assays.