Published ahead of print on May 8, 2003, doi:10.1165/rcmb.2002-0167OC
© 2004 American Thoracic Society DOI: 10.1165/rcmb.2002-0167OC Low Molecular Weight Hyaluronan from Stretched Lung Enhances Interleukin-8 ExpressionPulmonary and Critical Care Unit, Department of Medicine, Massachusetts General Hospital and Harvard Medical School, Boston; Division of Pulmonary and Critical Care, Tufts University School of Medicine, New England Medical Center, Boston, Massachusetts; and Division of Critical Care Medicine, Medellín General Hospital and Pontifical Bolivarian University, Medellín, Colombia Address correspondence to: Marcella M. Mascarenhas, Ph.D., Pulmonary and Critical Care Unit, Massachusetts General Hospital, 55 Fruit Street, Bulfinch-148, Boston, MA 02114. E-mail: mmascarenhas{at}partners.org
Mechanical ventilation has been shown to cause ventilator-induced lung injury (VILI), probably by overdistending or stretching the lung. Hyaluronan (HA), a component of the extracellular matrix, in low molecular weight (LMW) forms has been shown to induce cytokine production. LMW HA is produced by hyaluronan synthase 3 (HAS 3). We found that HAS 3 mRNA expression was upregulated and that LMW HA accumulated in an animal model of VILI. We hypothesized that stretch-induced LMW HA production that causes cytokine release in VILI was dependent on HAS 3 mRNA expression. We explored this hypothesis with in vitro lung cell stretch. Cell stretch induced HAS 3 mRNA expression and LMW HA in fibroblasts. Nonspecific inhibitors of HAS 3 (cyclohexamide and dexamethasone), a nonspecific inhibitor of protein tyrosine kinases (genistein), and a janus kinase 2 inhibitor (AG490) blocked stretch-induced HAS 3 expression and synthesis of LMW HA. Stretch-induced LMW HA from fibroblasts caused a significant dose-dependent increase in interleukin-8 production both in static and stretched epithelial cells. These results indicated that de novo synthesis of LMW HA was induced in lung fibroblasts by stretch via tyrosine kinase signaling pathways, and may play a role in augmenting induction of proinflammatory cytokines in VILI.
Abbreviations: acute respiratory distress syndrome, ARDS • chondroitin sulfate, CS • extracellular matrix, ECM • extracellular signal–regulated kinase, ERK • hyaluronan, HA • hyaluronan synthase 3, HAS 3 • Hanks' buffered saline solution, HBSS • high molecular weight, HMW • interleukin, IL • Janus kinase, JAK • low molecular weight, LMW • lipopolysaccharide, LPS • macrophage inflammatory protein, MIP • polymerase chain reaction, PCR • proteoglycans, PGs • reverse transcription, RT • ventilator-induced lung injury, VILI
Positive pressure mechanical ventilation is life-saving in patients with acute respiratory distress syndrome (ARDS), but it can lead to ventilator-induced lung injury (VILI). Because lung damage in ARDS is not homogeneous, some areas are less involved than others. These less involved areas are more compliant and thus can be overdistended by mechanical breaths of even moderate volume.
