Activity of growth factors in the IL-6 group in the differentiation of human lung adenocarcinoma

The role of the interleukin-6 (IL-6) group of cytokines in differentiation of two lung adenocarcinoma cell lines has been examined using induction of alkaline phosphatase and expression of surfactant protein A. Oncostatin M was the most active and potent for alkaline phosphatase in A549 cells, with IL-6 having similar activity but less potency. Neither cytokine induced alkaline phosphatase in NCI-H441 cells, although induction was obtained with lung fibroblast-conditioned medium. Surfactant protein A was induced in NCI-H441 cells by conditioned medium and dexamethasone and, to a much lesser extent, by oncostatin M or IL-6. Induction of alkaline phosphatase and surfactant protein A were both dexamethasone-dependent, though some induction of surfactant protein A was obtained with interferon-α in the absence of dexamethasone. The activity present in lung fibroblast-conditioned medium suggests paracrine control, but this appears not to be due to oncostatin M or IL-6 as disabling antibodies to either cytokine were not inhibitory, and, although alkaline phosphatase was induced in A549 by both cytokines, it was only induced by conditioned medium in NCI-H441 cells. Furthermore, surfactant protein A was induced in H441 by conditioned medium to a much greater extent than by oncostatin M or IL-6. These data demonstrate that cytokines of the IL-6 group have potential as differentiation inducers in lung adenocarcinoma cells and that there is an equivalent paracrine factor(s) in lung fibroblast conditioned medium. As the production of this factor by fibroblasts is not enhanced by glucocorticoid, although the response of the target cell is, it would appear to be distinct from the fibrocyte pneumocyte factor previously described by Post et al 1984 Nature308: 284–286. © 2000 Cancer Research Campaign

IL-6 is one of a group of growth factors operating through the Gp130 receptor sub-unit (Taga and Kishimoto, 1997). Others in the group include IL-11, ciliary neurotropic factor (CNTF), oncostatin-M (OSM) and leukaemia inhibitory factory (LIF). Granulocyte-colony stimulating factor (G-CSF) has been included in this group (Rose and Bruce, 1991;Bruce et al, 1992), and also acts via the JAK-STAT pathway (Akira, 1997) but most recent reviews exclude it (e.g. Taga and Hishimoto, 1997). G-CSF, LIF and IL-11 are differentiation and proliferation factors, active principally in the haematopoietic system (Taga et al, 1992;Katayama and Ogawa, 1994), and CNTF in the neural system (Lillien et al, 1990). LIF is an inhibitor of differentiation of embryonal stem (ES) cells (Boeuf et al, 1997;Niwa et al, 1998), and the other members of the group have similar, though less potent, activity on ES cell differentiation (Bruce et al, 1992).
In the present report we have confirmed the activity of IL-6 and shown that OSM is significantly more active and potent than IL-6 in the induction of A549 differentiation and, like all other factors previously investigated, with the exception of FGF-2, is dependent on the presence of DX for its activity. LIF and IL-11 are also active, though less so than OSM, and CNTF and G-CSF are inactive. Transforming growth factor (TGF)-β is inhibitory in this system (McCormick et al, 1995) and we now show that it blocks the inductive activities of OSM and IL-6.
The potential roles of IL-6 and OSM as paracrine factors produced by lung fibroblasts were tested by use of disabling antibodies. The results failed to show any reduction in the inductive effect of CM, in spite of demonstrable inhibition of induction of AP by recombinant human OSM (rhOSM) and rhIL-6. Furthermore, DX and CM induced AP and SPA in NCI-H441 cells, another Type II/Clara cell-derived tumour cell line, while OSM and IL-6 induced SPA but not AP.

Cell culture
A549 (CCL-185) and NCI-H441 (HTB-174) cells were obtained from the American Tissue Culture Collection (ATCC) and maintained in a 50:50 mixture of Dulbecco's modification of Eagle's medium and Ham's F10 (F10:DMEM) (Life Technologies), supplemented with 10% fetal bovine serum (FBS; Globepharm). MOG-LF113 cells, used as a source of conditioned medium, are normal human diploid lung fibroblasts derived from a first trimester embryo. They were used between generations 10 and 30. Cell lines were tested monthly for mycoplasma by fluorescence staining with Hoechst 33258 (Chen, 1977).

Conditioned medium
Fibroblast CM was prepared from LF113 cells, seeded at 2 × 10 4 cells ml -1 , grown to confluence, and grown on for 8 days in F10:DMEM with 0.5% FBS. To prepare CM, the medium was replaced with serum-free F10:DMEM, or, to prepare DXCM, with serum-free F10:DMEM containing 0.25 µM DX, for a further 3 days, collected and stored frozen. CM was thawed, filtered through a 0.22 µm Millipore GV filter, and used at 50% (unless otherwise indicated) in fresh serum-free F10:DMEM.

