Beta-adrenergic signalling in neoplastic lung type 2 cells: glucocorticoid-dependent and -independent defects.

Tumorigenic mouse lung-derived type 2 cell lines have large reductions in both beta-adrenergic-stimulated cAMP production and ligand binding to beta-adrenergic receptors. These tumorigenic cells are also relatively insensitive to glucocorticoids. Because glucocorticoids regulate both beta-adrenergic receptor expression and receptor coupling to the stimulatory guanine nucleotide binding protein Gs interactions between the glucocorticoid and beta-adrenergic signalling systems were examined. This study demonstrates that beta-adrenergic ligand binding and agonist sensitivity are increased in a tumorigenic cell line stably expressing a normal glucocorticoid receptor transgene. However, although the transfected tumour cells and non-tumorigenic cells have similar amounts and affinities of beta-adrenergic agonist and antagonist binding, similar amounts of Gs subunits and similar forskolin-stimulated adenylyl cyclase activities, the former remain much less isoproterenol responsive. Competition binding studies demonstrate that tumour cell beta-adrenergic receptors have both high- and low-affinity agonist binding but are functionally uncoupled from Gs. This uncoupling may involve an alteration in Gs, as guanine nucleotides exhibit a reduced ability to stimulate adenylyl cyclase. Thus, some aspects of tumorigenic cell dysfunction in beta-adrenergic signalling can be ameliorated by interactions with the glucocorticoid pathway, but additional defects are also involved.

,B-Adrenergic receptor (PAR) expression and P-adrenergic hormone sensitivity increase during lung development and glucocorticoids accelerate these increases (Cheng et al., 1980;Barnes et al., 1984). In some cell types, glucocorticoids increase PAR expression (Collins et al., 1988) or PAR coupling to G. (Davies and Lefkowitz, 1984), the heterotrimeric guanine nucleotide-binding protein that regulates PAR stimulation of adenylyl cyclase (Levitzki, 1988). When compared with two non-tumorigenic mouse lung alveolar type 2-derived cell lines, a number of tumorigenic cell lines of type 2 cell origin exhibit aberrant signal transduction including considerable reductions in sensitivity to both f,adrenergic (Droms et al., 1989;Lange-Carter et al., 1992) and glucocorticoid (Droms et al., 1993) hormones. As both f,adrenergic and glucocorticoid signals are major regulators of type 2 cell function (Ballard, 1986), the defects in these systems may be important aspects of the neoplastic progression of type 2 cells.
The heterotrimeric G protein G, consists of a, ,B and y subunits. Hormone binding to receptor induces G,a to exchange bound GDP for GTP, resulting in adenylyl cyclase activation. Hydrolysis of the terminal phosphate restores the resting state (Levitzki, 1988). The PAR exhibits two affinity states for agonist binding: one of higher relative affinity (kH) and the other of low affinity (kL) and glucocorticoids can affect both the relative proportions and affinities of these two sites (Davies and Lefkowitz, 1981). The proposed molecular basis of the high-affinity binding site is that agonist binding to receptor results in formation of a 'ternary complex' consisting of hormone, receptor and G protein (De Lean et al., 1980). Thus, the existence of high-affinity agonist binding sites is an indicator of the 'coupling' between receptor and G protein. Addition of guanine nucleotides destabilises this ternary complex, resulting in low-affinity hormone binding (Rodbell et al., 1971). The relative ability of guanine nucleotides to induce the low-affinity state is an indicator of functional coupling between receptor and G-protein Received 30 October 1995;revised 27 February 1996;accepted 5 March 1996 (Cheung et al., 1989). It is not clear whether glucocorticoid enhancement of PAR-G. coupling involves glucocorticoid effects on the PAR or G. or what the mechanisms of these effects might be. However, glucocorticoid treatment of cultured fibroblasts increased GTP-dependent activation of adenylyl cyclase in the absence of added hormone (Johnson and Jaworski, 1983), suggesting that some glucocorticoid effects can be exerted distal to hormone receptors.
