Kinetics and Thermodynamics of the Insulin-like Growth Factor II (IGF-II) Interaction with IGF-II/Mannose 6-phosphate Receptor and the function of CD and AB Loop Solvent-exposed Residues. Research Team: Oliver Zaccheo, Stuart Prince, David Miller, Christopher Williams, Lucy Catto, Matthew Crump Bassim Hassan (University of Bristol) James Brown, Yvonne Jones (University of Oxford), Fred Kemp (Biocentre Facility) Project Background: The mammalian cation-independent mannose 6-phosphate/insulin-like growth factor II receptor (IGF2R) is a type I integral membrane protein and P-type lectin, with multiple functions attributable to its wide variety of known ligands. The 270 kda glycosylated protein consists of an N-terminal signal sequence, 15 homologous extracytoplasmic repeating domains, a transmembrane region and a C-terminal cytoplasmic domain. Ligands of the IGF-II/mannose 6-phosphate receptor (IGF2R) include IGF-II and mannose 6-phosphate modified proteins, with the binding site for IGF-II being localised to domain 11. Disruption of the negative regulatory effects of IGF2R on IGF-II-induced growth can lead to embryonic lethality and cancer promotion. In order to define the functional basis of this critical biological interaction, we performed alanine mutagenesis of structurally determined solvent-exposed loop residues of the IGF-II-binding site of human domain 11, expressed these mutant forms in Pichia pastoris, and determined binding kinetics with human IGF-II using isothermal calorimetry and surface plasmon resonance with transition state thermodynamics. Two hydrophobic residues in the CD loop (F1567 and I1572) were essential for binding, with a further non-hydrophobic residue (T1570) that slows the dissociation rate. Aside from alanine mutations of AB loop residues that decrease affinity by modifying dissociation rates (e.g. Y1542), a novel mutation (E1544A) of the AB loop enhanced affinity by threefold compared to wild-type. Conversion from an acidic to a basic residue at this site (E1544K) results in a sixfold enhancement of affinity via modification principally of the association rate, with enhanced salt-dependence, decreased entropic barrier and retained specificity. These data suggest that a functional hydrophobic binding site core is formed by I1572 and F1567 located in the CD loop, which initially anchors IGF-II. Within the AB loop, residues normally act to either stabilise or function as negative regulators of the interaction. These findings have implications for the molecular architecture and evolution of the domain 11 IGF-II-binding site, and the potential interactions with other domains of IGF2R. Biocentre Contribution (Fred Kemp): Wildtype Domain 11 vs IGF-II: Following purification, the interaction of domain 11 with IGF-II was investigated by SPR. To reduce the heterogeneity of the active surface, we immobilised biotinylated IGF-II to the sensor chip surface by affinity capture to streptavidin. In order to minimise mass transport limitation, approximately 50 response units (RU) of biotinylated IGF-II was immobilised, giving a theoretical Rmax for the domain 11 products of 100 RU.
(1) Duplicate injections of domain 11 analyte were performed at varying concentrations and the resulting sensorgrams (Figure 1(a)) analysed. As equilibrium was achieved over the course of each injection, KD from steady-state binding was determined by plotting the response at equilibrium against analyte concentration and fitting to a 1:1 binding isotherm (Figure 1(b)). Figure 1: Kinetics of Wild type Domain 11 binding to immobilized IGF-II. The mean value of the dissociation affinity constant (KD± SEM) was 103.3(±4.4) nm with an observed Rmax (97.9(±0.7) RU) very close to the expected value of 100 RU, indicating that all immobilised IGF-II was active and accessible for binding domain 11. Kinetic parameters for the interaction were also determined, based on global fitting of the sensorgrams, providing values for the association (ka) and dissociation (kd) rate constants and the dissociation affinity constant (KD), calculated from kd/ka. Initial attempts to fit the sensorgrams to standard 1:1 Langmuir binding model indicated a slight but significant difference between the fitted model and the observed curves. Global fitting of sensorgrams to a two-state (conformational change) model produced an excellent fit to the observed data (χ2=0.6) (Figure 1(a)). The second, minor component calculated using this model had an extremely low apparent KD (5 M). Our preliminary structural NMR studies of domain 11 interacting with IGF-II have indicated binding-site loop movements upon ligand binding, supporting a conformational change that might account for a two-state model. As the second component made an insignificant contribution to the overall affinity, only the kinetic parameters of the major component were used for comparison between mutants. Using this approach, we determined mean kinetic values (±SEM) for wild-type domain 11 of ka =6.62(±0.13) 105 M 1 s 1 and kd=7.87(±0.29) 10 2 s 1. Mean values of KD calculated from the reported kinetic parameters (118.8(±3.5) nm) were in close agreement with those obtained from steady-state binding (103.3(±4.4) nm) (Table 1).
