Overlapping Multi-site Domains of the Muscarinic Cholinergic Hm1 Receptor Involved in Signal Transduction and Sequestration*

scanning of the intracellular portion of the human muscarinic cholinergic Hml receptor was performed to identify domains mediating agonist induced receptor sequestration. Using these multiple alanine point mutants of Hml, we had previ- ously identified several receptor domains in the intracellular loops i1-3 that play a role in coupling to phos- phatidyl inositol turnover, most notably, a lipophilic residue, Leu-131, in the conserved i2 loop domain DRYXXVXXPL (Moro, O., Lameh, J., Hogger, P., and Sadbe, W. (1993) J. Biol. Chern. 68624866). We now demonstrate that alanine substitutions in three of these domains, i.e. middle of the i2 loop and both junctions of the i3 loop, also result in defective sequestration (loss of surface receptor sites accessible to a polar tracer) in transfected human kidney U293 cells. The i2 loop was studied further by single point mutations. The strongest impairment of sequestration occurred with mutant L131A which was also highly defective in phosphatidyl inositol (PI) coupling. Substitution of Leu-131 with several distinct amino acids indicated that a bulky lipo- philic residue is required for sequestration in this position, as shown for coupling to PI turnover. Further, the double point mutation, V127A/L131A,

Alanine mutagenesis scanning of the intracellular portion of the human muscarinic cholinergic Hml receptor was performed to identify domains mediating agonist induced receptor sequestration. Using these multiple alanine point mutants of Hml, we had previously identified several receptor domains in the intracellular loops i1-3 that play a role in coupling to phosphatidyl inositol turnover, most notably, a lipophilic residue, Leu-131, in the conserved i2 loop domain DRYXXVXXPL (Moro, O., Lameh, J., Hogger, P., and Sadbe, W. (1993) J. Biol. Chern. 268,68624866). We now demonstrate that alanine substitutions in three of these domains, i.e. middle of the i2 loop and both junctions of the i3 loop, also result in defective sequestration (loss of surface receptor sites accessible to a polar tracer) in transfected human kidney U293 cells. The i2 loop was studied further by single point mutations. The strongest impairment of sequestration occurred with mutant L131A which was also highly defective in phosphatidyl inositol (PI) coupling. Substitution of Leu-131 with several distinct amino acids indicated that a bulky lipophilic residue is required for sequestration in this position, as shown for coupling to PI turnover. Further, the double point mutation, V127A/L131A, almost completely suppressed both sequestration and coupling of Hml. In the pa adrenoceptor, alanine substitution of the i2 residue Phe-139, equivalent to Leu-131 in Hml, also resulted in impaired coupling to adenylyl cyclase and sequestration, indicating a general role for this conserved i2 loop residue in both processes. The combined results show that the multi-site domain involved in signal transduction of Hml is similar to and overlaps with that involved in sequestration. However, three Hml mutants that were moderately deficient in stimulating PI turnover displayed normal sequestration, suggesting distinct mechanisms. We propose that cellular mediators of receptor sequestration are structurally similar or identical to the heterotrimeric G proteins.
Stimulation of cell membrane receptors with agonists often triggers rapid sequestration, first at the cell's surface and then into the cell's interior (internalization), followed by recycling or down-regulation. The mechanisms of these processes are poorly understood for G protein-coupled receptors (GPCRs),l whereas receptor domains mediating endocytosis have been defined for growth factor, low density lipoprotein, and transferrin receptors (see Refs. 1-3, and references therein). Mutational studies with several GPCRs, including the p2 adrenoceptor, have yielded receptor mutants that are sequestration defective (4-7); however, the nature of these mutants did not specify any discrete sequestration domains. Rather, mutations may have caused conformational changes that prevent other domains to mediate sequestration.