In rat models of VILI that employed whole lung stretch to mimic the regional alveolar overdistention seen in human patients with ARDS, large tidal volumes were shown to produce noncardiogenic pulmonary edema, accumulation of neutrophils, and elevated levels of chemoattractant cytokines such as macrophage inflammatory protein-2 (MIP-2) (1). Fibroproliferative changes were also observed in VILI models, and these may contribute further to changes in lung function (2–8). It has been demonstrated by some but not all investigators that in ex vivo experiments with rat and mouse lungs, positive pressure ventilation at very large tidal volumes of 40 ml/kg produced an outpouring of inflammatory cytokines and chemokines, including MIP-2, tumor necrosis factor-
The mechanism of lung inflammation in VILI is not well understood. Changes in the extracellular matrix (ECM) may be involved. Pulmonary fibroblasts, which are located in the lung interstitial space of capillary wall throughout the lung parenchyma and within the large vessels and airways, are uniquely situated to sense changes in mechanical force. These interstitial fibroblasts undergo significant cyclic stretch during mechanical ventilation, which may play a key role in contributing to ECM-induced inflammatory changes in VILI. Hyaluronan (HA), which is composed of disaccharide repeats, with alternating glucuronic acid and N-acetyl glucosamine, is an important component of the ECM. In its native state, HA exists as a high molecular weight (HMW) form, and serves as a structural scaffolding in tissue. HA also accumulates (11) in low molecular weight (LMW) forms (200 kD) at sites of inflammation, where it is implicated in multiple proinflammatory activities: monocyte activation, leukocyte adhesion to endothelium, smooth muscle cell migration after wound healing, and induction of chemokines such as macrophage chemoattractant protein-1 Cyclic stretch of cells in culture has been shown to activate a number of signal transduction pathways, including protein tyrosine kinases (PTK) (16), janus kinase (JAK) (17), and mitogen-activated protein kinases, extracellular signal–regulated kinase 1/2 (ERK1/2) (16). Other reports have implicated these pathways in remodeling of the ECM in response to cytokines and growth factors (18). These pathways have not previously been explored in stretch-induced remodeling of the ECM via HAS. In the present study, we hypothesized that stretch-induced LMW HA production that causes cytokine release in VILI is dependent on increased HAS 3 mRNA expression. To investigate this hypothesis, intact rats were ventilated for 2 h with a modestly increased tidal volume (VT) of 20 cc/kg at 85 breaths/min, followed by measurements of both HA quantity and HAS 3 mRNA expression in the lung. The mechanisms of stretch-induced HA production were explored using an in vitro model of stretch using primary fetal and adult lung fibroblasts. We find that LMW forms of HA are produced in both in vivo and in vitro models of stretch, and in both cases, the LMW HA is accompanied by increased levels of HAS 3 mRNA. We also find that in vitro stretch of fetal and adult lung fibroblasts upregulates expression of activated JAK 2 (p-JAK 2). Furthermore, we find that stretch-induced accumulation of LMW HA and upregulation of HAS 3 mRNA in vitro can be inhibited by cyclohexamide or dexamethasone, nonspecific inhibitors of HAS 3 mRNA. We also show that stretch-induced HAS 3 mRNA and accumulation of HA can be inhibited by the nonspecific PTK inhibitor genistein, and by the JAK 2 inhibitor AG490. Finally, we show that LMW HA produced by cyclic stretch of fibroblasts induces IL-8 production in static cells, and augments stretch-induced cytokine production in vitro.
Experimental Animals and Ventilator Protocol Sprague-Dawley virus-free rats weighing between 200 and 300 g were obtained from Charles River Laboratories (Wilmington, MA). Under general anesthesia with intraperitoneal ketamine (50 mg/kg; Bedford Laboratories, Bedford, OH) and diazepam (5 mg/kg; Elkins-Sinn, Cherry Hill, NJ), the animals were orally intubated with a 2.42-mm OD (1.67 ID) polyethylene catheter. They were then attached to a small animal ventilator model 683 (Harvard Apparatus Inc, Holliston, MA) set to deliver 20 cc/kg or 7 cc/kg VT at a rate of 85 breaths/min for 2 h on room air. The animals remained intact and with closed chests (19). A VT of 20 cc/kg was used to mimic the overdistention of normal lung thought to occur in uninvolved lung during mechanical ventilation in human ARDS. The tidal volume delivered by the ventilator was checked by fluid displacement from an inverted calibration cylinder. Intubation tubing was increased in length at VT 20 cc/kg to provide adequate dead space to maintain PCO2 at 30–40 mm Hg. All animals were ventilated without positive end-expiratory pressure (PEEP). After 2 h of ventilation, the animals were killed, and their lungs removed en bloc, flash frozen in liquid nitrogen, and stored at –70°C. Control nonventilated animals were anesthetized as above, killed, and their lungs harvested as done for ventilated animals.
Immunohistochemistry HA expression was quantified using microscopic image analysis. After staining slides for HA, the fields were chosen randomly from all sections and captured using a digital camera. Then intensity and area of HA in different groups was measured using IP imaging analysis lab software (Scanalytics, Inc., Fairfax, VA).
