Bombesin and substance P analogues differentially regulate G-protein coupling to the bombesin receptor. Direct evidence for biased agonism.

Substance P analogues including [d-Arg1,d-Phe5,d-Trp7,9,Leu11]substance P (SpD) act as "broad spectrum neuropeptide antagonists" and are potential anticancer agents that inhibit the growth of small cell lung cancer cells in vitro and in vivo. However, their mechanism of action is controversial and not fully understood. Although these compounds block bombesin-induced mitogenesis and signal transduction, they also have agonist activity. The mechanism underlying this agonist activity was examined. SpD binds to the ligand-binding site of the bombesin/gastrin-releasing peptide receptor and blocks the bombesin-stimulated increase in [Ca2+]i within the same concentration range that causes sustained activation of c-Jun N-terminal kinase and extracellular signal-regulated protein kinase (ERK). The activation of c-Jun N-terminal kinase by SpD and bombesin is blocked by dominant negative inhibition of G(alpha12). The ERK activation by SpD is pertussis toxin-sensitive in contrast to ERK activation by bombesin, which is pertussis toxin-insensitive but dependent on epidermal growth factor receptor phosphorylation. SpD does not simply act as a partial agonist but differentially modulates the activation of the G-proteins G(alpha12), G(i), and G(q) compared with bombesin. This unique ability allows the bombesin receptor to couple to G(i) and at the same time block receptor activation of G(q). Our results provide direct evidence that SpD is acting as a "biased agonist" and that this has physiological relevance in small cell lung cancer cells. This validation of the concept of biased agonism has important implications in the development of novel pharmacological agents to dissect receptor-mediated signal transduction and of highly selective drugs to treat human disease.

Neuropeptides have been implicated in the pathogenesis of a number of human disease states including inflammatory disease, cardiovascular disease, and cancer (1). Neuropeptides including bombesin are autocrine growth factors for a number of cancers including breast, prostate, and small cell lung cancer (SCLC) 1 (2)(3)(4)(5). In particular, neuropeptides and their receptors are the principle driving force behind one of the most clinically aggressive cancers, SCLC. SCLC cells sustain their growth in part as a result of multiple autocrine and paracrine loops involving calcium-mobilizing neuropeptides (6,7). SCLC provides a paradigm for the investigation of neuropeptide-mediated growth. Modulating neuropeptide-induced signal transduction may therefore have important implications in the treatment of a number of human diseases.
Neuropeptides are a structurally diverse group of hormones and neurotransmitters that bind to a related subfamily of Gprotein-coupled receptors. Predominately, these receptors couple to G q to elicit phospholipase C-␤ activation and subsequent production of diacylglycerol and phosphatidylinositol 1,4,5trisphosphate leading to protein kinase C activation and Ca 2ϩ release (8 -10). These receptors also couple to G 12 to elicit c-Jun N-terminal kinase (JNK) activation and Rho-dependent activation of stress fiber formation via tyrosine phosphorylation of a number of tyrosine kinases including FAK, paxillin, and p130 cas (11)(12)(13)(14). Bombesin and other neuropeptides have also been shown to activate the ERK pathway leading to the stimulation of immediate early genes and proliferation (10,15,16).
Analogues of substance P, [D-Arg 1 ,D-Phe 5 ,D-Trp 7,9 ,Leu 11 ]substance P (SpD) and [Arg 6 ,D-Trp 7,9 ,N me Phe 8 ]substance P (6 -11) can inhibit neuropeptide-stimulated Ca 2ϩ mobilization, tyrosine phosphorylation, and ERK activation (17)(18)(19). Crucially, SpD and [Arg 6 ,D-Trp 7,9 ,N me Phe 8 ]substance P inhibit SCLC cell growth in vivo and in vitro (7,20) and stimulate SCLC cell apoptosis (21,22). Substance P analogues are about to enter phase II clinical investigation for the treatment of SCLC and could provide a novel form of therapy for other neuroendocrine tumors in addition to SCLC (23). Hence understanding the mechanism of action of this class of compound is attracting considerable interest and is critical for future drug development.