Tumour-associated decreases in receptor-coupled adenylyl cyclase activity often result from reduced receptor affinity or number (Hunt and Martin, 1980). Indeed, the lack of sensitivity of mouse lung tumour cells to P-adrenergic agonists does involve decreased numbers of PAR (Lange-Carter et al., 1992). In addition, these tumour cells exhibited altered guanine nucleotide analogue binding to a 45 kDa membrane associated protein and enhanced cholera toxin responsiveness (Droms et al., 1989;Lange-Carter et al., 1992), suggesting the hypothesis that G-protein function might be altered in mouse lung tumours. As glucocorticoids enhance both PAR expression and PAR-G. coupling (Davies and Lefkowitz, 1984), both defects in the ,B-adrenergic signalling pathway could result from the loss of glucocorticoid responsiveness. Glucocorticoids may exert independent effects on the PAR and G. or, as even unoccupied PAR can influence G protein activity (Bond et al., 1995), the functional alteration in tumour cell G. may result directly from decreased PAR expression. Alternatively, the reduced GTPdependent coupling of receptors to adenylyl cyclase activation observed in hepatomas (Okamura and Terayama, 1976), suggests that independent defects in G. may also occur in tumours.
This study addresses the mechanisms of defective ,Badrenergic signal transduction in tumorigenic mouse alveolar type 2 cell lines and uncovers both glucocorticoid-dependent and independent mechanisms. Reduced ,BAR expression can be ameliorated by dexamethasone treatment of a tumorigenic cell line stably expressing a transfected glucocorticoid receptor gene. In response to dexamethasone, this transfected cell line exhibits the same number and affinities of PAR antagonist and agonist binding sites as non-tumorigenic cells. However, even though these tumour cells exhibit dexamethasone enhancement of isoproterenol-stimulated intracellular cAMP production, they remain much less isoproterenol responsive than non-tumorigenic cells. This relative insensitivity results from a loss of functional ,BAR-Gs coupling that may involve an intrinsic defect in the interaction of G, with guanine nucleotides.

Materials and methods Cell lines
The non-tumorigenic (C10) and tumorigenic (A5) type 2 cell lines were derived and cultured as described previously (Bentel et al., 1989;Droms et al., 1989Droms et al., , 1993. A5 cells were transfected by addition of media conditioned by VDG12P2 cells (supplied by Dr Gary Firestone, University of California, Berkeley, CA, USA). VDG12P2 cells release recombinant virus with the rat glucocorticoid receptor gene linked to neomycin resistance (Cook et al., 1988). After selection in 600 ,ug ml-1 geneticin (G418, Gibco, Grand Island, NY, USA), a single resistant colony remained. Cells from this colony were isolated and named A5GR1. A5GR1 cells express the transfected glucocorticoid receptor (Droms, 1995) and are routinely cultured with 200 Mg ml-' geneticin to maintain this expression. For all experiments, cells were plated on Corning tissue culture dishes in CMRL 1066 medium with 5% fetal bovine serum (FBS), 100 units ml-' penicillin, and 100 Mg ml-' streptomycin and allowed to attach overnight. On the following day cells were rinsed twice with 0.9% sodium chloride, and media containing 4% FBS from which endogenous steroids were removed with charcoal was added. At this time, 10 nM dexamethasone was added to the samples indicated in Results. Cells were cultured for an additional 4 days before experimental analyses.
