| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
-receptor interactions
,
,,1
* Division of Cell and Molecular Biology, Toronto General Research Institute, University Health Network, Toronto, and Departments of
Immunology,
Chemistry and
Pharmaceutical Sciences, University of Toronto, Toronto, Ontario, Canada
1Correspondence: Toronto General Research Institute, 67 College St., Rm. 424, Toronto, ON M5G 2M1 Canada. E-mail: en.fish{at}utoronto.ca
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
subtypes exhibit differences in biological potencies based on their affinity interactions with the IFN receptor subunits, IFNAR1 and IFNAR2. Using available three-dimensional structural information and computational biology, homology models of human IFN-1, human IFN-8, IFN alfacon-1, and murine IFN-4 were derived and docked with the extracellular region of human IFNAR2 to evaluate the behavior of potential interacting residue pairs and characterize the nature of the IFN-IFNAR2 binding interfaces. The data suggest that IFN afacon-1 has 9 optimal interactions with IFNAR2, comprising hydrophobic, electrostatic, and hydrogen bonding. Human IFN-2 exhibits 8 optimal interactions, human IFN-1, 7, and murine IFN-4 exhibits the least number of optimal interactions, at 5. A model of IFNAR1 was generated, taking into consideration the IFNAR1 extracellular domain interaction with cell surface glycosphingolipids, putative ligand interaction residues, and residues stabilizing the structural integrity of IFNAR. IFNAR1 was then docked with the various IFN-IFNAR2 complexes to describe the complete extracellular receptor pocket with bound IFN. These data provide insights into the species specificity of IFN-s: residues in murine IFN-4 that preclude strong affinity interactions with human IFNAR because of steric crowding and residues in human IFN-8 that resemble a receptor interactive domain in murine IFN-4, are described.—Kumaran, J., Wei, L., Kotra, L. P., Fish, E. N. A structural basis for interferon--receptor interactions.
Key Words: interferon • interferon receptor • homology modeling • docking
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
s and IFN-ß occupy central positions as cytokines, which exhibit antiviral, growth inhibitory and immunomodulatory activities (1)
. IFNs are encoded by nonallelic, intronless genes, and the 13 human IFN- subtypes share between 70 and 80% amino acid sequence identity. Each member is composed of 165/166 amino acid residues, with 2 conserved disulfide bonds: Cys1-Cys-99 and Cys 29-Cys139. Inferred from available three-dimensional structures, IFNs-/ß exhibit a conserved secondary structure comprising five alpha helices, designated A to E, which pack together as a helical bundle (Figs. 1
and 2
) (2
, 3
, 4
, 5
, 6)
.
|
|
The ubiquitously expressed IFN receptor, IFNAR, is composed of two single transmembrane spanning proteins, IFNAR1 and IFNAR2c (7
, 8
, 9)
. Human IFNAR1 is a 530 amino acid protein of molecular weight 63 kDa, which may vary from 120–130 kDa due to the 15 potential N-glycosylation sites. The extracellular portion of IFNAR1 comprises 409 amino acid residues, composed of four immunoglobulin-like domains.IFNAR2, the primary binding chain, exists as three isoforms: IFNAR2a (soluble) is 239 amino acids (24 kDa), IFNAR2b (short) is 331 amino acids (34 kDa), and IFNAR2c (long) is 515 amino acid (55 kDa). All three isoforms have identical extracellular domains, composed of 2 immunoglobulin-like domains but diverge in their intracellular domain sequences.
