Intracellular localization of the M1 muscarinic acetylcholine receptor through clathrin-dependent constitutive internalization is mediated by a C-terminal tryptophan-based motif

G-protein-coupled receptors (GPCRs) are a family of seven-pass transmembrane receptors that transduce extracellular signals to the intracellular signaling cascade. Cell surface localization of GPCRs is tightly regulated through synthesis, receptor trafficking, internalization and degradation, all of which contribute to the magnitude and duration of signaling and to the abundance of cell surface versus internalized receptor. GPCR internalization is well characterized and occurs through a clathrin- and dynamin-dependent process involving β-arrestin binding to the agonist-stimulated receptor (von Zastrow, 2003). In this process, β-arrestin generally binds to GPCR that has been phosphorylated by a GPCR kinase. This complex is then recruited into clathrin-coated pits through interaction with clathrin and the adaptor protein-2 (AP-2) complex. However, recent studies have revealed multiple pathways for GPCR internalization, including agonist-independent constitutive internalization (Wolfe and Trejo, 2007). For example, proteinase-activated receptor 1 (PAR1) directly binds to the μ2 subunit of AP-2 through the C-terminal cytoplasmic tail of PAR1 and is constitutively trafficked into the cell by the clathrin machinery (Paing et al., 2006).

Muscarinic acetylcholine receptors (mAChRs), a representative family of GPCRs, are involved in many important physiological functions such as cognitive processes, gland secretion and smooth muscle contraction. The five mAChR subtypes (M1, M2, M3, M4 and M5, encoded by CHRM1–CHRM5, respectively) are generally subdivided into two further groups based on their mechanism of signal transduction. The M1, M3 and M5 receptors are all coupled to phospholipase C (PLC) through G_q/11 and result in mobilization of intracellular Ca²⁺, whereas the M2 and M4 receptors are negatively coupled to adenylate cyclase through G_i/o and act to reduce cAMP levels (Caulfield and Birdsall, 1998; Felder, 1995). All of the subtypes are expressed in the central nervous system but, among them, M1 is the major subtype in cortical and hippocampal neurons (Levey et al., 1995; Levey et al., 1991). Trafficking of mAChRs is regulated in a subtype- and/or cell-type-specific manner (Nathanson, 2008; Reiner and Nathanson, 2012). Interestingly, recent immunohistochemical and pharmacological studies have shown stable intracellular localization of M1-mAChR, especially in neuronal cells (Anisuzzaman et al., 2013; Wang et al., 1994; Yamasaki et al., 2010). In contrast, M2 and M4 subtypes exist predominantly on the cell surface (Liste et al., 2002). Thus, it has been suggested that there is an M1-subtype-specific mechanism for its intracellular localization. Additionally, we recently reported that intracellular M1-mAChRs are physiologically functional receptors whose activation leads to upregulation of mitogen-activated protein kinase (MAPK) activity and contributes to regulation of synaptic plasticity in hippocampal neurons (Anisuzzaman et al., 2013; Uwada et al., 2011). Therefore, the intracellular localization of M1-mAChR is important for not only the downregulation of signals from cell surface receptors, but also for the upregulation of functional intracellular receptors. However, the mechanisms whereby M1-mAChR is localized to intracellular sites are largely unknown.

Internalization of M1-mAChR is currently understood for the canonical agonist-dependent process. Agonist treatment induces M1-mAChR internalization through a dynamin- and clathrin-dependent mechanism, although the contribution of β-arrestin is controversial (Lee et al., 1998; Vögler et al., 1999). However, it is unclear whether M1-mAChR can internalize constitutively without agonist treatment. Therefore, in this study, we examined the internalization machinery for both agonist-mediated and constitutive internalization of M1-mACh and evaluated the contribution of these mechanisms to the intracellular localization of the receptor.