Substance P analogues were characterized originally as "broad spectrum neuropeptide antagonists" (7,(17)(18)(19). However, the precise mechanism of action of these compounds seems more complex and remains unclear and contentious (23). Studies in Swiss 3T3 cells suggested that substance P analogues competitively inhibit the binding of neuropeptides to their receptors, accounting for the inhibition of neuropeptidestimulated signal transduction (17,18). Substance P analogues were thought to inhibit SCLC cell growth by competitively inhibiting the mitogenic effects of autocrine neuropeptides (17)(18)(19)(20). However, the SCLC cell growth inhibitory and proapoptotic activities of substance P analogues are not reversed by supramaximal saturating concentrations of neuropeptides (21). Furthermore, substance P analogues themselves activate JNK and potentiate bradykinin-induced edema formation in rabbit skin (21,23,24). Further studies have shown that although SpD irreversibly blocks bombesin-induced phospholipase C-␤ activation and mitogenesis and reversibly inhibits ERK activation by bombesin (25), SpD augments Raf-1 and ERK activation by high concentrations of bombesin (25). This suggests that the GRP receptor may still be capable of signaling even when bombesin-induced phospholipase C activation is fully blocked. Thus in addition to its well described antagonist activity, SpD also has agonist activity. The ability of a receptor to have more than one active state and interact with multiple G-proteins to produce cellular responses has been described in a variety of receptor systems and has been termed "agonist-receptor trafficking" (26,27). Jarpe et al. (24) hypothesized that SpD may act as an agonist at GRP receptors, activating the G 12 family of guanine-nucleotide binding proteins while blocking signal transduction via G q and proposed a novel pharmacological term, "biased agonism," to describe this. Agents acting by the mechanism of biased agonism that selectively activate G-proteins would have enormous pharmacological and clinical importance and could become valuable tools in the dissection of signal transduction pathways downstream of receptors. However, the hypothesis of biased agonism was challenged by Sinnett-Smith et al. (28), who concluded that substance P analogues acted primarily as antagonists of neuropeptide receptors, blocking signal transduction via both G 12 and G q , and that any agonist activity was caused by partial agonism.
The experiments presented here were designed to examine the mechanism underlying the agonist activity of SpD and to investigate specifically the validity of the biased agonist hypothesis.
Measurement of Intracellular Calcium-[Ca 2ϩ ] i was measured using the fluorescent Ca 2ϩ probe Fura-2-AME as described previously (27). Briefly, cells in 90-mm dishes were made quiescent overnight in DMEM containing 0.1% (v/v) FCS, washed once, and trypsinized. The cells were washed and suspended in calcium-free Hanks' modified electrolyte solution and incubated with 2 M Fura-2-AME for 10 min at 37°C. The cells were suspended in Hanks' modified electrolyte solution containing calcium and transferred to a cuvette maintained at 37°C. Agonists were added as described in the figure legends. Ratiometric fluorescence was monitored in a PerkinElmer fluorometric spectrophotometer with dual excitation wavelengths of 340 and 380 nm and an emission wavelength of 510 nm. The [Ca 2ϩ ] i was calculated according to the equation where F is the ratio of the unknown sample, F max is the ratio after the addition of 0.2% Triton X-100, and F min is the ratio after Ca 2ϩ chelation with 10 mM EGTA. K is the dissociation constant for Fura-2, which is 224 nM (29).
Measurement of ERK Activity-Quiescent cell cultures were treated as described in the figure legends and lysed at 4°C in 0.5 ml of lysis buffer containing 25 mM HEPES, pH 7.4, 0.3 M NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5% Triton X-100, 20 mM ␤-glycerophosphate, 0.5 mM dithiothreitol, 1 mM sodium orthovanadate, and protease inhibitors (the protease inhibitor mixture was from Roche Molecular Biochemicals and prepared as per manufacturer instructions). The lysates were clarified by centrifugation, equilibrated for protein, and ERK 2 -immunoprecipitated using 3 g of anti-ERK 2 polyclonal antibody coupled to Sepharose (Santa Cruz Biotechnology). An aliquot of lysate was retained and boiled in Laemmli SDS-PAGE loading buffer for future Western analysis. Immune complexes from cell lysates containing 400 g of protein were washed three times at 4°C in 20 mM HEPES, pH 7.4, containing 50 mM NaCl, 2.5 mM MgCl 2 , and 0.1 mM EDTA by repeated centrifugation and once at 4°C in kinase buffer (20 mM HEPES, pH 7.4, containing 0.5 mM NaF, 7.5 mM MgCl 2 , 0.2 mM EGTA, 2 mM dithiothreitol, 10 mM ␤-glycerophosphate, and 0.5 mM sodium orthovanadate). Kinase activity was estimated in 25 l of kinase buffer containing 100 M ATP, 1 Ci of [␥-33 P]ATP (3000 Ci/mmol), and 10 g of myelin basic protein. The reaction was carried out for 20 min at 30°C and terminated by spotting the supernatant onto P81 phosphocellulose paper. The papers were washed three times in 0.5% (v/v) phosphoric acid and dried briefly in acetone. The results are expressed as specific disintegrations/min bound or -fold increase over basal.
Measurement of JNK Activity-Cell lysates were generated as described for ERK 2 . After immunoprecipitation of JNK 1 from whole-cell lysates, kinase activity was carried out as described (21) using 20 M ATP, 1 Ci of [ 32 P]ATP (3000 Ci/mmol), and 1 g of glutathione Stransferase-c-Jun (79) substrate. Phosphorylated c-Jun was identified from autoradiographs of Coomassie Blue-stained SDS-PAGE gels and quantified by phosphorimaging. HA-JNK 1 was precipitated from transiently transfected COS-7 cells using 2 g of anti-HA antibody coupled to Sepharose (Santa Cruz Biotechnology). JNK activity was determined as described above.