Through a comprehensive analysis of how structural features in the IFN-/ß molecules, specifically critical clusters of amino acids, affect the sensitivity of target cells to IFN-induced biological responses, we (10
, 11
, 15)
and others (12
13
14
, 16
, 17)
have identified regions on the exposed surface of the IFN molecule that are associated with receptor recognition. Residues in the IFN-s that may mediate receptor binding have been mapped to the AB loop connecting helices A and B, the B and C helices, the D helix, the DE loop connecting helices D and E and the E helix. Our data demonstrated that the biological potency of a particular IFN-/ß subtype is determined at the level of receptor recognition, dictated by the nature of the interaction between the IFN subtype and the receptor complex (10
, 18
, 19)
. Our earliest studies focused on three regions of the IFN molecule, associated with residues 29–35, 78–95, and 123–140 in human IFN-2a, referred to as IFN receptor recognition peptides 1, 2, and 3 (IRRP1, IRRP2, and IRRP3), respectively. IRRP1 and IRRP3 are spatially in close proximity in the native protein and are kept in an active biological conformation by the Cys29 to Cys139 disulfide bond. The IRRP1 sequence is a loop structure between helices A and B, and the IRRP3 sequence is located in the DE loop and helix D. Residues within the AB loop and E helix (Fig. 2I
) contribute most of the binding energy for interactions with the extracellular IFNAR2 domain of the receptor. Residues in IFNAR2 contributing to binding to IFNs have been localized to the hinge region of the extracellular domain (10
, 11
, 12
, 13)
. The three-dimensional structure of the extracellular region of IFNAR2 associated with IFN-2a was solved by NMR spectroscopy recently (13)
. Additionally, in vitro binding studies of IFNAR2 and IFN-2a revealed a number of nuclear Overhauser effect-like constraints representing specific ligand-receptor residue pair interactions. Studies that have focused on the interactions of IFNs with IFNAR1 suggest that aromatic residues in the SD100B and SD100A domains of IFNAR1 function as the binding residues (14
, 15
, 16
, 17)
.
Despite the relatively high amino acid identity among the IFN- subtypes and the conserved secondary structures of IFNs, they bind with varying affinities to IFNAR. Human IFN-1 exhibits the weakest affinity for IFNAR and the lowest potency of all the human IFN- subtypes in vitro (14
, 18)
. In contrast, human IFN-2 binds with greater affinity to IFNAR and is the most potent of the naturally occurring IFN-s. Notably, IFN alfacon-1, an expanded-spectrum cytokine that was engineered to contain the most frequently occurring amino acids among the nonallelic IFN- subtypes, demonstrates increased potency compared to naturally occurring IFN-s in cell culture models and exhibits the highest affinity for IFNAR (18
, 19)
.
In this report, we describe a structure-based approach to rationalize the affinities of various IFNs toward IFNAR. Using available three-dimensional structural information, homology models of human IFN-1, human IFN-8, human IFN alfacon-1, and murine IFN-4 were generated and analyzed in the context of biological and biochemical data. Models of the complexes of the different IFNs with IFNAR2 were analyzed to evaluate the behavior of interacting residue pairs and the nature of the IFN-IFNAR2 binding interfaces. Subsequently, IFNAR1-human IFN-2-IFNAR2, and IFNAR1-murine IFN-4-IFNAR2 complexes were modeled to describe the complete extracellular receptor pocket contacting these IFN molecules.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
the MODELLER software (21)
, or the ClustalW software (22)
. Energy minimizations, molecular dynamics simulations, and trajectory analyses were carried out using the AMBER suite of programs version 7.0 (23)
. For all other tasks, Sybyl molecular modeling program was used. The following 3D structures were obtained from the Research Collaboratory for Structural Bioinformatics (http://www.rcsb.org): the NMR structures of human IFN-2 (1ITF), and the human IFNAR2 extracellular domain (1N6V), and the X-ray crystal structures of ovine IFN-
(1B5L), and human IFN-ß (1AU1). Human IFN-1 and IFN alfacon-1 were modeled using the respective three-dimensional structures (1ITF, 1B5L, and 1AU1). Murine IFN-4 and human IFN-8 were modeled using the structures, 1ITF and 1B5L.
Homology modeling
First, multiple sequence alignment was performed using the primary sequences of the IFNs human IFN-1, human IFN-8, IFN alfacon-1, and murine IFN-4 (Fig. 1A
). These alignments were further used for the homology model building.