Alkaline phosphatase activity
Cell monolayers in microtitration plates were washed in 0.15 M sodium chloride (NaCl) and frozen and thawed 3 × in 10 µl 0.15 M NaCl. AP was assayed by the dephosphorylation of p-nitrophenyl phosphate at pH 8.5 (Sigma kit #104). The concentration of coloured product, p-nitrophenol (PNP), was determined by absorbance at 310 nm, with reference to a standard curve. Enzyme activity is expressed as µmol PNP produced h -1 10 -5 cells (counted in replicate wells). The reaction was shown to be linear over the range of activities measured (data not shown).

Growth factors, antibodies and other reagents
IL-6 was obtained from Boehringer Mannheim, OSM from NBS Biologicals, LIF and IL-11 from Genetics Institute, G-CSF from Amgen, CNTF from Pepro Tech, TGF-β from British Biotechnology, anti-IL-6 antibody from NIBSC and anti-OSM from R&D. DX was obtained from Sigma (#D1756) and used at 0.25 µM, unless indicated otherwise. Actinomycin D (#A1410) and cycloheximide (#C6255) were obtained from Sigma and used at 5.0 µg ml -1 , at which concentration no loss of viability was observed using naphthalene black as a viability stain (Freshney, 2000).

Assay for exposure to growth factors and CM
A549 cells were seeded at 5 × 10 4 cells ml -1 , grown for 3 days and the medium changed to serum-free F10:DMEM. Growth factors or CM were added for a further 3 days and the AP activity measured as above.

Exposure to antibodies
CM or growth factors were exposed to antibodies at a range of concentrations, including zero, in polypropylene microtubes (Eppendorf) for 2 h at 37°C, immediately before addition to assay plates, and the assay continued as above.

Northern blots
Northern blots were prepared by standard procedures (Sambrook et al, 1989). Blots were probed with Probes 1255 and 2305, obtained from Riken.

RESULTS
Incubation of A549 cells for 3 days with DX alone, in the absence of serum, induced AP three-fold at 0.25 µM (P < 0.0001) and 4.8fold at 10 µM (P < 0.0001), but not at all at 50 nM ( Figure 1). IL-6 added for 3 days in the presence of 0.25 µM DX, gave a dosedependent increase in AP activity, reaching a maximum at 2.5 µg ml -1 , amounting to 2.3 × DX alone in 0.25 µM (P < 0.0001), and 2 × DX alone in 10 µM (P < 0.0001).
OSM, under similar conditions, induced AP 2.6-fold at 20 ng ml -1 (P = 0.0036), and was, therefore, both more potent and more active than IL-6 ( Figure 2). Like IL-6, it required DX for activity; it was inhibitory on its own. LIF gave 50% induction at 200 U ml -1 (P < 0.0001) and IL-11 60% induction at 1000 U ml -1 (P = 0.0001), in the presence of 0.25 µM DX (Figure 3), and were inactive in the absence of DX. CNTF and G-CSF gave no significant induction (P > 0.05). The combination of optimal concentrations of IL-6 and OSM showed about 15% more activity than either IL-6 or OSM alone but was not additive (data not shown).
TGF-β was shown previously to be inhibitory for the induction of AP in A549 cells (McCormick et al, 1995) and was able to obliterate the effects of either IL-6 or OSM in the present series ( Figure 4).
As previously reported (McCormick et al, 1995), CM from a lung fibroblast cell line, LF113, induced AP activity in the presence of DX. Unlike the induction of pulmonary surfactant (Speirs et al, 1991), prior exposure of the fibroblasts to DX did not increase the AP-inductive capacity of CM, and, in fact, reduced its potency, perhaps due to down regulation of IL-6 (McCormick et al, 1995) ( Figure 5). The induction by CM was completely blocked by 5.0 µg ml -1 cycloheximide and 5.0 µg ml -1 actinomycin D ( Figure 6).
In order to determine whether IL-6 or OSM were potential candidates for paracrine differentiation factors present in CM, CM was treated with disabling antibodies to IL-6 and OSM. While both antibodies inhibited their respective cytokines in positive controls (Figure 7 B, D), no inhibition of activity in CM was observed (Figure 7 A, C).
AP activity is used in these experiments as a marker of differentiation. To confirm that differentiation was being induced, expression of mRNA for SPA was determined in A549 and NCI-H441 cells. There was little induction of SPA by DX alone in A549. CM+DX, DXCM, OSM and IL-6 all showed less SPA expression than the serum-free control (SF) or DX alone, although the DX values were always higher than CM or cytokines alone. In H441 cells, SPA and AP were both induced by DX and CM (Figures 8 and  9). DX (P = 0.02), CM+DX (P = 0.0003) and DXCM (P = 0.003) induced AP, with greatest activity in CM+DX, but OSM and IL-6 showed no AP induction over DX alone. IFN-α, previously shown to induce AP in A549 cells (McCormick et al, 1995), also induced AP in H441 (P = 0.06). Preliminary data from Northern analysis suggest that CM+DX and DXCM induce SPA expression in H441, while OSM, IL-6, insulin and interferon, in the presence of DX, also induced SPA relative to DX alone (Figure 9), although expression was never as great as with CM+DX.