Receptor binding studies Cells were plated at a density of 1.3 x 10 -5 x 105 cells per 100 mm culture dish. Membranes were prepared as described previously (Lange-Carter et al., 1992) and stored for up to 9 days at -80°C before use. For saturation binding studies 5-10 Mg of membrane protein per sample was incubated for 90 min at 37°C with (-)'25I-labelled cyanopindolol (125ICYP, 2200 Ci mmol-1; DuPont NEN, Boston, MA, USA) ranging in concentration between 10 and 200 pM as described previously (Lange-Carter et al., 1992). Protein concentration was determined by the Lowry method (Lowry et al., 1951). For isoproterenol competition studies, membranes were incubated with 30 pM 125ICYP and isoproterenol ranging in concentrations between 1 nm and 1 mM as described previously (Valverius et al., 1987). In some cases, GTP was included at the concentrations indicated. Membranes were harvested on glass fibre filters (Whatman GF/C) and counted in a gamma counter. Non-specific binding ranged between 27% and 32% for all conditions and was determined by adding 1 jiM of the unlabelled antagonist propranolol. Nonspecific binding has been subtracted from all data presented. Competition binding data were analysed with the Ligand program (Munson and Rodbard, 1983). cAMP radioimmunoassay CIO cells were plated at a density of 5 x 104 cells per 60 mm culture dish and A5 and A5GR1 cells at 105 cells per 60 mm dish. Cell lysates were harvested in 10% trichloroacetic acid (TCA) after 1 min treatment with 1 guM isoproterenol in buffer (130 mM sodium chloride, 5 mM potassium chloride, 1 mm calcium chloride, 1 mm magnesium sulphate, 1 mM potassium hydrogen phosphate, 6 mM glucose, 1 mM ascorbic acid, 25 mM Hepes, pH 7.4) at 37°C. After three extractions with diethyl ether, the TCA-soluble fraction was assayed for cAMP using a radioimmunoassay kit (Amersham, Arlington Heights, IL, USA). The TCA-insoluble fraction was solubilised with 0.2 N sodium hydroxide and assayed for protein by the Lowry method (Lowry et al., 1951).
Western blots Equal amounts of membrane protein (40 jig) from dexamethasone-treated ClO and A5GR1 cells were separated by 10% sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) as described previously (Droms et al., 1989). Resolved proteins were electrophoretically transferred to BA85 nitrocellulose (Schleicher and Schuell, Keene, NH, USA) in buffer containing 190 mM glycine, 20% methanol and 25 mM Tris, pH 8.3. Nitrocellulose membranes were subsequently incubated for 1.5 h in block buffer, consisting of 5% bovine serum albumin and 0.1% NP-40 in PBS (140 mM sodium chloride, 4 mM disodium hydrogen phosphate, 3 mM potassium chloride, 1.5 mM potassium hydrogen phosphate, 0.5 mM calcium chloride, 0.5 mM magnesium chloride, pH 7.4), rinsed 3 x 10 min in PBS and incubated for 12-16 h at 4°C with gentle rocking in block buffer with either a 1:16 000 dilution of an anti-G,Lx rabbit polyclonal antibody (UBI, Lake Placid, NY, USA) or with a 1: 40 000 dilution of anti-human GA common rabbit polyclonal antiserum (UBI). After rinsing 3 x 10 min in PBS, membranes were incubated  determined by saturation binding of the antagonist 1251CYP (Figure 1). Culturing the A5 cell line with dexamethasone did not increase 1251CYP binding ( Figure 1). However, when cultured with 10 nM dexamethasone, the A5GR1 cells exhibited similar numbers of,AR as CIO cells (Figure 1). There were no significant differences in kD or Bmax between CIO cells cultured without dexamethasone, ClO cells with 10 nM dexamethasone or A5GR1 cells with 10 nM dexamethasone [kD=23+2.6 (s.e.m.), 35+9.6 and 30+12 pM, P> 0.1; Bmax= 88 + 10, 91 + 11, 80 + 7.7 fmol per mg protein, P> 0.5; n =6, 4 and 5 respectively]. Statistical analyses were by one-way ANOVA. 12±4.0 370±23 Intracellular cAMP production after 1 min treatment with 1 uM isoproterenol (iso) was significantly (P<0.01) enhanced if CIO or A5GR1 cells were cultured for 4 days with 10 nM dexamethasone (dex), whereas A5 cell isoproterenol responsiveness was unaffected. Basal intracellular cAMP was also greater in CIO than in A5 or A5GR1 cells, but was not affected by dexamethasone treatment.