The three-dimensional structure of the IFNAR1 domain (Fig. 3
) of IFNAR was modeled using the three-dimensional structure of the IFN-
receptor. First, the two IFNAR1 cytokine receptor homology (CRH) domain sequences were aligned with the human IFN- receptor 1 extracellular sequence using the ClustalW alignment algorithm in the MODELLER software, and each CRH domain was threaded into its three-dimensional model and energy-minimized (Fig. 3)
. To generate a complete model of the IFNAR1 domain, the membrane distal CRH domain was placed relative to the membrane proximal CRH domain, such that the IFNAR1 extracellular domain was aligned with it (vide infra) (24)
. The two CRH domains were connected at the hinge region by introducing a peptide bond between Pro203 and Pro204. Conformations of the backbone of the Pro residues in the hinge region were set so that the postulated bent structure of IFNAR1 due to the galabiosylceramide (Gb2) and/or globotriaosylceramide (Gb3) interactions (24)
could be reflected in the three-dimensional model of IFNAR1.
|
Docking of IFNs with the IFNAR2 extracellular region
Each IFN, i.e., human IFN-1, human IFN-2, IFN alfacon-1, human IFN-8, and murine IFN-4, was docked onto the extracellular region of IFNAR2. During this docking process, NOE constraints were used to guide the fitting of the IFN to the extracellular region of IFNAR2 (residues Met46, Lys48, His76, Glu77 in IFNAR2 interact with Arg144, Asp35, Ser152, and Arg149 in IFN-2, respectively) (13)
. The pairs of interacting atoms from the residue side-chains were positioned to optimize interactions. These distance constraints confirmed the overall orientation for the binding of IFNs at the binding site on IFNAR2. These complexes were then subjected to energy minimization followed by molecular dynamics simulations (vide infra).
IFN-IFNAR complex
Accumulating evidence suggests that the primary interaction at the cell surface is between IFN and IFNAR2, followed by the IFNAR2-bound IFN interacting with IFNAR1. IFNAR2-IFN-IFNAR1 interactions lead to receptor activation and intracellular signal transduction. Accordingly, complexes of IFN-1, IFN-2, IFN-8, and IFN alfacon-1 with extracellular (EC) IFNAR2 were derived as described in the previous section, then molecular dynamics simulation trajectories employed to generate the average structures for each of these IFN-EC IFNAR2 complexes. The energy-minimized structure of the IFNAR1 EC domain was then docked onto the energy-minimized average structure of the complex of a human IFN-IFNAR2. Docking was guided by the helical structural elements and associated interactions in IFNAR1 and IFN-2 for the best fit, due to the lack of reliable experimental data on specific interactions between IFN and IFNAR1. This IFNAR2-IFN-IFNAR1 complex was then energy minimized. Using this procedure, complete ligand-receptor complexes were generated with human IFN-2, IFN alfacon-1, and human IFN-1. All complexes were then energy-minimized using the standard protocol described below. Residue constraints based on NMR experiments were used to manually dock human IFN-1, human IFN-2, human IFN-8, and IFN alfacon-1 with the EC domain of human IFNAR2 (13)
. Each of these docked structures was subjected to energy minimizations and molecular dynamics simulations. The RMSD plots of the protein backbone C atoms and the total energy plots derived following MD simulations were used to establish the equilibrium and the trajectories for further analysis of these molecular systems (Supplementary Material).
Energy minimizations and molecular dynamics simulations
Energy minimizations and molecular dynamics simulations were performed using Amber version 7.0 suite of programs. The disulfide bridges were built using the information from Swiss-Prot database and the homologous positions of the cysteine residues. An all-atom force field and the default atomic charges were used as implemented in Amber software package. Each molecular system was solvated by a rectangular TIP3P water box that is at least 8 Å from the surface of the protein. During the energy minimization, nonbonded interactions were computed for up to a distance of 12.0 Å. The molecular systems were first energy-minimized using the steepest decent method for 200 cycles and then using conjugate gradient method up to 20,000 cycles.
