Dual Role of a Dileucine Motif in Insulin Receptor Endocytosis*

Two leucines (Leu986 and Leu987) have recently been shown to take part in the control of human insulin receptor (HIR) internalization (Renfrew-Haft, C., Klausner, R. D., and Taylor, S. I. (1994) J. Biol. Chem. 269, 26286–26294). The aim of the present study was to further investigate the exact mechanism of this control process. Constitutive and insulin-induced HIR internalizations were studied biochemically and morphologically in NIH 3T3 cells overexpressing either a double alanine (amino acid residues 986–987) mutant HIR (HIR AA1) or HIR truncated at either amino acid residue 981 (HIR Δ981) or 1000 (HIR Δ1000). Data collected indicate that: (a) the three mutant HIR show a reduced association with microvilli as compared with HIR wild-type; (b) the two receptors containing the dileucine motif (HIR WT and HIR Δ1000) show the highest propensity to associate with clathrin-coated pits, independently of kinase activation; (c) the two receptors lacking the dileucine motif but containing two tyrosine-based motifs, previously described as participating in clathrin-coated pit segregation, associate with these surface domains with a lower affinity than the two others, (d) in the presence of the kinase domain, an unmasking of the tyrosine-based motifs mediated by kinase activation is required. These results indicate that the dileucine motif is not sufficient by itself, but participates in anchoring HIR on microvilli and that another sequence, located downstream from position 1000 is crucial for this event. This dileucine motif also plays a role in HIR segregation in clathrin-coated pits. This latter function is additive with that of the tyrosine-based motifs but the role of the dileucine motif predominates. Eventually, the clathrin-coated pit anchoring function of the dileucine motif is independent of receptor kinase activation in contrast to the tyrosine-based motifs.

The internalization of many signaling receptors (i.e. insulin, epidermal growth factor receptors, CD4 . . . ) is a ligand-dependent process which requires the ligand-mediated activation of a tyrosine kinase, intrinsic to or associated with the cytoplasmic domain of the receptor (1)(2)(3)(4)(5). In the case of the insulin receptor, activation of this tyrosine kinase releases a constraint maintaining the unoccupied receptor on cytoskeleton-rich microvilli (1,2,6). The activated receptor then moves in the plane of the plasma membrane, and gains access to the non-villous domains of the cell surface, where it is internalized via the clathrin-coated pits. Very little information is available regarding the receptor domain(s) involved in the retention of unoccupied receptors on microvilli or about the cellular component(s) that anchor(s) the insulin receptor on microvilli. The mechanism(s) allowing kinase activation to induce the release of the insulin receptor to migrate away from microvilli are also unresolved. In contrast, some light was shed on the subsequent surface steps. The speed of the surface shift has been shown to depend at least in part on the structure of the receptor transmembrane domain (7)(8)(9) similar to what was observed in the case of other receptors (10). Furthermore, the last surface event, the association with clathrin-coated pits, has been shown to require the integrity of two tyrosine-based signals present in the juxtamembrane cytoplasmic domain of the insulin receptor (GPLY and NPEY) (11)(12)(13)(14). Similar motifs forming a ␤ turn and exposing an aromatic amino acid (preferentially a tyrosine) have been described as participating in the internalization of various receptors including receptors for which internalization is ligand-independent (3,(15)(16)(17)(18)(19).
Recently, a dileucine motif present within exon 17 (Leu 986 and Leu 987 ) has been shown to participate in the control of insulin receptor internalization and sorting inside the cell (20). Mutation of these two leucines for alanines indeed results in a decreased insulin-induced internalization of the mutant receptor. In addition, the direct fusion of this dileucine motif (together with four neighboring amino acids) to the carboxyl terminus of Tac to form a chimeric molecule identified this motif as a lysosomal sorting sequence (20). The present work was designed to further investigate the role of this dileucine motif encoded by exon 17. To that end, two truncated receptors (at positions 981 and 1000, respectively) were prepared in addition to the dileucine mutant receptor previously described. Constitutive as well as insulin-induced internalization processes were analyzed both biochemically and morphologically at the electron microscopic level. Results collected indicate that the sequence anchoring the unoccupied insulin receptors on microvilli is located downstream from Glu 1000 and that the dileucine motif at positions 986 -987 participates in the control of this anchoring. In addition, our observations reveal that this dileucine motif plays a role in the segregation of the insulin receptor in clathrin-coated pits. This latter function of the dileucine motif is additive with that of the tyrosine-based motifs present in exon 16 but the role of the dileucine motif predominates. Receptor tyrosine kinase is not required for this function of the dileucine motif. In contrast, the function of tyrosine-based motifs to segregate receptors in clathrin-coated pits requires kinase activation.