Table 1. Kinetic data obtained from SPR analysis of domain 11 mutants interacting with IGF-II Kinetic parameters k a 1 k d 1 K D 1 Steady state, K D Relative Loop Mutation ( 10 5 M 1 s 1 ) ( 10 2 s 1 ) ( 10 9 M) ( 10 9 M) affinity WT 6.62±0.13 7.87±0.29 118.8±3.5 103.3±4.4 1.00 AB Y1542A 3.74±0.23 25.20±1.40 675.2±6.8 633.0±6.0 0.16 S1543A 4.80±0.14 13.30±0.10 277.5±7.2 262.3±2.8 0.39 E1544A 12.30±0.52 5.58±0.18 45.5±1.0 37.0±2.6 2.79 K1545A 3.73±0.13 11.77±0.84 315.2±15.0 289.3±9.6 0.36 CD F1567A Q1569A 7.16±0.10 10.46±0.38 146.2±7.4 129.7±2.7 0.80 T1570A 4.27±0.23 42.80±2.00 1004±12 929.7±12.0 0.11 I1572A I1572T FG S1596A 6.21±0.58 11.80±0.50 191.6±8.9 170.0±3.6 0.61 Figure 2. Representative sensorgrams depicting the range of observed affinities obtained from SPR analysis of mutated IGF2R domain 11 constructs binding to IGF-II. Using the same approach, binding of each mutant construct to IGF-II was assessed by SPR and compared to the wild-type domain 11. Again, sensorgrams were generated using biotinylated IGF-II immobilised to the cell surface and domain 11 mutant proteins as the free analyte (Figure 2). As before, kinetic parameters and steady-state affinity constants were determined by fitting sensorgrams to a two-state (conformational change) model (Table 1).
Domain 11 E1544 mutants vs IGF-II: Having established that replacement of glutamate at position 1544 with alanine caused an almost threefold increase in IGF-II affinity, via both on and off rate modification, further mutations were generated at this position in order to further evaluate this observation. Sitedirected mutagenesis was used to generate constructs in which E1544 was replaced with all possible amino acid residues. SPR analysis was conducted, using the purified domain 11 constructs, mutated at position 1544, as analyte. Where possible, kinetic parameters and steady-state affinity data were obtained from the resultant sensorgrams, as before (Figure 3a). Figure 3: Histogram showing steady-state affinities (KD) obtained from domain 11 constructs specifically mutated to replace the glutamate residue at position 1544 with the indicated amino acid. Characterisation and Thermodynamics of the E1544K enhanced affinity mutant: To investigate the relative contribution of electrostatic interactions to the binding of wild-type domain 11 and the high-affinity mutant E1544K to IGF- II, binding of each was assessed in conditions of various ionic concentrations. Kinetic parameters and steady-state affinity data were obtained from the resulting sensorgrams. For each construct, the log of 1/K D, from steady-state analysis, was plotted against the log of the concentrations of NaCl (Figure 3b). The gradient of the straight line obtained from the wild-type and the E1544K mutant was 1.0 and 1.5, respectively, suggesting that electrostatic forces make only a minor contribution to this interaction. The increased influence of ionic strength on IGF-II binding of the E1544K mutant indicates that replacement of the acidic glutamate residue at position 1544 with a basic lysine residue does indeed increase the electrostatic component of the interaction.
Figure 4. (a) A van't Hoff plot, (b) Eyring association and (c) dissociation plots showing the linear transformation of affinity (K D ) and kinetic (k a and k d ) parameters varying with reaction temperature for wild-type domain 11 (continuous line) and E1544K (broken line). In order to further understand the nature of the affinityenhancing E1544K mutation, transition-state thermodynamic data were obtained for both domain 11 and the E1544K mutant by analysis of SPR data obtained at a range of reaction temperatures (20-42C). For each construct, the relationship between K D and temperature was represented on a van't Hoff plot (ln(k D) versus 1/T, were T is the temperature in Kelvin, (Figure 4a) and the relationship between k a and k d and temperature on Eyring plots (ln(k a/t) versus 1/T and ln(kd/t) versus 1/T, (Figure 4b&c). In all cases, the transformed data fit well to a straight line. and TΔS ), as summarised in Figure 5. The resulting fitted straight lines were used to obtain values for changes in Gibbs free energy, enthalpy and entropy at equilibrium (ΔG, ΔH, and TΔS, respectively) and for the transition from unassociated to associated reactants (ΔG, ΔH
Figure 5. Summary of change in (a) Gibbs free energy (ΔG ), (b) enthalpy (ΔH ) and (c) entropy ( TΔS ) for wild-type domain 11 (continuous line) and E1544K (broken line) interacting with IGF-II, for unassociated reactants (A+B), transition from unassociated to associated binding partners (A B) and at equilibrium (AB). The activation energy that must be overcome during complex formation (Figure 5a A-B) showed only a minor difference between wild-type domain 11 and E1544K, as might be expected for interactions with only a sixfold difference in affinity. However, when the activation free energy of association (ΔG ) was resolved into enthalpy and entropy, a significant difference in the activation pattern of domain 11 versus E1544K was revealed. The activation enthalpy (Figure 5b A- B) of the E1544K mutant was elevated above that of the wildtype. This increased energy requirement might represent that needed to form the additional electrostatic interactions, evident from the enhanced effect of ionic concentration. However, the increased activation enthalpy is compensated for by the E1544K mutant having a considerably reduced activation entropy compared to the wild-type, thereby reducing the entropic barrier that must be overcome in order for the interaction to occur. The interpretation of this would be that either E1544K is less structurally constrained in the transition state, able to move rotate or vibrate more freely, or that E1544K is more able to displace solvent from the binding interface. Output: Oliver J. Zaccheo, Stuart N. Prince, David M. Miller, Christopher Williams, C. Fred Kemp, James Brown, E. Yvonne Jones, Lucy E. Catto, Matthew P. Crump and A. Bassim Hassan. Kinetics of Insulin-like Growth Factor II (IGF-II) Interaction with Domain 11 of the Human IGF-II/Mannose 6-phosphate Receptor: Function of CD and AB Loop Solvent-exposed Residues. Journal of Molecular Biology, Volume 359, Issue 2, 2 June 2006, Pages 403-421