Energy-minimized structures were used as the starting conformations for molecular dynamics simulations. A constant volume constraint (NVT) was used for the first 10 ps, and then the system was equilibrated at a constant pressure of 1 bar (NPT) until the total energy of the system was stable. All simulations were performed at 300 K, and a time step of 2 fs was used. During molecular dynamics simulations, the cutoff for the nonbonded interactions was set at 12 Å. Periodic boundary conditions were imposed during the molecular dynamics simulations and Ewald summation was used for electrostatics, as implemented in the Amber software. The snapshots from the molecular dynamics simulations were collected at every picosecond for up to 1 nanosecond. Total energy and the root-mean squared deviation (RMSD) for the C atoms were plotted against time (Supplementary Material). On the basis of the total energy profiles for the four complexes of IFN-IFNAR2, trajectories from 550 ps-1 ns for human IFN-2, human IFN-8, IFN alfacon-1, and from 650 ps-1 ns for human IFN-1, were used in subsequent analyses.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
s1, human IFN-8, IFN alfacon-1, and murine IFN-4, a structure-based approach to delineate the architecture of ligand-receptor interactions was used. At the outset, the primary amino acid residue sequences for each of these IFNs were compared to those of IFNs with experimentally solved three-dimensional structures, namely human IFN-2, ovine IFN-, and human IFN-ß (Fig. 1A
). Table 1
illustrates the high percentage of primary sequence identity scores. IFN alfacon-1, which is a synthetic IFN, showed the highest identity with all three solved X-ray crystal structures. Amino acid residue composition in the structurally conserved regions (highlighted by red bars in Fig. 1A
) are almost identical among all four IFNs depicted. Thus, three-dimensional structural similarities are also anticipated. Additionally, the two disulfide bridges are located at identical positions, thus providing reasonable confidence that the three-dimensional folds will be quite similar for each of these IFNs (Fig. 1A
). The only exception is murine IFN-4, where a gap was observed between the residues Gln102-Glu103, which would result in a shorter loop region between helices C and D in comparison to that in human IFN-2 (Fig. 2)
.
|
Modeling IFNAR1
While the primary amino acid sequence alignments demonstrate the considerable identity among these IFNs, subtle structural differences associated with residue differences are anticipated to modulate receptor affinity, resulting in differential receptor activation. Accumulating evidence suggests that the ligand first binds to IFNAR2, and a complete receptor-ligand complex is then formed by recruitment of IFNAR1 (25
, 26)
. Notably, earlier studies showed that the cell membrane glycosphingolipids, galabiosylceramide (Gb2), and/or globotriaosylceramide (Gb3), are required for appropriate presentation of IFNAR1 at the cell surface (24)
. The Gb2/Gb3 interactions with IFNAR1 are believed to contribute to the proper orientation of the receptor chain, so that IFN molecules are bound efficiently to induce receptor activation and signaling. There are five N-glycosylation sites in the first fibronectin type III domain of IFNAR1 which, if glycosylation occurs, may also be mediating Gb2/Gb3 interactions with IFNAR1 through carbohydrate residue interactions. To accommodate the glycosphingolipid-IFNAR1 interaction, the EC domain of the IFNAR1 chain must fold over itself, resulting in proper presentation of the IFNAR1 binding surface to the IFN ligand. Thus, the three-dimensional model of IFNAR1 generated in this investigation (Fig. 3)
has taken into account data demonstrating the requirement for Gb2/Gb3 interactions for activity (24)
.
Specific residues in IFNAR1 that may be important in mediating binding and signaling have been identified using site-directed mutagenesis, namely Lys66, Tyr70, and Trp129 (17)
. In the context of the IFNAR1 model proposed here (Fig. 3)
, Lys66 and Trp129 would be capable of interacting with cognate receptor interactive residues on the IFN molecule, whereas Tyr70 would not. Mutation of Tyr70 in IFNAR1 abolished more than 80% of the binding affinity for human IFN-2 and human IFN alfacon-1. Cells expressing this mutated receptor were refractory to IFN-inducible transcription factor activation and antiviral activity. When Tyr70 is mapped onto the three-dimensional structure of IFNAR1, it is located away from the ligand binding surface of IFNAR1 (Figs. 3
and 5)
. This aromatic residue may be involved in stabilizing the cytokine receptor homology (CRH) subdomain. Tyr70 may not be playing a direct role in ligand binding but contributing to the structural integrity of IFNAR1 for optimal interactions with Gb2/Gb3 molecules. Tyr70 is in close proximity to Phe96 in the same subdomain and Trp183 in the adjacent subdomain of the N-terminal CRH domain of IFNAR1, allowing for putative interactions between these three aromatic residues. All three aromatic residues are solvent exposed in this model, and entropically driven hydrophobic interactions could develop in this region, contributing to the stabilization of the CRH domain. Furthermore, if a Tyr70-mutant disrupts the structure of IFNAR1, it is likely that glycosphingolipid interactions are also disrupted leading to unproductive ligand binding (17)
. A neutralizing antibody, 64G12, targets the region containing Tyr70 and affects ligand binding and biological activity, (27)
perhaps by interfering with Gb2/Gb3 interactions.
