Skip to main content
Proceedings of the National Academy of Sciences of the United States of America logoLink to Proceedings of the National Academy of Sciences of the United States of America
. 2006 Feb 13;103(8):2788–2793. doi: 10.1073/pnas.0511161103

Crystal structure of the IL-2 signaling complex: Paradigm for a heterotrimeric cytokine receptor

Deborah J Stauber *,†,, Erik W Debler *,, Patricia A Horton *,, Kendall A Smith §, Ian A Wilson *,
PMCID: PMC1413841  PMID: 16477002

Abstract

IL-2 is a cytokine that functions as a growth factor and central regulator in the immune system and mediates its effects through ligand-induced hetero-trimerization of the receptor subunits IL-2Rα, IL-2Rβ, and γc. Here, we describe the crystal structure of the trimeric assembly of the human IL-2 receptor ectodomains in complex with IL-2 at 3.0 Å resolution. The quaternary structure is consistent with a stepwise assembly from IL-2/IL-2Rα to IL-2/IL-2Rα/IL-2Rβ to IL-2/IL-2Rα/IL-2Rβ/γc. The IL-2Rα subunit forms the largest of the three IL-2/IL-2R interfaces, which, together with the high abundance of charge–charge interactions, correlates well with the rapid association rate and high-affinity interaction of IL-2Rα with IL-2 at the cell surface. Surprisingly, IL-2Rα makes no contacts with IL-2Rβ or γc, and only minor changes are observed in the IL-2 structure in response to receptor binding. These findings support the principal role of IL-2Rα to deliver IL-2 to the signaling complex and act as regulator of signal transduction. Cooperativity in assembly of the final quaternary complex is easily explained by the extraordinarily extensive set of interfaces found within the fully assembled IL-2 signaling complex, which nearly span the entire length of the IL-2Rβ and γc subunits. Helix A of IL-2 wedges tightly between IL-2Rβ and γc to form a three-way junction that coalesces into a composite binding site for the final γc recruitment. The IL-2/γc interface itself exhibits the smallest buried surface and the fewest hydrogen bonds in the complex, which is consistent with its promiscuous use in other cytokine receptor complexes.

Keywords: common γ chain, cooperativity, IL-2 receptor, receptor assembly, structure–activity relationship


In 1976, lymphocyte-conditioned medium was found to support the long-term growth of T lymphocytes (T cells) (1), but the active component was not characterized as a variably glycosylated 15.5-kDa protein until 1981 (2). Subsequently, IL-2 was purified to homogeneity (3), and its cDNA was cloned in 1983 (4). The discovery of IL-2 permitted the creation of monoclonal T cells (5), which were instrumental in the characterization of many aspects of T cell biology. IL-2 is the primary cytokine responsible for the rapid expansion, differentiation, and survival of antigen-selected T cell clones during an immune response, but is also important for B cell, natural killer (NK) cell (6), and regulatory T cell (Treg) (7) function.

IL-2 was the first cytokine found to mediate its effects via a cell surface binding site with all of the characteristics of classic hormone receptors, including high affinity, stereospecifity, and saturability (8). Kinetic binding studies revealed that the high affinity binding (Kd ≈ 10 pM) to the IL-2 receptor (IL-2R) is due to a rapid rate of association (kon ≈ 107 M−1·s−1), but relatively slow dissociation rate (koff ≈ 10−4 s−1) (9). The high-affinity IL-2R comprises three separate, noncovalently linked chains, termed α (IL-2Rα, p55, CD25) (10), β (IL-2Rβ, p75, CD122) (11, 12), and γ (γc, IL-2Rγ, p65, CD132) (13), the latter being a common receptor component for many cytokines, including IL-4, IL-7, IL-9, IL-15, and IL-2 (14). IL-2Rα contributes the rapid association rate of IL-2 binding, whereas IL-2Rβ is responsible for its slow dissociation rate (9).

A critical number of IL-2Rs must be triggered before an individual T cell will make the irrevocable, quantal (all-or-none) decision to pass through the G1 restriction point so as to undergo DNA replication and subsequent mitosis (15, 16). However, exactly how the cell senses this threshold of triggered IL-2Rs has remained obscure. A clue to this issue may then reside in the structure–activity relationship (SAR) of the IL-2/IL-2R interaction.

Here, we present the crystal structure of the fully assembled, human IL-2 receptor ectodomains in complex with its ligand at 3.0 Å resolution. The molecular architecture of this heterotrimeric receptor can now be analyzed to decipher the assembly, signaling, and disassembly mechanisms of this enigmatic cytokine receptor complex, and provide insights into the cooperativity of high-affinity binding, as well as the promiscuous use of γc in other cytokine receptor complexes.

Results

Assembly of the Quaternary IL-2 Signaling Complex.

The high-affinity complex of IL-2 with its extracellular IL-2Rα, IL-2Rβ, and γc receptor domains was assembled from the individual receptor ectodomains expressed in insect cells. Throughout purification of γc, this receptor subunit existed as a stable homotrimer, as determined by gel filtration. When incubated at 37°C in the presence of IL-2Rα, IL-2Rβ, and IL-2, the γc trimer could dissociate and incorporate into the stable quaternary IL-2 signaling complex. However, upon chemical deglycosylation, γc aggregated and could not be dissociated for engagement in the high-affinity complex. Thus, self interaction through trimerization and surface carbohydrate sites on γc may serve an important solubility role, masking epitopes for IL-2 and IL-2Rβ recognition until formation of a productive signaling complex.