Numbers represent the mean ± s.e.m. cAMP produced (pmol cAMP per mg protein). Statistical analyses were by t-tests. n=6 for each condition. Dexamethasone-treated A5GRJ cells are much less isoproterenol responsive than CIO cells Intracellular cAMP production stimulation by the ,Badrenergic agonist isoproterenol did increase approximately 5-fold when A5GR1 cells were cultured with 10 nM dexamethasone, whereas no such increase was observed in A5 cells (Table I). However, although CIO and A5GR1 cells cultured with dexamethasone had similar number of 125ICYP binding sites, A5GR1 cells remained much less responsive to isoproterenol both in whole cells (Table I) and in crude membrane fractions (Table II). The dexamethasone-induced increase in ,B-adrenergic responsiveness in A5GR1 cells is likely to result from the increase in ,BAR expression. The approximate 30% increase in isoproterenol responsiveness in CIO cells cultured with dexamethasone compared with control CIO cells was not associated with increased PAR expression and involves enhanced coupling of the PAR to Gs (KA Droms, unpublished). CIO cells also exhibited a greater basal level of intracellular cAMP than did A5 or A5GR1 cells (Table I) and an increased basal adenylyl cyclase activity compared with A5GR1 cells (Table II). The increased basal adenylyl cyclase activity in CIO cells may be due to a difference in the activity of Gs between cell lines (see below).
CJO and A5GR1 cells have similar forskolin-stimulated adenylyl cyclase activities As both basal and isoproterenol-stimulated adenylyl cyclase activities are lower in ASGR1 than in CIO cells, a difference between cell lines in the activity of adenylyl cyclase is possible. However, when membranes prepared from CIO and ASGR1 cells cultured with dexamethasone were stimulated with 100 gM forskolin, which activates adenylyl cyclase directly (Downs and Aurbach, 1982), only a slight difference in adenylyl cyclase activity was observed ( Figure 2). This difference in forskolin-stimulated adenylyl cyclase activity is much less than the isoproterenol sensitivity difference between cell lines. In fact, as the basal activity is also lower in ASGR1 cells, the fold stimulation by forskolin is greater than in CIO membranes. dexamethasone do exhibit a slight difference in forskolinstimulated adenylyl cyclase activity, but this difference is much less than the difference between cell lines in isoproterenol sensitivity. Basal adenylyl cyclase activity is also lower in A5GR1 cells. Data from two assays are combined, n =6 for each condition. 2, Basal activity; *, + 100 gM forskolin. (lanes 2) cells were cultured with 10nM dexamethasone and membranes prepared as described in Materials and methods.
Representative blots probed with anti-Gsx or anti-G,B are shown.
No clear difference between cell lines in the amount of these subunits has been observed. Both the 45 kDa and 52 kDa molecular weight forms of Gsa are observed in CO0 and A5GR1 cells. For comparison, S49wt type cells, which express both G,a isoforms, and S49cyc-cells, which do not express G,a (Harris et al., 1985), are also shown. The blot shown is representative of at least three blots for each subunit.

CIO and A5GRI cells have similar levels of Goc and Gf subunit expression
In some cases, the relative levels of expression of G protein subunits can regulate receptor-G protein coupling (Blumer and Thorner, 1990). Therefore, G,cC and G3 subunit expression were examined on Western immunoblots ( Figure  3). For each subunit, three individual blots were quantified as indicated in Materials and methods and the signal from CIO cells was set to 1 (signals from the 45 kDa and 52 kDa forms of G,ca were combined). A5GRl cells did not differ significantly (P>0.5 for both subunits) from CIO for either G,aO (1.1 +0.31 s.e.m.) or Gfl (0.91 +0.33) expression. Thus, no clear differences between cell lines were observed in the amounts of any of the components of the ,BAR-coupled adenylyl cyclase system that might explain their differences in hormone responsiveness.