We used N1E-115 neuroblastoma cells as a model of central nervous system (CNS) neurons because the M1-mAChR subtype is predominantly expressed in this neuroblastoma cell line, as well as in brain neurons. However, the density of endogenous mAChRs in N1E-115 cells is extremely low in contrast to the brain (∼100 versus 3000 fmol/mg protein) (Anisuzzaman et al., 2013; Uwada et al., 2011). Therefore, we transfected exogenous mAChRs in order to more readily analyze receptor dynamics. To distinguish exogenous mAChRs from the endogenous receptors, we constructed tagged mAChRs, in which c-Myc was added to the N-terminus of the five different mAChR subtypes (Myc–M1-, Myc–M2-, Myc–M3-, Myc–M4- and Myc-M5-mAChR). Immunofluorescence staining of the Myc epitope showed that transfected Myc–M1-mAChR is localized not only at the cell surface, but also at intracellular sites (white and yellow arrows, respectively, in Fig. 1A), whereas Myc–M2-, Myc–M3-, Myc–M4- and Myc–M5-mAChRs predominantly locate on the cell surface (Fig. 1A). Furthermore, the intracellular M1-mAChR distribution overlapped with that of the trans-Golgi network protein, TGN46 (Fig. 1B). These results, together with our previous demonstration of both the cell surface and intracellular distribution of endogenous M1-mAChR (Uwada et al., 2011), suggest that the intracellular distribution is specific for the M1 subtype, regardless of its origin. In addition, our data indicate the presence of M1-selective trafficking machinery in N1E-115 neuroblastoma cells.

Internalization is a common route for the intracellular localization of GPCRs. We considered that the intracellular distribution of M1-mAChR might result in part from a dynamic equilibrium between its localization to surface and intracellular sites. Thus, we first examined the internalization of all mAChR subtypes. Cell surface Myc-tagged mAChRs were first labeled in situ with anti-Myc antibody at 4°C. After washing out excess antibody, the cells were incubated with or without the muscarinic agonist carbachol at 37°C to follow both agonist-induced and agonist-independent constitutive internalization. Before incubation at 37°C (at 0 min in Fig. 2A, left), antibody signals for all five mAChR subtypes (M1–M5) were detected on the cell surface. After 30 min incubation without carbachol at 37°C, anti-Myc-antibody-labeled M2-, M3-, M4- and M5-mAChRs were retained on the cell surface, whereas M1-mAChR appeared at intracellular sites (Fig. 2A, center). In contrast, incubation with carbachol triggered the internalization of all the mAChR subtypes (Fig. 2A right). Fig. 2B shows time courses for internalization, where the signals of surface Myc-labeled M1- or M3-mAChRs were measured by cell-based ELISA of intact (non-permeabilized) cells. In contrast with the persistent localization of M3-mAChRs at the cell surface, the percentage of membrane-localized M1-mAChR gradually decreased during simple (no agonist) incubation at 37°C (88.7%±7.7 at 15 min, 58.6%±6.0 at 30 min and 46.1%±2.3 at 60 min, as compared with the level at 0 min, mean±s.e.m.). The reduction of surface M1-mAChR was accelerated in the presence of carbachol (71.8%±10.6 at 5 min, 68.7%±7.2 at 15 min, 49.0%±7.6 at 30 min and 45.0%±2.3 at 60 min), and finally reached the same level as simple incubation at 30 and 60 min (Fig. 2B). To exclude the possible stimulation of cell surface M1-mAChR by endogenous agonists in the medium, the cells were treated with muscarinic antagonists (subtype-nonselective atropine, and M1- and M4-selective pirenzepine). Atropine or pirenzepine at 100 µM did not affect the spontaneous internalization of M1-mAChR (Fig. 2C,D), although atropine at this concentration completely inhibited carbachol-induced internalization in the other mAChR subtypes (data not shown). Internalized M1-mAChR in the absence of carbachol was located in the trans-Golgi network (Fig. 2E). From these results we conclude that M1-mAChR undergoes agonist-independent constitutive internalization.

