Low molecular weight hyaluronan , via AP-1 and NF-κB signalling , induces IL-8 in transformed bronchial epithelial cells

List of abbreviations used ANOVA: Analysis of Variance; AP-1: Activator Protein-1; BEGM: Bronchial Epithelial Cell Growth System; DNA: deoxyribonucleic acid; ELISA: Enzyme-linked Immunosorbent Assay; GAG: Glucosaminoglycans; HA: Hyaluronan; HAdase: Hyaluronydase; HAS: Hyaluronan Synthase; HMW-HA: High Molecular Weight Hyaluronan; IFN: Interferon; IL-8: Interleukin-8; JNK: c-Jun NH2-terminal Kinase; ERK 1/2: Extracellular Signal-regulated Kinase 1/2; JAK/STAT pathway: Janus Kinase/Signal Transducers and Activator of Transcription pathway; LDH: Lactate dehydrogenase; LMW-HA: Low Molecular Weight Hyaluronan; NF-κB: Nuclear Factor kappa-light-chain-enhancer of activated B cells; MAPK: Mitogen-activated Protein Kinase; TLR: Toll-like Receptors; TNF: Tumour Necrosis Factor; VALI: Ventilator-associated Lung Injury.


Introduction
Hyaluronan (HA) -first discovered in 1934 -is a nonsulphated linear glycosaminoglycan (GAG) composed of repeating disaccharides (β, 1-3 D-N-acetylglucosamine, β, 1-4 D-glucuronic acid) [1].HA normally occurs as a part of the extracellular matrix in a high molecular weight (HMW) polymer form (>1000 kDa) [2,3].Low molecular weight HA (LMW, defined as <500 kDa) can be produced by the breakdown of HMW-HA to LMW-HA, or by de novo synthesis by hyaluronan synthases.Traditionally, it was believed that the physiological function of HA was only structural [4].However, HA serves as a pro-inflammatory molecule.Recent studies revealed increased amounts of total HA and LMW-HA in animal models of acute lung injury [5][6][7][8].Moreover, LMW-HA can induce the expression of several pro-inflammatory cytokines, including interleukin-8 (IL-8) [1,[9][10][11][12].New evidence has demonstrated that high tidal volume mechanical ventilation results in substantial bronchial airway mechanical strain [13].Our laboratory has investigated the role of HA in ventilator-associated lung injury (VALI) [1,2,4,6].Using in vitro and in vivo models of VALI, we found that cyclic stretch produced LMW-HA though the janus kinase/signal transducers and activator of transcription pathway (JAK/STAT) signalling pathway, and that LMW-HA induced IL-8 expression.However, the downstream mechanism by which LMW-HA induces IL-8 expression in the lung is unclear.Moreover, whether LMW-HA plays a role in bronchial injury induced mechanical strain is unknown.To this end, we evaluated IL-8 production in human bronchial epithelial cells in response to LMW-HA synthesised by stretched human lung fibroblasts.In order to determine the response of IL-8 expression by HA size and species, we compared the results of LMW-HA synthesised by stretched human lung fibroblasts to LMW-HA from bovine vitreous and HMW-HA from rooster comb.It was found that newly synthesised LMW-HA increases IL-8 production via activator protein-1 (AP-1) and NF-κB signalling pathways.