Maintenance of Cell Cultures and Stretching Fibroblasts were seeded at 2 x 106 cells/dish onto sterile silicone membranes mounted on a polypropylene dish for use in mechanical strain device. On the day of stretch, plates were washed with Hanks' buffered salt solution (HBSS) and serum-free medium was added. The plates were placed on a mechanical deforming device. This stretching device was custom built and provided by Martha Gray, Ph.D. (Massachusetts Institute of Technology, Cambridge, MA) and has been previously been described in detail (20, 21)· This device provides sinusoidal, spatial homogenous and isotropic biaxial strain to cultured cells. Cells were subjected to stretch at 60 cycles/min with 15% strain for 0.5, 1.0, 2.0, or 4.0 h. These parameters were selected to approximate the conditions used in the ventilated rats. More intensive stretch was not possible with this apparatus. Parallel dishes of control cells were seeded identically and maintained for the same intervals without mechanical deformation. A549 cells were also stretched as described above, for measurement of IL-8 production. The cell density was 3 x 106 cells/plate. Cells were stretched for 4 h at 20 cycles/min followed by a 2-h static culture to allow time for protein production. We have used this stretch protocol in A549 cells extensively and have established it as a reproducible model of stretch-induced IL-8 production (10). LMW HA and HMW HA were added to the A549 cells with and without stretch.
Extraction of Proteoglycans For DEAE anion exchange chromatography, the dialysate as applied directly to a 15 cm x 2 cm column packed with Whatman DE-52 anion exchange resin and equilibrated with Buffer A. Fractions were eluted in sequence with 250 ml Buffer A, followed by 250 ml of a linear gradient of 0–0.2 M NaCl in Buffer A. HA eluted first from the column, due to its lower charge compared with the remaining PGs, which were highly sulfated, carry a higher charge, and therefore bind more firmly to the column. Therefore, the remaining PGs elute at a higher salt concentration. To quantify the PG content in column fractions, uronic acid was determined by the method of Bitter and Muir (22). Pooled uronic acid–containing peaks were dialyzed and lyophilized at -70°C at 10 mm Hg to complete dryness. The column conditions used with all extracts were identical. Uronic acid measurements were repeated after lyophilization and redissolution. For the measurement of PGs from the cultured cells, culture supernatants from stretched cells and control cells were dialyzed at 4°C. The PGs were then extracted using 4 M guanidinium HCl containing protease inhibitors as described above for PG extraction and quantification from rat lung tissue.
Identification of PGs by Cellulose Acetate Electrophoresis The PG identified as CS on cellulose acetate plates was confirmed as CS by Condroitin ABCase (0.1 U/100 µg in Tris acetate buffer pH 8.0) for 30 min, followed by inactivation at 100°C. Condroitin ABCase enzyme have been shown to be specific for the breakdown of CS. The Condroitin ABCase treated fraction was dialyzed against distilled water, freeze-dried, and analyzed on cellulose acetate plates as done for the HAdase-treated sample.