DISCUSSION
Previous results indicated that IL-6 was an active inducer of AP in A549 cells in the presence of DX (McCormick et al, 1995), prompting the investigation of other members of this family of cytokines. Among those studied, the highest activity and greatest potency were found with OSM, with 2.6-fold stimulation at 10 ng ml -1 , compared with 2.3-fold stimulation with IL-6 at 2.5 µg ml -1 . Activity was also seen with LIF (50% at 200 U ml -1 ) and IL-11 (60% at 1000 U ml -1 ). The response to IL-11 may have been crossreaction with one of the other receptors, as the concentration of ligand was quite high, but the response to LIF, although modest, may have been through the LIF receptor, known to be expressed on A549 (Piquet-Pellorce et al, 1994). TGF-β stimulates A549 cells to produce IL-11 and DX inhibits this stimulation (Wang et al, 1999) so, even if IL-11 receptors are present on A549 the stimulatory role of TGF-β and inhibitory role of DX would conflict with the effects seen here with AP induction. No significant activity was found with G-CSF and CNTF. Activities may be ranked as follows: OSM > IL-6 > LIF > IL-11 > G-CSF > CNTF.
This group of cytokines has been shown to share activities in other systems, e.g. inhibition of differentiation in ES cells, where LIF has the greatest effect (Smith et al, 1988). It has been suggested that the shared activity results from a common interaction with the Gp130 subunit of the receptor, which they all share, while specificity is determined by the α-subunit which is unique to each cytokine (Gearing et al, 1992;Taga and Kishimoto, 1997). Low concentrations of the active cytokines are additive, while high concentrations are not (data not shown), again suggesting a shared receptor. Activity was only seen in the presence of DX at a minimum concentration of 0.25 µM. The effect of DX is likely to be multifactorial. Glucocorticoids are known to influence gene transcription when the nuclear receptor binds to the glucocorticoid response element (GRE) (Reichardt and Schutz, 1998) and modulate genes activated by OSM (Haselmann and Goppelt-Struebe, 1997). In addition, they can influence signal transduction by interaction with the IL-6 signal transduction pathway, JAK tyrosine kinase and the STAT transcriptional activator (Takeda et al, 1998), and by inhibition of protein kinase C (PKC) action downstream of PKC activation (Tian et al, 1999). They can also affect cytokine receptor presentation (Wiegers and Reul, 1998), inhibit extracellular proteases (Freshney, 1985) which, in turn, can block activation of TGF-β (Cai et al, 1997).
DX has also been shown to alter the extracellular matrix (Mackie et al, 1988) and extracellular matrix has been shown to be important in promoting the expression of the type II phenotype in culture (Liu and Mautone, 1996;Alcorn et al, 1997). In the current system DX has been found to increase the synthesis of a fraction of heparan sulphate present in the medium of DX-treated A549 cells. This fraction is heparinase and heparitinase-sensitive and can replace DX in the induction of AP in A549 cells by CM, IL-6 and OSM (Yevdokimova and Freshney, 1997). CNTF, a member of the IL-6 family, has been shown to interact with matrix in the differentiation of type 2 astrocytes (Lillien et al, 1990), and heparan sulphate proteoglycan (HSPG) has been implicated in receptor dimerization (Schlessinger et al, 1995). It would appear that, if induction of an HSPG is the main role of DX in this system, then either the HSPG shares affinity for a wide range of growth factors, including IF-α, IF-β, OSM, IL-6 and insulin, all of which have been shown to have DX-dependent activity (McCormick et al, 1995, and this report), or they all induce a secondary autocrine factor to be produced by the target cells, A549 or H441, and the specificity of the HSPG is directed towards that factor.
The activities of both OSM and IL-6 were inhibited by TGF-β, consistent with the inhibitory role proposed for TGF-β in type II cell maturation (Torday and Kourembanas, 1990;Beers et al, 1998), where TGF-β in the lung remains high during embryonic development but falls just before parturition when hydrocortisone levels rise, inhibit TGF-β3 gene expression and induce surfactant production (Jaskoll et al, 1996). Although TGF-β is low at parturition, it can be induced by hydrocortisone in embryonic rat lung fibroblasts and may be implicated in the morphogenesis of bronchial epithelial cells (Wang et al, 1995). Preparation of fibroblast-conditioned medium in the presence of DX decreased its inductive effect on A549 cells (see Figure 5) which would be compatible with induction of TGF-β in the fibroblasts. The effect of TGF-β on SPA expression was not investigated in this study, but inhibition in H441 has been reported (Whitsett et al, 1992).
