Molecular basis for the role of disulfide-linked αCTs in the activation of insulin-like growth factor 1 receptor and insulin receptor

  1. Jie Li
  2. Jiayi Wu
  3. Catherine Hall
  4. Xiao-chen Bai
  5. Eunhee Choi

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    This is a valuable manuscript that addresses an important question and provides interesting mechanistic insights into the roles of specific regions of the IR and IGF1R in their activation. While many of the data convincingly support the conclusions, in some areas the data are incomplete so we are left with an unfinished picture of the mechanisms of activation of the IR and IGF1R, and why they differ.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1, Reviewer #2 and Reviewer #3 agreed to share their names with the authors.)

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

The insulin receptor (IR) and insulin-like growth factor 1 receptor (IGF1R) control metabolic homeostasis and cell growth and proliferation. The IR and IGF1R form similar disulfide bonds linked homodimers in the apo-state; however, their ligand binding properties and the structures in the active state differ substantially. It has been proposed that the disulfide-linked C-terminal segment of α-chain (αCTs) of the IR and IGF1R control the cooperativity of ligand binding and regulate the receptor activation. Nevertheless, the molecular basis for the roles of disulfide-linked αCTs in IR and IGF1R activation are still unclear. Here, we report the cryo-EM structures of full-length mouse IGF1R/IGF1 and IR/insulin complexes with modified αCTs that have increased flexibility. Unlike the Γ -shaped asymmetric IGF1R dimer with a single IGF1 bound, the IGF1R with the enhanced flexibility of αCTs can form a T -shaped symmetric dimer with two IGF1s bound. Meanwhile, the IR with non-covalently linked αCTs predominantly adopts an asymmetric conformation with four insulins bound, which is distinct from the T -shaped symmetric IR. Using cell-based experiments, we further showed that both IGF1R and IR with the modified αCTs cannot activate the downstream signaling potently. Collectively, our studies demonstrate that the certain structural rigidity of disulfide-linked αCTs is critical for optimal IR and IGF1R signaling activation.

Article activity feed

  1. Version published to 10.7554/elife.81286 on eLife
  2. eLife

    Author Response

    Reviewer #1 (Public Review):

    This paper follows several innovative articles from the authors exploring the molecular mechanisms of insulin and IGF1 receptors activation by their ligands using cryo-electron microscopy. Here the authors explore the role of an alpha helical C-terminal segment (called the alpha-CT motif) of a disordered disulfide-linked insert domain in the FnIII-2 module of the insulin and IGF1 receptors (at the end of the alpha subunit), in the mechanism of ligand binding, negative cooperativity and receptor activation.

    Biochemical data gathered over several decades have suggested that insulin and IGF1 use two separate binding sites, site 1 and site 2, to bind to two distinct domains (sites 1 and 2, and 1'and 2') on each protomer of the homodimeric receptors, disposed in an antiparallel symmetry. This disposition was corroborated by the early x-ray crystallographic studies of the unliganded insulin receptor ectodomain (apo-receptor). A subsequent somewhat surprising finding was that the insulin receptor site 1 is in fact a composite, made of the beta surface of the L1 module of one protomer, and of the alpha-CT motif of the other protomer which binds perpendicularly to the L1 surface (a "tandem binding element"), with insulin binding more to the alpha-CT motif than to L1.

    Previous work from the authors showed that the subsaturated insulin receptor has an asymmetric configuration while the receptor saturated with 4 insulins has a symmetric T-shaped configuration. In contrast, the IGF1R shows only one IGF1 bound to an asymmetric configuration, indicating according to the authors a stronger negative cooperativity. This is attributed to a more rigid and elongated conformation of the alpha-CT motives that restricts the structural flexibility of the alternate binding site.

    To test this hypothesis, the authors determined the cryo-EM structure of IGF1 bound to IGF1R with a mutated alpha-CT motif elongated by four glycine residues. Strikingly, a portion of these constructs adopt a T-shaped symmetric structure.

    Conversely, they show that the cryo-EM structure of insulin bound to an insulin receptor with non-covalently bound alpha-CTs insert domains by mutation of the cysteines to serine adopts asymmetric conformations even at saturated insulin concentrations. They conclude that the alpha-CTs in disulfide-linked insert domains of the insulin receptor play an important role in the structural transition from asymmetric to symmetric during the insulin-induced insulin receptor activation.