Size Fractionation of Rat Lung HA on 0.5% Agarose Gel Electrophoresis
Size Fractionation and Purification of HA from Fibroblasts by Sepharose CL-4B
RT-PCR for HAS RNA was reverse transcribed (RT) using Gene Amp polymerase chain reaction (PCR) core kit (Perkin Elmer) following the manufacturer's instructions. The 20 µl RT mixture contained 1 µg total RNA, 1 mM dNTP, 2.5 µM random hexamers, 5 mM Mg Cl2, and 2.5 U/µl reverse transcriptase (RTase). The RT was performed at 42°C for 45 min, after which enzyme was inactivated by a 5-min incubation at 99°C. Murine HAS sequences, which are 90–99% homologous with rat, were used for this analysis. The primers had the following sequences: 5'-AGTATACCTCGCGCTCCAGA-3' up, 5'-AGCAGCAGTAGAGCCCAGAG-3' down; AACAGGGTGTTGAGTCTGGG-3' up, 5'-TAAACCACACGGACACTGGA-3' down; and 5'-CGGGTGAAGGAGAGACAGAG-3' up, 5'-GCAATGAGGAAGAATGGGAA-3' down. These resulted in reaction products of 481 bp for HAS 1, 502 bp for HAS 2, and 712 bp for HAS 3. PCR for GAPDH was performed as a control, using the following primers: 5'-AATGCATCCTGCACCACCAA-3' up and 5'-GTAGCCATATTCATTGTCATA-3' down, which gives a 516-bp product. The human primers used for studies in cultured fibroblasts were: 5'-TCATGGTGGTGGATGGCAACCGC-3' and 3'-CTAAGCCACCTGATGTACGTCCA-5', which gave rise to a 283-bp reaction product with HAS 3 and 324-bp product with HAS 1. For HAS 2, primers used were 5'-ATTGTTGGCTACCAGTTTATCCAAACGG-3' and 3'-GGCCACTCTGTCTACTCAGTGTATTTCTTT-5', which gives 409 bp reaction product. PCR for tubulin-ß was performed as a control using: 5'-CTCCATCCTTCACCACCCACA-3' and 5'-CAGGGTCACATTTCACACCATCT-3', which gives a 365-bp product. The 25-µl PCR reaction contained 0.14 µM each primer (Sigma Chemical Co.), 2.5 U/µl AmpliTaq DNA polymerase, 2 mM MgCl2, and 0.2 µM dNTP. The following cycling parameters were used: denaturation at 94°C for 2 min, followed by 35 cycles of 95°C for 1 min, annealing at 58°C for 2 min, and extension at 72°C for 2 min, with a terminal extension at 72°C for 7 min.
Inhibition of HA Synthase
Immunoblot Analysis for JAK-2
Stimulation of A549 Cells with HA On the day of stimulation, cultures of A549/ BEAS-2B bronchiolar epithelial cells were washed with HBSS and then treated with LMW HA or HMW HA for 6 h at 37°C, at concentrations of 10, 50, 100, 150, and 200 µg/ml media. Contamination of HA samples by lipopolysaccharide (LPS), which could result in artifactual production of IL-8, was checked in two ways. First, 10 µg/ml polymyxin B was included in all cultures. Second, LPS was quantified in cultures by a Limulus endotoxin assay (Sigma Chemical Co.). Less than 0.03 endotoxin units (EU)/ml (a negative) was detected in any sample: HMW HA, LMW HA, or media with/without Polymyxin B. For HA stimulation of cells in stretch culture, on the day of stimulation, cultures were washed with HBSS and then treated for 6 h with LMW HA and HMW HA at a dose of 100 µg/ml media and 10 µg/ml polymixin B to inhibit the production of IL-8 by contaminating LPS in sample. LPS was monitored by the Limulus endotoxin assay (Sigma Chemical Co.). Less than 0.03 EU/ml (negative test) was detected in HMW HA, LMW HA, and media with/without Polymixin B. In static cells, four controls were also run along with the treated cultures: (i) control with no added HA; (ii) HAdase-treated HA (Streptomyces hyaluronidase [0.05 U/2-µg sample] at 60°C for 72 h followed by inactivation at 65°C for 10 min, which destroys all the HA in the sample); (iii) heat-treated HA (100°C for 10 min, which denatures any possible protein contamination in the fraction); and (iv) DNase treated HA (10 U DNase I for 1 h at 37°C followed by inactivation at 65°C for 10 min, which removes trace amounts of DNA contamination in the sample).