With the use of multiple point and deletion mutations, we demonstrated that an S/T rich domain in the middle of the i3 loop of Hml, 2, and 3 receptors is involved in agonist induced sequestration (1,2). Replacement of these S/T residues with alanine caused a profound defect in receptor sequestration. However, numerous i3 loop deletions encompassing the S/T domain yielded a range of H m l sequestration defects, from nearly normal to complete, and no single i3 loop domain could be identified that was consistently associated with strongly defective sequestration (1). We therefore proposed that the S/T rich domains, possibly upon phosphorylation, regulates sequestration by affecting the i3 loop conformation. Thus, the i3 loop is thought to act as a molecular switch, permissive for sequestration only in a conformation that enables access of any sequestration factor to the requisite domains elsewhere on the receptor. Similar results were obtained previously with the p2 adrenoceptor. Whereas a small domain in the COOH tail containing Ser and Thr residues was shown to affect coupling and sequestration (61, truncation mutations indicated that the COOH tail is not required per se (5). Further, truncation of the extended COOH tail of the avian pz receptor unmasks its ability to internalize (8). These results support the hypothesis that the overall conformation of the highly variable i3 loops and COOH tails play a permissive role for receptor function. I t further implies that the receptor domains directly mediating sequestration remain unknown.
Because both G protein activation and sequestration are dependent upon agonist stimulation, several studies have addressed the relationship between these two processes. Earlier mutational analysis of the p2 receptor demonstrated that structural features involved in receptor activation are also essential to receptor sequestration, suggesting that G protein activation and sequestration depend on the same receptor domains (9). However, several mutants of Hml and Hm3 were able to stimulate fully PI turnover, but were completely resistant to sequestration (1,2,10). Conversely, an Hml& adrenoceptor chimera had lost the ability to couple to G proteins, but yet continued to undergo sequestration (11). In this study (111, the i3 loop region 222-229 located near the 5'-junction of p2 was either deleted or replaced with the equivalent sequence 220-230 of Hml. Whereas deletion of domain 222-229 abolished both Gs coupling and sequestration, introduction of the Hml domain 220-230 restored sequestration but not G. coupling (11) suggesting a specific role for the Hml 220-230 domain in sequestration. However, deletion of 219-231 from the Hml receptor had no 6651 NH.  (1)(2)(3)(4)(5)(6)(7)(8)(9) were selected to avoid overlap with previous mutations (14). Filled circles indicate point mutations already shown not to affect significantly G protein coupling and sequestration (1, 2, E), whereas the largest deletions of the i3 loop that still yielded full coupling efficiency (1,2,10) are shown by the arrows. The new double point mutation V127AL131A caused highly deficient agonist induced coupling and sequestration. effect on sequestration (l), arguing against a direct role of this domain. Further, p2 receptors of S49 cells that are functionally uncoupled from adenylyl cyclase because of genetic lesions of G, still were able to internalize normally (12,13). These results support distinct and independent mechanisms for G protein activation in signal transduction and sequestration.

Sequestration of Muscarinic Receptors
To identify the receptor domains involved in coupling to PI turnover, we performed alanine mutagenesis scanning covering the intracellular portion of H m l (14). Residues shown by earlier point and deletion mutations not to affect coupling and sequestration (see Fig. 1) (1, 2, 10, 15) were excluded, and the remaining receptor portions were analyzed by distributing 2 4 alanine mutations per construct throughout loops il-3 and the COOH tail (mutants 1-9). The i l and i2 loops and both junctions of the i3 loop were found to play a role in G protein signal transduction of H m l (14). Further, a bulky lipophilic amino acid in the conserved i2 domain DRYMNXXPL (where L is any lipophilic amino acid) was shown to play a key role in G protein coupling of Hml, Hm3, and the p2 adrenoceptor (14). In the present study, a similar expanded set of point mutants was used to compare domains involved in signal transduction and sequestration. Alanine substitution in three domains, i.e. both junctions of the i3 loop, and most notably, the conserved i2 domain, were found to affect sequestration, suggesting the hypothesis that overlapping multi-site domains mediate both signal transduction and sequestration.

EXPERIMENTAL PROCEDURES
Material~-[~H]NMS (specific activity 80 Cilmmol), L3H1QNB (specific activity 40 Ci/mmol), and [3HlCGP12177 (specific activity 46 Ci/ mmol) were obtained f r o m h e r s h a m Corp. and DuPont NEN. All other reagents were of analytical grade quality. Restriction enzymes were from either Boehringer Mannheim or Life Technologies, Inc.