Western Blotting-Whole-cell lysates were normalized for protein concentration (typically 1.0 -2.0 mg/ml) and denatured by boiling (5 min) in SDS-PAGE loading buffer. 20 l of lysate/lane was resolved on 12% SDS-PAGE gels and electroblotted onto nitrocellulose membranes. The membranes were blocked in 5% nonfat milk in PBS containing 0.05% Tween 20. ERK 1/2 and JNK 1/2 phosphorylation was determined using 1:1000 dilution of the primary antibody followed by the appropriate horseradish peroxidase-labeled goat IgG (DAKO) diluted 1:5000. Bands were visualized using enhanced chemiluminescence (ECL, Amersham Pharmacia Biotech).
[ 125 I]-GRP Receptor Binding-[ 125 I]-GRP receptor binding was carried out in confluent and quiescent cultures of BOR-15 cells exactly as described (10). The binding parameters K d and B max were calculated according to the method described by Scatchard et al. (30). The IC 50 (concentration of drug displacing 50% specific binding) was converted to the inhibitory constant

SpD Blocks Bombesin-induced Ca 2ϩ
Mobilization-The mechanism underlying the agonist activity of SpD is unclear and controversial. SpD can bind to multiple characterized and uncharacterized G-protein-linked receptors. To define precisely the mechanism underlying the agonist activity of SpD, we used a rat-1a fibroblast model system (into which the bombesin receptor is transfected) to mediate SpD activity. Bombesin and other Ca 2ϩ -mobilizing neuropeptides (vasopressin, neurotensin, bradykinin, and gastrin), fail to mobilize intracellular Ca 2ϩ in native rat-1a fibroblasts, suggesting the absence of these receptors in this cell line (Fig. 1A). In rat-1a cells stably expressing the bombesin/GRP receptor (BOR-15), bombesin induces a marked and rapid increase in [Ca 2ϩ ] i , which was blocked by SpD (Fig. 1B). In addition, SpD at concentrations as high as 100 M failed to mobilize intracellular calcium in rat-1a or BOR-15 cells (Fig. 1C). In BOR-15 cells, bombesin induced a concentration-dependent increase in [Ca 2ϩ ] i with EC 50 ϭ 1.3 Ϯ 0.2 nM. These results are similar to those seen for bombesin stimulation of [Ca 2ϩ ] i in Swiss 3T3 cells (7,17). SpD (30 M) shifted the bombesin concentration response curve to the right (Fig. 1D, inset) with a K i for the inhibition of bombesin-stimulated Ca 2ϩ flux of 6.4 Ϯ 0.7 M. These results are similar to those seen for SpD inhibition of both bombesin-induced stimulation of [Ca 2ϩ ] i and DNA synthesis in Swiss 3T3 cells 2 (7,17) and for the inhibition of [ 125 I]-GRP receptor binding (Fig. 5).
SpD Stimulates ERK at Concentrations That Inhibit Bomb-esin-stimulated Ca 2ϩ Release-The addition of SpD for 10 min causes a marked concentration-dependent increase in ERK activity in quiescent rat-1a fibroblasts expressing the bombesin receptor (BOR-15 cells), in contrast to the untransfected rat-1a cells (Fig. 2). In BOR-15 cells, bombesin stimulation of ERK activity was seen in the nM range, maximal at 30 nM (EC 50 ϭ 5.9 Ϯ 1.8 nM, n ϭ 4). SpD-induced stimulation of ERK activity in BOR-15 cells was evident at 3 M, maximal at 10 M (EC 50 ϭ 4.19 Ϯ 0.6 M, n ϭ 3) (Fig. 2). To ensure that this effect was not a result of clonal selection, cells were transiently transfected with the mouse bombesin receptor, and similar results were obtained (results not shown). The stimulation of ERK occurred within the same concentration range at which SpD inhibits bombesin-stimulated Ca 2ϩ release. The time course of ERK activation by SpD is different from that of bombesin (Fig. 3). ERK 1/2 phosphorylation as an indication of ERK activity was assessed by immunoblotting with a phosphorylation state-specific monoclonal antibody to ERK 1/2 and showed quantitatively similar results to that of the immunoprecipitation kinase assay (data not shown). SpD-induced ERK 1/2 phosphorylation was evident at 5 min, reached a maximum at 10 min, and persisted for over 60 min. The time course of ERK 1/2 activation by SpD was slower in onset and more sustained in comparison with that of bombesin, which returned to control levels by 30 min (Fig. 3). This suggests that SpD and bombesin differentially regulate signal transduction pathways leading to ERK activation. , and lysates were prepared as described under "Experimental Procedures." Immunoprecipitated ERK 2 was assayed by the phosphorylation of myelin basic protein incorporated onto p81 phosphocellulose paper. Activity is expressed as disintegrations/min bound and represents the mean Ϯ S.E. of at least four independent experiments performed in duplicate. Basal ERK 2 activity was 100 Ϯ 30 dpm bound for rat-1a cells and 350 Ϯ 110 dpm bound for BOR-15 cells.