Analysis of IL-8 Protein
Statistical Analysis
Large Tidal Volume Ventilation Increases LMW HA Production and HAS3 mRNA Synthesis To determine the effect of lung stretch on HA synthesis and HAS mRNA expression, rats were mechanically ventilated for 2 h with VT 20 cc/kg. Total HA was in lung tissue sections from ventilated rats was localized with bHABP and compared with lung tissue from control, nonventilated animals. With high VT ventilation, HA was deposited in the interstitium of the lung parenchyma and in airway epithelium as compared with the control lung tissue (Figure 1). Because bHABP binds both HMW and LMW HA, bHABP staining recognizes total HA accumulation. HAdase pretreatment of the lung section abolished HA staining, confirming the specificity of HA staining (Figure 1). Extent of HA staining was measured by microscopic image analysis. The HA staining in animals ventilated at VT 20 cc/kg was 336,602 arbritary units (AU) ± 55,575 compared with control nonventilated animals, which showed 142,997 AU ± 31,076, and to HAdase treated sections (negative control, no HA staining) P < 0.02. To identify the forms of HA present in ventilated rat lungs, we performed column chromatography to separate HA from the sulfated proteoglycans and extraneous proteins in the lung tissue sample. The pattern of 25 fractions that eluted with a salt gradient of 0–0.2 M NaCl containing HA (total of 100 fractions were collected) is shown in Figure 2. HA from rats ventilated at VT 20 cc/kg (Figure 2A) was eluted from the column at a wider range of conductivity of 9–15 and in multiple peaks compared with the HA from rats ventilated at 7 cc/kg (Figure 2B) and control nonventilated animals (Figure 2C) that was eluted at a narrow range of conductivity of 12–13 and in a single peak. These data showed the greater heterogeneous nature of the HA isolated from rats ventilated at VT 20 cc/kg. The amount of HA in the pooled column fractions that were freeze-dried from the rat lung was determined (Figure 3A). There was a significant increase in the amount of total HA in the rats ventilated at VT 20 cc/kg as compared with rats ventilated at 7 cc/kg and nonventilated animals.
Using cellulose acetate electrophoresis, we confirmed the identity of the major GAG molecule within the column fraction as HA by comparing its mobility to that of standard HA and HAdase treatment of the fraction. Because the HAdase treatment fraction was completely digested, as seen by the disappearance of the HA band, the GAG in the fraction from the ventilated lungs and controls was confirmed to be HA (Figure 3B). HA standards run on agarose gel electrophoresis was used to determine the molecular weight of HA that accumulated in the animals ventilated at VT 20 cc/kg, animals ventilated at VT 7 cc/kg and control nonventilated animals. A standard graph was plotted between log molecular weight versus relative mobility (Figure 4A). The molecular weight of the HA in VT 20 cc/kg was extrapolated from a standard curve (Figure 4B). In VT 20 cc/kg rat lungs, two LMW (MWs 180 kD and 370 kD) forms and one HMW (MW 3,100 kD) form of HA accumulated compared with animals ventilated at 7 cc/kg (MW 2,730 kD) and control nonventilated animals (MW 3,100 kD) in which only the HMW form accumulated.
HAS 3 and HAS 2 mRNA expression were measured by RT-PCR on rat lung tissue from animals ventilated with VT 20 cc/kg and VT 7 cc/kg, and from nonventilated animals. Animals ventilated at VT 20 cc/kg showed an increase in HAS 3 mRNA compared with animals from the other two groups (Figure 3C). There was no increase in HAS 2 mRNA expression (data not shown). Low levels of HAS 2 and HAS 3 were observed in animals with VT 7 cc/kg and nonventilated animals.
Mechanical Stretch–Induced LMW HA Is Dependent on HAS 3 Expression in Human Fibroblasts
To confirm the stretch response observed in human fetal fibroblasts, we used normal adult human fibroblasts. In those cells, cyclic stretch also caused a significant time-dependent increase in PGs, which was maximal at 4-h stretch in adult lung fibroblasts (Figure 6B). The PGs extracted from the supernatant of normal adult fibroblasts were analyzed by cellulose acetate electrophoresis. An HA-like fraction was identified based on its mobility compared with standard HA, and its disappearance following HAdase treatment, whereupon it was completely degraded and was replaced by a PG identified as CS. The HA-like band was completely digested by HAdase and Condroitin ABCase. Therefore, the major PGs upregulated by stretch in normal human adult lung fibroblasts are HA and CS (Figure 6A). We did not find CS in fetal fibroblasts. The PGs produced by in vitro stretched adult fibroblasts were identified as LMW (219 kD) forms compared with nonstretched cells, which produced HMW forms only (1,600 kD) (Figure 4B).