Construction of Truncated and Mutant Human Insulin Receptors by
Mutagenesis-The plasmids containing either the mutant insulin receptor with 2 alanines substituted for the Leu-Leu pair at positions 986 -987 or the insulin receptor truncated at position 981 were previously described (20 -22). The receptor truncated at position 1000 was constructed as follows. A fragment containing nucleotides 2289 -3265 of the human insulin receptor cDNA (23) was amplified by polymerase chain reaction using the full-length insulin receptor cDNA cloned into pSP64 vector (Promega Corp., Madison, WI) as template and the following oligonucleotide primers: 5Ј-GCCGAGGACCCTAGGCCATCTC-GGA-3Ј (nucleotides 2289 -2313 of the sense strand with an AvrII restriction site (underlined)) and 5Ј-AAGACTTAAGCTAATACACCAT-GCC-3Ј (nucleotides 3138 -3162 of the antisense strand with a BfrI restriction site (underlined) and a stop codon (bold)). The polymerase chain reaction product was digested with AvrII and BfrI to yield a 858-base pair fragment. This fragment was substituted for the AvrII/ BfrI segment (2298 -4326) of the wild-type insulin receptor cDNA. This construction was verified by nucleotide sequencing and subcloned into a bovine papilloma virus-based expression vector pBPV (Pharmacia Biotech Inc.), in which insulin receptor cDNA expression was driven by the murine metallothionin promoter.
Cell Expression of Mutant Insulin Receptors-NIH 3T3 cells (ϳ750,000 cells) in Petri dishes (10 cm in diameter) were transfected using 40 l of Lipofectin (Life Sciences Technologies Inc.) and a mixture of 1 g of pRSV-Neo (a plasmid encoding neomycin resistance) and 5 g of pBPV-HIR ⌬1000. After selection for resistance to the antibiotic G418 (800 g/ml) (Life Technologies, Inc.), stable transfectants were isolated by FACS (fluorescence-activated cell sorter). Cells were recovered by EDTA treatment and incubated for 30 min with 500 l of biotinylated anti-HIR 1 antibody (1:500 in phosphate-buffered saline, 1% bovine serum albumin) at 4°C, After three washes, 100 l of streptavidin Rphycoerytherin conjugate (1:20) (Caltag Laboratories Inc., San Franscico, CA) were added for 30 min in the dark. Flow cytometry was performed on a Becton Dickinson FACScan. Cell surface expression was verified by 125 I-insulin binding.
Biochemical Analysis of Constitutive and Insulin-induced Internalizations-Confluent monolayers of NIH 3T3 cells expressing either the wild-type or mutant receptors were first incubated for 2 h at 4°C in the presence of 5 pM 125 I-mAb 83-14 (constitutive internalization) or 50 pM 125 I-insulin (insulin-induced internalization) then transferred to 37°C for various periods of time. At each time point studied, the unbound 125 I-ligand was removed and cells were washed three times with cold phosphate-buffered saline. Cells were then subjected to three 5-min washes with cold phosphate-buffered saline at pH 1.5 to release the 125 I-ligand bound at the cell surface. Finally, cells were lysed in 1 N NaOH and all samples (acid washes and lysates) were counted in a ␥-counter to determine the internalized radioactivity versus total radioactivity.