Construction of Vectors Expressing Hrnl Point Mutants-The construction of Hml in vector pSG5 was described previously (1,2), having EcoRI and BarnHI restriction sites at the 5' and 3' ends, respectively. The point mutations were introduced using the "unique site elimination" method (USE, TransformerTM site-directed mutagenesis kit, Clonetech) (see Ref. 14). All mutants were sequenced before further use.
Construction of p2 Adrenoceptor Mutant-The hamster pz adreno-ceptor expression vector (pCDM8) was provided by Dr. Diane Barber in the Department of Stomatology and Surgery, University of California, San Francisco. The point mutation F139A was introduced using the unique site elimination method.
Zkansfection of Human Embryonal Kidney Cells (U293)"The cells were transfected with use of the calcium phosphate precipitation method as previously described (1, 2). Transient expression yielded -900 fmol/mg of protein for Hml, and -4000 fmoVmg of protein for p2.
Receptor Binding and Sequestration-These procedures were described previously (1, 2, 10). Briefly, the transfected cells were replated onto 12-well cell culture dishes and allowed to attach overnight. In the case of Hml, the cells were then incubated with or without 1 mM carbachol for up to 2 h. For the pz adrenoceptor, 1 p~ isoproterenol was used. After drug treatment, the cells were washed three times with ice-cold phosphate-buffered saline to remove residual carbachol or isoproterenol and incubated in a n isotonic buffer containing 1.5 r m [3H]NMS or 3 nM [3H]CGP12177 a t 12 "C (to prevent receptor recycling) for 90 min. At the end of the incubation, the cells were filtered though glass-fiber filters (SS 321, followed by three rapid rinses with phosphate-buffered saline. Six replicate independent samples were assayed for each data point unless noted otherwise. To ascertain that loss of surface L3H1NMS binding was not paralleled by a loss of total receptor binding in the cells, PHIQNB binding was assayed similarly in H m l transfected cells. There was no detectable loss of PHIQNB binding after 2 h of carbachol (1 mM) treatment (1,2). This result indicates that, during the observation period, no measurable down-regulation of H m l occurred in U293 cells.

RESULT AND DISCUSSION
A series of nine scanning mutants with multiple alanine substitutions (nos. 1-9) (Fig. 1) was tested for receptor expression, carbachol-induced PI turnover (141, and sequestration. Two of the scanning mutants (3 and 9) failed to yield measurable expression, possibly because of folding defects. The expression levels and PI-coupling efficiency of mutants with detectable receptor yield were previously reported (141, and these results are reproduced in Table I for comparison with the sequestration values. Several mutants showed defective coupling to PI turnover (mutants 1, 4, 5, 6, and 7; below 60% coupling efficiency compared to wild-type Hml, at 1 mM carbachol), and we proposed a multi-site domain of H m l to be involved in G protein coupling (14).  To test whether any of these mutated residues also play a role in carbachol induced sequestration of Hml, we studied the sequestration behavior of each construct, in comparison to the wild-type receptor serving as a control. The polar tracer [3HJNMS was used to label Hml receptor sites accessible on the surface of intact U293 monolayers, and the carbachol induced loss of tracer binding was taken as a measure of sequestration (1,2, 10). This procedure does not distinguish between sequestration at the surface and internalization into the cell. Muscarinic receptor localization by immunohistochemistry in our own laboratory2 and elsewhere (16) indicates that agonist stimulation causes rapid receptor clustering at the surface and true internalization. We use the term sequestration here to account for both processes.
Sequestration of Alanine Scanning Mutants-Three of the nine alanine scanning mutants with multiple substitutions (nos. 4, 6, and 7) were defective in carbachol induced sequestration (Table I). Each of these mutants was also strongly defective (by 7045%) in stimulating PI turnover Table I) (14). On the other hand, mutants 1 and 5 were defective only in PI coupling but not internalization (Table I). These results implicate domains of the i2 loop and both junctions of the i3 loop in receptor sequestration.
