Introduction

Top Summary Introduction Procedures Results Discussion References

In unstimulated cells, G protein-coupled receptors are found predominantly at the plasma membrane. After agonist stimulation, the receptors are often efficiently endocytosed and delivered to intracellular compartments, from where they may be either recycled to the plasma membrane or transported to lysosomes for degradation. In many cases, receptor function has been shown to be regulated by this intracellular transport (Pippig et al., 1995; Bogatkewitsch et al., 1996), and so the unraveling of the mechanisms involved in receptor endocytosis is crucial to our understanding of receptor function.

The molecular and cellular processes involved in receptor endocytosis and recycling remain poorly understood. Normally, agonists, but not antagonists, cause an increase in the rate of endocytosis of receptors, suggesting that the activated conformation of the receptor, which couples to G proteins, is also responsible for initiating receptor endocytosis. The involvement of G protein activation and second messenger generation in the triggering of endocytosis, however, remains controversial (Clark et al., 1985; Cheung et al., 1990; Thompson et al., 1991; Benya et al., 1994; Hunyady et al., 1994). Furthermore, studies of the relationship between the activity of agonists and their ability to cause receptor internalization have generated conflicting results. Indeed, it has recently been reported that an antagonist at the cholecystokinin receptor can stimulate internalization (Roettger et al., 1997). Part of the problem may relate to confusion of the terms "endocytosis," which refers to the actual process of removal of receptors from the plasma membrane, and the process that is often measured, internalization (the reduction in number of receptors at the cell surface), which is a function of the rate constants for both endocytosis and recycling (Koenig and Edwardson, 1997).

In the present study, we compared the ability of a series of muscarinic ligands acting at m3 muscarinic receptors in human neuroblastoma SH-SY5Y cells (Wall et al., 1991) to generate second messenger and to modify receptor trafficking. Using a two-compartment model for receptor trafficking (Koenig and Edwardson, 1994), we analyzed the effects of these ligands on the kinetics of receptor endocytosis and recycling. We demonstrate that there is a linear relationship between the increase in the rate constant for receptor endocytosis and the intrinsic activity of the ligand, but no effect of the stimulating ligand on the rate constant for recycling. Furthermore, using mathematical simulations we show that the level of internalization elicited by a drug is dependent not only on drug efficacy, but also varies with cell type-specific factors, such as the kinetics of receptor endocytosis and recycling. This analysis demonstrates that, under certain conditions, compounds with extremely low efficacy can cause significant receptor internalization.

	Experimental Procedures

Top Summary Introduction Procedures Results Discussion References

Cell culture. SH-SY5Y cells (passage 5-25) were grown at 37° in Dulbecco's modification of Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum in 5% CO₂/95% humidified air. Serum was heat-inactivated by incubation at 56° for 30 min. Cells were detached from the culture flasks for passaging every 3-5 days by brief (<2 min) incubation with trypsin (0.5 mg/ml) and EDTA (0.2 mg/ml) in phosphate-buffered saline.

Measurement of Ins(1,4,5)P₃ mass. Confluent cultures of SH-SY5Y cells were harvested and reseeded at an equivalent density in 24-well multidishes. After overnight incubation to allow cell attachment, cells were washed twice with 250 µl of medium at 37° and left for 10 min at 37°. Experiments were initiated by the removal of the wash buffer and its immediate replacement with 150 µl of buffer with or without agonist. Incubations were terminated by the addition of 150 µl of ice-cold 1 M trichloroacetic acid. Multiwell plates were then put on ice for 15 min. A 160-µl aliquot of the incubate was added to 40 µl of 10 mM EDTA, followed by 200 µl of a 1:1 (v/v) mixture of tri-n-octylamine and 1,1,2-trichlorotrifluoroethane. After vortexing, the samples were left at room temperature for 15 min and subsequently centrifuged at 13,000 × g for 3 min. The upper aqueous phase (100 µl) was taken, and 50 µl of 25 mM NaHCO₃ were added. Samples were then stored at 20° until assay of Ins(1,4,5)P₃.

