In vivo coupling of insulin-like growth factor II/mannose 6-phosphate receptor to heteromeric G proteins. Distinct roles of cytoplasmic domains and signal sequestration by the receptor.

We examined the signaling function of the IGF-II/mannose 6-phosphate receptor (IGF-IIR) by transfecting IGF-IIR cDNAs into COS cells, where adenylyl cyclase (AC) was inhibited by transfection of constitutively activated G alpha i cDNA (G alpha i2Q205L). In cells transfected with IGF-IIR cDNA, IGF-II decreased cAMP accumulation promoted by cholera toxin or forskolin. This effect of IGF-II was not observed in untransfected cells or in cells transfected with IGF-IIRs lacking Arg2410-Lys2423. Thus, IGF-IIR, through its cytoplasmic domain, mediates the Gi-linked action of IGF-II in living cells. We also found that IGF-IIR truncated with C-terminal 28 residues after Ser2424 caused G beta gamma-dominant response of AC in response to IGF-II by activating Gi. Comparison with the G alpha i-dominant response of AC by intact IGF-IIR suggests that the C-terminal 28-residue region inactivates G beta gamma. This study not only provides further evidence that IGF-IIR has IGF-II-dependent signaling function to interact with heteromeric G proteins with distinct roles by different cytoplasmic domains, it also suggests that IGF-IIR can separate and sequestrate the G alpha and G beta gamma signals following Gi activation.

Insulin-like growth factor II (IGF-II) 1 promotes growth, mainly in fetal development. In cultured cells, it exerts mitogenic and metabolic stimulation by binding to cell surface receptors. IGF-IIR is a high-affinity receptor for IGF-II (1)(2)(3). It is also a receptor for M6P (3). However, these two distinct ligands bind to different sites in IGF-IIR, which has been indicated by competition experiments and by the fact that only mammalian IGF-IIR can bind IGF-II (4). For several reasons (for review, see Ref. 4), it remains unclear whether the IGF-IIR executes signaling functions in response to IGF-II. Nonetheless, there are multiple lines of independent evidence that IGF-IIR has signaling function activated by IGF-II. In multiple culturedcell systems, IGF-II evokes cellular responses, most likely through IGF-IIR (3,(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15). We and another group independently showed that IGF-II stimulation of IGF-IIR promotes calcium influx through G i , a member of the heteromeric G protein family, in Balb/c3T3 or CHO cells (12)(13)(14)(15). In reconstituted vesicles, purified IGF-IIR directly couples to G i in response to IGF-II (16 -18), and human IGF-IIR has a cytoplasmic 14residue region at Arg 2410 -Lys 2423 , which can directly activate G␣ i (18,19). In cell-free systems, this region most likely functions as the effector domain of IGF-IIR for G i coupling (18 -20). Although failure of IGF-IIR coupling to G proteins in cell-free systems was once reported (21), a subsequent paper (15), with two of the same authors, suggested the G i coupling function of IGF-IIR, based on the observation that IGF-II stimulates Ca 2ϩ influx via a pertussis toxin-sensitive G protein in a manner resistant to tyrosine kinase inhibitors.
Intensive studies of the molecular signaling function of IGF-IIR have so far been conducted only on cell-free experimental systems, which have serious limitations. The present study was conducted to establish a more physiological system, where one can investigate the signaling functions of IGF-IIR. Here we report that IGF-II links recombinant IGF-IIR to the G i /AC system in living cells. Furthermore, to the extreme C terminus of IGF-IIR, we assigned a novel function of inactivating G␤␥, which is another component of heteromeric G proteins. This study not only offers further evidence for the interaction of IGF-IIR with heteromeric G proteins, it provides a novel insight into the differential regulation of G protein subunit signals by receptors. EXPERIMENTAL PROCEDURES G␣ t and gip2 (G␣ i2 Q205L) cDNAs were provided by Dr. H. R. Bourne. Wild-type G␣ i2 cDNA, human IGF-IIR cDNA, ⌬2410 -2423 cDNA, and the construction method of IGF-IIR mutants were described previously (20,22). Oligonucleotide-directed mutagenesis was done to construct ⌬CT41 and ⌬CT28 according to the Kunkel method (23). Oligonucleotides used were GAGCGTGAGGACGATTGATGAAGGGTGGGGCTG-GTC for ⌬CT41, and GCGAGGAAAGGGAAGTGATGATCCAGCTCT-GCACAG for ⌬CT28.