Analysis of i2 Loop Residue Leu-131-The marked coupling and sequestration defect of the triple alanine mutant 4 suggested a role of the i2 loop. As previously shown for PI coupling (14), the sequestration defect of mutant 4 was again solely attributable to alanine substitution of Leu-131, whereas alanine substitution of the other two amino acids failed to affect coupling and sequestration (Table I). Substitution of L131 with alanine did not affect carbachol affinity (14). Residue Leu-131 is located at the 3' end of the consensus motif DRYXXVXYPI, (Fig. I) Table I).
On the other hand, L-131 substitution with M, present in the equivalent position of the thrombin receptor, caused moderately impaired coupling (14), and it also moderately impaired agonist induced sequestration ( Table I). Because of these parallel changes of coupling efficiency and sequestration behavior, we propose that a bulky lipophilic residue in the position equivalent to Leu-131 in Hml plays a key role in both processes.
Residue Phe-139 in the p2 Adrenoceptor--To evaluate the general importance of the lipophilic residue L-131 of Hml in the sequestration of other G protein coupled receptors, we also tested the equivalent F139A mutation in the p2 receptor. Alanine substitution of Phe-139 did not affect binding affinity to the polar tracer L3H]CGP12177 (14). However, marked defects of the alanine mutant were observed for both stimulation of CAMP production by isoproterenol (14) and for agonist induced receptor sequestration (Table 11). These results support a general role for this conserved lipophilic amino acid in both coupling and sequestration.
Other Residues in the i2 Loop Motif DRKKWXXPL of Hml-Because the quadruple mutant 3, scanning the 5' junction of the i2 loop, did not yield measurable tracer binding, single alanine substitutions were introduced throughout the conserved motif DRYXXVXXPL. No effect on coupling and sequestration was observed by substituting the highly conserved P130 with alanine (Table I). Since alanine occurs naturally in Zieu of proline ina few GPCRs, e.g., the a2 adrenoceptors (171, alanine in this position appears to be permissive for receptor function. As residue Val-127 is also very highly conserved (only V or I are found in 67 out of 70 mammalian GPCR sequences listed in Ref. 17), the single point mutant V127A was constructed.

TABLE I1
Effect of F139A point mutation of the p2 adrenoreceptor on expression, agonist-induced stimulation of CAMP production, and sequestration For second messenger activation, transiently transfected cells were incubated 30 min with 1 isoproterenol to stimulate production of CAMP, and for 2 h to induce sequestration. The polar tracer L3HlCGP12177 measures surface accessible receptor sites only (13). All data are mean 2 S.D. Three independent experiments are shown for sequestration.

PHIGCP12177
Stimulation of CAMP Numbers in parentheses = number of data points.
Whereas PI coupling was measurably impaired (14), any effect on sequestration was absent or too small for detection (Table I).
To test further the role of V127A, the double mutant V127A/ L131A was newly constructed. A profound defect in both coupling and sequestration (Table I) demonstrates that this portion of the conserved i2 loop motif is essential for both processes. The fact that the V127A mutation alone did not measurably change internalization suggests that conformational changes of i2 caused by alanine substitution at this site contribute to the deficiency.
The less conserved residue Phe-125 was also exchanged without effect on PI turnover (14), but with even stronger sequestration than observed with the wild-type receptor ( Table I).
The reason for this enhanced internalization remains unknown, but it emphasizes the relevance of the i2 loop in sequestration.
Residue Asp-122 was previously shown to play a role in Hml mediated stimulation of PI turnover (18). Thus, mutation D122N decreased carbachol potency, although maximum effects on PI turnover were barely affected (18). We therefore tested the response of D122N to 1 m~ carbachol (18). As seen in Table I, both stimulation of PI turnover and sequestration of D122N were moderately impaired in comparison to wild-type Hml, and similar data were obtained at 100 1.1~ carbachol (not shown). Hence, mutations throughout domain DRYXXVXXPL affected coupling and sequestration, indicating its relevance to both processes.