Internalization of Biotinylated Insulin Receptors-Confluent monolayers of NIH 3T3 cells were grown up in Petri dishes (10 cm in diameter). Cell surface proteins were biotinylated as described in Ref. 24. Briefly, cells were incubated in phosphate-buffered saline containing 0.5 mg/ml N-hydroxysuccinimide long chain biotin (Pierce) for 30 min at 4°C. Cells were then incubated in 5 ml of binding buffer in the absence or presence of 10 Ϫ7 M insulin for 15 min at 37°C. Cell surface proteins were subjected to Pronase digestion (2.5 mg/ml, Boehringer Mannheim) for 1 h at 4°C. Thereafter, cells were washed and solubilized in 500 l of lysis buffer containing protease inhibitors. Insulin receptors were immunoprecipitated with a mixture of rabbit antibodies (B7 and B10) or with a monoclonal antibody (83-14) directed against the ␣-subunits of the human insulin receptor. After SDS-polyacrylamide (6.5%) gel electrophoresis and electroblotting, nitrocellulose sheets were probed with horseradish peroxidase-linked streptavidin (Amersham) at a dilution of 1:500. After extensive washes, enhanced chemiluminescence (ECL) detection was performed according to the manufacturer's instructions (Amersham). Signals were analyzed by densitometry (Molecular Dynamics).
Quantitative EM Autoradiography-After incubation at 37°C in the presence of tracer amount of 125 I-insulin (10 Ϫ11 M), cells were fixed, dehydrated, and quantitated as described previously (1). For each time point studied, three Epon blocks were prepared and sectioned. About 450 -600 grains were analyzed from morphologically intact cells. Grains within a distance of 1 Ϯ 250 nm from the plasma membrane were considered associated with cell surface; grains inside the cytoplasm and Ͼ250 nm from the plasma membrane were considered internalized. Grains present at the plasma membrane fell into the following classes: microvilli, clathrin-coated pits, non-villous nonclathrin-coated pit segments, and uninterpretable. Grains were considered associated with microvilli or clathrin-coated pits if the center was Ͻ250 nm from the surface domains.

Expression and Autophosphorylation of Normal and Mutant Human Insulin Receptors in NIH 3T3
Fibroblasts-NIH 3T3 fibroblast cell lines were transfected with an expression vector encoding wild-type or one of three mutant insulin receptors: HIR with a pair of alanines in place of the dileucines normally present at positions 986 and 987 (AA1) (20), HIR truncated at position 981 (⌬981) or truncated at position 1000 (⌬1000) (Fig.  1). HIR AA1 exhibited close to normal ␤-subunit phosphorylation as well as insulin-induced phosphorylation of insulin receptor substrate-1 (20), while, as expected, HIR ⌬981 and HIR ⌬1000, were not autophosphorylated (data not shown).

Constitutive and Insulin-stimulated Internalizations of Insulin Receptors in NIH 3T3 Fibroblasts Expressing Normal or Mutant Insulin
Receptors-HIR cytoplasmic domain contains several dileucine motifs, one of which (leucines 986 -987) has been proposed to play a role in HIR internalization (20). In an attempt to clarify the exact function of this dileucine motif in the internalization process, the internalization previously reported defect was studied by comparing constitutive and insulin-induced insulin receptor internalizations in HIR AA1 and HIR WT cells by various methodological approaches. First, comparison of the constitutive internalization (i.e. in the absence of insulin) of the two receptors was carried out by using a monoclonal anti-human insulin receptor antibody (83-14) coupled to 125 I as a ligand (25). Under these conditions, constitutive HIR internalization observed in AA1 cells and in WT cells were identical (Fig. 2). Second, insulin-induced internalization was compared in WT and AA1 cells by tracking 125 Iinsulin internalization both by the acid wash technique and by quantitative EM autoradiography. Both techniques confirmed an inhibition of the internalization rate and a decrease in the total amount of radioactivity recovered inside AA1 cells at each time point studied ( Fig. 2 and data not shown). Third, we compared constitutive and insulin-induced internalization in the same experiment by analyzing internalization of biotinylated receptors. According to this procedure, cell surface proteins were labeled with biotin, and cells further incubated at 37°C in the presence or absence of insulin (10 Ϫ7 M) for 15 min to promote receptor internalization. At the end of this incubation, biotinylated receptors at the cell surface were digested with Pronase. Internalized insulin receptors were protected from digestion. After cells were solubilized, biotinylated receptors were immunoprecipitated with anti-insulin receptor antibody (mAb 83-14). Biotinylated receptors were quantitated in a blotting experiment using horseradish peroxidase-labeled streptavidin (Figs. 3, A and B). Under these circumstances, the slow constitutive internalization of AA1 receptors was accelerated by insulin, but this stimulation was smaller in magnitude than that observed in WT cells, where more than 70% of biotinylated receptors became resistant to Pronase digestion by 15 min of incubation at 37°C (Fig. 3). Taken together, these internalization studies demonstrate that insulin-stimulated HIR internalization is inhibited when the two leucines present at positions 986 and 987 are replaced by alanines, while constitutive internalization is relatively normal in this mutant receptor.