SpD Acts at the Ligand-binding Site of the Bombesin Receptor to Induce ERK Activation-SpD-and bombesin-induced ERK activity in BOR-15 cells was inhibited by two specific
critical for high affinity bombesin binding and that the R288H mutation does not affect GRP receptor expression or conformation and otherwise the R288H mutant GRP-receptor is fully functional (33,34). 5ET4 or R288H cells were stimulated with bombesin (3 nM) or SpD (0.3-30 M) for 10 min. Bombesin (3 nM) caused a 2.6-fold stimulation of ERK in 5ET4 cells. However, despite good expression of the R288H mutant bombesin receptor in Balb 3T3 cells (Ref. 30 and results not shown), the mutated receptor was no longer able to respond to bombesin (Fig. 4B). SpD (3 M) induced a 2.5-fold stimulation of ERK activity in 5ET4 cells. Crucially, SpD was also no longer able to stimulate an increase in ERK 1/2 phosphorylation or ERK 2 activity in cells expressing the R288H mutant, although these cells are still able to respond to lysophosphatidic acid (results not shown). This confirms that SpD acts at the agonist-binding domain of the GRP receptor to stimulate ERK activation but does not mobilize Ca 2ϩ .
GRP Receptor Desensitization-Previous experiments have shown that GRP receptors undergo desensitization upon exposure to bombesin (35) and that differences in receptor desensitization and down-regulation can result in altered agonist responses. Given the differences in time course for bombesin-and SpD-induced ERK activation we determined whether this could be explained by differential GRP receptor desensitization.  (Fig. 5A), which is in good agreement with its ability to activate ERK (Fig. 2) and JNK (Fig. 6) and confirms that SpD acts via the GRP receptor. [ 125 I]-GRP binding was measured after exposure to maximal inhibitory concentrations of bombesin (30 nM) or SpD (30 M, Fig. 5B). As expected, pre-exposure to bombesin produced a decrease in receptor binding with a 54 Ϯ 10% reduction after 2 h. In contrast, SpD produced only a 15 Ϯ 7% reduction, which was nonsignificant over this time scale. SpD therefore does not induce receptor desensitization to the same extent as bombesin. This will have important implications for ERK and JNK activation.
SpD-induced JNK Stimulation Is Blocked by Dominant Negative G ␣12 -We and Jarpe et al. (24) have shown previously that substance P analogues activate JNK (21,25). Evidence suggests that JNK activation may be mediated by members of the G 12 family of G-proteins such as G ␣12 and G ␣13 (36, 37).
Hence it was hypothesized that SpD acted as an agonist for the G 12 family of G-proteins. However, Sinnett-Smith et al (28) recently showed that substance P analogues block bombesinstimulated assembly of focal adhesion and actin stress fiber formation in Swiss 3T3 cells. G ␣12 and G ␣13 activation can induce these Rho-mediated events, suggesting that SpD acts as an antagonist for members of the G 12 family of G-proteins (28). We therefore went on to examine the effect of directly blocking G ␣12 and G ␣13 on SpD activation of JNK. We used G ␣12 and G ␣13 dominant negative constructs that have been shown to block G ␣12 -and G ␣13 -mediated stress fiber formation in fibroblasts and to have no effect on G q -or G i -mediated events (14,38). The constructs contain a G3 A substitution in the nucleotide-binding pocket, which blocks activation. COS-7 cells, which express endogenous GRP receptors, were transiently transfected with HA-JNK 1 . After 48 h, the cells were washed, stimulated with peptide for 5 min, and lysed. JNK activity was determined from anti-HA precipitates. SpD caused a concentration-dependent stimulation of JNK activity in COS-7 cells (Fig. 6A). This was evident at 0.3 M, maximal at 30 M with  (Fig. 6, B and C) but not G ␣13 (Fig. 6B), and thus SpD activates JNK via a stimulation of G 12 . Measurement of ERK activity, assessed by Western blot analysis of phosphorylated ERK 1/2 , showed that as expected, concentration response curves to bombesin-and SpD-induced ERK activation was completely unaffected by the expression of the dominant negative G ␣12 (Fig. 6C). This suggests that G ␣12 specifically inhibits JNK activation, which is not caused by a general inhibition of receptor function by G ␣ subunits. These results show that within the same concentration range that inhibits a bombesin-induced increase in [Ca 2ϩ ] i and activates ERK, SpD stimulates G ␣12 -dependent JNK activation.