Erne Course of ZnternaZizution-To test whether the Hml mutants displayed changed kinetics of sequestratiodinternalization-recycling, we measured the time course of carbachol induced loss of [3H]NMS binding sites from the cell surface (Fig. 2). The double mutant V127A/L131A did not measurably sequester at any time point. Each of the defective mutants tested rapidly sequestered over the initial 30 minutes, but only to an intermediate level. Because of the rather low degree of sequestration, we were unable to measure accurately initial sequestration rates for the mutants. Nevertheless, the time to reach equilibrium was not prolonged compared to wild-type Hml. This result suggests that the rate of sequestration is reduced, rather than the rate of recycling, which would be expected to delay equilibrium between surface accessible and sequestered receptor sites. This result differs from the slower equilibration of H m l sequestration observed previously with several i3 loop deletions, e.g. d247-304 (1). The different kinetics suggests different mechanisms affecting the sequestration behavior of the point mutations studied here and the i3 loop deletion mutations, consistent with the distinct function proposed for the large i3 loop in regulating sequestration.
Comparison of Signal Bansduction and Sequestration-The processes of signal transduction and sequestration maybe related or independent of each other. Fig. 3 shows a plot comparing degree of sequestration and coupling efficiency to PI turnover for Hml and the mutants listed in Table I. There is a  Table I) are plotted against each other. The linear regres-after carbachol treatment for Hrnl wild-type and mutants (data obsion line reflects the results shown as filled circles ( R = 0.89). The three mutants shown as open circles were measurably defective only in coupling to PI turnover, but not sequestration. Inclusion of these data with the regression analysis lowers the correlation coefficient to R = 0.79.
striking correlation between coupling to PI turnover and sequestration, indicating that both functions require very similar domains. However, three mutants, i.e. scanning mutants 1 ( i l loop) and 5 (5' junction of i2), and V127A, are measurably defective only in stimulating PI turnover, but not in sequestrhtion. As the observed changes in coupling efficiency are rather small, one could argue that a threshold activation of second messenger (50430% of maximal wild-type stimulation) is required above which sequestration proceeds at a normal rate. However, the ability of an HmlIP-2 chimeric receptor to sequester normally while being unable to couple to adenylyl cyclase via G proteins (11) argues against any dependence of sequestration on second messenger coupling.
Our results support the hypothesis that the Hml receptor binding pocket of the G protein that stimulates PI turnover is similar but not identical to that of a factor responsible for receptor sequestration, implying that G proteins and any putative sequestration protein are structurally similar. Indeed, several small GTP binding proteins with some structural similarity to the heterotrimeric G coupling proteins have been shown to play a role in protein receptor trafficking (19)(20)(21), and a pl00 protein related to both G, proteins and adaptins was suggested to play a role in receptor trafficking (22). Further, growing evidence indicates a role for heterotrimeric G proteins in cellular protein trafficking (23). It is possible that G proteins not only mediate signaling of GPCRs, but also their sequestration, and other cellular trafficking pathways. Indeed, Thompson et al. (24) proposed that sequestration of muscarinic receptors in SK-N-SH cells (containing mostly Hm3) requires the involvement of a GTP binding protein but not phosphoinositide-derived second messengers. Since both processes are dependent upon agonist stimulation, a common mechanisms of GPCR activation is postulated, with formation of a generic binding pocket in the active receptor state which yields productive signal transduction via G proteins and also subserves other receptor functions such as sequestration. This hypothesis can account for previous results obtained with mutagenesis experiments, where selected mutations may either affect coupling, or sequestration, or both. We can now interpret the unpredictable effects of mutations of the large Hml i3 loop on sequestration (1, 2): conformational changes of the i3 loop, acting as a molecular switch, determine access to a multi-site binding domain located in the i2 loop and the junctions of the i3 loop. However, one cannot exclude the possibility that the alanine point mutations, particularly in the hinge regions of the i3 loop, also cause conformational changes, rather than disrupt direct binding to target molecules.
The postulated multi-site binding domain of H m l respon-sible for agonist induced sequestration differs fundamentally from those of other classes of membrane receptors where a single contiguous domain was found to mediate internalization (e.g. for low density lipoprotein, transferrin, and epidermal growth factor receptors) (3). Thus, strategies for isolating the cellular factors mediating GPCR cellular trafficking must reflect the need for maintaining the structural integrity of the multi-site binding domain in the native receptor.