To confirm the role of HAS 3 in stretch-induced LMW HA-form production, we also measured the expression of HAS 3 mRNA by RT-PCR in adult lung fibroblasts (Figure 6B) as we had done in fetal fibroblasts (Figure 5B). Stretch-induced upregulation of HAS 3 at 30 min of stretch; this was sustained through the 4-h stretch. These data suggest that the transient response of fetal fibroblasts to stretch-induced injury differs from the more sustained response of adult fibroblasts.
Stretch-Induced HAS 3 mRNA Expression Is Dependent on JAK 2 To confirm that JAK2 is part of the signal transduction pathway in stretch-induced upregulation of HAS 3 and accumulation of LMW forms of HA in adult fibroblasts, we added AG490, a specific JAK2 inhibitor, and then measured activation of p-JAK 2 using Western blot analysis. AG490 blocked stretch-induced accumulation of LMW HA forms and upregulation of HAS 3 in the adult fibroblasts (Figures 6A and 6B), confirming the findings made with fetal fibroblasts. Here, too, p-JAK2 was upregulated as early as 5 min stretch, and the signal was sustained over at least a 30-min stretch (Figure 6C). These results are consistent with the conclusion that stretch-induced production of LMW HA by lung fibroblasts is dependent on new HAS 3 mRNA expression regulated by Jak-2 kinase.
LMW HA Induced IL-8 Expression in Epithelial Cells and Augmented Stretch-Induced IL-8 Production
We confirmed these findings in human bronchiolar epithelial cells (BEAS-2B), another cell type in the lung capable of producing IL-8. BEAS-2B cells were exposed to HA at 100 µg/ml with 10 µg/ml Polymixin B for 6 h. Only the LMW HA (370 kD) caused a significant increase in IL-8 production (LMW: 820.857 ± 237.3) as compared with the HA of higher molecular weight (2,300 kD) (HMW: 37 ± 12.7 and Control: 3.430 ± 1.2). IL-8 production was measured in A549 cells stimulated with HA fractions of molecular weight 370 kD (LMW HA; 272.31 ± 37.687) and 2,300 kD (HMW HA; 30.75 ± 9.5) with and without HAdase treatment (LMW HA; 5.55 ± 3.03 and HMW HA; 0.75 ± 0.5), heat treatment (LMW HA; 266.15 ± 9.5 and HMW HA; 30.5 ± 9.5) and DNase treatment (LMW HA; 280 ± 10.5 and HMW HA; 32.3 ± 0.5). Inactivation of HA with HAdase treatment confirmed that LMW HA induced IL-8 production. Neither heat nor DNase treatments altered LMW HA stimulation of IL-8 production. Therefore, it is the LMW HA in this sample that induces IL-8, and not contaminating protein or DNA. Although LPS could potentially have contaminated these samples, it was not a factor here: the endotoxin content of all the samples was < 0.03 EU/ml. To explore the role of LMW HA in IL-8 production in cells undergoing cyclic stretch, standard HAs described above were used to stimulate A549 cells undergoing cyclic stretch. Commercially purchased LMW HAs and not HMW HA augmented stretch-induced IL-8 production (Figure 7C). The LMW HA from the stretched fibroblasts and the HMW HA from the nonstretched fibroblasts were used to confirm the role of LMW HA in cytokine production in VILI. The purified LMW HA (178 kD) and HMW HA (> 1,600 + 1,600 kD) from fetal fibroblasts also induced IL-8 production in statically cultured A549 cells, and augmented stretch-induced IL-8 production (Figure 7C). Our results suggest that LMW HA is an additional mechanism for increased production of IL-8 observed in VILI.
