The rationale for this work was to improve understanding of how the subunits of the multimeric M6P/IGF2R participated in binding the two main classes of ligands, IGF-II and phosphomannosylated glycoproteins. In mammals, binding of IGF-II by the M6P/IGF2R is thought to contribute to growth homeostasis. Previous work has shown that the receptor operates optimally as a dimer and we wanted to determine what effect the dimeric structure may have on IGF-II binding. It has been suggested that IGF-II binding to the M6P/IGF2R requires contributions of repeats 11 and 13, but only within a single polypeptide chain [20
]. It is well established that the receptor binds Man-6-P ligands in a multivalent fashion [1
]. However, because the receptor has two Man-6-P binding domains within a single polypeptide chain, it remains uncertain whether this bivalent binding activity is a property of a single monomeric receptor or the result of cooperative interaction between the two subunits of a dimeric receptor. There is strong evidence that, in the cell, the preferred mode of binding is through a dimeric structure as shown by York et al
., who found that a multivalent phosphomannosylated ligand cross-bridged the dimeric receptor to promote optimal internalization [24
]. This conclusion was reinforced by the work done by Byrd et al
], who analyzed mutant receptors bearing a substitution of Arg for Ala at position 1325 that knocks out Man-6-P ligand binding to the repeat 9 site. Scatchard plot analysis showed that these mutant receptors were still able to bind bivalent Man-6-P ligands with high-affinity, leading to the conclusion that high-affinity binding in that case must be due to alignment of two repeat 3 Man-6-P binding domains on paired monomers. Furthermore, Olson et al
] demonstrated via x-ray projection models that the closest distance between the two Man-6-P binding sites of one monomeric receptor is 45–70 Å, indicating that a single diphosphorylated oligosaccharide, with a maximum distance of ~30 Å [35
], could not bind to the Man-6-P binding domains of both repeats 3 and 9 simultaneously [16
]. The present study was designed to test if ligand binding by the dimeric receptor is cooperative, based on the hypothesis that IGF-II binds independently to cognate sites on both monomers of the dimeric receptor, but that Man-6-P ligand binding would require cooperation of both monomers. For this purpose we developed a quantitative assay for heterodimer formation that was based on immunoprecipitation of differentially epitope-tagged receptors. The availability of the I1572T mutant allowed us to address the initial question whether a non-functional IGF-II binding site would interfere with the function of a wild-type binding site when paired in a single heterodimeric structure.
Association assays indicated that immunoprecipitation between differentially epitope-tagged mini-receptors was feasible. These assays also indicated that immunoprecipitation was not preferential to the epitope used to tag the receptor, as there was no discernable difference between the Myc- and HA-tagged receptors in co-immunoprecipitation with the FLAG-tagged receptor. As expected, we observed that essentially all of the FLAG-tagged mini-receptors were precipitated by incubation with M2 resin. In experiments with the FLAG-tagged receptor as the bait, ~50% of the Myc-tagged mini-receptors were co-precipitated from a cell lysate prepared from cells co-transfected with equal amounts of the tagged mini-receptor cDNA. This strongly suggests, but does not prove, that the mini-receptors associated in a 1:2:1 relationship: 25% FLAG homodimers, 50% FLAG-Myc heterodimers, and 25% Myc homodimers (which would not precipitate in this assay). The simplest interpretation of these data relative to the structure of the receptor is that the mini-receptors were in the form of dimers. Byrd et al
] showed via mutational analysis that receptors with only one functional Man-6-P binding site exhibited high-affinity binding of Man-6-P-containing ligands. Given that high-affinity binding of a bivalent ligand is due to cooperative interaction with two or more receptor binding sites [35
], these data suggested that oligomerization of the receptor contributes to high-affinity binding. In addition, native gel electrophoresis demonstrated that the receptor could be separated into monomeric and dimeric forms in the presence or absence of Man-6-P-containing ligands [25
]. York et al
] demonstrated by sucrose gradient sedimentation and gel filtration that the receptor bound to a multivalent Man-6-P-containing ligand, β-glucuronidase, exhibited a sedimentation coefficient and Stokes radius that were consistent with a complex of two receptor molecules plus one molecule of ligand. However, when these experiments were performed with IGF-II, the receptor appeared to exist as a monomer. Internalization experiments performed with 125
I-IGF-II in the presence of β-glucuronidase revealed that β-glucuronidase accelerated the rate of IGF-II uptake, suggesting that intermolecular cross-linking of receptors enhanced receptor endocytosis [24
]. Thus, all the available data are consistent with the conclusion that the M6P/IGF2R functions as a dimer in high-affinity binding of Man-6-P-bearing ligands, but possibly not for IGF-II. Our work with soluble forms of the receptor ([25
] and the present study) indicates that the ectodomain of the receptor is capable of dimer formation in the absence of ligand.