    All in all, this is a very interesting and well-designed study that represents an advance in the knowledge of the insulin/IGF1 receptor systems, although the details of the structural interpretations deserve some discussion.

    This is very clear and succinct summary of our work. We thank Dr. Pierre De Meyts for the positive assessment of our manuscript, and we greatly appreciate the constructive comments which we have addressed.

    Reviewer #2 (Public Review):

    Li et al build upon recent observations that the alphaCT peptide is a key element in the IGF-1R and IR that regulates negative cooperativity and receptor activation. The use of IGF-1R and IR mutants builds upon previous observations with these mutants by Li et al (IGF-1R) and Weis et al. (IR).

    Here they determined the structures of the IGF-1R mutant, IGF-1R-P673G4, which has a 4 glycine motif inserted at residue P673 at 4Å resolution. By introducing structural flexibility in the alphaCT the IGF-1R is able to bind 2 IGFs and adopts a symmetric T conformation, which is in contrast to the single IGF bound WT IGF-1R that adopts an asymmetric conformation. The ability to bind two IGFs is taken as a sign that negative cooperativity has been affected and confirms the importance of the alphaCT in constraining the IGF-1R into the asymmetric conformation. The increased flexibility of the alphaCT linkage between the two receptor monomers results in reduced ability of IGF-I to activate the IGF-1R and Erk leading to reduced IGF-1R internalisation. This is consistent with previous reports that effective Erk signalling is dependent on endosomal signalling. A second mutant, IGF-1R -3CS, was also used where a cysteine triplet in the alphaCT is mutated to serine to perturb the disulfide bonding between the alphaCTs of the two monomers. IGF-1R activation and signalling by this mutant was also reduced.

    In addition, the structure of an equivalent insulin receptor mutant, IR-3CS, was determined with complexes formed with excess insulin. Again, the increased flexibility of the alphaCT altered the structural rearrangement upon ligand binding. Three conformations with 4 insulins bound were detected, two unique asymmetric (4.5 Å and 4.9 Å) and one symmetric (3.7 Å), whereas WT IR:insulin complex predominantly forms a symmetric T 4 insulin bound structure. This suggests the IR alphaCT is important in stablising the active T structure. In contrast to the IGF-1R -3CS, the IR-3CS has the same affinity for insulin as WT IR, is more potently activated (pY1150/Y1151) by insulin but has a reduced signalling response. This demonstrates the role of alphaCT in the activation.

    Whilst the symmetric IR-3CS:insulin complex structure is compared with the WT IR: insulin complex, no comparisons were made between the asymmetric conformations described here and those previously reported. Is the ligand binding in the asymmetric conformations different to the asymmetric binding seen when WT IR:insulin complexes were generated at low insulin concentrations? It would be interesting to see these overlaid. How do these asymmetric conformations relate to the existing asymmetric conformations reported by Nielsen (10.1016/j.jmb.2022.167458) and Xiong (DOI 10.1038/s41589-022-00981-0)?

    Thank you for the good suggestions. We have now prepared a new Figure 4-supplement 2 that compares the structure of asymmetric IR-3CS/insulin with that of asymmetric IR bound with subsaturated insulin previously published by us and others. All asymmetric structures of IR bound with subsaturated insulin have similar structural features, i.e., in one half of the complex, one insulin bound at site-1 also contacts site-2 from adjacent protomer, or vice versa. However, in the asymmetric structure of IR-3CS/insulin, two insulins were bound at the hybrid site in the middle of the IR-3CS/insulin complex. To accommodate the binding of two insulins, the L1/αCT together with bound site-1 insulin move outward as compared to the asymmetric structure of IR bound with subsaturated insulin. This is the major structural difference between these asymmetric structures. We have discussed this in the revised manuscript.

    What is the distance between the FnIII-3 domains of the IR:insulin asymmetric conformations and in the symmetric structure? Does this correlate to activity as is seen for the IGF-1R-P673G4? It would be good to comment on this, particularly as there is an interesting disconnect between the receptor activation and downstream signalling activity. Why is there greater pY1150/Y1151 activation than for the WT IR and how can the lower downstream signalling activity be explained?