Dynamin is a key constituent of several internalization machineries, including in the clathrin-dependent pathway where it is necessary for the scission of newly formed vesicles in the cell membrane. To examine the possible involvement of dynamin-dependent machinery in the constitutive internalization of M1-mAChR, a dominant-negative mutant of dynamin-2 (K44A) was co-expressed with Myc–M1-mAChR and the distribution of surface receptors in the absence of agonist was followed as described above. The inhibitory effect of the dynamin-2 dominant-negative mutant was verified by observing the uptake of a conventional endocytosis marker, transferrin, which is mediated through a dynamin- and clathrin-dependent mechanism (supplementary material Fig. S1). Expression of wild-type dynamin-2 did not modify the surface and intracellular distribution of M1-mAChR during a 30-min incubation at 37°C, but co-transfection of the dominant-negative mutant K44A, inhibited the constitutive internalization of M1-mAChR (Fig. 3A,B). Next, we examined the involvement of the clathrin-dependent pathway. Constitutive internalization of M1-mAChR was not observed after treatment with hypertonic sucrose or following K⁺ depletion (Fig. 3C,D), both of which are known to inhibit clathrin-dependent internalization (Hansen et al., 1993; Luttrell et al., 1997). The inhibition of clathrin-dependent endocytosis by each treatment was also confirmed using a transferrin uptake assay (supplementary material Fig. S2). The clathrin adaptor AP-2, a heterotetrameric complex composed of α, β2, μ2 and σ adaptin subunits, mediates the association of receptors with the clathrin-coated pit, resulting in internalization from the cell surface to the cytosolic compartment. In this process, the μ2 subunit of AP-2 (AP-2 μ2) plays a central role in receptor recognition. Expression of small hairpin RNA (shRNA) for mouse AP-2 μ2 reduced endogenous AP-2 μ2 expression in co-transfected mCherry-positive cells (Fig. 3E, lower panel), in contrast to negative control shRNA expression (Fig. 3E, upper panel). Reduction of AP-2 μ2 protein by shRNA abolished constitutive internalization of M1-mAChR (Fig. 3F) and resulted in retention of the M1-mAChR at the plasma membrane even after 30 min incubation at 37°C (Fig. 3G). Therefore, constitutive internalization of M1-mAChR appears to involve the AP-2-μ2-mediated clathrin pathway.

Arrestins are multifunctional adaptor proteins known to associate with the clathrin endocytic machinery for agonist-stimulated GPCRs. In some studies, β-arrestin has also been shown to mediate constitutive internalization of a number of GPCRs (Barak et al., 2001; Galliera et al., 2004). Therefore, we tested the possible contribution of β-arrestin to the constitutive internalization of M1-mAChR by using a dominant-negative β-arrestin-2 C-terminal mutant (β-arrestin-2 Ct). Like wild-type β-arrestin-2, constitutive internalization of M1-mAChR was observed in the cells expressing β-arrestin-2 Ct (Fig. 3H, left panel, Fig. 3I). Given that agonist-dependent and agonist-independent internalization were both elicited for M1-mAChR, the effectiveness of β-arrestin-2 Ct was examined for the subtype M3-mAChR and we found that for this protein the mutant completely inhibited carbachol-induced internalization (Fig. 3H, right panel, Fig. 3I). Thus, our data suggest that constitutive internalization of M1-mAChR is mediated through the pathway dependent on dynamin, clathrin and its adaptor protein AP-2, but not on β-arrestins.

Clathrin-dependent internalization of many membrane proteins, including GPCRs, is regulated by their C-terminal cytosolic tails (Pandey, 2010). C-terminal amino acid sequences of the mAChR subtypes are shown in Fig. 4A. First, we made chimeric mutants of M1 and M3 subtypes, in which the C-terminal sequences were exchanged between the two subtypes. Unlike wild-type M1-mAChR (Fig. 2A), the M1 mutant containing the M3 C-terminal tail (amino acids 543–590) lost its constitutive internalization capacity (Fig. 4B left). By contrast, the M3 mutant with the M1 C-terminal tail (amino acids 417–460) acquired constitutive internalization activity (Fig. 4B, right), unlike for wild-type M3-mAChR (see Fig. 2A). These results show that the C-terminal tail of M1-mAChR is essential for the constitutive internalization process.