Affinity cross-linking of 125
I-IGF-II to a mutant repeat 11 mini-receptor revealed that the I1572T mutation completely abolished IGF-II binding [31
]. This finding was confirmed by Linnell et al
] by utilizing surface plasmon resonance of a truncated receptor containing extracytoplasmic repeats 10–13, which contained the I1572T mutation in repeat 11. However, the mechanism by which this mutation abrogates IGF-II binding is still not clear. Structural analysis of repeat 11 identified the presumptive IGF-II binding site in a hydrophobic pocket at the end of a β-barrel structure [36
was found to lie near, but not directly within, this putative IGF-II binding site. This mutation involves substituting a polar residue, Thr, for a bulky, nonpolar residue, Ile, which might have altered the IGF-II binding pocket by inducing a conformational change that reduces binding energy or makes the site less hydrophobic. In any case, this type of effect should be regional and have minimal influence on the wild-type IGF-II binding site on an adjacent mini-receptor within a dimer. Our experiments support this prediction, showing that the pairing of wild-type and 11572T mutant IGF-II binding sites between two dimerized mini-receptors had no effect on the function of the contralateral binding site. The mutant site does not prevent the wild-type site from binding IGF-II and pairing with a wild-type subunit does not repair the defect in the mutant site inducing it to bind IGF-II. This indicates that IGF-II binding to each side of the dimer is independent. Symmetric heterodimers (having identical epitope tags, both of which are tethered to the resin bead) achieve the predicted amount of binding as described above. Tethering of both sides of the dimer likely mimics the structure obtained when anchored in the membrane, in accordance with the notion that this structure is the receptor's normal functional state.
The most interesting and unexpected finding of the present study is that asymmetric heterodimers (having different epitope tags, of which only one is tethered to the resin bead) demonstrate complex binding behavior. Dimers of this type exhibit the predicted amount of binding only if the heterodimer is tethered by the wild-type partner. In contrast, we found that if the heterodimer is tethered to the resin by the mutant partner, the amount of binding observed is substantially less than expected. This complex binding behavior seen with IGF-II binding must only be a local effect, as binding of PMP-BSA, which binds to other sites in the ectodomain of these heterodimers, resulted in the predicted amount of ligand binding. This loss of binding function observed with asymmetric heterodimers may be due to deformation of the dimeric structure. One possible explanation for failure to form a dimer of correct structure could be steric hindrance between the Myc tag and the M2 resin bead. This is envisioned to cause the Myc-tagged receptor partner to be bent outward away from the tethered FLAG-tagged partner, potentially resulting in distortion of the IGF-II binding pocket and consequent reduced ability to bind IGF-II. This effect is likely not due to reduced contact between repeats 11 and 13, as repeat 11 is capable of binding IGF-II even in the absence of repeat 13 [20
]. These data suggest that the failure to tether the tail of the extracytoplasmic domain results in the inability to form appropriate contacts between dimeric partners. Follow-up experiments using structural approaches will be needed to address this possibility.