Determination of receptor binding affinities. The binding affinities of the muscarinic ligands were determined by the displacement of 0.25 nM [³H]-NMS in SH-SY5Y cells in Dulbecco's modification of Eagle's medium/HEPES buffer containing 0.5 M sucrose to block receptor internalization. Incubations were for 1 hr at 37°. Displacement curves were analyzed by EBDA/LIGAND (Elsevier Biosoft, Cambridge, UK) to obtain Hill slopes and binding affinities (K_i). The dissociation constant (K_d) for [³H]NMS was obtained under the same conditions as the agonist binding assays.

Analysis of receptor internalization. Changes in surface muscarinic receptor number were quantified through the binding of the polar radioligand [³H]NMS to adherent SH-SY5Y cells grown in 24-well plates, as described previously (Koenig and Edwardson, 1996). In experiments to determine the concentration dependence of receptor internalization, curves were fitted to a standard four-parameter logistic equation using SigmaPlot (SPSS Inc., Chicago, IL).

Mathematical modeling of receptor trafficking. In a two-compartment model, receptors cycle between the cell surface and endosomes. The rate of change of the number of surface receptors is given by the rate of receptors arriving by recycling less the rate leaving by endocytosis

(1)

where R_s is the number of surface receptors, R_e is the number of endosomal receptors and k_r and k_e are the recycling and endocytotic rate constants, respectively. The total number of receptors, R_s + R_e, is equal to the number of receptors at time t = 0; i.e., R_s + R_e = R_s0 + R_e0. Substituting R_e = R_s0 + R_e0  R_s, we can rewrite eq. 1 as which is now of the form dy/dt = b  ay. This can be integrated to ay  b = Ce^at, where b = k_r(R_s0 + R_e0); a = k_e + k_r and C is the constant of integration. Rearrangement results in an equation describing the number of surface receptors as a function of time:

(2)

Receptor recycling curves make use of eq. 2 with k_e and R_s0 set to zero. Then, R_s as a function of time is given by

(3)

Thus, k_r can be determined from a plot of R_s against t. At large values of t, the exponential terms in eq. 2 approach zero, and therefore the steady state number of surface receptors (R_s,ss) may be written as

(4)

Eq. 4 allows experimentally determined changes in R_s,ss to be converted to changes in the endocytotic rate constant k_e, if k_r is known.

Materials. [³H]Ins(1,4,5)P₃ (43 Ci/mmol) was from Amersham (Buckinghamshire, UK) and [³H]NMS (85 Ci/mmol) was from DuPont-NEN (Stevenage, UK). Oxotremorine-M was from Research Biochemicals International (St. Albans, UK). McN-A-343 and propyl-benzilylcholine mustard (PrBCM) were generous gifts of Dr. J. M. Young (Department of Pharmacology, University of Cambridge, UK). All other materials were from Sigma (Poole, UK).

	Results

Top Summary Introduction Procedures Results Discussion References

Intrinsic activity and potency of muscarinic ligands. The intrinsic activity and potency of the eight muscarinic ligands used in this study were assayed through their ability to stimulate the production of Ins(1,4,5)P₃. Addition of muscarinic agonists caused a biphasic Ins(1,4,5)P₃ response, with a rapid initial increase that reached a peak 10 sec after addition of the drug, followed by a reduction over the next 120 sec to a lower but sustained phase (data not shown) (Willars and Nahorski, 1995). Dose-response curves measuring peak Ins(1,4,5)P₃ responses were constructed for all eight muscarinic ligands (Fig. 1 and Table 1). Carbachol, methacholine, and oxotremorine-M were all full agonists, whereas bethanechol, arecoline, and pilocarpine were partial agonists, eliciting 51.6 ± 11.0%, 26.1 ± 3.9%, and 8.4 ± 1.2% of the maximal response to carbachol, respectively. Both N-methylatropine and McN-A-343 were antagonists and without effect in this assay.

View larger version (24K): [in this window] [in a new window]   Fig. 1.   Dose-response relationships for peak changes in Ins(1,4,5)P3 mass (determined at 10 sec) in adherent SH-SY5Y cells. Data are expressed as a percentage of the maximal response seen with carbachol. Resting [Ins(1,4,5)P3] was 3.8 ± 1.5 pmol/106 cells, and peak [Ins(1,4,5)P3] after a maximal dose of carbachol was 46.4 ± 7.1 pmol/106 cells. McN-A-343 and N-methylatropine did not stimulate the production of Ins(1,4,5)P3, and the data for these compounds are not shown for the purpose of clarity.