|
Determinants of species-specificity
Murine IFN-4 has a high degree of sequence identity with human IFN-2 and ovine IFN- (59.3% and 42.6%, respectively) (Fig. 1A
). The three-dimensional structural model of the complex of human IFN-2 and IFNAR2 (Fig. 4
A) was used to investigate murine IFN- interactions with IFNAR2. Specifically, muIFN-4 C atoms were fit to the C atoms of human IFN-2 on the interaction surface, based on homology alignment (huIFN-2: R12-Q20, C138-N156 and muIFN-4: K12-E20, C134-N152). A network of electrostatic interactions is observed between huIFN-2 and IFNAR2, along with hydrophobic interactions and hydrogen bonds (Fig. 4A
). Most of the residues involved in these interactions are concentrated on the helices A, E, and the AB-turn (Fig. 4A
and Table 2
). Table 2
outlines the key interactions observed at the binding interface of IFNAR2 and the ligand, in each of the complexes with huIFN-1, huIFN-2, IFN alfacon-1, and muIFN-4. The data suggest that IFN alfacon-1 has the highest number of optimal interactions with IFNAR2: hydrophobic, electrostatic, and hydrogen bonding. MuIFN-4 exhibits the least number of optimal interactions (Table 2)
.
|
|
An alignment of primary amino acid residue sequences reveals that mouse IFN-4 is missing residues 103–107 that are present in the human IFN-s. These missing amino acids are located on the CD loop of the human IFNs (Fig. 1A
). As a result, the CD loop in murine IFN-4 shifts outward, positioning itself to contact the IFNAR1 backbone between residues Thr123 and Met-128 (Fig. 2
, Panel II). The C atom of Met-101 on IFN-2 is as close as 2.6 Å to that of Leu-124 of IFNAR1, and the distance between the C of Gln-102 on murine IFN and the C of Leu-124 of IFNAR1 is 2.9 Å. The side chains in this region do not permit interactions between murine IFN-4 and human IFNAR1, due to steric crowding, contributing to poor interactions between these two proteins.
At the interface of human IFN-2 and IFNAR2, eight interaction sites were identified from the structural model of the corresponding ligand-receptor complex. In contrast, only five of these were conserved in the modeled murine IFN-4 interaction with IFNAR2 (Fig. 5
). Additionally, it appears that Trp144 in murine IFN-4 would sterically hinder the approach of murine IFN-4 toward the ligand binding surface IFNAR2, affecting other stronger interactions such as Glu77:IFNAR2·Arg145:murine IFN-4 (Fig. 5)
, resulting in poor binding to IFNAR2. This reduced binding affinity would then result in an overall weak or null response. Furthermore, because of conservation of residues among the murine IFN-s, these observations focused on murine IFN-4 bound to IFNAR2, which will likely be reflected in the context of other murine IFN subtypes, providing an explanation for the inactivity of murine IFNs in human cells.
Another important residue, Ala19, is conserved among human IFNs and may interact with Met46 and His76 on IFNAR2. Glu19 at this position on murine IFN-4, however, compromises the ability of murine IFN-4 to bind to IFNAR2 due to the charged side chain. The conserved Gln20 residue in human IFNs undergoes hydrogen bonding interactions with Glu189 and Gln191 on IFNAR2. Murine IFN-4 presents Glu20 at this position, which could result in charge-charge repulsion with Glu189 on IFNAR2, further contributing to the inactivity of murine IFN-4 in the context of the human IFNAR receptor complex.
Residues Glu159 and Lys164 on human IFN-2 create a salt bridge network with Lys159 and Glu132 on IFNAR2. On murine IFN-4, Ala155 disrupts this network affecting the potential binding of murine IFN-4 to IFNAR2. A conserved residue, Glu78, on huIFN-2 interacts with Lys240 on IFNAR1, an interaction which is eliminated by the presence of Ala69 in muIFN-4. Viewed altogether, these altered residue-level interactions are sufficient to affect the binding of murine IFN-4 to human IFNAR2. Thus, despite the high degree of homology and three-dimensional structure similarities, murine IFN-4 presents unfavorable binding interactions at the binding interface with IFNAR2, which explains its poor affinity for the human IFNAR.