Mechanical stretch of lung fibroblasts
The IMR-90 cells were stimulated with cyclic stretch to produce LMW-HA, as previously described [6].Fibroblasts were seeded at 2 × 10 6 cells/dish onto sterile silicone membranes mounted in a polypropylene 100 mm dish.This stretching device was custom built and provided by Martha Gray Ph.D (Massachusetts Institute of Technology, Cambridge, MA) [14].This stretch apparatus provides a sinusoidal, spatially homogeneous, and isotropic biaxial strain to cultured cells.Cells were subjected to stretch in 37 °C and 5% CO 2 at 60 cycles/minute with 15% strain for 4 hours.In addition, it was previously estimated that a change from 42% to 64% of total lung capacity corresponded to a 15% of linear strain [15].This protocol was chosen because it induced maximal hyaluronan production and did not affect lactate dehydrogenase (LDH) release from stretched cells.LDH was measured in static cells supernatant vs. stretched cells supernatant as described by our group before [16].
Extraction and Analysis of HA HA was extracted from pooled cell-free culture supernatants under dissociative conditions using 4 M guanidine HCl, as previously described ( [6].The size of the HAs used has been previously reported and were determined using column chromatography [6].Briefly, conditioned media form stretched fibroblasts were lyophilised against water.HA presence was confirmed by proteoglycan electrophoresis as described earlier [1].HA was purified using Sepharose CL-4B size exclusion chromatography (150 cm × 2 cm) equilibrated in 0.05 M sodium acetate buffer pH 6.7.Calibration of the column occurred as follows: void volume (Vo) was measured using Dextran blue 2000; total volume (Vt) was measured using uronic acid; HA sizes were determined by relative elution volume (Kaw) compared with three standards of known molecular weight (1600 kDa; 370 kDa; and 178 kDa).All samples were analysed by the same method and purified for further analysis.
We also confirmed the purity of HA in the column fraction of stretched fibroblasts by FT-IR spectroscopy analysis.We have previously reported that the molecular weight of HA produced by stretched fibroblasts is <500 kDa with an average weight of 178 kDa [1].

Stimulation of BEAS-2B cells with HA
LMW-HA (370 kDa) from bovine vitreous humor was purchased from Sigma Chemical Co.(St.Louis, MO) and ICN Biochemicals (Costa Mesa, CA), and HMW-HA (1600 kDa) from rooster comb was purchased from Sigma Chemical Co (St. Louis, MO).Static cultures of BEAS-2B cells were washed with HBSS and then treated with LMW-HA, HMW-HA, or IMR-90 HA for 6 hours at 37 °C, at a concentration of 100 µg/ml in serum-free media.Possible contamination of HA samples by lipopolysaccharide (LPS) was controlled for by adding 10 µg/ml of polymyxin.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 the HA samples.
A dose of LMW-HA at 100 µg/ml was chosen to maximise IL-8 production.In a previous report, it was found that the maximum IL-8 production in alveolar type II-like cells was induced by LMW-HA at 100 µg/ml [1].
To rule out effects of contaminating proteins in HA samples, cells were treated with heat inactivation (100 °C for 10 minutes) and the protease trypsin as described before [16].To rule out effects of contaminating DNA in HA samples, HA was treated with DNAse (10 Units of DNAse I for one hour at 37 °C followed by inactivation at 65 °C for 10 minutes, to remove trace amounts of DNA).To confirm LMW-HA induction of IL-8, HA samples were treated with 0.05 Units/2 µg sample HAdase (Streptomyces HAdase, Sigma Chemical Co., St. Louis, MO) at 60 ºC for 72 hours followed by inactivation at 65 °C for 10 minutes to remove HA.

Inhibitors of mitogen-activated protein kinase (MAPK) and NF-κB/IκB pathways
To determine the role of the MAPK and NF-κB/IκB pathways in LMW-HA induced IL-8 production, we used the extracellular signal-regulated kinase 1/2 (ERK 1/2) inhibitor U0126 (25 µM) the JNK II inhibitor SP600125 (10 µM), the NF-κB peptide inhibitor SN 50 and the NF-κB control peptide SN 50 inactive control (EMD Biosciences, La Jolla, CA).Cells were pre-incubated with inhibitors for one hour in serum-free media, followed by stimulation with HA for 6 hours.An LDH-based toxicology assay (Sigma Chemical Co, St. Louis, MO) was performed to rule out toxic effects of the inhibitors.No toxicity was found.