SpD-induced ERK Activation Is Inhibited by Pertussis Toxin-We examined the G-protein dependence of SpD-induced ERK activation. Fig. 7 shows that ERK stimulation by SpD in BOR-15 cells is inhibited by 24-h pretreatment with pertussis toxin (100 ng/ml). Pertussis toxin, however, had no effect on bombesin-stimulated ERK in accordance with previously reported data (17,39). These results suggest that unlike bombesin, SpD-induced ERK activation occurs via an activation of G-proteins of the G i/o subtype. SpD therefore has the unique ability to inhibit responses mediated by G q (Ca 2ϩ mobilization) within the same concentration range that activates G i -mediated responses.
It has been reported that many G-protein-linked receptors such as bradykinin receptors activate ERK via the transactivation of growth factor receptors such as the epidermal growth factor (EGF) receptor (40,41). Fig. 8 shows that bombesininduced activation of ERK in BOR-15 cells was blocked completely by prior treatment for 30 min with the EGF receptor kinase inhibitor AG1478, whereas SpD-induced activation was completely insensitive. These results suggest that bombesin stimulates ERK activation via transactivation of the EGF receptor in contrast to SpD. This confirms that bombesin and SpD differentially modulate G-protein signal transduction via the bombesin receptor.

FIG. 6. SpD-induced JNK activation is inhibited by a dominant negative G ␣12 .
A, COS-7 cells were grown to 80% confluency in 100-mm dishes and serum-starved (0.1% FCS) for 24 h prior to experimentation. Cells were stimulated with SpD for 15 min at 37°C. Lysates were prepared, and JNK 1 was assayed from immunoprecipitated JNK 1 phosphorylation of 1 g of glutathione S-transferase-c-Jun(79). Phosphorylated proteins were resolved by SDS-PAGE, and incorporated radioactivity was measured by phosphorimaging. The results represent the mean Ϯ S.E. of three independent experiments. A representative autoradiograph is shown. B, COS-7 cells were grown to 70% confluency and transfected with the pCIS vector containing lacZ (control), the dominant negative G ␣12 (G228A), or the dominant negative G ␣13 (G225A) and HA-tagged JNK 1 . COS-7 cells were stimulated with SpD (30 M) or bombesin (3 nM). HA-JNK 1 was precipitated with an anti-HA antibody, and kinase activity was measured by phosphorylation of glutathione S-transferase-c-Jun (79) and phosphorimaging. The results are expressed as the -fold increase over unstimulated cells and represent the mean Ϯ S.E. of at least four independent experiments performed in duplicate. *, p Ͻ 0.05, analysis of variance. C, aliquots of cell lysate transfected with the dominant negative G ␣12 or lacZ and stimulated with bombesin or SpD were resolved by SDS-PAGE and transferred onto nitrocellulose. The blots were probed with a monoclonal antibody raised to the dually phosphorylated form of JNK 1 or ERK 1/2 . The results are representative of at least three independent experiments.
SpD Acts as a Biased Agonist in SCLC Cells-Substance P analogues have been developed for use as anticancer agents in SCLC. We therefore examined whether biased agonism has functional relevance in SCLC cells. Bombesin/GRP has been shown to be a principal autocrine growth factor for the SCLC cell line H345, with bombesin receptor blockade by the antagonist [Leu 13 (CH 2 NH)-Leu 14 ]bombesin or the monoclonal antibody 2A11 inhibiting growth in vitro and in vivo (3)(4)(5)42). In the H345 cells, bombesin induced a marked and rapid increase in [Ca 2ϩ ] i , which was blocked by SpD (Fig. 9A). In addition, SpD at concentrations as high as 100 M failed to mobilize intracellular calcium in SCLC cells (data not shown). Bombesin induced a concentration-dependent increase in [Ca 2ϩ ] i in H345 cells, with EC 50 ϭ 9.2 Ϯ 4.2 nM (n ϭ 5). SpD shifted the [Ca 2ϩ ] i response curve to bombesin to the right (Fig. 9A, insert). The inhibitory effect of SpD on bombesin (10 nM) is seen first at 3 M, maximal at 100 M, with IC 50 ϭ 12.7 Ϯ 2.5 M (Fig. 9A). The addition of SpD inhibited the growth of H345 SCLC cells in liquid culture. Growth inhibitory effects were seen first at 3 M, maximal at 50 M with IC 50 ϭ 29.5 Ϯ 5.5 (Fig. 9B). This is in keeping with previously published results (7,17). We further show that within the concentration range in which SpD inhibits bombesin-stimulated increase in [Ca 2ϩ ] i and growth in H345 SCLC cells, SpD stimulates ERK (Fig. 9C) and JNK activation (Fig. 9D) (EC 50 ϭ 3.9 Ϯ 1.7 M and 1.2 Ϯ 2.3, respectively) without increasing [Ca 2ϩ ] i . Thus in the physiologically relevant in vitro cell system, SpD exhibits both antagonist and agonist activity at the same concentrations, consistent with biased agonism. C, the effect of SpD on ERK activity. H345 SCLC cells were quiesced overnight in serum-free medium and washed prior to the assay. Cells were plated at 5 ϫ 10 6 /ml in RPMI 1640 medium and stimulated with SpD (1-100 M). Immunoprecipitated ERK 2 activity is expressed as a percentage of maximal response to SpD and represents the mean Ϯ S.E. of three independent experiments performed in duplicate. A representative Western blot of two independent experiments of phoshpo-ERK 1/2 in the presence of SpD is shown. D, the effects of SpD on JNK activity. H345 SCLC cells were quiesced overnight in serum-free medium and washed prior to assay. Cells were plated at 1 ϫ 10 7 /ml in RPMI 1640 medium and stimulated with SpD (1-100 M). Immunoprecipitated JNK 1 activity is expressed as a percentage of maximal response to SpD and represents the mean Ϯ S.E. of three independent experiments performed in duplicate. A representative autoradiogram of three independent experiments is shown.