In contrast to other human IFN-s, human IFN-8 exhibits activity in mouse cells, implying productive interactions with the murine receptor complex and/or structural similarities between human IFN-8 and murine IFN-4 in those regions that interact with the mouse receptor. Indeed, residues in the IRRP-I region (helix A and AB loop; contact region with IFNAR2) of human IFN-8 show a higher degree of identity with those in murine IFN-4 than those in human IFN-2 (Fig. 1B
). Residues 11 to 53 in human IFN-8 are 49% identical with the corresponding region in murine IFN-4, but only 35% identical with that in human IFN-2. This region on human IFN-2 forms a large contact surface with human IFNAR2. We infer, therefore, that the activity of human IFN-8 in murine cells resides in its ability to interact with murine IFNAR2 in a similar manner to murine IFN-4.
IFN--IFNAR2 interactions
The IFNAR2 gene encodes a soluble form, IFNAR2a, which may function in down-regulation of an IFN response by sequestering IFN molecules away from membrane IFNAR2c, thereby prevent signaling. Alternatively, IFNAR2a may present IFNs to IFNAR1 to enable signaling in the absence of IFNAR2c. Docked IFN-IFNAR2 models are therefore biologically relevant models with which to study IFN ligand interactions with IFNAR2.
The three-dimensional structures of human IFN-1, human IFN-8, and human IFN alfacon-1 were compared with that of human IFN-2, and their potential interactions within the IFNAR2 binding pocket were analyzed. These four human IFN- subtypes were docked in the putative binding site of IFNAR2 to evaluate their potential interactions with this receptor subunit. Recent studies examining the interaction between human IFN-2 and the IFNAR2-EC domain have suggested that IFN-2 does not undergo significant changes in conformation on binding to IFNAR2 (28)
. Mutagenesis and NMR cross-saturation experiments have identified a number of residues in human IFN-2 that contribute significant free energy to binding (13)
. In our three-dimensional model of human IFN-2-IFNAR2-EC, the following residues in human IFN-2 interact with IFNAR2 residues: Leu26, Phe27, Arg33, Asp35, Glu146, and Arg149. The current three-dimensional structure model has also identified IFN-2 residues involved in IFNAR2 interactions not reported in the NMR studies, i.e., Arg12, Gln20, Lys164, Glu159 (see Table 2
). The importance of the latter residues in human IFN-2-IFNAR2 interactions requires further consideration. As indicated in Table 2
and Fig. 4
, two major interaction regions (b and c regions, Fig. 4
) were identified between human IFN-2 and IFNAR2. These interactions are localized to the helices A, AB loop, and helix E on human IFN-2. A hydrophobic pocket on human IFN-2 accommodates Pro-4 of IFNAR2, and a number of electrostatic interactions were observed that maintain the close interactions between these two protein surfaces (Table 2)
.
Among all of the naturally occurring human IFN-s, huIFN-1 has the weakest affinity toward its receptor and is also the least biologically potent IFN (29)
. IFN-2 is the most potent of the naturally occurring type I IFN subtypes. The IRRP1 sequence in IFN-1 differs by one amino acid compared to IFN-2: Cys29-Leu-Met-Asp-Arg-His-Asp vs. Cys29-Leu-Lys-Asp-Arg-His-Asp (Fig. 1A
). The nonconservative substitution at position 31 (Lys31Met) may likely be responsible for the disruption of the 310-helix occurring at this site. It is plausible that the Met31 in huIFN-1 is not able to contribute the necessary side-chain interactions that result in proper formation of the 310-helix, subsequently leading to improper presentation of key binding residues.
Because of the nature of the secondary structure of the AB loop in IFN-1, there is a disruption in the spatial conformation of the "hot spot" residues within this loop, preventing appropriate interactions with IFNAR2. In fact, the binding pocket structure recognized by IFNAR2 could not be found in the IFN-1 model. Apparently, the disordered IFNAR2 binding region in IFN-1 may contribute to the low binding affinity exhibited by this subtype. It is reasonable to postulate that residue substitutions in the binding regions in other Type I IFN molecules also result in changes in local secondary structure, thereby affecting receptor affinities.