The PAR and Gs are functionally uncoupled in A5GRI cells When A5GR1 and CIO cells are cultured with dexamethasone they express similar amounts of,AR and G, and have similar forskolin-stimulated adenylyl cyclase activities, yet the A5GRl cells are much less responsive to isoproterenol. Therefore, the interactions between the components of the ,BAR-coupled adenylyl cyclase system were examined. PAR-G. coupling was analysed by isoproterenol competition for nucleotide. When 8pM GTP was included, CIO had only lowaffinity isoproterenol binding sites. In constrast, even 40/M GTP only minimally reduced isoproterenol affinity in A5GR1 membranes. Membranes were prepared from cells cultured with 10nM dexamethasone and incubated with 30pM 125ICYP and the indicated concentrations of unlabelled isoproterenol. Data were analysed using the Ligand program (Munson and Rodbard, 1983). Each point is the mean of duplicate determinations from one assay, whereas each line is the best fit from Ligand analysis of the combined data from at least three independent assays, except for A5GR1 -GTP for which two assays were performed (see Table I). Open symbols, minus GTP; closed symbols, + GTP (8pM for CIO, 40pM for A5GR1). 3 Membranes from cells cultured with 10 nM dexamethasone were incubated with 125ICYP, varying isoproterenol concentrations and the GTP concentrations indicated. The kDs for high (kH) and low (kL) affinity isoproterenol binding and the percentage of high-affinity sites (%RH) are reported. Numbers are the mean±s.e.m. of parameters estimated from Ligand analysis of individual assays. The number of individual assays for each condition is also reported (n). '25ICYP binding in the presence or absence of GTP. In such assays, fl-adrenergic receptors exhibit two affinity states for agonist binding: one of higher affinity (kH) and the other of lower affinity (kL). As demonstrated in Figure 4 and Table  III, both CIO and A5GRl cells exhibit high-and low-affinity isproterenol binding in the absence of GTP. The presence of high-affinity sites in both cell lines indicates that there is fAR-Gs 'coupling' in both cell lines (De Lean et al., 1980). However, inclusion of GTP in the incubations is much more effective at reducing high-affinity isoproterenol binding in CIO than in A5GR1 cells, indicating greater functional PiAR-Gs coupling in the former (Rodbell et al., 1971;Cheung et al., 1989). It is also interesting to note that kL is GTP produced a concentration-dependent activation of adenylyl cyclase in membranes prepared from C10 cells that were cultured with dexamethasone, but was completely ineffective in A5GR1 cells cultured with dexamethasone. (b) The non-hydrolysable analogue Gpp(NH)p activated adenylyl cyclase in both cell lines, but to a lesser extent in A5GR1 than in CIO cells. Stimulation of adenylyl cyclase by Gpp(NH)p in A5GR1 cells that were not cultured with dexamethasone and, thus expressed very few ,BAR, was the same as for A5GR1 cells cultured with dexamethasone. Points represent the combined data from three independent assays for which triplicate determinations were done, with the exception of A5GR1 without dexamethasone, for which one assay was performed. 0, CIO; *, A5GRI; open symbols, minus dexamethasone; closed symbols, + 10 nM dexamethasone. much greater in CIO than in A5GR1 cells in the absence of GTP (Table III). This high kL is typically observed when flAR-Gs coupling is enhanced by glucocorticoid treatment (Davies and Lefkowitz, 1981), and is consistent with the increased isoproterenol responsiveness that was observed in CIO cells cultured with dexamethasone (Table I). The kL for CIO cells (140+22 nM) that were not cultured with dexamethasone is similar to that for A5GR1 cells (230 + 20 nM).
Guanine nucleotides stimulate adenylyl cyclase more effectively in C10 than in A5GRI cells Neither GTP nor the non-hydrolysable analogue 5-guanylyimidio-diphosphate [Gpp(NH)p] activated adenylyl cyclase as effectively in A5GR1 cell membranes as in ClO membranes ( Figure 5). In fact, GTP at concentrations as high as 100 gM produced no stimulation of adenylyl cyclase above basal in A5GR1 cells. Although Gpp(NH)p was also much less effective in A5GT1 than in ClO cells, 15 gM Gpp(NH)p did stimulate adenylyl cyclase approximately 3-fold above basal activity in the former. As adenylyl cyclase was stimulated by Gpp(NH)p to a similar extent in A5GR1 cells that either were or were not cultured with dexamethasone ( Figure 5), a defect in Gs that is independent of PAR expression is implicated.