As IL-6 and OSM are both potential candidates for the inducing activity of CM, blocking antibodies were used to determine whether they could account for all or part of the activity of CM. Antibodies to IL-6 and OSM, selected for their ability to inhibit the pure cytokine, were unable to inhibit CM. This suggests that the major active component of CM is neither IL-6 nor OSM, although it is acknowledged that native IL-6 or OSM in conditioned medium may be in a different form from the recombinant cytokines to which the antibodies were raised, e.g. bound to a cofactor such as a proteoglycan (Yevdokimova and Freshney, 1997) or by post-translational modification. However, CM, in the presence of DX, induced both AP activity and SPA gene expression in H441 cells, while OSM and IL-6 induced SPA but not AP, although some AP-inducing activity was seen with IFN-α. This suggests that the activity in CM is not IL-6 or OSM. Experiments with anti-gp130 antibody would confirm that activity is due to a member of this family.
It is possible that activity is due to a combination of factors. OSM, LIF, and/or another member of the IL-6 group, combined with IFN-α or IFN-β could generate AP-inducing activity in A549 cells equivalent to that of CM, but would not account for the induction of AP in H441 cells. A combination of optimal concentrations of IFN-α, IL-6 and insulin were shown previously to have equivalent activity to conditioned medium (McCormick et al, 1995), but this would still show inhibition with anti-IL-6, subject to the reservations above. Keratinocyte growth factor (KGF, FGF-7) has been shown to induce alveolar type II cell maturation (Shiratori et al, 1996;Chelly et al, 1999) but as it had no activity in AP induction with A549 (data not shown) it was not included in this series. It could, however, have an important role in surfactant induction, if not in A549 cells, perhaps in H441.
Induction of SPA expression confirmed the differentiating activity of CM+DX in H441 cells and suggested that, unlike AP, SPA is induced by OSM, IL-6 insulin and IFN in the presence of DX. IF was also active, though less so, in the absence of DX. This was distinct from the response of A549 cells, where all conditions, other than DX alone, decreased SPA expression, though less so in the presence of DX. It would appear that the regulation of AP and SPA may be different and that A549 clearly responds differently from H441 and may be progressing along a different lineage. Induction of AP suggests that this is not towards a type I pneumocyte, but it may be towards a Clara cell or other secretory phenotype. As Clara cells also make SPA (Kalina et al, 1992) the second may be more likely. Alternatively, inappropriate type II Effect of fibroblast conditioned-medium (CM) and cytokines on AP activity in NCI-H441 cells. Mid-log phase H441 cells were incubated in microtitration plates in serum-free medium for 3 days with or without 0.25 µM DX, 50% CM, 10 ng ml -1 OSM, 2.5 µg ml -1 IL-6, 2.5 µg ml -1 insulin and 20 ng ml -1 IF-α, and assayed for AP as described in the Methods.  Figure 9 Effect of fibroblast conditioned-medium (CM) and cytokines on expression of SPA mRNA in A549 and NCI-H441 cells. Radioactive cDNA was prepared from Probe 155 or 2305, specific for SPA mRNA, and hybridized to RNA extracted from A549 or NCI-H441 cells after treatment for 3 days with 50% CM, 10 ng ml -1 OSM, 2.5 µg ml -1 IL-6, 2.5 µg ml -1 insulin, or 20 ng ml -1 IF-α, in the presence and absence of 0.25 µM DX Relative values can be compared within one, but not between two experiments H441 were due to different hybridization characteristics of the two probes or differences in the regulation of expression of mRNA for SP-A1 and SP-A2 in the two cell lines (Scavo et al, 1998).
In conclusion, IL-6 and OSM are among the most active cytokines in the induction of AP activity in A549 cells, but the lack of inhibition of OSM and IL-6 by blocking antibodies, and their lack of AP-inducing activity with H441 cells, suggests that the active paracrine factor(s) produced by lung fibroblasts have not yet been identified. Purification or cloning of the factor or factors present in conditioned medium, plus the recreation of the correct microenvironment with extracellular matrix constituents such as laminin and proteoglycans, and the presence of cells or conditioned medium from primary cultures from the type II lineage, may throw further light on this.