In this study we found that high tidal volume ventilation of rat lungs caused changes in their production of HA (Figures 1 and 3), a component of the ECM. LMW HA in the size ranging from 180–370 kD accumulated in rats ventilated at VT 20 cc/kg, but not in rats ventilated at VT 7 cc/kg (Figure 4), and this LMW HA accumulation was accompanied by an increase in the expression of HAS 3 mRNA (Figures 5 and 6). We used an in vitro model of lung cell stretch to explore the mechanism of stretch-induced production of LMW HA and the potential role of LMWHA in cytokine production. We identified a potential role for the JAK 2 signal transduction pathway in stretch-induced production of the LMW HA (Figures 5 and 6) and documented an increase in the proinflammatory cytokine IL-8 following exposure of human type II–like A549 cells to this LMW HA in an in vitro model of lung cell stretch (Figure 7). In addition to HA, other changes in ECM have been hypothesized to contribute to the pathogenesis of VILI. Berg and coworkers (5) described increased levels of procollagen and fibronectin mRNAs after ventilation with high levels of PEEP. Parker and colleagues (4) found that ventilation at high peak airway pressures also led to increased mRNA expression of these ECM components. In vitro studies by Xu and associates (2, 3) and others (6, 7) have demonstrated the direct effects of mechanical strain on ECM synthesis and secretion. Jamal and coworkers (8) have shown that in vivo mechanical strain selectively induces increased synthesis and protein expression of PGs. Thus, the ECM appears to play an important role in VILI. Three isoforms of HA synthase have previously been identified and cloned (25). The product of HAS 3 is HA of size 200–300 kD; HAS 1 and HAS 2 make larger HA, with sizes up to 2,000 kD (26). Although each form of HAS protein can catalyze hyaluronan biosynthesis in eukaryotic cells, their enzymatic properties differ. These distinct enzymatic properties may drive distinct physiologic functions. HAS 3 is intrinsically more catalytically active than HAS 2, which in turn is intrinsically more active than HAS 1. HAS 3 has a higher Vmax than HAS 1 or HAS 2 and can produce large amounts of LMW HA in short periods of time (26). Thus in VILI, LMW HA was found in the same size range as produced in mammalian cell culture via HAS 3, suggesting that de novo synthesis of LMW HA by HAS 3 is one possible mechanism that can cause accumulation of LMW HA in the animal model of VILI. LMW HA may also be produced by breakdown of HMW HA. In vivo, HA fragmentation has been shown to result from regulated expression of secreted hyaluronidase (27) or from oxidation by reactive oxidant species (28). Hyaluronidase (Hyal-2) generated HA fragments of 10–20 kD (29). Therefore in VILI, LMW HA may be produced by multiple processes. To explore the role of HAS 3 in lung cell stretch we used human fetal and normal adult lung fibroblasts exposed to cyclic stretch in vitro. Both fetal and adult lung fibroblasts produced similar sizes of LMW HA in response to stretch, and this production paralleled an increase in the mRNA of the HAS 3 enzyme. Cyclohexamide and dexamethasone, which have diverse chemical structures, blocked stretch-induced upregulation of HAS 3 mRNA and stretch-induced accumulation of LMW HA. Inhibition of HAS 3 mRNA transcription by cyclohexamide suggests that de novo protein synthesis was required for induction of the HAS 3 gene. We found that one protein involved in the stretch-induced expression of HAS 3 was JAK2. The mechanism of dexamethasone inhibition of HAS 3 gene expression is unknown. These data suggested that stretch-induced production of LMW HA is in part dependent on HAS 3. Cyclic stretch has been shown to activate signal transduction pathways including JNK, ERK1/2, p38 and tyrosine kinases. We found that a PTK inhibitor (genistein) and a JAK 2 (AG490) inhibitor, but not a MEK 1/2 inhibitor, blocked HA production. These data suggest that stretch-induced production of LMW HA is dependent on HAS 3 that may be regulated by JAK 2 kinase. Angiotensin II remodeling of the ECM has also been shown to be regulated by JAK 2 kinase. These data are consistent in supporting an important role of the JAK-STAT pathway in the regulation of the ECM composition (18).