    We thank Dr. Briony Forbes for raising this point. Asymmetric IR-3CS/insulin, asymmetric IR/insulin and symmetric IR/insulin have similar distances between their membrane-proximal regions (approximately 30 – 35 Å). This indicates that the distances between the membrane-proximal regions within these complexes are all short enough to allow the intracellular kinase to undergo efficient autophosphorylation, in contrast to IGF1R.

    As indicated by Dr. Forbes, our cellular functional assays showed that the IR-3CS has higher levels of autophosphorylation, but lower levels of downstream signaling activity and a defect in endocytosis. Although the distances between the membrane-proximal regions are similar, the relative positions and orientations between the two membrane proximal regions are significantly different between asymmetric IR-3CS and symmetric IR. Given the fact that the FnIII-3 domain is connected to the transmembrane domain by a short linker containing four residues, we speculate that the structural differences in the extracellular domains may lead to both differential dimeric assembly of transmembrane and intracellular domains, as well as the stable interaction between the intracellular IR domains and downstream adaptors and effectors. This could in part explain why IR-3CS can still undergo robust autophosphorylation but its downstream signaling becomes defective. Similar hypothesis has been proposed in the EGF and TGF-α induced activation of EGFR (PMID: 34846302). The endocytosis defects of IR-3CS might be the result of reduced IR signaling, but it is tempting to speculate that less endocytosis of IR-3CS may cause defective downstream signaling. The structure of transmembrane and intracellular domains in the context of the entire full-length/insulin complex needs to be further investigated. We have included new analysis and expanded the discussion.

    It would be good to reword the opening statement that "IGF1 only has one type of ligand binding site (site-1)" to acknowledge that two binding sites on IGF-I have been detected through analysis of competition binding studies which are fitted to a two-site sequential model and detect both high affinity and low affinity binding (Kiselyov). Site directed mutagenesis studies of both IGFs have detected two binding surfaces analogous to insulin's site 1 and site 2 (Gaugin et al and Alvino et al). Furthermore, binding assays with mini-IGF-1R (L1, CR, L2 fused to alphaCT, ie site 1 only) clearly demonstrated that IGF-II site 2 residues do contribute to overall binding affinity (Alvino et al). Perhaps we are yet to capture site 2 of IGF-1R as it is not in the same location as IR site 2? It would be good to comment on this.

    Point accepted. Gauguin L. et al. demonstrated that alanine mutagenesis in IGF1, including E9A, D12A, F16A, D53A, L54A, and E58A, markedly reduced IGF1R-binding affinity. With the exception of IGF1 E9 (Site-1b of IGF1R, the same position in IGF2, E12), none of IGF1 D12, F16, D53, L54, and E58 are involved in the binding to site-1, suggesting that IGF1 has an additional site that maximizes the binding to the receptor. Despite saturated IGF1 levels, however, our previous and current structural studies did not reveal the putative site-2 of IGF1R-IGF1 binding. We speculate that IGF1 binds to site-2 transiently, which might be important for IGF1-induced activation of IGF1R. We have revised the manuscript and expanded the discussion.

    Reviewer #3 (Public Review):

    Li et al. present cryo-EM structures of the insulin receptor (IR) and insulin-like growth factor-1 receptor (IGF1R), exploring the functional roles of the disulfide-linked alphaCT regions in ligand binding and receptor activation.

    Cryo-EM structures of mutants of IGF1R and IR designed to increase the flexibility between disulfide-linked alphaCT regions revealed conformational states that were distinct from those of the wild-type (WT) receptors. Mutant (P673G4) IGF1R displayed conformations in which two IGF1 molecules were bound, rather than the 1:1 ligand:receptor state observed previously for WT IGF1R. Mutant (3CS) IR displayed asymmetric conformations with four insulin molecules bound, as well as the symmetric T conformation with four insulin molecules bound observed previously for WT IR. In each case, the mutant receptor was shown in cells to be poorly activated by its respective ligand.