The AP-2 complex can bind to some membrane proteins directly, causing their constitutive endocytosis, and several short peptide sequences have been identified as recognition motifs for AP-2 subunits (Pandey, 2010). The tyrosine-based YxxΦ motif, where Y is tyrosine and Φ is a bulky hydrophobic residue, represents a consensus sequence for direct binding to the μ2 subunit of AP-2. M2- and M4-mAChR did not show significant internalization without agonist stimulation (see Fig. 2A), though these subtypes contain the YxxΦ motif in their C-terminal tail (Y⁴⁵⁹KNI⁴⁶² and Y⁴⁷²RNI⁴⁷⁵, respectively). The [D/E]xxxLL motif is also a common endocytosis signal, which interacts mainly with the β-subunit of AP-2. Among the five mAChR subtypes, only M1-mAChR contains a DxxxLL motif in its C-terminal tail (D⁴²⁷TFRLL⁴³²). Leucine residues at this site are important for forward trafficking, and mutation causes the protein to accumulate in the ER and shows no cell surface expression (Sawyer et al., 2010; also see Fig. 7B in this paper). Therefore, we only checked the effect of the mutation of D⁴²⁷ to A (D427A) in M1-mAChR. However, the D427A mutation had no apparent effect on the constitutive internalization of the receptor (Fig. 4C,D). Next, we introduced a single point mutation with an alanine residue in the amino acid sequence to replace W⁴³⁷ to G⁴⁵⁰. Among these mutants, only W442A and I445A showed a significant loss of constitutive internalization (Fig. 4C,D). Tryptophan is an aromatic amino acid, as is tyrosine, and isoleucine is a canonical hydrophobic amino acid. Importantly, the neonatal Fc receptor and the synaptic vesicle membrane protein, synaptotagmin 1, have been reported to use a WxxΦ motif instead of the YxxΦ sequence for AP-2 μ2 binding and internalization (Jarousse et al., 2003; Wernick et al., 2005; Wu and Simister, 2001). Therefore, the ability of the M1-mAChR C-terminal WxxΦ motif to bind to the AP-2 μ2 subunit was examined in a GST pulldown assay. A GST-fused RRW⁴⁴²RKI⁴⁴⁵PK peptide effectively pulled down His-tagged AP-2 μ2 (Fig. 5A). Constitutive-internalization-deficient mutations (W442A and I445A) abolished the binding affinity for AP-2 μ2. Interestingly, a tyrosine substitution (W442Y) led to more effective interaction with AP-2 μ2, indicating that a tyrosine-based motif is a more effective target of constitutive internalization than a tryptophan-based motif. In addition, a competition test using a structural analogue of tyrosine was conducted. Tyrphostin AG18 has been described as an internalization inhibitor due to competitive binding to the tyrosine-based-binding site of AP-2 μ2 (Banbury et al., 2003). AG18 inhibited the binding of His–AP-2-μ2 to the RRW⁴⁴²RKI⁴⁴⁵PK peptide in a concentration-dependent manner (Fig. 5B). Furthermore, AG18 effectively blocked the constitutive internalization of M1-mAChR (Fig. 5C,D). Taken together, our data indicate that the C-terminal tryptophan-based motif of M1-mAChR is essential for its constitutive internalization and binding to AP-2 μ2.

Next, we checked whether the C-terminal tryptophan-based motif is involved in agonist-stimulated internalization of M1-mAChR. In contrast to agonist-free incubation at 37°C (Fig. 4C,D), 1 mM carbachol caused significant internalization of the M1-mAChR mutants W442A and I445A (Fig. 6A,C). This internalization was inhibited by dominant-negative β-arrestin-2 Ct, but not by wild-type β-arrestin-2 (Fig. 6B,C). These results show that M1-mAChR is internalized through two distinct pathways: β-arrestin-dependent internalization by agonist-stimulation, and clathrin-dependent but β-arrestin-independent constitutive internalization through its C-terminal tryptophan-based motif.