COS cells were grown in DMEM plus 10% calf serum and streptomycin/penicillin. For stable expression of G␣ t , COS cells were transfected by the calcium phosphate method using 10 g of G␣ t cDNA and 0.3 g of pBabe/Puro, a puromycin resistance gene. Cells were then selected with 3 g/ml puromycin and tested for immunoblot analysis with AS/7. The COS cell line used here expresses G␣ t at an approximately half the level of endogenous G␣ i .
Plasmids were transfected by the lipofection method as described (22). Intracellular accumulation of cAMP was measured as described (24). A day before transfection, 5 ϫ 10 4 cells were seeded on a 12-well plate. Unless otherwise specified, cDNAs encoding IGF-IIRs or G␣ i2 / gip2 (0.5 g/ml each) were transfected with 1 l/ml LipofectAMINE (Life Technologies, Inc.) and incubated for 24 h in a serum-free culture. After washing cells with fresh media, cells were labeled with 3 Ci of [ 3 H]adenine for another 23 h 30 min. It should be emphasized that cells were then washed rigorously with solution containing M6P, as follows. DMEM-Hepes (DMEM containing 25 mM Hepes/NaOH, pH 7.4) containing 10 mM M6P was added to cells after discarding media. Cells were then incubated for 15 min at room temperature and washed four times with DMEM-Hepes. These procedures, which dissociate M6P and M6P-containing proteins from IGF-IIR, ensured reproducibility for inhibition of AC by IGF-IIR stimulation. This was reasonable because M6P binding to IGF-IIR impaired the action of IGF-II to inhibit AC in cells transfected with IGF-IIR cDNA (see "Results"). Cells were then treated with 2.5 g/ml CTX (Calbiochem) and 1 mM isobutylmethylxanthine with or without IGF-II (or IGF-I) in DMEM-Hepes at 37°C for 30 min. Reactions were terminated by aspiration and the immediate addition of 5% ice-cold trichloroacetic acid (1 ml/well). Acid-soluble nucleotides were separated on two-step ion-exchange columns as described (24), and specific accumulation of cAMP is expressed as (cAMP/ ADP ϩ ATP) ϫ 10 3 .
For binding assay of IGF-IIR, cells (3 ϫ 10 6 /dish) were transfected with 10 g of IGF-IIR cDNAs and 20 l of LipofectAMINE in 5 ml of DMEM plus streptomycin/penicillin. Twenty-four hours after transfection, the medium was renewed to DMEM plus 10% calf serum and streptomycin/penicillin. By scraping cells 48 h after transfection, membranes were prepared and IGF-II binding assay was performed as described (20). Specific binding was calculated by subtracting nonspecific binding, the binding in the presence of 100 nM IGF-II. All other materials were obtained from commercial sources. Data were analyzed with Student's t test.

RESULTS AND DISCUSSION
We initially examined whether our COS cells were appropriate to see the effects of G␣ i . Indeed, COS cells have not frequently been used to examine the effects of G i or G i -coupled receptors, although Bell and co-workers (25) have described the G i -coupled effect of somatostatin receptors using COS cells. For this reason, we tested the effect of transfection of wild-type G␣ i2 or constitutively activated G␣ i2 mutant gip2 cDNA on AC activity. As shown in Fig. 1A, transfection of gip2 resulted in dose-dependent inhibition of CTX-stimulated cAMP accumulation, whereas that of wild-type G␣ i2 had no effect. Therefore, our COS cells seemed to be suitable for examining the G icoupling function of receptors with transient transfection of cDNAs.