DISCUSSION
The novel findings in this paper are: 1) SpD binds to the ligand-binding site of the bombesin/GRP receptor and this binding critically depends on amino acid Arg-288. 2) SpD increases ERK and JNK activity via the activation of the bombesin/GRP receptor within the same concentration range that blocks bombesin receptor-mediated increases in [Ca 2ϩ ] i . 3) The activation of JNK by SpD and bombesin depends on G 12 activation. 4) SpD unlike bombesin does not cause GRP receptor desensitization. 5) ERK activation by SpD is pertussis toxinsensitive unlike ERK activation by bombesin, which is pertussis toxin-insensitive but depends on EGF receptor tyrosine kinase activity.
SpD is a decapeptide analogue of substance P. Compounds of this type (e.g. [Arg 6 , D-Trp 7, 9 , N me Phe 8 ]substance P (6 -11) and [D-Arg 1 ,D-Trp 5,7,9 ,Leu 11 ]substance P) have been shown to inhibit the growth of SCLC cells in vitro and in vivo and stimulate SCLC cell apoptosis (22,23). Traditionally, substance P analogues were thought to inhibit SCLC cell growth by competitively inhibiting the effects of autocrine and paracrine mitogenic neuropeptides (7,(17)(18)(19)(20). However, substance P analogues potentiate bradykinin-induced edema formation in rabbit skin (21). This suggests that physiologically these compounds are not simply acting as competitive neuropeptide antagonists. Our data demonstrate that SpD has the unique ability to allow the bombesin receptor to couple to G i and G 12 , leading to subsequent ERK and JNK activation, respectively, while at the same time blocking receptor activation of G qmediated calcium release. SpD is the first compound shown to be capable of activating and inactivating different arms of the signal transduction pathways activated by a single receptor. Thus SpD is acting as a biased agonist at the GRP receptor and our results provide a formal validation of this novel pharmacological term.
Studies with synthetic bombesin-like peptides have demonstrated that full biological activity requires more than seven but no more than nine N-terminal amino acids (WAVGHLM-N); on this basis it is proposed that these N-terminal residues interact with the bombesin receptor ligand-binding pocket defined by residues Gln-121, Arg-288, Ala-308, and Pro-199 (33). The R288H GRP receptor mutant shows a 1000-fold reduction in GRP/bombesin affinity relative to the wild-type mouse GRP receptor (33). SpD-induced ERK activity was inhibited by two specific GRP receptor antagonists and was abrogated in cells expressing the mutant (R288H) GRP receptor. Thus the biased agonist activity of SpD is mediated via binding to the ligandbinding site of the bombesin/GRP receptor. Substitution of D-Phe for Gln at position 5 in SpD results in a complete loss of activity against the bombesin receptor (result not shown). Hence a large nonpolar hydrophobic residue D-Phe in SpD and Trp in bombesin at a position of seven amino acids from the N terminus may be critical for optimal binding to the ligandbinding pocket.
We show that despite binding to the agonist-binding domain of the bombesin/GRP receptor, SpD is unable to produce a Ca 2ϩ response but can inhibit bombesin-induced Ca 2ϩ mobilization, which is mediated via G q -dependent stimulation of phospholipase C-␤. Bombesin on the other hand stimulates JNK and ERK and increases [Ca 2ϩ ] i . However, SpD-induced ERK stimulation is prolonged compared with bombesin. These kinetics are similar to that observed for SpD-induced activation of JNK, which also has a protracted time course (21,24). This suggests that SpD, upon binding to the ligand-binding site, differentially regulates signal transduction pathways downstream from the bombesin/GRP receptor compared with the natural agonist. The mechanisms for this are unclear, but desensitization stud-ies showed that SpD is much less efficient at desensitizing the receptor compared with bombesin. This has important functional implications. It could suggest that bombesin cannot fully activate all possible receptor-G-protein interactions because it causes rapid desensitization. Previous evidence suggests that GRP receptors cause rapid ligand degradation, which can account for much of the observed desensitization caused by bombesin (35). SpD therefore may stabilize an active conformation not activated by bombesin (e.g. G i ) because of a longer-lived association with the receptor.