Finally, the contribution of internalization mediated by the tryptophan-based motif to the intracellular localization of M1-mAChR was examined. Immunofluorescence observation using antibody for the N-terminal Myc epitope tag showed predominant expression on the cell surface and a significant loss of trans-Golgi network localization for W442A or I445A mutant M1-mAChR (Fig. 7A), similar to that observed for other mAChR subtypes (see Fig. 1A). These changes were confirmed by non-epitope-tagged M1-mAChR mutants, which were detected with an M1 antibody for an endogenous receptor epitope sequence (data not shown). Thus, a tryptophan-based motif is essential for the trans-Golgi network localization of M1-mAChR. Interestingly, a tyrosine substitution mutation (W442Y) caused most receptors to localize in the trans-Golgi network, with a resulting loss of cell surface localization. An estimate of the constitutive internalization of the W442Y mutant, as illustrated in Fig. 2A, was difficult owing to its extremely low level cell surface expression. Therefore, we assessed the internalization of the W442Y mutant by monitoring anti-Myc antibody uptake activity at 37°C (Fig. 7B). N1E-115 cells expressing wild-type and mutants of M1-mAChR were incubated with anti-Myc antibody at 37°C for 30 min. After washing to remove excess antibody, cells were fixed and permeabilized to label the receptor–Myc-antibody complexes with secondary antibody. Wild-type and the W442Y mutant of M1-mAChR showed intracellular signals corresponding to the uptake of anti-Myc antibody through receptor-mediated internalization. However, the L432A+L433A mutants, which were retained in the ER, did not show any intracellular anti-Myc antibody signal due to a lack of cell surface expression (Fig. 7B, left panels). This distribution of M1-mAChR was confirmed by staining with anti-M1-mAChR (Fig. 7B, right panels). Thus, the results indicate that the W442Y mutant localizes exclusively at an intracellular site through more active constitutive internalization compared with forward trafficking to the cell surface.

Surface and intracellular mAChRs can be distinguished pharmacologically by using two muscarinic ligands, hydrophilic N-methyl scopolamine (NMS) and hydrophobic 3-quinuclidinyl benzilate (QNB), based on their distinct membrane permeabilities. To evaluate the relative abundance of intracellular versus membrane-localized M1-mAChR, we measured the binding of [³H]NMS (membrane-located receptors) and [³H]QNB (membrane plus intracellular receptors) with intact N1E-115 cells at 4°C (Fig. 7C). Whole-cell binding of these drugs to cells expressing the wild-type M1-mAChR showed that the binding of [³H]NMS to the membrane receptors was ∼50% of the total receptor binding estimated by [³H]QNB binding. By contrast, the binding of [³H]QNB to the internalization-deficient mutant M1-mAChRs (W442A and I445A) was decreased relative to the binding of [³H]NMS. These results are consistent with the immunofluorescence studies and represent the reduced intracellular localization of M1-mAChR due to the lower activity of constitutive internalization. In contrast, the W442Y mutation that leads to an intracellular accumulation of receptor caused a significant reduction of [³H]NMS binding, indicating a loss of cell surface expression. Taken together, our data indicate that the tryptophan-based motif in the C-terminal tail of M1-mAChR is important for its proper distribution between cell surface and intracellular sites.

The main finding of our study is that M1-mAChR undergoes constitutive internalization and intracellular accumulation through a mechanism involving its C-terminal tryptophan-containing WxxΦ motif, which interacts with the μ2 subunit of the AP-2 complex and leads to clathrin-dependent internalization.

Constitutively intracellular localization was observed only in the M1 subtype among the five mAChR subtypes and depended on an M1-specific C-terminal sequence. Such intracellular localization has also been found for endogenous M1-mAChR in N1E-115 cells (Uwada et al., 2011), and in the cortex and hippocampal neurons of rats, mice and humans (Anisuzzaman et al., 2013; Yamasaki et al., 2010). More recently, we observed constitutive internalization of M1-mAChR that was dependent on a WxxΦ motif in exogenously transfected HEK293 and HeLa cells (our unpublished observations). Thus, it is likely that intracellular localization of M1-mAChR can occur in a wide variety of cells and tissues. Of note, M3-mAChR has been reported to internalize constitutively in HeLa cells through clathrin-independent endocytosis (Scarselli and Donaldson, 2009). However, we could not detect constitutive internalization of M3-mAChR in the N1E-115 cells. The reasons for these differences between HeLa and N1E-115 cells for M3-mAChR internalization remain to be determined.