    This study demonstrates the importance of the disulfide-coupled alphaCT regions in the IR and IGF1R for ligand binding and receptor activation. What is not resolved in this study is whether differences in the alphaCT regions of these two highly related receptors contribute to their disparate active states - asymmetric for IGF1R (and 1:1 IGF1:IGF1R) vs. symmetric (T) for IR (and 4:1 insulin:IR).

    We thank Dr. Stevan Hubbard for the positive assessment of our manuscript, and we greatly appreciate the constructive comments which we have addressed.

  3. eLife

    Evaluation Summary:

    This is a valuable manuscript that addresses an important question and provides interesting mechanistic insights into the roles of specific regions of the IR and IGF1R in their activation. While many of the data convincingly support the conclusions, in some areas the data are incomplete so we are left with an unfinished picture of the mechanisms of activation of the IR and IGF1R, and why they differ.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1, Reviewer #2 and Reviewer #3 agreed to share their names with the authors.)

  4. eLife

    Reviewer #1 (Public Review):

    This paper follows several innovative articles from the authors exploring the molecular mechanisms of insulin and IGF1 receptors activation by their ligands using cryo-electron microscopy. Here the authors explore the role of an alpha helical C-terminal segment (called the alpha-CT motif) of a disordered disulfide-linked insert domain in the FnIII-2 module of the insulin and IGF1 receptors (at the end of the alpha subunit), in the mechanism of ligand binding, negative cooperativity and receptor activation.

    Biochemical data gathered over several decades have suggested that insulin and IGF1 use two separate binding sites, site 1 and site 2, to bind to two distinct domains (sites 1 and 2, and 1'and 2') on each protomer of the homodimeric receptors, disposed in an antiparallel symmetry. This disposition was corroborated by the early x-ray crystallographic studies of the unliganded insulin receptor ectodomain (apo-receptor). A subsequent somewhat surprising finding was that the insulin receptor site 1 is in fact a composite, made of the beta surface of the L1 module of one protomer, and of the alpha-CT motif of the other protomer which binds perpendicularly to the L1 surface (a "tandem binding element"), with insulin binding more to the alpha-CT motif than to L1.

    Previous work from the authors showed that the subsaturated insulin receptor has an asymmetric configuration while the receptor saturated with 4 insulins has a symmetric T-shaped configuration. In contrast, the IGF1R shows only one IGF1 bound to an asymmetric configuration, indicating according to the authors a stronger negative cooperativity. This is attributed to a more rigid and elongated conformation of the alpha-CT motives that restricts the structural flexibility of the alternate binding site.

    To test this hypothesis, the authors determined the cryo-EM structure of IGF1 bound to IGF1R with a mutated alpha-CT motif elongated by four glycine residues. Strikingly, a portion of these constructs adopt a T-shaped symmetric structure.

    Conversely, they show that the cryo-EM structure of insulin bound to an insulin receptor with non-covalently bound alpha-CTs insert domains by mutation of the cysteines to serine adopts asymmetric conformations even at saturated insulin concentrations. They conclude that the alpha-CTs in disulfide-linked insert domains of the insulin receptor play an important role in the structural transition from asymmetric to symmetric during the insulin-induced insulin receptor activation.

    All in all, this is a very interesting and well-designed study that represents an advance in the knowledge of the insulin/IGF1 receptor systems, although the details of the structural interpretations deserve some discussion.

  5. eLife

    Reviewer #2 (Public Review):

    Li et al build upon recent observations that the alphaCT peptide is a key element in the IGF-1R and IR that regulates negative cooperativity and receptor activation. The use of IGF-1R and IR mutants builds upon previous observations with these mutants by Li et al (IGF-1R) and Weis et al. (IR).

    Here they determined the structures of the IGF-1R mutant, IGF-1R-P673G4, which has a 4 glycine motif inserted at residue P673 at 4Å resolution. By introducing structural flexibility in the alphaCT the IGF-1R is able to bind 2 IGFs and adopts a symmetric T conformation, which is in contrast to the single IGF bound WT IGF-1R that adopts an asymmetric conformation. The ability to bind two IGFs is taken as a sign that negative cooperativity has been affected and confirms the importance of the alphaCT in constraining the IGF-1R into the asymmetric conformation. The increased flexibility of the alphaCT linkage between the two receptor monomers results in reduced ability of IGF-I to activate the IGF-1R and Erk leading to reduced IGF-1R internalisation. This is consistent with previous reports that effective Erk signalling is dependent on endosomal signalling. A second mutant, IGF-1R -3CS, was also used where a cysteine triplet in the alphaCT is mutated to serine to perturb the disulfide bonding between the alphaCTs of the two monomers. IGF-1R activation and signalling by this mutant was also reduced.