Measurement of muscarinic receptor recycling rates after removal of agonist. Upon agonist stimulation, muscarinic receptors are internalized and cycle continuously between the cell surface and intracellular compartments (Koenig and Edwardson, 1994, 1996). The quantification of muscarinic receptor trafficking therefore requires the determination of rate constants for both the endocytosis of receptors from the plasma membrane (k_e) to an intracellular compartment and their recycling to the cell surface (k_r). It is possible to measure the rate of recycling of endocytosed receptors in isolation from the effect of receptor endocytosis after removal of agonist, because in the absence of agonist k_e is so small that it may be considered insignificant (Koenig and Edwardson, 1994; Morrison et al., 1996). k_r was determined by treating cells with carbachol (1 mM; 30 min) to achieve maximal endocytosis (Koenig and Edwardson, 1996). Cells were then washed free of agonist and exposed to the muscarinic receptor alkylating agent PrBCM (100 nM) to alkylate any receptors still remaining at the cell surface. After removal of PrBCM, the rate of receptor reinsertion into the plasma membrane was measured through the binding of the polar radioligand [³H]NMS (Fig. 2 and Table 2). Modeling of these data using eq. 3 indicates that muscarinic receptors were recycled to the cell surface with first order kinetics, giving a rate constant (k_r) of 0.054 ± 0.004 min¹.

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Delivery of muscarinic receptors to the plasma membrane. Cells were stimulated with 1 mM carbachol for 30 min at 37°C to induce endocytosis of muscarinic receptors. After washing, cells were incubated with PrBCM (100 nM for 30 min at 15°C) to alkylate remaining surface receptors, washed again, and then incubated for various times at 37°C before determination of [3H]NMS binding. Recovery is expressed as a percentage of control specific binding (i.e., binding in untreated cells), with the residual binding after PrBCM subtracted. Data are means ± standard error from three separate experiments. The recycling rate constant, kr, was 0.054 ± 0.004 min1.

Receptor endocytosis initiated by full and partial muscarinic agonists. To establish whether agonist intrinsic activity plays a role in determining the rate of receptor endocytosis, we measured receptor internalization in response to increasing concentrations of eight muscarinic ligands of varying intrinsic activity. Cells were stimulated for 30 min with the appropriate agonist, after which time receptor cycling to and from the plasma membrane has reached steady state (data not shown) (Koenig and Edwardson, 1996), and then thoroughly washed before quantifying the amount of receptors left at the cell surface (Fig. 3A and Table 1). The full agonists carbachol, methacholine, and oxotremorine-M all induced maximal receptor internalization, causing the loss of ~90% of surface receptors to intracellular compartments at maximal concentrations. Bethanechol, arecoline, and pilocarpine, which were partial agonists in the Ins(1,4,5)P₃ mass assays, also exhibited partial agonism in internalization assays, causing the maximal removal of 82.0 ± 0.6%, 78.4 ± 2.3%, and 18.7 ± 0.7%, respectively, of surface receptors at steady state, whereas McN-A-343 and N-methylatropine were again without effect. All agonists except the very weak partial agonist pilocarpine caused internalization with EC₅₀ values that were significantly less than their K_i values (Table 1), suggesting that maximal receptor internalization does not require the presence of saturating concentrations of agonist.

View larger version (23K): [in this window] [in a new window]   Fig. 3.   A, Dose-response curves for the internalization of muscarinic receptors in SH-SY5Y cells. Cells were treated for 30 min at 37°C with the indicated concentration of drug and then washed thoroughly before determination of the number of muscarinic receptors remaining at the cell surface using [3H]NMS. Data are means of three or four separate experiments. Error bars have been omitted for clarity. B, Transformation of steady state cell surface receptor number to increases in the endocytotic rate constant ke. Best fit lines generated in A were transformed using eq. 4.

View larger version (18K): [in this window] [in a new window]   Fig. 4.   Correlation between the ability of muscarinic ligands to stimulate the production of Ins(1,4,5)P3 and their ability to A, cause internalization of muscarinic receptors, and B, increase ke.