Evidence suggests that JNK activation may be mediated by members of the G 12 family of G-proteins such as G 12 and G 13 . Hence it was hypothesized that SpD acts as an agonist for the G 12 family of G-proteins (25,36,37). Recently, Sinnett-Smith et al. showed that substance P analogues block bombesin-stimulated assembly of focal adhesions and actin stress fiber formation and only showed activation of these responses at high concentrations, suggesting that they were acting as partial agonists (28). G ␣12 and G ␣13 activation can induce these Rhomediated events, suggesting that SpD acts as an antagonist for members of the G 12 family of G-proteins. However, Jarpe et al. (24) showed that substance P analogues could stimulate the assembly of focal adhesion and an increase in actin stress fibers in Swiss 3T3 cells. We show that in the same samples, SpD activates JNK and ERK at the same concentrations that inhibit bombesin-stimulated Ca 2ϩ mobilization (EC 50 for JNK and ERK activation is 4.2 and 3.2 M, respectively, and IC 50 for inhibition of bombesin-induced Ca 2ϩ mobilization ϭ 3.7 M). These data would not be consistent with partial agonism but suggest that SpD can induce a conformational state that favors coupling and activation of some but not all G-protein subtypes. We examined the effect of directly blocking G ␣12 and G ␣13 on SpD activation of JNK. Both bombesin and SpD activation of JNK is blocked by dominant negative G ␣12 but not by a dominant negative G ␣13 . Neither of these dominant negative Gproteins affected ERK activation. Therefore we have shown that JNK stimulation by SpD is mediated via G ␣12 activation, and crucially, this occurs within the same concentration range, which inhibits bombesin-induced G q activation.
The mechanisms by which neuropeptides activate ERK are diverse and cell type-dependent (41). In this study we showed that bombesin-stimulated ERK activation was inhibited by the EGF receptor tyrosine kinase inhibitor AG1478 but was pertussis toxin-insensitive. This finding is consistent with previous studies on G-protein-coupled receptor-induced ERK activation in rat-1a cells, which demonstrate that transactivation of growth factor receptors is required for ERK activation by Gprotein-coupled receptors (41). Previous results have also shown that bombesin inhibits [ 125 I]-EGF binding to its receptor (43), implicating EGF receptor activation in bombesin-mediated signaling. However, we also showed that SpD-induced ERK activation was not blocked by AG1478 but was blocked by pertussis toxin. ERK activation by G i receptors in many cases has been shown to be caused by the liberation of ␤␥ subunits from G i that can directly activate phosphatidylinositol 3-kinase, leading to the activation of Ras and the Raf/MEK/ERK cascade (44). However there is evidence in Jurkat T lymphocytes that G i proteins can activate ERK via a Ras and phosphatidylinositol 3-kinase-independent pathway that is mediated by the G ␣i subunit (45). Whatever the case, our data suggest that SpD has the unique ability to differentially activate signal transduction pathways at the agonist-binding site, allowing bombesin receptors to couple to G i proteins, leading to subsequent ERK activation while at the same time blocking receptor activation of G q -stimulated phospholipase C activation and Ca 2ϩ release. SpD is the first compound shown to be capable of simultaneously activating and inactivating different arms of the signal transduction pathways normally activated by a full agonist. This is not caused by SpD favoring activation of G-proteins with faster rates of GDP/GTP exchange, because dissociation of GDP from G 12 is 10 -20-fold slower than other ␣ subunits (46). Furthermore, the prolonged kinetics of both JNK and ERK activation and the reduced ability of SpD to desensitize the GRP receptor compared with bombesin support the concept that SpD is bound to the receptor for a long enough period to activate all potential family members of G-proteins. It would seem likely from our observations that SpD is not acting as a partial agonist. Partial agonists bind to the receptor and compete with the agonist for binding but have lower efficacy than a full agonist; however, they should activate all the same signal transduction pathways as the natural ligand with the same time course. Our data therefore suggest that SpD, upon binding to the bombesin receptor, causes a conformational change that differentially activates G-proteins.
For agonist-receptor trafficking (26,27) it has been suggested that structurally different agonists may occupy the receptor, causing different active receptor conformations that selectively activate G-proteins (47,48). Therefore, the diversity of agonist responses in different cell types may not be solely caused by cellular differences in receptor/G-protein stoichiometry. The cubic ternary complex model (Fig. 10) allows for the existence of multiple active and inactive receptor states, with agonists stabilizing the formation of multiple states via affinity factors ␣, ␥, and ␦ (47). Although designed to describe the efficacies of different agonists, this model can be extended to explain equally well the action of a biased agonist that promotes the existence of activation states that stimulate certain G-proteins (e.g. G i and G 12 for SpD) but not others or even stabilizes an inactive state (inhibition of G q ). In this model the final response given by an agonist will be governed by its promotion of active states and the relative amounts of G-protein available for stimulus-response coupling. Biased agonism therefore extends the idea of agonist-receptor trafficking to include compounds that activate and inactivate different receptor/G-protein conformations.