Analysis of IL-8 protein
Cell culture supernatants were removed and centrifuged at 1200 g for 15 minutes to remove cellular debris.An enzyme-linked immunosorbent assay (ELISA) was used to measure IL-8 protein levels (R&D, Minneapolis, MN).

Statistical analysis
Statistical analyses were performed using Statview 4.5 (Abacus Concepts, Inc., Berkeley, CA).The concentration of IL-8 in cell supernatants was compared by analysis of variance (ANOVA) and then subsequent multiple comparisons were performed by the Scheffe-test.All values were expressed as mean ± standard error of the mean.Significance was determined by p <0.05.

LMW-HA increased IL-8 production in BEAS-2B cells independent of contaminating LPS, protein or DNA
To address the hypothesis that LMW-HA production is proinflammatory, LMW-HA from the stretched IMR-90s was used, as previously described [1].We previously showed that un-stretched IMR-90s only produce HMW HA [1].In the current study, LMW-HA from bovine vitreous humor (fig.1A) and LMW-HA isolated from stretched IMR-90 fibroblasts (fig.1B), but not HMW-HA from rooster comb, significantly increased IL-8 protein in BEAS-2B cells (fig.1A).HAdase treatment of HA preparations significantly blocked IL-8 production, whereas treatment of HA preparations with heat inactivation or protease to remove contaminating protein, or DNAse to remove contaminating DNA did not affect IL-8 production.
LMW-HA activated JNK, ERK and NF-κB/IκB pathways, but not the p38 pathway Both forms of LMW-HA, from bovine vitreous (fig.2A) and from stretched IMR-90s (fig.3A), but not HMW HA from rooster comb (fig.2A), increased phosphorylation of JNK (JNK 1 and JNK 2) as early as 15 minutes after exposure.This effect was sustained for up to 60 minutes with no change in the total JNK (JNK 1 and JNK 2) (fig.2A and 3A).Both forms of LMW-HA (fig.2B and 3B), but not HMW-HA (fig.2B), increased phosphorylation of ERK 1 and ERK 2 with no change in total ERK.Both forms of LMW-HA (fig.2C and 3C), but not HMW-HA (fig.2C), also increased NF-κB expression in the nuclear fraction and decreased IκB expression in the cytosol, suggesting NF-κB activation.To determine the specificity of action of LMW-HA in the activation of MAPK kinase and NFκB/IκB pathways, HAdase treated HMW-HA from rooster comb and LMW-HA from bovine vitreous were included as controls.HAdase treated LMW-HA and HMW-HA did not increase JNK or NF-κB activation (fig.2D).LMW-HA and HMW-HA did not activate the p38 pathway (data not shown).

IL-8 production by LMW-HA depended on the JNK and NF-κB/IκB pathways and, to a lesser extent, on the ERK pathway
Treatment with the JNK inhibitor SP600125, a reversible ATP-competitive inhibitor with greater than 20-fold selectivity relative to other kinases in inhibiting the action of p-JNK, but not the expression of p-JNK, significantly inhibited bovine vitreous LMW-HA-induced IL-8 production (fig.4A).Inhibition of NF-κB activity with NF-κB inhibitor SN50, which inhibits translocation of the NF-κB active complex into the nucleus [17], significantly blocked the LMW-HA-induced IL-8 production compared to the NF-κB control peptide (fig.4A).We observed only a partial inhibition of IL-8 production with ERK inhibition (fig.4A).HMW-HA from rooster comb did not induce IL-8 production and this was not affected by inhibition of JNK or NF-κB activity (fig.4A).Similar results were obtained regardless of whether the LMW-HA was derived from bovine vitreous humor (fig.4A) or was collected from stretched IMR-90 fibroblasts (fig.4B).