    In addition, the structure of an equivalent insulin receptor mutant, IR-3CS, was determined with complexes formed with excess insulin. Again, the increased flexibility of the alphaCT altered the structural rearrangement upon ligand binding. Three conformations with 4 insulins bound were detected, two unique asymmetric (4.5 Å and 4.9 Å) and one symmetric (3.7 Å), whereas WT IR:insulin complex predominantly forms a symmetric T 4 insulin bound structure. This suggests the IR alphaCT is important in stablising the active T structure. In contrast to the IGF-1R -3CS, the IR-3CS has the same affinity for insulin as WT IR, is more potently activated (pY1150/Y1151) by insulin but has a reduced signalling response. This demonstrates the role of alphaCT in the activation.

    Whilst the symmetric IR-3CS:insulin complex structure is compared with the WT IR: insulin complex, no comparisons were made between the asymmetric conformations described here and those previously reported. Is the ligand binding in the asymmetric conformations different to the asymmetric binding seen when WT IR:insulin complexes were generated at low insulin concentrations? It would be interesting to see these overlaid. How do these asymmetric conformations relate to the existing asymmetric conformations reported by Nielsen (10.1016/j.jmb.2022.167458) and Xiong (DOI 10.1038/s41589-022-00981-0)?

    What is the distance between the FnIII-3 domains of the IR:insulin asymmetric conformations and in the symmetric structure? Does this correlate to activity as is seen for the IGF-1R-P673G4? It would be good to comment on this, particularly as there is an interesting disconnect between the receptor activation and downstream signalling activity. Why is there greater pY1150/Y1151 activation than for the WT IR and how can the lower downstream signalling activity be explained?

    It would be good to reword the opening statement that "IGF1 only has one type of ligand binding site (site-1)" to acknowledge that two binding sites on IGF-I have been detected through analysis of competition binding studies which are fitted to a two-site sequential model and detect both high affinity and low affinity binding (Kiselyov). Site directed mutagenesis studies of both IGFs have detected two binding surfaces analogous to insulin's site 1 and site 2 (Gaugin et al and Alvino et al). Furthermore, binding assays with mini-IGF-1R (L1, CR, L2 fused to alphaCT, ie site 1 only) clearly demonstrated that IGF-II site 2 residues do contribute to overall binding affinity (Alvino et al). Perhaps we are yet to capture site 2 of IGF-1R as it is not in the same location as IR site 2? It would be good to comment on this.

  6. eLife

    Reviewer #3 (Public Review):

    Li et al. present cryo-EM structures of the insulin receptor (IR) and insulin-like growth factor-1 receptor (IGF1R), exploring the functional roles of the disulfide-linked alphaCT regions in ligand binding and receptor activation.

    Cryo-EM structures of mutants of IGF1R and IR designed to increase the flexibility between disulfide-linked alphaCT regions revealed conformational states that were distinct from those of the wild-type (WT) receptors. Mutant (P673G4) IGF1R displayed conformations in which two IGF1 molecules were bound, rather than the 1:1 ligand:receptor state observed previously for WT IGF1R. Mutant (3CS) IR displayed asymmetric conformations with four insulin molecules bound, as well as the symmetric T conformation with four insulin molecules bound observed previously for WT IR. In each case, the mutant receptor was shown in cells to be poorly activated by its respective ligand.

    This study demonstrates the importance of the disulfide-coupled alphaCT regions in the IR and IGF1R for ligand binding and receptor activation. What is not resolved in this study is whether differences in the alphaCT regions of these two highly related receptors contribute to their disparate active states - asymmetric for IGF1R (and 1:1 IGF1:IGF1R) vs. symmetric (T) for IR (and 4:1 insulin:IR).

  7. Version published to 10.1101/2022.07.06.498988v1 on bioRxiv