Intact IGF-IIR cDNA was transfected into these COS cells, which were treated with IGF-II before cell lysis. Parental COS cells expressed 3.6 fmol/g of endogenous IGF-II binding sites having the K d of 0.90 nM. With cDNA transfection, these cells expressed recombinant IGF-IIRs with comparable affinities by severalfold of the endogenous binding level (Fig. 1B, B max and K d were 7.3 fmol/g and 1.3 nM in intact IGF-IIR transfection, 9.5 fmol/g and 1.4 nM in ⌬CT41 transfection, 7.1 fmol/g and 0.85 nM in ⌬CT28 transfection, and 6.0 fmol/g and 0.88 nM in ⌬2410 -2423 transfection, respectively). The endogenous IGF-II binding site appears to be virtually IGF-IR, because only a ϳ110-kDa protein was cross-linked with radioactive IGF-II in an IGF-II-inhibitable manner in parental COS cell membranes under the same condition as in the IGF-II binding assay (data not shown). This assessment is not only consistent with the report of Steele-Perkins et al. (26) that IGF-IR exhibits considerably high affinity for IGF-II, but is also strongly supported by the report of Oshima et al. (27) showing that endogenous IGF-IIR is scarce in COS cells.
In the cells transfected with either gene, 2.5 g/ml CTX constantly increased AC activity by ϳ5-fold over the basal level. In cells transfected with intact IGF-IIR cDNA, IGF-II significantly impaired the CTX-stimulated AC activity in a dose-dependent manner (Fig. 1C). IGF-II also inhibited forsko-lin-stimulated AC activity (Fig. 1D). It is underscored that no inhibition of AC was observed by IGF-II without transfection of IGF-IIR cDNA (Fig. 2). Thus, the effect of IGF-II was attributable to the recombinant IGF-IIR. In accord with this idea, 10 nM IGF-I did not reproduce the effect of IGF-II (Fig. 2). Consistent with our previous study (12,13,16,17,20), this inhibitory effect of IGF-II was abolished by PTX and by 10 mM M6P (Table I). These data indicate that IGF-II triggers the signaling function of IGF-IIR and couples it to the G␣ i /AC system in living cells.
In cell-free systems, the Arg 2410 -Lys 2423 region of IGF-IIR FIG. 1. Effects of activated G␣ i2 and intact IGF-IIR on AC activity in COS cells. A, dose effect of transfection of either gip2 cDNA or wild-type G␣ i2 cDNA on CTX-stimulated AC activity in COS cells. AC activity is indicated as a percentage of CTX-stimulated activity in mock-transfected COS cells, which was 27.2 Ϯ 2.9 (cAMP/(ADPϩATP) ϫ 10 3 ). Each value represents the mean Ϯ S.E. of single determinants done with four independent transfections. *, p Ͻ 0.05; **, p Ͻ 0.01 versus no cDNA. B, specific IGF-II binding to the membranes of transfected COS cells. COS cells were transfected with each IGF-IIR or mutant cDNA. Forty-eight h after transfection, membranes were prepared and specific IGF-II binding was measured. The results are representative of four independent transfections, which yielded similar results. C, dose effect of IGF-II on CTX-stimulated AC activity in IGF-IIR-transfected COS cells. COS cells were transfected with 0.5 g/ml IGF-IIR cDNA. Cells were stimulated by 2.5 g/ml CTX in the presence of various concentrations of IGF-II. AC activity was assessed by measuring (cAMP/(ADP ϩ ATP)) ϫ 10 3 . Each value represents the mean Ϯ S.E. of single determinants done with four independent transfections. Thus, the effect of IGF-II was highly reproducible across transfections. *, p Ͻ 0.05; **, p Ͻ 0.01 versus no IGF-II. D, effect of IGF-II