Discussion
Ventilator-associated lung injury (VALI) is acute lung injury that develops while a critically ill patient is mechanically ventilated [18].Alveolar over-distention and cyclic atelectasis are known as the principal triggers of alveolar injury during positive pressure ventilation [19][20][21].Therefore, all current ventilatory and pharmacologic strategies aim to reduce the incidence of VALI by preventing alveolar damage.However, even with these strategies, the mortality associated with VALI remains unacceptably high [22].This suggests that other segments in the airway besides the alve- oli might be affected by high-tidal volume ventilation and could contribute to VALI.Indeed, new evidence showed that high tidal volume mechanical ventilation -one of the culprits in VALI -results in substantial bronchial airway mechanical strain [13].Our own observations had suggested a similar phenomenon [1].The choice of transformed bronchial epithelial cells in the current study reflects our intent to understand the molecular and signalling underpinnings of the effects of high tidal volume mechanical ventilation in bronchial epithelium and its possible role in VALI.
To this end, we have studied the pro-inflammatory properties of HA, a common lung extracellular matrix component [26].HA is synthesised by HA synthases (HAS) [27][28][29] and it exists in a HMW form (>1000 kDa).However, it is the low molecular weight form (LMW-HA) that is pro-inflammatory and recent evidence revealed that LMW-HA is a pro-inflammatory molecule in the lung: LMW-HA (200 kDa) isolated from the serum of patients with acute lung injury, stimulated MIP-2 production in macrophages through binding to Toll-like receptors (TLR) 2 and 4 [12].In addition, the over-stretched lung produces LMW-HA that upregulates IL-8 production in a HAS 3 dependent manner [6].In some animal models of lung injury, the production of HA starts a host innate immunity response that leads to the production of more inflammatory molecules and the mobilisation of cellular components that are critical for the resolution of the injury.This however, could lead to more tissue damage.The size of HA that induces cytokine production appears to vary and may depend on the structure of the HA produced.It was found that LMW-HA of 180 kDa and 370 kDa was produced in a rat model of VALI [6].The measurement of the size of HA can be difficult and the measured size represents an average size, since the fragments of HA in a sample are not uniform.Despite any variation that may have occurred in our measurement of the size of the HA isolated from the stretched lung fibroblasts, there was a clear difference in LMW-HA <500 kDa and HMW-HA >1000 kDa.Since LMW-HA can be generated by degradation of HMW-HA forms or by the direct synthesis of HAS 3, we cannot rule out that the breakdown of HMW-HA to LMW-HA may also occur in vivo.However, HAS 3 deficient mice did not produce LMW-HA when exposed to high tidal volume ventilation, indicating that at least in this form of acute lung injury the breakdown of HMW-HA to LMW-HA was not important [6].
The downstream signalling mechanism of LMW-HA involves the binding to HA receptors, called hyaladherins [30].The hyaladherins include CD44, the receptor for HAmediated motility (RHAMM), TLR2, TLR4, layilin, LYVE-1 and HARE [31].CD44 is a trans-membrane glycoprotein and it is the major receptor for HA [32].LMW-HA cross-links with CD44 and activates the signal transduction pathway that causes activation of transcription factors and chemokine production.In the current study, it was shown that LMW-HA synthesised by human lung fibroblasts, induced IL-8 production in human bronchial epithelial cells, via JNK, ERK1/2, and NF-κB/IκB path-ways leading to AP-1 and NF-κB translocation to the nucleus.
We have previously shown -in vitro and in vivo -that JNK activation is critical in the pathogenesis of VALI and the current observations confirm the role played by the extracellular signal-regulated kinases in its pathogenesis [15,33].Therefore, the findings presented here suggest that two mechanisms of JNK and ERK activation might be in play during VALI.Firstly, JNK and ERK are activated due to their mechano-sensitive nature, and secondly, JNK and ERK could perpetuate lung damage when turned on by stretch-induced LMW-HA.
The synchronised activation of AP-1 and NF-κB by the same effecter is in line with previous observations.The concurrent mobilisation of both transcription factors was reported in human hepatoma cells in response to hepatitis B virus [34], in mouse dendritic cells in response to nickel sulphate [35], and in primary human lung fibroblasts in response to Tumour Necrosis Factor (TNF)-α stimulation [36].Similarly, the induction of IL-8 production and other pro-inflammatory cytokines by the simultaneous nuclear translocation of AP-1 and NF-κB has been reported in the past.Yasumoto reported that TNF-α and interferon (IFN)-γ synergistically induced IL-8 production in a human gastric cell line through the activation of AP-1 and NF-κB [37].
In relation to HA, other investigative groups have reported activation of AP-1 and NF-κB by HA.Boodoo et al.
showed that HA fragments induced production of the IL-8 and IP-10 in human mucoepidermoid carcinoma cells (NCI-H292) and was dependent on the activation AP-1 and NF-κB, respectively [9].Our findings may differ from those of Boodoo et al. for very specific reasons.Firstly, the cell lines used to investigate AP-1 and NF-κB activation in both experiments were different (NCI-H292 v. BEAS-2B).
Secondly, the size of the HA fragments might have affected the end-point.The current study reports HA fragments of <500 kDa in size (that we called LMH-HA), while the size of the HA fragments on the aforementioned study was never reported (although it is understood to be LMW-HA).Finally, the measurement of IP-10 was never an end-point in the current study.Despite these differences, both studies are in agreement with the premise that LMW-HA can induce a pro-inflammatory cascade in the lung through AP-1 and NF-κB activating pathways.