Although several GPCRs are known to internalize constitutively, there are multiple mechanisms that mediate this process. For example, the constitutive internalization of β2 adrenergic receptor and M3-mAChRs is clathrin independent (Scarselli and Donaldson, 2009). PAR1, G-protein-coupled estrogen receptor (GER) and chemokine decoy receptor D6 (D6) internalize through clathrin-dependent machinery. However, the internalization of the D6 receptor, unlike the internalization of PAR1 and GER, involves an interaction with β-arrestin (Cheng et al., 2011; Galliera et al., 2004; Paing et al., 2006). Constitutive internalization of PAR1, in keeping with the mechanism we describe for M1-mAChR, is mediated by direct binding of its C-terminal YxxΦ motif to the μ2 subunit of AP-2. Thus far, the YxxΦ motif is thought of as a canonical sequence for AP-2 μ2 binding. In this motif, it has been suggested that the hydroxyl group of tyrosine is essential for recognition by AP-2 μ2 (Owen and Evans, 1998). However, the neonatal Fc receptor and synaptotagmin 1 have been found to internalize through direct binding of their tryptophan-containing motif, WxxΦ, to AP-2 μ2 (Jarousse et al., 2003; Wernick et al., 2005; Wu and Simister, 2001). These studies suggest that aromatic amino acids in this sequence are sufficient for AP-2 μ2 recognition. In addition, positively charged residues between W and Φ, such as R and K, might also contribute to AP-2 μ2 recognition (Boll et al., 1996). Interestingly, our study shows that a YxxΦ sequence interacts more effectively with AP-2 μ2 than the WxxΦ motif (Fig. 5A). Consistent with this result, the W442Y mutation causes greater internalization of cell surface M1-mAChR than the wild-type sequence (Fig. 7). Therefore, a tryptophan-based motif as opposed to a tyrosine-based sequence might provide for a unique balance of mAChRs distributed between the cell surface and intracellular sites.

As well as agonist-independent constitutive internalization, agonist-triggered M1-mAChR internalization is thought to proceed through dynamin- and clathrin-dependent pathways (Shmuel et al., 2007; Tolbert and Lameh, 1996). Although some studies suggest that this process involves β-arrestin recognition, other reports have indicated that β-arrestin-independent internalization can also occur (Lee et al., 1998; Vögler et al., 1999). In the present study, we demonstrate that agonist-stimulated internalization of M1-mAChR is indeed dependent on β-arrestin, whereas the constitutive receptor internalization process does not depend on β-arrestin. Therefore, there are at least two mechanisms for M1-mAChR internalization.

Internalized GPCR could be sorted to and equilibrated by distinct pathways: recycling to the cell surface, degradation in lysosomes or trafficking to specific cellular domains. Interestingly, the ELISA experiment shown in Fig. 2B indicates that the amount of anti-Myc-antibody-labeled M1-mAChR at the cell surface was relatively stable during a 30- to 60-min incubation. If it is assumed that the speed of internalization of M1-mAChR is constant, this result indicates recycling of anti-Myc-antibody-labeled M1-mAChR to the cell surface. Additionally, our pharmacological-ligand-binding study showed that the cell surface concentration of wild-type M1 was at the same level as that of constitutive internalization deficient mutants (W442A and I445A), whereas the total amount of M1-mAChR was higher for wild-type compared with the mutants. Therefore, in N1E-115 cells, internalized M1-mAChR seems not to undergo substantial degradation. These results indicate that the cell surface and intracellular M1-mAChR level is maintained by the balance between internalization and forward trafficking to the cell surface.

We do not know what the physiological role of the constitutive internalization of M1-mAChR might be. One simple possibility is the regulation of receptor signaling through a reduction of cell surface expression of M1-mAChR. A more attractive hypothesis is that this constitutive internalization process maintains a key level of functional intracellular M1-mAChR that is important for intracellular signaling. We recently reported that intracellular M1-mAChR in neuronal cells has a distinct signaling role compared with cell surface M1-mAChR (Anisuzzaman et al., 2013; Uwada et al., 2011). Metabotropic glutamatergic receptor 5 (mGluR5) also has an intracellular function in striatal neurons (Jong et al., 2005). Recently, the constitutive internalization of mGluR5 has been reported, although the physiological function of these intracellular receptors has not yet been established (Trivedi and Bhattacharyya, 2012). Thus, the relationship between constitutive internalization and intracellular function of GPCRs represents an important issue for future work.

In summary, our data show that a geographically distinct subcellular distribution of M1-mAChR is regulated by constitutive internalization. Given that cell surface and intracellular M1-mAChR could contribute to synaptic plasticity through distinct signaling cascades in the hippocampus (Anisuzzaman et al., 2013), the regulation of constitutive internalization might play a role in muscarinic cognitive processes and thus have an important physiological and pathophysiological function.