To collect further information on the role of the dileucine per se, internalization studies analogous to the one described above, were carried out with two truncated receptors. These two receptors (HIR ⌬981 and HIR ⌬1000) have in common the absence of the kinase domain but they differ by the presence (HIR ⌬1000) or absence (HIR ⌬981) of the dileucine motif. Following biotinylation of the cell surface, constitutive internalization of HIR ⌬981 and HIR ⌬1000 was more rapid than in the case of HIR WT (Fig. 3). Moreover, insulin did not accelerate internalization of either truncated receptors, confirming that these two mutant receptors were internalized in a constitutive way (Fig. 3). Based on these observations, further studies of these two truncated receptors were performed by measuring receptor-mediated endocytosis of 125 I-mAb 83-14.
In both cases, the internalization rate as well as the maximal amount of radioactive material internalized were significantly higher than those recorded in WT cells (Fig. 4). The highest values were obtained in ⌬1000 cells, where Ͼ55% of total cell associated radioactivity was found inside the cells by 15 min of incubation at 37°C (Fig. 4). In the case of ⌬981 cells, both the internalization rate and the maximal amount of radioactivity internalized were intermediate between those obtained in ⌬1000 cells and in WT cells (Fig. 4). Thus, HIR ⌬1000 contains, in its native and unoccupied state, all the elements allowing the optimal internalization of HIR while HIR ⌬981 appears to lack one or more of these elements. Normal and Mutant Insulin Receptors-To further understand why the two truncated receptors differ in terms of their internalization capacities and in an attempt to shed some light on how the 986 -987 dileucine motif participates in HIR internalization, a morphological ultrastructural analysis of the surface distribution of the above described mutants was carried out.
As determined by quantitative EM autoradiography analysis, in conditions revealing the localization of unoccupied receptors (2 h of incubation at 4°C in the presence of 125 I-mAb 83-14), HIR WT preferentially associated with microvilli on the surface of NIH 3T3 cells (Fig. 5). In contrast, in the case of receptors truncated at position 981 and 1000 and constitutively internalized, such preferential initial localization was not observed (Fig. 5). These observations confirm and extend the previously described lack of association with microvilli of a receptor truncated at position 965 (6). HIR AA1 behaved similarly to the truncated receptors: at the end of a 4°C incubation in the presence of either 125 I-mAb 83-14 or 125 I-insulin, it did not preferentially associate with microvilli (Figs. 5 and 6).
As a function of incubation time at 37°C, 125 I-insulin moved in the plane of the plasma membrane of WT cells so that by 30 min of incubation at 37°C, Ϸ70% of the autoradiographic grains were scored as associated with non-villous domains of the cell surface, indicating a preferential association with these surface regions (Fig. 6A). In the case of AA1 cells, although HIR AA1 did not preferentially associate with microvilli (see above), a surface redistribution of 125 I-insulin was observed, attesting to a preferential association of the mutant receptor with the non-villous domains of the cell surface (Fig. 6A). In contrast, the two receptors containing an active kinase (HIR WT and HIR AA1), did not redistribute in the sole presence of 125 I-mAb 83-14 (Fig. 6) although these two receptors did undergo inter-nalization in response to insulin binding. In the case of the two truncated receptors tagged with 125 I-mAb 83-14, a redistribution reflecting a progressive concentration of the receptors in the non-villous domains of the cell surface was noted (Fig. 6B).
Taken together with the biochemical observations (see above), these morphological data demonstrate that the receptor signal sequence responsible for HIR anchoring on microvilli is located downstream from amino acid 1000. The dileucine motif present at positions 986 -987 also participates in this anchoring process although it is not sufficient to maintain the unoccupied receptor on microvilli.