on forskolin-stimulated AC activity in COS cells transfected with IGF-IIR. After transfection of IGF-IIR cDNA, cells were treated with increasing concentrations of forskolin with or without 10 nM IGF-II. AC activity is indicated as n-fold of basal activity in these cells, which was 0.345 Ϯ 0.10. The S.E. of AC activity in the presence of 1 M forskolin stimulation without IGF-II was 2.4. Each value represents the mean Ϯ S.E. of single determinants done with four independent transfections. **, p Ͻ 0.05 versus no IGF-II has been implicated in its G i coupling function (18 -20). To examine whether this is the case in living cells, we constructed mutant IGF-IIRs lacking the C-terminal 41 residues after Arg 2410 (⌬CT41) or the 28 residues after Ser 2424 (⌬CT28) and lacking Arg 2410 -Lys 2423 (⌬2410 -2423). Despite remarkable expression of ⌬CT41 (Fig. 1B), IGF-II failed to inhibit CTXstimulated AC activity in cells transfected with this mutant (Fig. 2), indicating an essential role of the C-terminal 41 residues for AC suppression. We unexpectedly found a novel AClinked function of ⌬CT28. In ⌬CT28-transfected COS cells, CTX augmented AC activity to the same level as in COS cells transfected with other IGF-IIRs; however, only in the ⌬CT28transfected cells, did IGF-II further potentiate CTX-stimulated AC activity (Fig. 2). This effect of IGF-II depended on the amount of ⌬CT28 cDNA used for transfection (Fig. 3A). In the same ⌬CT28-transfected cells, IGF-II did not affect AC activity without CTX (not shown). These suggest that AC potentiation by ⌬CT28 is mediated by the G␤␥ subunit of heteromeric G proteins (28,29).
To confirm the G␤␥ mediation, we examined the effect of G␣ t on the function of ⌬CT28 (Fig. 3B). In COS cells overexpressing G␣ t (COS/G␣ t ), CTX stimulated AC activity with a similar fold of the basal activity to that observed in parental COS cells and ⌬CT28-transfected COS cells. In COS/G␣ t cells transfected with ⌬CT28, IGF-II hardly potentiated the CTX-stimulated AC activity, indicating that G␣ t impaired this function of ⌬CT28. The reason why ⌬CT28 could not significantly inhibit AC activity in COS/G␣ t cells was likely that G␣ t expression was not sufficient to totally absorb the G␤␥ released by ⌬CT28 stimulation. Since G␣ t acts as a specific inhibitor of G␤␥ without affecting AC (29), these data confirm that G␤␥ mediates the IGF-II-triggered potentiation of AC by ⌬CT28.
As high concentrations of G␤␥ are required to enhance AC activity, the source of G␤␥ should be G i in non-neuronal cells (30). It was thus likely in our COS cells, that ⌬CT28 releases G␤␥ by activating G i . To confirm the source of G␤␥, we examined the effect of PTX on this function of ⌬CT28. As in Fig. 3C,

FIG. 2. Effect of IGF-IIR and cytoplasmic mutants on AC activity in COS cells. COS cells were transfected with recombinant IGF-IIR
cDNAs or pECE vector (each 0.5 g/ml). Cells were then stimulated by 2.5 g/ml CTX in the presence or absence of 10 nM IGF-II, and AC activity was measured. As a control, the effect of 10 nM IGF-I was examined in the IGF-IIR-transfected COS cells. Each value represents the mean Ϯ S.E. of single determinants done with three independent transfections. AC activity is indicated as a percentage of CTX-stimulated activity in mock-transfected COS cells, which was similar to that in the left panel C. ***, p Ͻ 0.005. n. s., not significant. Inset, illustration of IGF-IIR mutants.