Compounds purchased from commercial sources were as follows: [³H]NMS (specific activity 3.00 TBq/mmol) and [³H]QNB (specific activity 1.81 TBq/mmol) from Amersham Biosciences (Buckinghamshire, UK), atropine sulfate from Nacalai Tesque (Kyoto, Japan), carbachol and pirenzepine from Sigma-Aldrich (St Louis, MO), TMB One Component HRP Microwell substrate from SurModics (Eden Prairie, MN) and Tyrphostin AG18 from Cayman Chemical (Ann Arbor, MI). Antibodies were: mouse monoclonal anti-Myc antibody from Wako (Tokyo, Japan), horseradish peroxidase (HRP)-conjugated anti-mouse IgG and anti-rabbit IgG antibodies from Sigma-Aldrich, anti-His₆ antibody from Roche (Mannheim, Germany), anti-AP-2 μ2 antibody from BD Biosciences (San Jose, CA), anti-M1-mAChR antibody from Frontier Institute (Hokkaido, Japan), and anti-mouse IgG conjugated to DyLight488 and anti-rabbit IgG conjugated to Cy3 from Jackson ImmunoResearch (West Grove, PA).

DNA encoding human M1-, M2-, M3-, M4- and M5-mAChRs were subcloned into the EcoRI site of pCMV-myc vectors. Mouse TGN46 and dynamin-2 were constructed into the EcoRI/SalI site of pmCherry-N1 vectors. Wild-type and N-terminal (amino acids 1–318) truncated mouse β-arrestin-2 were inserted into the SalI site of pmCherry-N1 vectors. M1-mAChR with point mutations in the cytoplasmic C-terminal tail, and the K44A mutation of dynamin-2, were constructed by ‘mega-primer’ PCR. DNA encoding peptide, representing the C-terminal 440–447 amino acids of M1-mAChR (RRWRKIPK) and mutants (RRARKIPK, RRWRKAPK and RRYRKIPK) were subcloned into pGEX-6P1. DNA encoding the C-terminal region of human AP-2 μ2 (amino acids 160–435) was subcloned into pET-28(b). TRC2-pLKO-puro-shAP-2 μ2, which encodes short hairpin RNAs (shRNAs) targeting mouse AP-2 μ2 (TRCN0000287805), was purchased from Sigma-Aldrich. Empty vector was used as a negative control. All sequences were verified by the dideoxy chain termination method.

N1E-115 cells, passages 5–10, were maintained in high glucose Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (v/v) fetal bovine serum (FBS) and antibiotics (penicillin and streptomycin). The cells were grown under 5% CO₂ at 37°C. Transient transfection was performed using X-tremeGENE9 reagent (Roche, Mannheim, Germany) according to the manufacturer's instructions.

For normal immunofluorescence (Fig. 1A,B; Fig. 7A), N1E-115 cells were transfected with Myc–M1-mAChR wild-type or mutants for 36 h, fixed with 4% paraformaldehyde in PBS for 15 min at room temperature, then permeabilized with 0.2% Triton X-100 in PBS for 5 min. For AP-2 μ2 staining, cells were fixed in methanol at −20°C for 5 min (Fig. 3E). After permeabilization, the cells were blocked with 0.2% BSA and incubated with anti-Myc antibody for 1.5 h, followed by incubation with secondary antibody for 2 h.

For the internalization analysis, N1E-115 cells were grown on poly-L-lysine-coated glass coverslips. Following 36 h incubation after transfection with Myc-M1-mAChR wild-type or mutants (72 h incubation for coexpression with shRNA), the cells were washed once with PBS and pre-labeled with anti-Myc monoclonal antibody diluted in DMEM containing 10% FBS for 1 h at 4°C. After three washings, the cells were incubated in medium at 37°C for 30 min in the presence or absence of 1 mM carbachol, 400 mM sucrose or 1 mM Tyrphostin AG18. K⁺ depletion was performed as previously described (Hansen et al., 1993). In Fig. 2C, cells were treated with 100 µM atropine or 100 µM pirenzepine for 15 min at 4°C before 37°C incubation. Thereafter, cells were fixed with 4% paraformaldehyde in PBS for 15 min at room temperature. In Fig. 7B, cells were incubated with anti-Myc antibody at 37°C for 30 min and washed three times to remove excess antibody before 4% paraformaldehyde treatment. After fixing, the cells were permeabilized with 0.2% Triton X-100 in PBS for 5 min. After blocking with 0.2% BSA, the cells were incubated with secondary antibody for 2 h. The coverslips were mounted on glass slides using glycerol-DABCO. Images were obtained with a confocal microscope (TCS-SP2-AOBS; Leica Microsystems, Wetzlar, Germany).