View larger version (17K): [in this window] [in a new window]   Fig. 5.   Theoretical internalization curves for a weak partial agonist generated from eq. 4, assuming that the ke(max) is related linearly to the intrinsic activity of the agonist. A, Effect of varying the rate constant for recycling; B, effect of varying the rate constant for endocytosis. Curves, theoretical surface receptor number at steady state after stimulation with a weak partial agonist which has an intrinsic activity of only 3% of that of a full agonist. ke is increased by the partial agonist with an EC50 of 1 µM and a Hill slope of 1. In A, the maximal value of ke in the presence of a maximal dose of a full agonist (ke(max)) was set at 0.4 min1, and kr was varied between 0.01 and 0.3 min1. In B, the value of kr was set at 0.05 min1, and ke(max) was varied between 0.01 and 0.6 min1.

	Discussion

Top Summary Introduction Procedures Results Discussion References

We have shown previously (Koenig and Edwardson, 1997) that the number of receptors internalized in response to agonist stimulation depends not only on the rate constant for endocytosis, k_e, but also on the rate constant for recycling, k_r. In the present study, we compared the abilities of various muscarinic ligands, ranging from full agonists through partial agonists to antagonists, to stimulate Ins(1,4,5)P₃ production and receptor internalization. Ins(1,4,5)P₃ production was used as an index of agonist intrinsic activity because this was considered to be the parameter that would most closely reflect receptor activation, being the first measurable point in the signal transduction cascade in whole cells.

We analyzed the internalization data to dissect out the effects of the ligands on k_e and k_r, and found that the intrinsic activities of the ligands correlate directly with their ability to increase k_e. In contrast, k_r is independent of the ligand that caused the endocytosis. The constitutive nature of the recycling process is in agreement with a previous report that the recycling rate of ₂-adrenergic receptors is not affected by the presence of agonist (Morrison et al., 1996). It has also been shown (Mayor et al., 1993) that transferrin receptors and membrane lipids pass through the endocytotic system with indistinguishable kinetics, suggesting that recycling of membrane proteins is the default pathway in the absence of specific lysosomal sorting motifs.

To measure k_r as accurately as possible, recycling was determined after alkylation of all receptors remaining at the surface with PrBCM at a relatively high concentration (100 nM). The extent of recycling observed under these conditions (approximately 25% of the number of receptors initially present at the surface) is much smaller than the loss of receptors from the surface in response to agonist stimulation (approximately 90%). In fact, the protocol used in this study results in an underestimation of the true extent of recycling, because PrBCM is apparently able to alkylate some internal receptors. Consequently, in these experiments we are underestimating the extent of recycling to measure k_r accurately. If the PrBCM step is omitted, the extent of recycling is approximately 70% of the initial cell-surface value (Szekeres PG, Koenig JA, and Edwardson JM, unpublished observations). Even in the absence of the PrBCM treatment, therefore, recycling is incomplete. The shortfall cannot be accounted for by receptor degradation,¹ indicating that some of the receptors are being translocated into a compartment from which they recycle very slowly or not at all. These findings suggest that a two-compartment model is an oversimplification, and we are currently developing a model that will more accurately reflect the real situation.

The correlation between intrinsic activity of the ligand and its effect on k_e extends the observations of a previous study (Thompson and Fisher, 1990), and strongly suggests that, at least in this case, the same receptor conformation is responsible for coupling to G proteins and stimulating endocytosis. Further evidence for this concept comes from the identical potency profiles observed for both production of Ins(1,4,5)P₃ and stimulation of the endocytotic rate constant, k_e. Usually, agonists but not antagonists cause internalization of G protein-coupled receptors. However, a recent report (Roettger et al., 1997) indicated that antagonists at the cholecystokinin receptor may also cause receptor internalization, implying that different receptor conformations are required to stimulate second messenger production and endocytosis. Further evidence for separate receptor conformations has been obtained from recent observations of the µ-opioid and -opioid receptors in transfected cells and enteric neurons (Sternini et al., 1996; Keith et al., 1996), where enkephalins and etorphine stimulate receptor endocytosis, but the full agonist morphine does not. One interpretation of these results is that the opioid receptors contain multiple agonist recognition domains (Wang et al., 1995; Kong et al., 1993), leading to the possibility of multiple activated conformations upon treatment with different agonists. In contrast, muscarinic agonists are generally thought to interact with a conserved agonist binding site (Hulme et al., 1990; Wess et al., 1991). Our study certainly provides no evidence for the involvement of separate receptor conformations in initiating signal transduction and endocytosis of muscarinic receptors.