I-mAb 83-14 and I-Insulin Association with Clathrincoated Pits on the Non-villous Surface of NIH 3T3 Fibroblasts Expressing Normal and Mutant Insulin
Receptors-In the presence of insulin, normal insulin receptors are classically segregated in the internalization gates present on the nonvillous surface domains: the clathrin-coated pits (26 -28). A rapid concentration of HIR WT in clathrin-coated pits was similarly observed in NIH 3T3 fibroblasts in the presence of insulin (Fig. 7B). As described previously (25), even in the absence of insulin ( 125 I-mAb 83-14 experiments) a progressively increasing proportion of HIR WT was associated with clathrin-coated pits (Fig. 7A). However, association with clathrin-coated pits remained significantly lower than that observed when 125 I-insulin was used as ligand (Fig. 7, A and B). Under similar experimental conditions ( 125 I-mAb 83-14 experiments), HIR AA1 association with clathrin-coated pits remained low at all time points studied (Fig. 7A). In the presence of 125 I-insulin, HIR AA1 progressively associated with clathrin-coated pits but I-insulin (B) and then fixed, dehydrated, and processed for EM autoradiography. Results are expressed as mean Ϯ S.E. of the analysis of three different Epon blocks from two individual experiments (n ϭ 6). They represent the percentage of the total number of grains associated with the cell surface (Ϯ250 nm from the plasma membrane) whose centers were within a distance of 250 nm from a microvillus.  6). They represent the percentage of the total number of grains associated with the cell surface (Ϯ250 nm from the plasma membrane) whose centers were within a distance of 250 nm from a microvillus. with a lower affinity for these surface domains than HIR WT under the same conditions (Fig. 7B). HIR ⌬981, which undergoes ligand-independent internalization (see above and Fig. 3), also associated with clathrin-coated pits but with a low affinity, similar to that of HIR AA1 in the presence of insulin (Fig. 7B). In contrast, HIR ⌬1000, which also undergoes constitutive internalization but at a higher rate than HIR ⌬981 (see above and Fig. 3), showed the highest association with clathrincoated pits at all time points studied. By 2 h of incubation at 4°C, 13% of the radioactive ligand was already present within these surface domains (in contrast to Ϸ5% in all other cell lines), and this value reached a maximum of 23% by 30 min of incubation at 37°C) (Fig. 7A).
To better discriminate the propensity of the various HIR to associate with clathrin-coated pits independently of their capacity to anchor to microvilli, the morphological quantitative analysis was limited to the autoradiographic grains present on the non-villous domains of the cell surface. In this analysis, both HIR WT (ϩinsulin) and HIR ⌬1000 showed the highest affinity for clathrin-coated pits (Fig. 8). In contrast, the two mutant receptors which lack the 986 -987 dileucine motif (HIR AA1 (ϩinsulin) and HIR ⌬981), showed a comparable propensity to associate with these surface invagination gates but their affinity for clathrin-coated pits was half that noted for the two first receptors (Fig. 8). Based on these observations, it can be concluded that the dileucine motif is involved in HIR anchoring on clathrin-coated pits. Moreover, since a significant residual association with clathrin-coated pits is detected in both HIR AA1 (ϩinsulin) and HIR ⌬981, it is evident that (an)other motif(s) is(are) additive to the dileucine motif in this function. The GPLY and NPEY motifs, present in the juxtamembrane domain of HIR and retained in both HIR AA1 and HIR ⌬981, have been demonstrated previously to play such a role (11)(12)(13)(14). Present data indicate that this second set of motifs is masked in the absence of insulin since in the presence of 125 I-mAb 83-14, which does not activate HIR, the association of HIR AA1 with clathrin-coated pits is dramatically reduced as compared with that observed when 125 I-insulin is used as a ligand (Fig. 8). In contrast, such unmasking mediated by insulin binding does not seem required in the case of the dileucine motif since in its inactivated form (i.e. tagged with 125 I-mAb 83-14), HIR WT is able to efficiently associate with clathrin-coated pits despite the masking of the juxtamembrane domain (Fig. 8). DISCUSSION Cell surface receptors taken up by clathrin-coated pits can be subdivided in two categories. Class I receptors, including transport protein receptors (i.e. low density lipoprotein, transferrin, and asialoglycoprotein receptors) are spontaneously segregated in clathrin-coated pits and are continuously internalized and recycled even in the absence of ligand (29). In contrast to these nutrient receptors, signaling receptors (class II receptors, i.e. insulin, epidermal growth factor receptors) must first bind their respective ligand to have access to clathrin-coated pits which mediate their internalization (29). Present data demon- strate that the truncation of the HIR cytoplasmic tail, either at position 981 or at position 1000, transforms this typical class II receptor into a class I receptor. Indeed, as demonstrated either by tagging the receptor with an anti-insulin receptor antibody (that does not activate the receptor tyrosine kinase (25)) or through biotinylation of the cell surface proteins, internalization of these two truncated receptors shows the characteristics of constitutive internalization. It occurs at a high rate, independent of insulin binding. These observations are similar to the one obtained with a HIR truncated at position 965 which, as determined by different techniques, appeared also to be constitutively internalized (6). They extend, however, these former observations by eliminating 35 amino acids (amino acids 965-999) from the domain which is the candidate to determine the ligand-specificity of HIR internalization. Taken together with previous observations that the last 95 amino acids of the COOH-terminal tail of HIR are not required for insulininduced HIR internalization (1), it can thus be concluded that the HIR domain(s) governing insulin-induced HIR internalization is (are) contained between amino acids 1000 and 1248.