HAS isoforms have been found to be independently expressed and activated in mammalian cells (15, 26). Both fetal and adult lung fibroblasts express HAS 2 and HAS 3. Dowthwaite and colleagues (30) also found that HAS 3 is stretch-responsive, and that HAS 3 levels are correlated with accumulation of HA in embryonic fibrocartilage cells. It has been found that interferon- In our in vivo model of VILI, we have previously shown the development of noncardiogenic pulmonary edema with neutrophil influx that was dependent on stretch-induced MIP-2 production (9). However, the mechanisms of stretch-induced cytokine production and inflammation have not been fully characterized and may involve many factors. Others have shown that under physiologic conditions, HA exists in excess of 1,000 kD. However, the present study shows that following tissue injury, HA fragments of lower molecular mass accumulated. The LMW HA of the size range that accumulated in our model of VILI also induced IL-8 production in a dose-dependent manner in human type II–like alveolar epithelial A549 cells. High molecular weight HA was incapable of this effect. These findings with commercially obtained HA were confirmed by LMW HA and HMW HA purified from the supernatants of fetal fibroblasts. LMW HA from stretched fibroblasts induced IL-8 in static A549 cells and augmented stretch-induced IL-8 production by these cells. These data are consistent with the hypothesis LMW HA produced in VILI can serve as an active proinflammatory stimulant in vivo, and may provide an alternative mechanism to increase cytokine production. HA has been shown to produce distinct biological effects depending on the molecular weight and cellular source such as macrophages and endothelial cells. HA with molecular size of 250 kD but not HMW precursors induces the expression of inflammatory genes (12–15). Similar results have been shown with renal tubular cells, T-24 carcinoma cells and eosinophils. In contrast, a different size requirement is observed to affect dendritic cells, where 6–20 oligosaccharides induce inflammatory gene expression. HMW HA activated protein tyrosine kinases to a limited extent in endothelial cells; this activation is enhanced several fold in the presence of small HA fragments. In 1989 West and Kumar (33) showed that oligomers of 8–16 disaccharides prepared by enzymatic digestion of native HA-induced angiogenesis in the chick corneal assay, whereas native high molecular weight polymer did not. Other authors have suggested that the observed induction of cytokines is due to impurities in the HA fragments, and not to the HA itself (34). Like Balazas and coworkers (35), we have ruled out contaminating proteins, DNA, and LPS in the proinflammatory properties of these forms of LMW HA. In contrast, we have shown that LMW HA is responsible for these effects, because inactivation of LMW HA with HAdase successfully blocked the stimulation. Thus, LMW HA, and not contaminants, are responsible for the induction of cytokines. In adult fibroblasts we found increased expression of both HA and CS. In normal lung tissue, HA is not covalently associated with proteins, although HA does noncovalently associate with versican, the CS-containing PG in the lung. After we treated the GAGs isolated from the supernatants of the stretched adult fibroblasts with HAdase, the HA was degraded and then the CS became visible. Others have shown that stretch increased the production of versican mRNA and versican and may be the source of CS in cyclic stretch (8). In conclusion, large VT ventilation causes production of LMW HA, which is accompanied by increased HAS 3 mRNA expression. In vitro, cyclic stretch of fibroblasts also induces production of LMW HA, which is dependent on HAS 3 activity and is regulated by the JAK 2 pathway. Stretch-induced LMW HA causes production of IL-8 in type II–like cells. LMW HA production may play an important role in VILI. Elucidation of the molecular mechanism of accumulation of HA and HA-induced cytokine production may lead to the development of new strategies to interrupt the exuberant inflammatory response characteristic of ARDS and VILI.
The authors thank Dr. Charles Underhill, Department of Oncology, EG19A NRB, Washington, DC, for the generous gift of bHABP. R.M.D. is a recipient of American Heart Association Scientist Development Grant and American Lung Association Grant. D.A.Q. is supported by NIH #HL03920–01. M.M.M. is supported by NIH #5T32 HL07874–05. C.A.H. is supported by NIH #HL39150–13 and Shriners Hospital for Burns #8620. The authors thank Susannah W. Wood for her generous support of this work and Kathie Sweeney Laing for her assistance in preparation of the manuscript. Received in original form August 23, 2002 Received in final form April 23, 2003
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Abstract
Introduction
, and interleukin (IL)-6. We previously showed that the stretch-induced release of MIP-2 led to the infiltration of leukocytes, which are part of the tissue response to injury in VILI (
, tumor necrosis factor-