TABLE I Effects of PTX and M6P on the IGF-II action in
IGF-IIR-transfected COS cells COS cells were transfected with 0.5 g/ml intact IGF-IIR cDNA for 24 h, then treated with PTX (10 ng/ml) or M6P (10 mM) for another 24 h. During the last 30 min, cells were stimulated by 2.5 g/ml CTX in the presence of 1 mM isobutylmethylxanthine, and AC activity of the cells was measured. Values represent means Ϯ S.E. of three independent experiments done with single transfection. The results presented here were reproduced by at least two independent transfections.

induced AC stimulation in COS cells overexpressing G␣ t (COS/G␣ t ). COS/G␣ t cells were transfected with
⌬CT28 cDNA (or vector), and stimulated by 2.5 g/ml CTX with or without 10 nM IGF-II. AC activity is indicated as a percentage of CTX-stimulated activity in mock-transfected COS/G␣ t cells, which was 22.7 Ϯ 2.6. Each value represents the mean Ϯ S.E. of four independent experiments. Note that IGF-II augmented CTX-stimulated AC activity by ϳ200% in parental COS cells that received ⌬CT28 transfection under the same condition. C, effect of PTX on IGF-II/⌬CT28-induced augmentation of CTX-stimulated AC activity. COS cells were transfected with ⌬CT28 for 24 h and treated with 10 ng/ml PTX for another 24 h. During the last 30 min, cells were stimulated by 2.5 g/ml CTX with or without 10 nM IGF-II. AC activity is indicated as a percentage of CTX-stimulated activity in ⌬CT28-transfected COS cells. Each value represents the mean Ϯ S.E. of four independent experiments. D, effect of IGF-IIR⌬2410 -2423 on AC activity in COS cells. COS cells were transfected with ⌬2410 -2423 cDNA and then stimulated by 2.5 g/ml CTX in the presence or absence of 10 nM IGF-II, and AC activity was measured. Each value represents the mean Ϯ S.E. of four independent experiments. AC activity is indicated as percentage of the CTXstimulated activity in ⌬2410 -2423-transfected COS cells, which was 20.7 Ϯ 2.8.
24-h treatment of 10 ng/ml PTX blocked the stimulatory effect of 10 nM IGF-II in COS cells transfected with ⌬CT28, while the CTX response was not changed by PTX. These data indicate that the action of ⌬CT28 on G␤␥ is through G i , again suggesting that the Arg 2410 -Lys 2423 region is interactive with G i .
We confirmed the inability of ⌬2410 -2423 to affect AC (Fig.  3D). In COS cells transfected with ⌬2410 -2423, IGF-II could neither inhibit nor augment AC activity, despite the expression of this mutant comparable to that of intact IGF-IIR (Fig. 1B). ⌬2410 -2423 is mutant IGF-IIR that lacks Arg 2410 -Lys 2423 but retains the extreme C-terminal 28 residues that ⌬CT28 lacks. Therefore, this finally demonstrates that the domain that is essential for the interaction with G i is not the extreme C terminus but the Arg 2410 -Lys 2423 region.
We have herein established a whole-cell system in which IGF-II-triggered signaling function of IGF-IIR can be examined. Using this system, multiple lines of evidence show that recombinant human IGF-IIR activates G␣ i and suppresses AC in response to IGF-II. IGF-IIR transfection was required to observe the effect of IGF-II on AC, consistent with the scarceness of endogenous IGF-IIR in COS cells. IGF-I could not reproduce the effect of IGF-II. In addition, M6P treatment of transfected COS cells blocked the effect of IGF-II, reproducing our in vitro data (16). It is thus emphasized that insufficient removal of lysosomal enzymes from IGF-IIR precludes this receptor from responding to IGF-II. Furthermore, the Arg 2410 -Lys 2423 region is shown here to be essential for AC suppression by IGF-IIR, as predicted by our in vitro study (18 -20).