In apparent contradiction with these conclusions are the results obtained with the mutant insulin receptor in which the two leucines at positions 986 -987 have been replaced by alanines. Indeed, although the AA1-mutant insulin receptor retains the normal amino acid sequence downstream from amino acid 1000, the unoccupied receptor exhibits a decreased association with microvilli, suggesting that the dileucine motif plays a role in this association. However, the presence of this dileucine motif within ⌬1000 receptors does not improve their ability to associate with microvilli as compared with ⌬981 receptors, indicating that the insulin receptor domain centered on leucines 986 -987 is not sufficient to allow HIR anchoring on microvilli. It must rather act in conjunction with another major domain present downstream from position 1000.
To maintain unoccupied normal HIR on microvilli requires not only a specific anchoring sequence present in the cytoplasmic domain of HIR ("receptor side" component discussed above) but also a "cellular side" partner with which the receptor should interact. No information is available regarding this "cellular side" component of the brake, but the association of unoccupied HIR with microvilli suggests an implication of cytoskeleton elements which are particularly enriched in these regions. Various surface proteins including epidermal growth factor receptors and L-selectin interact with cytoskeleton elements (30 -32), but, to our knowledge, the mechanism responsible for their positioning on microvilli remains unknown. Members of the ERM (ezrin, radixin, moesin) family of cytoskeletal proteins, abundant in microvilli and becoming tyrosine-phosphorylated in response to the formation of various ligand-receptor complexes (i.e. epidermal growth factor receptors and CD4), represent good candidates for such a function (33,34).
Despite the fact that the unoccupied AA1-mutant HIR was not retained efficiently on microvilli, the constitutive internalization of this mutant receptor was not increased. On the other hand, the insulin-stimulated internalization of HIR AA1 remained lower than that of HIR WT. Thus, both constitutive and insulin-stimulated HIR AA1 internalizations are relatively less efficient than corresponding HIR WT internalization processes. Present data demonstrate that defects in internalization of HIR AA1 were caused by a decreased propensity of the mutant receptor to associate with clathrin-coated pits. This is the first direct demonstration of the involvement of dileucine motifs in the segregation of HIR in clathrin-coated pits. It provides the clue to the inhibition of internalization previously described for a series of receptors with a mutation of dileucine motif(s) pres-ent in their cytoplasmic tail, including CD4, CD3 (␥ and ␦ chains), interleukin-6 receptor, IgG Fc receptor, insulin receptor, Iip 31 invariant chain which associates with major histocompatibility complex II molecules, and interferon-␥ receptor (20,(35)(36)(37)(38). It remains an open question how dileucine motifs mediate receptor segregation in clathrin-coated pits. Recent studies making use of a yeast two-hybrid system have failed to demonstrate a direct interaction with the chains of the APcomplexes similar to the one observed in the case of tyrosinebased internalization motifs (i.e. a short four-amino acid sequence containing a tyrosine at position 1 followed by two basic amino acids and ending with a hydrophobic residue (39,40)). However, these results do not rule out the possibility of either an interaction of dileucine motifs with another chain of the AP-complex or the participation of a connecting molecule, intermediate between the dileucine motif and the AP-complex, as recently suggested in the case of CD4 (41). 2 Two tyrosine-based motifs present in the juxtamembrane domain of the cytoplasmic tail of HIR (GPLY and NPEY) have previously been demonstrated to participate in HIR segregation in clathrin-coated pits (1,2,6,13,14). Although different from the 4-amino acid stretches that interact directly with AP-2, similar motifs forming a ␤ turn and exposing an aromatic amino acid (preferentially a tyrosine) have been described as participating in the internalization of various receptors via clathrin-coated pits (43)(44)(45)(46). The high constitutive internalization rate of HIR ⌬981, which includes the two tyrosine-based motifs but does not contain the dileucine motif, and the ability of this receptor to preferentially associate with clathrin-coated pits confirm these observations. The comparison of the behavior of this truncated HIR with that of HIR ⌬1000 reveals, however, that the optimal anchoring is obtained when, in addition, the dileucine-based motif of exon 17 is present. Thus, at least three signals are required in the cytoplasmic domain of HIR for maximally efficient association with clathrin-coated pits: the two tyrosine-based motifs encoded by exon 16 and a dileucine-based motif encoded by exon 17.