Whether G protein activation and second messenger production are themselves involved in triggering endocytosis is a question that has been addressed by many groups. The most extensively studied G protein-coupled receptor is the ₂-adrenergic receptor, which couples to G_s and activates adenylyl cyclase. Because ₂-adrenergic receptors are internalized in cells that do not express G_s (Clark et al., 1985) and mutant ₂-adrenergic receptors that do not effectively couple to G_s are internalized to the same extent as wild-type receptors (Cheung et al., 1990), it is likely that, at least for this receptor, G protein coupling and/or activation is not a prerequisite for receptor internalization.

It has been reported previously that, in permeabilized SH-SY5Y cells, both U-73122, an inhibitor of phospholipase C-dependent processes, and guanosine-5'-O-(2-thiodiphosphate), which blocks G protein activation, inhibit agonist-mediated muscarinic receptor internalization, but that omission of either ATP or Ca²⁺, both of which are required for stimulation of phosphoinositide hydrolysis, was without effect (Thompson et al., 1991; Slowiejko et al., 1994). These results indicate that G protein activation is required for internalization of the m3 muscarinic receptor but that second messenger production is not. Conflicting results have been obtained for other G protein-coupled receptors. For example, Benya et al. (1994) produced two mutant gastrin-releasing peptide receptors and showed that one of these mutant receptors maintained intact G protein coupling and showed a robust internalization in response to agonist, whereas the other mutant was no longer coupled to G protein and did not internalize, suggesting a functional link between the two processes. On the other hand, the coupling of the thyrotropin-releasing hormone receptor to the G protein G_q was not required for receptor internalization (Petrou et al., 1997), and there was no correlation between G protein coupling and internalization in a series of mutants of the angiotensin II receptor constructed by Hunyady et al. (1994), suggesting that endocytosis of these receptors is independent of G protein coupling. Finally, mutant muscarinic receptors have been constructed that show impaired internalization but normal G protein coupling (Moro et al., 1993). Overall, the results of the above studies leave us with the possibility that there is no general rule governing the behavior of all members of the G protein-coupled receptor superfamily, and that the determinants of internalization may depend on both receptor type and cell type. One potentially confusing factor, which is never addressed in studies based upon receptor mutagenesis, is that the mutant receptors may have altered recycling rates. This may cause apparent differences in their internalization properties, because the number of receptors internalized will depend on both the rates of endocytosis and recycling. The results of these studies, therefore, need to be interpreted with some caution.

An important implication of our results is that the removal of receptors from the plasma membrane is not a linear function of k_e, and in addition will depend on both k_r and k_e(max), parameters that are cell type-specific (Koenig and Edwardson, 1996). Furthermore, agonist regulation of k_e, but not k_r, provides a means by which very weak stimulation of the receptor can result in substantial internalization. In the simulations shown here (Fig. 5), we consider the effects of a partial agonist, which possesses only 3% of the intrinsic activity of a full agonist. In our view, the detection of such a low level of agonist activity would be unlikely in most second messenger assays, and therefore this compound could easily be classified as an antagonist. Our analysis shows that when k_e and k_r are varied over a range often seen for G protein-coupled receptors in mammalian cells (Koenig and Edwardson, 1994, 1996; Barak et al., 1995; Pippig et al., 1995; Morrison et al., 1996), significant internalization of receptors can be achieved with a maximal dose of drug, under conditions where either k_r is low or k_e(max) is high, or both. Such values for the rate constants would give rise to a system where stimulation with a full agonist results in rapid internalization of a large proportion of surface receptors. It is interesting to note that in the one published example of antagonist-induced receptor internalization (Roettger et al., 1997), full agonist occupation of the same receptor gave rise to a rapid internalization of 90% of the receptors, whereas antagonist caused the internalization of only 37% of receptors. Thus, the amplification effect of small agonist-induced increases in k_e producing large amounts of receptor internalization may provide an alternative explanation of apparent antagonist-triggered receptor internalization. Our results also raise the possibility that cells which have a low value of k_r and a high value of k_e(max) for a particular receptor type might provide a useful system in which to use receptor internalization to detect weak partial agonist activity.