Localization of the two Man-6-P binding domains was previously reported by Westlund et al
], who subjected the M6P/IGF2R to partial proteolytic digestion using subtilisin. They determined that repeats 1–3 and 7–10 can independently bind Man-6-P-containing ligands. Dahms et al
] further defined the location of Man-6-P binding sites by using mutational analysis to establish the importance of specific Arg residues in the function of both Man-6-P binding sites. They determined that Arg426
in repeats 3 and 9, respectively, are essential components of the receptor's high-affinity Man-6-P binding sites. The structure of repeats 1–3 of the bovine M6P/IGF2R in the presence of Man-6-P was solved by Olson et al
]. This work revealed key amino acid residues in the binding site of repeat 3 that were important for Man-6-P binding. In particular it was found that the guanidinium group of Arg435
(corresponding to Arg111
of the CD-MPR and Arg426
of the human M6P/IGF2R) forms one or two critical hydrogen bonds with the 2'-hydroxyl group of the mannose ring but these represent only two out of more than a dozen non-covalent interactions between ligand and receptor within the binding pocket. This leads to the question as to whether the R→A mutation actually causes a loss of binding energy sufficient to abrogate Man-6-P binding or whether the mutated binding pocket becomes distorted, preventing Man-6-P binding. Glucose 6-phosphate (Glc-6-P) is unable to bind to the two Man-6-P binding sites of the receptor [17
]. The only difference between Glc-6-P and Man-6-P is the position of the 2'-hydroxyl group. The equatorial 2-hydroxyl group of Glc-6-P is not in the correct position to make the necessary hydrogen bonds with the critical Arg; however, the axial 2-hydroxyl of Man-6-P is. Therefore it seems that the binding energy of the site must be decreased with the R→A mutation preventing Man-6-P from binding to the Man-6-P binding sites.
Based on this analysis and the need for cooperative interaction between two binding sties, we expected the wild-type/R2A heterodimers to show equivalent Bmax
with reduced affinity; however, ligand binding analysis (data not shown) revealed a 50% decrease in Bmax
while the affinity remained unchanged. This surprising finding led us to hypothesize that the large multivalent Man-6-P-based ligands such as PMP-BSA can bind bivalently to wild-type subunits of heterodimeric receptors juxtaposed on the resin. Whether these represent nearby heterodimers or receptors interacting in large oligomeric clusters is not yet clear. It is known from earlier work that the M6P/IGF2R is concentrated in coated pits on the plasma membrane and that there is a greater than 60-fold enrichment of receptor in clathrin-coated pits when compared to microsomes [37
]. During receptor-mediated endocytosis when there is a high accumulation of receptors into one location, it is possible they can interact in higher-order oligomeric structures. This would explain why we do not see a decrease in affinity of phosphomannosyl ligands between wild type and R2A mutant receptors. This accumulation of receptors may be mimicked on the resin bead during an immunoprecipitation such that the multivalent ligand binding selects binding-competent partners, promoting preferential cross-bridging between wild-type receptors and forming higher oligomeric structures on the resin bead. As was seen with the IGF-II binding mutant, asymmetric heterodimers that are tethered by mutant receptors show an unexpected decrease in PMP-BSA binding. It thus appears that tethering of the C-terminal end of the ectodomain is important for both IGF-II and PMP-BSA binding.
In summary, the major findings of this work are consistent with a dimer model for M6P/IGF2R oligomerization as all co-immunoprecipitations resulted in a predicted outcome of pull-down or ligand binding irrespective of the tag or mutant. IGF-II was found to bind independently to sites on each monomeric partner while high-affinity binding of multivalent Man-6-P ligands was proportional to the number of wild-type binding sites available in a mixture of mutant/wild-type receptors. Tethering to the resin of asymmetric heterodimers suggests the importance of anchoring the tail of the ectodomain for both IGF-II and Man-6-P ligand binding. Receptor binding on resin resembles a patchwork model where a multivalent ligand can cross-bridge between nearby wild-type sites to achieve high affinity binding. Additionally, further studies are needed to obtain more information on the structure of the M6P/IGF2R ectodomain particularly toward the C-terminal repeat 15 and its interaction with the membrane to assess if there are important associated proteins or receptor-lipid interactions that cannot be mimicked using epitope-tagged portions of the ectodomain. Finally, experiments in whole cells must address the possibility that the receptor can associate in patches during endocytosis.