Mutagenesis studies have identified that the carboxy-terminal unstructured tail region in the human IFN-s may play a role in contributing to their differential binding and biological activities (30)
. Accumulating evidence suggests that this tail region acquires a specific conformation on binding to IFNAR2 in the groove below residues 132–134 that constitute a loop in this receptor subunit. Further refinement of the IFN-IFNAR models proposed herein therefore requires consideration of potential interactions between IFNAR2 and the carboxyl-termini of different IFNs.
Close scrutiny of interacting residues between IFN-2 and IFNAR2 identify specific contact sites that are precluded with mouse IFN-4. (Table 3
, Fig. 5B
). For example, Trp144 in IFN-4 may interfere with binding to IFNAR2, in addition to various other unfavorable interactions with IFNAR2 (vide supra). Table 3
depicts other structurally conserved residues at the interface of the ligand-IFNAR interaction.
|
Modeling the complete IFN-IFNAR complex
As discussed, a model of human IFNAR1 was generated, taking into consideration the IFNAR1 EC domain interaction with glycosphingolipids. Specifically, IFNAR1 was configured to adopt a bent conformation to accommodate interactions with membrane bound Gb2/Gb3 molecules. Residues that mediate IFNAR1 receptor binding identified using site-directed mutagenesis were found to be solvent exposed and part of the IFNAR1-IRRP2 binding interface. Tyr70 has been implicated as a binding residue based on binding and signaling data (17)
. However, the IFNAR1 model generated in the current study suggests that Tyr70 has a stabilizing function, ensuring proper folding of the membrane distal CRH domain by participating in intra- and interdomain hydrophobic interactions. To model the complete IFNAR complex in association with IFN-2, the energy-minimized IFNAR1 model was docked with the IFN-2-IFNAR2 complex model. In this structural model, Tyr89 in IFN-2 is in proximity to Trp129 and Tyr157 in IFNAR1. The phenyl rings of Tyr89 and Tyr157 can stack against each other and stabilize the side-chains of these two solvent-exposed hydrophobic residues. The phenyl ring of Tyr248 in IFNAR1 is positioned to fit into a pocket formed between Glu78 and Ser-73 in IFN-2. A putative salt bridge between the side-chain of Lys296 from IFNAR1 and Asp82 from IFN-2 was also observed in the model of the corresponding complex. Another putative hydrogen bond forming between Lys240 of IFNAR1 and Glu78 of IFN-2 was observed. On the basis of this model, it is plausible that both hydrophobic interactions and salt bridges stabilize the complex of IFN-2 and IFNAR1.
A number of interactions between IFNAR1 and IFN-2 may be extrapolated from the three-dimensional model of the complex. The BC loop and part of the C helix in IFN-2 has a large contact surface with IFNAR1 (Fig. 5)
. Residues Ala-74 and Ala-75 on the BC loop of IFN-2 form a hydrophobic surface, and this surface interacts with Tyr-248 on IFNAR1. There is a hydrogen bond network in this region that appears to be critical for the binding of human IFN to its receptor (Fig. 5B
and Table 3
).
The critical determinants of receptor activation that result in signal transduction are the high-affinity interactions between the IFN molecule and the extracellular regions of IFNAR2 and IFNAR1. Herein, we utilized existing three-dimensional structural information coupled with various three-dimensional structural models to shed light on the structural basis for the interactions of different IFNs with IFNAR1 and IFNAR2. The proposed models support biochemical data that IFNAR1-IFN-2 interactions are mediated by specific aromatic residues, which are perhaps further stabilized by salt bridges. This model of the IFN-2-IFNAR1 interaction is reminiscent of the IFN-2-IFNAR2 interaction, where hydrophobic and hydrophilic interactions are both important in mediating receptor-ligand interactions. Finally, we propose models for IFNAR1-IFN-2-IFNAR2 interactions and IFNAR1-IFN-4-IFNAR2 interactions that go part way toward explaining the species-specificity of different IFN- subtypes.
|
ACKNOWLEDGMENTS |
|---|
Received for publication March 19, 2007. Accepted for publication April 19, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
/ß receptor interactions to biologic outcomes: understanding the circuitry. J. Interferon. Cytokine. Res. 22,835-845[CrossRef][Medline]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