There is a structural basis to support the concept that different conformational states result in selective G-protein activation by a receptor. Deletion of part of the seventh transmembrane domain of the calcitonin receptor favors G s coupling over G q (49). Mutations in the thyrotropin receptor uncouple the receptor from the G-protein that mediates phospholipase C activation (most likely G q ) while maintaining coupling to G s (50). Furthermore, mutations in G-protein-coupled receptors result in receptors that are in an active conformational state in the absence of ligand (51).
We provide functional relevance for these observations in one of the most important biological systems in which neuropeptides play a role: SCLC. Our results show that SpD stimulates a sustained activation of JNK and ERK (EC 50 ϭ 3.9 Ϯ 1.7 M and 1.2 Ϯ 2.3, respectively) over the same concentration range that inhibits SCLC cell growth (IC 50 ϭ 29.5 Ϯ 5.5) and GRPstimulated increase in [Ca 2ϩ ] i (IC 50 ϭ 12.7 Ϯ 2.5 M). At these concentrations SpD also induces apoptosis (21,23). We propose that biased agonism at the GRP receptor causes discordant signaling (ERK and JNK activation in the absence of Ca 2ϩ mobilization), which results in the full growth inhibitory effect of substance P analogues. Our results suggest that the biased agonist activity of SpD subverts the cancer cells growth factor receptors from stimulating proliferation to initiating programmed cell death in addition to blocking the effects of mitogenic neuropeptides. This is an exciting new concept in cancer chemotherapy.
The results presented here provide a unifying mechanism to explain the previously published observations using substance P analogues. SpD activates the GRP/bombesin receptor in such a way as to cause association with only a subset of possible ␣ subunits; this is the first agonist shown to act in this manner. SpD binds to the ligand-binding site of the GRP receptor and will therefore inhibit ligand-receptor binding (7,17,18) and act as a competitive GRP antagonist for certain responses mediated via G q (7,17,19,20). However, SpD also has activity that cannot be explained by competitive antagonism alone. In Swiss 3T3 cells where SpD and [D-Arg 1 ,D-Trp 5,7,9 ,Leu 11 ]substance P inhibit bombesin-stimulated ERK activity (presumably via blocking G q (18,25)), SpD also augments raf and ERK activity at higher bombesin concentrations (25), which we propose results from the activation of G i . In different cells the degree of stimulation of G i -mediated ERK activation is likely to depend on the level and ratio of receptor and G-protein expression (41). The SCLC cell growth inhibitory and proapoptotic activities of substance P analogues are not reversed by supramaximal saturating concentrations of neuropeptides (21). In terms of conventional pharmacology, this is incompatible with competitive  (52)) accounts for both active (R a ) and inactive (R i ) forms of the receptor interacting with G-protein (G). The factors ␣, ␤, and ␥ define different affinities that the receptor has for G-protein when it has agonist (A) bound, where L is the ratio [R a ]/[R i ] and K a and K ␥ are the equilibrium dissociation constants for the agonist/receptor and the receptor/G-protein complexes respectively. Differences in the affinity factors ␣, ␤, and ␥ could allow for ligand-specific receptor active and inactive states. Right, the biased agonist hypothesis showing the interaction of bombesin (Bn) and SpD with the GRP receptor (GRPR). Although bombesin induces receptor conformations that activate G 12 and G q (R a 1 and R a 2 ), it does not activate G i to produce a stimulation of ERK (R a   3 ). SpD on the other hand activates G 12 and G i but does not activate and in fact blocks the effect of bombesin on G q . antagonism. However, for transient signaling events such as intracellular calcium mobilization, SpD acts as a competitive antagonist and is reversible with bombesin. Longer term effects lead to the activation of programmed cell death, which is irreversible. Thus, once apoptosis has been initiated by SpD this cannot be reversed by excess agonist.
Our characterization of the mechanism of action of SpD can be used further for future drug development. It also offers the opportunity to modulate neuropeptide receptor signaling to activate some signal transduction pathways while blocking others. Bombesin/GRP receptors are potential therapeutic targets for the treatment of a number of human disease states including obesity, inflammation, and cardiovascular disease as well as cancer. Thus modulation of neuropeptide signaling has important clinical implications. The majority of pharmacological work for the treatment of human disease has focussed on the development of highly specific receptor antagonists to treat diseases with minimal side effects; however, even specific receptor blockade can lead to deleterious side effects. With the establishment of the concept of biased agonism much more specific pharmacological agents could be generated that will selectively affect specific receptor downstream signal transduction pathways developing both receptor-specific and signal transduction-specific pharmacological agents.