Both AA1 and ⌬981 receptors contain the two juxtamembrane tyrosine-based motifs participating in anchoring in clathrin-coated pits, and lack the dileucine-based motif. Nevertheless, in the absence of kinase activation, HIR AA1 differs from HIR⌬981 in its ability to anchor on these internalization gates since: (a) constitutive internalization of insulin receptors occurs at a higher rate in ⌬981 cells than in AA1 cells and (b) ⌬981 receptors show a higher propensity to associate with clathrin-coated pits than AA1 receptors, in the course of constitutive internalization. However, in the presence of insulin, AA1 receptors increase their affinity for clathrin-coated pits so that it becomes comparable to that of the ⌬981 truncated receptors (Fig. 7). These observations suggest that the tyrosinebased signal sequences present in the juxtamembrane domain are mostly inaccessible in the inactivated receptor and that these motifs are unmasked by kinase activation as previously suggested (13). In contrast, the dileucine motif does not seem to be masked in the inactivated receptors. Indeed, as described previously in Chinese hamster ovary cells (10), kinase-inactive insulin receptors (with masked tyrosine-based motifs, see above), present on the non-villous surface domains of NIH 3T3 fibroblasts, show a high propensity to associate with clathrincoated pits. In these conditions, where the kinase is inactive, the implication of the dileucine motif is evidenced by the reduced ability of HIR AA1 receptors to associate with these surface domains as compared with that of HIR WT receptors.
Although two motifs have been implicated in the anchoring of the insulin receptor on clathrin-coated pits, the dileucine motif appears to play the most important role. Indeed, when association with clathrin-coated pits was quantitated exclusively in terms of the radioactive ligands present on the nonvillous domains of the cell surface (Fig. 8), several lines of evidence suggested the major role of the dileucine motif in this anchoring function: (a) inactivated HIR WT (acting via the dileucine motif) showed a higher propensity than activated HIR AA1 (acting through the tyrosine motifs) to associate with clathrin-coated pits; (b) for both HIR AA1 and HIR WT, only small differences were noted between the incubations carried out in the presence or absence of insulin (which reflected the role of the unmasking of tyrosine domains), (c) major differences (reflecting the role of the dileucine motif) separated HIR ⌬981 and HIR ⌬1000 receptors, (d) in the presence of insulin, significative differences distinguished HIR WT receptors (tyrosine-based and dileucine motifs exposed) and HIR AA1 receptors (only tyrosine-based motifs exposed), and (e) similarly, in the absence of insulin, HIR WT receptors (dileucine motif present) showed a significantly higher propensity to associate with clathrin-coated pits than HIR AA1 receptors (dileucine motif absent). This major role of the dileucine motif in HIR association with clathrin-coated pits could help to resolve the dilemma regarding the potential dual function of the juxtamembrane tyrosine-based motifs involved not only in association with clathrin-coated pits but also representing the site of insulin receptor substrate-1 and Shc binding to the insulin receptor (42,47,48).
In conclusion, the dileucine-based motif present at positions 986 -987 of the cytoplasmic domain of the ␤-subunit of the insulin receptor plays a dual role in the control of HIR internalization. The first function consists in participating in the anchoring of the unoccupied receptor on microvilli. Its second role resides in the anchoring of the receptor (occupied or not) on the internalization gates: the clathrin-coated pits. This latter function is independent of receptor kinase activation in contrast with the tyrosine-based motif participation in this same function.