In this study, an IGF-IIR mutant has pointed to a novel function of the receptor C terminus. ⌬CT28 enhanced cAMP production in response to IGF-II. Multiple lines of evidence indicate that this response was mediated by G␤␥, the source of which was the activated G i . In contrast, intact IGF-IIR activated G i and mainly generated the signal of G␣ i (inhibition of AC) in response to the same stimulation. These indicate that the C-terminal Ser 2424 -Ile 2451 region of IGF-IIR can inactivate G␤␥. This inactivation suggests direct or indirect interaction of this C-terminal region with G␤␥. In further support of this idea, the Ser 2424 -Ile 2451 region is homologous to a part of the PH domains (31) of multiple proteins including ␤-adrenergic receptor kinase (Fig. 4), which are proven to bind G␤␥ (32). It has also been shown that the isolated PH domain of ␤-adrenergic receptor kinase inactivates the action of G␤␥ (33). These findings suggest that this region of IGF-IIR inactivates G␤␥ through interaction. Because of technical difficulty, we were not able to examine the effects of the isolated regional peptides on AC augmentation by ⌬CT28 observed in this whole-cell system.
Among known AC subtypes, no single AC that responds to both G␣ i (inhibitory) and G␤␥ (stimulatory) has been specified. However, among AC types I-VI, our COS cells express type VI, which is inhibited by G␣ i , and type IV, which is stimulated by G␤␥ (data not shown). It is thus reasonable to assume that the whole response of AC to G i in intact COS cells is the sum of the respective effects of G␣ i and G␤␥ on these AC subtypes, thus allowing the total AC activity to respond to both G-protein subunits.
In summary, this study shows that IGF-IIR, in living cells, activates G i and affects AC through differential actions of multiple cytoplasmic domains of its own. In our cells, it activates G i through Arg 2410 -Lys 2423 and inactivates G␤␥ through Ser 2424 -Ile 2451 , resulting in the predominant action of G␣ i . Therefore, the distinct roles played by multiple domains of IGF-IIR separate and sequestrate the G␣ and G␤␥ signals following G i activation. This is potentially a very interesting mechanism that allows a receptor to differentially activate G␣ and G␤␥ and selectively turn on each subunit-specific pathway. With this novel mechanism, there is no longer necessity that a receptor must always turn on both subunit pathways by activating one heteromeric G protein complex. It is thus important to investigate whether a similar mechanism is possessed by other receptors.
It is also conceivable that this novel property of IGF-IIR may contribute to its unique signaling function in vivo. Multiple effects of G i depend on G i -released G␤␥ (34). Thus, IGF-IIRinduced G i activation may lack some of the G i outputs induced by conventional receptors. It is also conceivable that the G␤␥ inactivating effect of the C terminus of this receptor may be affected by the amount of free G␤␥ inside the cell. There may be an intracellular free G␤␥ pool with different sizes in different cells (35). Excess G␤␥ may thus occupy the C terminus and attenuate its inhibitory effect. In accord with this idea, IGF-II binding to IGF-IIR can potentiate AC stimulation in human fibroblasts (36) but not in COS cells (this report) and can stimulate PI turnover in renal cells (5) but not in Balb/c3T3 (12) or CHO cells (15). Alternatively, the G␤␥-linked function of IGF-IIR might be involved in its trafficking function as an M6P receptor, as one of the established functions of G␤␥ is translocation of target proteins (37). This is, however, less likely, because the residues essential for IGF-IIR trafficking have been mainly localized near the N terminus of the cytoplasmic domain, particularly before Arg 2410 (38). This possibility is further lowered by the fact that the cation-dependent M6P receptor, another trafficking receptor for M6P, has no cytoplasmic regions homologous to the PH-like domain in the extreme C terminus of IGF-IIR. In conclusion, this study demonstrates the coupling of IGF-IIR with heteromeric G proteins in native cell environments. While calcium influx is one of its most likely outputs (12)(13)(14)(15), it is important to determine which cellular function is executed by the demonstrated IGF-IIR interaction with the G proteins.