Discussion
Mouse lung tumour cells have considerably reduced sensitivity to P-adrenergic stimulation (Droms et al., 1989;Lange-Carter et al., 1992) and this loss of sensitivity involves reduced PAR expression (Lange-Carter et al., 1992). The present work indicates that glucocorticoid treatment of A5GR1 cells, a tumour cell line stably expressing a glucocorticoid receptor transgene, does allow restoration of PAR expression to a level similar to that observed in the non-tumorigenic ClO cells. The newly expressed PAR is capable of high-affinity ternary complex formation with agonist and G, as indicated by the observation that 60% of isoproterenol binding sites are high affinity (Table III). Reduced tumour cell ,BAR expression is unlikely to result directly from loss of glucocorticoid stimulation, as CIO cells retain a high level of PAR expression even in the absence of glucocorticoids in the culture medium. Thus, although glucocorticoid treatment does restore PAR expression in A5GR1 cells, the cause of reduced expression in tumour cells is unknown. Mouse lung tumour cells also exhibited reduced guanine nucleotide analogue binding to a 45 kDa membraneassociated protein and enhanced cholera toxin responsiveness (Droms et al., 1989;Lange-Carter et al., 1992), suggesting the hypothesis that G protein function is altered in the mouse lung tumours. A functional alteration in tumour cell Gs could result from the reduced PAR expression rather than from a direct defect in G, itself, as even unoccupied ,BAR can influence G protein activity (Bond et al., 1995). However, the present study indicates that functional flAR-G, coupling is considerably reduced in A5GR1 cells even when these cells express as many PAR as CIO cells. This lack of functional PAR-G. coupling in A5GR1 cells is indicated by a relative inability of GTP to destablise the high-affinity ternary complex and is associated with decreased effectiveness of guanine nucleotides to activate adenylyl cyclase. As A5GR1 cells exhibit a decrease in non-hydrolysable guanine nucleotide activation of adenylyl cyclase that is independent of glucocorticoid treatment and, consequently, ,BAR expression, an independent defect in tumour cell Gs is implicated.
One potential mechanism of the reduced ability of guanine nucleotides to activate G, in tumorigenic mouse lung-derived cell lines is that guanine nucleotide exchange is reduced (Droms et al., 1989). The relative inability of GTP to 0 ll 437 destabalise the ternary complex in A5GRl cells is consistent with this hypothesis. Additionally, the difference between CIO and A5GR1 cells in the effectiveness of GTP at activating adenylyl cyclase is even greater than the difference between cell lines when a non-hydrolysable analogue is used. Thus, the GTPase activity of Gs may also be enhanced in the tumour cells. A reduction in the intrinsic guanine nucleotide exchange rate with an increase in the GTPase activity of Gs in tumour cells would be predicted to produce a decrease in basal adenylyl cyclase activity, consistent with the 2to 3-fold lower basal intracellular cAMP and adenylyl cyclase activity observed in A5 and A5GRl cells compared with CIO cells.
Although a clear role for reduced intracellular cAMP in neoplasia has not been established, agonist-stimulated adenylyl cyclase activity is reduced in many tumours. One possibility is that cAMP interferes with mitogenic signal transduction, as observed in Ratl cells (Cook and McCormick, 1993) and human small-cell lung cancer cells (Viallet et al., 1990). Reduced activity of the cAMPdependent protein kinase is also a requirement for mitogenesis in some cells (Lamb et al., 1991). In addition, cAMP inhibits growth in soft agar of mouse lung tumour cell lines (KA Droms, unpublished observation). Tumourassociated reductions in cAMP often result from decreases in the number or affinity of receptors coupled to adenylyl cyclase activation (Hunt and Martin, 1980). Alternatively, defects in the G-proteins that couple receptors to adenylyl cyclase may also occur. For example, reduced GTP-dependent coupling of adrenergic receptors to adenylyl cyclase had been observed in hepatomas (Okamura and Terayama, 1976). As there are differences between ClO and A5GRl cells in G, interactions with both the ,BAR (guanine nucleotide destabilisation of the ternary complex) and adenylyl cyclase (guanine nucleotide activation of cAMP production), the structure of one or more of the subunits of G, may be altered in mouse lung tumour cells. This possibility is currently being investigated.