Conclusion
The current findings confirmed a role for LMW-HA in the biotrauma theory of VALI.Therefore, we conclude that LMW-HA produced in an in vitro model of VALI induces IL-8 production via AP-1 and NF-κB transcriptional mechanisms.LMW-HA plays an important role in lung inflammation and could provide a new target for the prevention and treatment of VALI [4].
Wang-Fang Hospital, Taipei Medical University, Taipei, Taiwan, for their technical assistance.

Figure 1 IL- 8
Figure 1 IL-8 production was stimulated by LMW-HA but not by HMW-HA in human transformed bronchial epithelial cells (BEAS-2B cells) BEAS-2B cells were incubated with LMW-HA, HMW HA, or no HA in serum-free media for 6 hours.A. Effect of LMW-HA from bovine vitreous humor and HMW HA from rooster comb on IL-8 production (n = 5); B. Effect of LMW-HA isolated from IMR-90 fibroblasts (n = 3) on IL-8 production.HAdase treatment completely blocked LMW-HAinduced IL-8 production.Protease treatment, heat treatment and DNAse pre-treatment had no effect on LMW-HA-induced IL-8 production.*p <0.05 vs HMW HA exposure and no HA control.

Figure 2 LMW
Figure 2 LMW-HA from bovine vitreous humor, but not HMW HA or HAdase treated LMW-HA from rooster comb, activated the JNK, ERK and NF-κB/IκB pathways in human BEAS-2B bronchial epithelial cells.Western blot analysis of effects of LMW-HA on BEAS-2B human transformed bronchial epithelial cells: A. c-Jun NH 2 -terminal kinase, B. ERK1 and ERK-2, and C. NF-κB in nuclear fraction / IκB in cytosol fraction, D. c-jun N-terminal kinase, NF-κB and IκB expression with HAdase treated HMW and LMW-HA.Each blot is representative of 3 separate experiments, and was quantified using densitometry.Results are expressed as foldchange compared with control in arbitrary units.* p <0.05 vs control.‡ p <0.05 vs HMW-HA
Funding / potential competing interests: This work was partially supported by grants from the Susannah Wood Fund and AHA EIA 0440146N (D.A.Q), AHA 0526046H (C.D.O), and HL007874, and HL39150 (C.A.H).D.A.Q. is now employed by Novartis Pharmaceuticals.Novartis did not support this work and was not involved in the planning or execution of this work.
Swiss Medical Weekly • PDF of the online version • www.smw.ch