N1E-115 cells were seeded onto poly-L-lysine-coated 24-well plates. After overnight incubation, expression plasmids encoding the N-terminal Myc-tagged muscarinic receptors were transfected into the cells. After 36 h, the cells were washed with ice-cold medium and then incubated with mouse anti-Myc antibody in 10% serum containing DMEM at 4°C for 60 min. Cells were washed with cold serum-free medium to remove excess antibody, suspended in pre-warmed 37°C medium with or without 1 mM carbachol, 100 µM, 400 mM sucrose or 1 mM Tyrphostin AG-18, and incubated at 37°C for the indicated time (0 min, 5 min, 15 min, 30 min or 60 min). In Fig. 2D, cells were treated with or without 100 µM atropine or 100 µM pirenzepine for 15 min at 4°C before 37°C incubation. K⁺ depletion was performed as previously described (Hansen et al., 1993). Then, cells were fixed in 4% paraformaldehyde in PBS for 20 min at room temperature and incubated with blocking buffer containing 0.2% BSA and 1% low-fat milk for 30 min at room temperature, then incubated for 2 h in HRP-conjugated anti-mouse IgG diluted in blocking buffer. After washing, the cells were incubated with TMB One Component HRP Microwell substrate at room temperature for 30 min. The reaction was stopped by addition of the same volume of 1 M HCl. The optical density of an aliquot was measured at 450 nm using a Molecular Devices SpectraMax Plus spectrophotometer (Sunnyvale, CA). Non-specific binding of antibodies was estimated from the results from non-transfected cells. Each value was normalized against the amount present before incubation at 37°C.

Expression constructs for the GST-fused M1-mAChR C-terminal sequences (pGEX-6P1 -RRW⁴⁴²RKI⁴⁴⁵PK, -RRA⁴⁴²RKI⁴⁴⁵PK, -RRW⁴⁴²RKA⁴⁴⁵PK and -RRY⁴⁴²RKI⁴⁴⁵PK) and the construct for His₆-tagged C-terminal (amino acids 160–435) mouse AP-2 μ2 in pET-28(b) were transformed into Escherichia coli (E. coli) BL21 (DE3). The expression of recombinant proteins and the GST pulldown assay was carried out according to Uchimura et al. with some modifications (Uchimura et al., 2006). Briefly, bacterially-expressed GST and GST fusion proteins were immobilized on glutathione–Sepharose beads (Amersham Biosciences, Buckinghamshire, UK) and incubated with His₆-tagged AP-2-μ2-expressing E. coli lysates in a buffer (500 µl) consisting of 25 mM HEPES pH 7.4, 120 mM NaCl, 4.7 mM KCl, 1.2 mM KH₂PO₄, 1.2 mM MgSO₄, 2 mM CaCl₂, 11 mM glucose and 0.5% Triton X-100 for 30 min at 4°C. For the competition experiments with AG-18, His₆-tagged AP-2-μ2-expressing E. coli lysates were preincubated with 0.02 volumes of the amount of AG18 solution in DMSO for 15 min at 25°C. The precipitated proteins were eluted with SDS-sample buffer and subjected to western blotting as described previously (Uwada et al., 2011).

N1E-115 cells were transiently transfected 24 h after plating with pCMV-myc-M1 wild-type or mutants. After 36 h of culture, the cells were subjected to whole-cell binding assays at 4°C as described previously (Sathi et al., 2008).

We thank Morley D. Hollenberg (Department of Physiology and Pharmacology, University of Calgary, Canada) for helpful discussions and comments concerning the writing of our manuscript. We also gratefully acknowledge Reiko Inaki for her secretarial assistance in the development of this study.