Structure Determination of the Quaternary IL-2 Signaling Complex.

The 3.0-Å crystal structure was determined by molecular replacement by using the coordinates from an independently assembled IL-2/IL-2R complex (17), in which mutations at Asn-17βGln and Asn-45βGln eliminated two glycosylation sites and which crystallized in a different space group (C2). A brief comparison of the IL-2/IL-2R complexes in the triclinic (here) and the monoclinic crystal form (17) is presented in Supporting Materials and Methods, which is published as supporting information on the PNAS web site. The x-ray structure was refined to Rfree and Rcryst of 26% and 22%, respectively (Table 2, which is published as supporting information on the PNAS web site). Two independent complexes in the asymmetric unit offered a valuable cross validation of key architectural features of the signaling complex (Table 1). In the following analyses, however, only values pertaining to the first complex (chain IDs A, B, C, and D under PDB ID code 2ERJ) are cited unless structural differences are explicitly highlighted or the values differ significantly.

Table 1.

Properties of the IL-2/IL-2Rα/IL-2Rβ/γc interfaces

IL-2 IL-2Rα IL-2Rβ γc (adjacent molecule)
IL-2Rα
    Total buried surface area, Å2 1590/1690 1150/1100
    % main chain buried surface area 20/22 17/17
    % polar buried surface area 67/68 67/63
    No. van der Waals contacts 94/91 70/80
    No. hydrogen bonds 6/6 9/9
    Shape correlation (Sc) 0.71/0.64 0.68/0.78
    Affinity constant (Kd), from ref. 28 10 nM ND
IL-2Rβ
    Total buried surface area, Å2 1160/1150 0
    % main chain buried surface area 12/12 0
    % polar buried surface area 69/70 0
    No. van der Waals contacts 67/76 0
    No. hydrogen bonds 10/10 0
    Shape correlation (Sc) 0.68/0.74 0
    Affinity constant (Kd), from ref. 28 144 nM 278 nM
γc
    Total buried surface area, Å2 970/980 0 1640/1610
    % main chain buried surface area 18/16 0 17/14
    % polar buried surface area 86/63 0 74/74
    No. van der Waals contacts 2/2 0 97/90
    No. hydrogen bonds 71/74 0 14/14
    Shape correlation (Sc) 0.76/0.72 0 0.64/0.63
    Affinity constant (Kd), from ref. 28 ND* ND ND

ND, not detected.

*However, weak binding (Kd > 50 μm) was detected in other assays, e.g. ref. 40.

Architecture of the Quaternary IL-2 Receptor Signaling Complex.

The assembly of the IL-2Rβ and γc ectodomains bound to IL-2 recapitulate architectural features of the prototypic, human growth hormone (GH) receptor dimer in that a four-helix bundle cytokine is clamped between the elbow regions of two receptor subunits that are roughly related by pseudo-2-fold symmetry, but with a slight translation of γc relative to IL-2Rβ (Fig. 1) (18). The interdomain region of γc contacts IL-2 helices A and D (classical binding site I), and the elbow region of IL-2Rβ contacts IL-2 helices A and C (classical binding site II) (18). IL-2Rα interacts mainly with the long connection between helices A and B that include helices A′ and B′ of IL-2, in a region recently termed “site IV” (19). IL-2Rα surprisingly does not contact IL-2Rβ or γc, and the three interaction sites on IL-2 generally do not overlap with each other, except for a small, but significant, region (see below). The IL-2 molecule in this quaternary complex undergoes only minimal conformational changes, the most significant of which is the ordering of the BC loop (residues 74–81) due to crystal-packing interactions. The IL-2Rβ and γc constructs terminate just 6 and 7 residues from the transmembrane domains, respectively (20). Thus, we can estimate the distance from the observed C terminus (Gly-165α) of IL-2Rα to the membrane to be ≈40 Å, consistent with the predicted unstructured 54-residue linker between Gly-165α and the putative transmembrane domain (20).

Fig. 1.

Fig. 1.

Architecture of the human IL-2R signaling complex. Shown are side (A) and top (B) views of the quaternary IL-2 signaling assembly. IL-2 binds to the elbow regions of IL-2Rβ and γc. IL-2Rα docks on top of this assembly without forming any contacts with the other two receptor subunits. Six N-linked carbohydrates (S1 to S6) are displayed as ball-and-stick models. S1 is wedged between D1 and D2 of IL-2Rβ and, thus, contributes to the stabilization of a specific D1/D2 interdomain angle.

Structure of IL-2Rα and Its Interface with IL-2.

IL-2Rα folds into two “sushi-like” domains D1 and D2, which form five-stranded β-sheet sandwiches. These domains are disposed at ≈75° to each other, as in the crystal structure of IL-2Rα complexed with IL-2 (19). D1 accounts for the majority (82%) of the total buried surface area (1,590 Å2) between IL-2Rα and IL-2, whereas D2 contributes only to a minor extent. The nature of the IL-2/IL-2Rα interface reveals a striking dichotomy of a hydrophobic center dominated by IL-2Rα residues Leu-2α, Met-25α, Leu-42α, and Tyr-43α and IL-2 residues Phe-42IL-2, Phe-44IL-2, Tyr-45IL-2, Pro-65IL-2, and Leu-72IL-2, and a polar periphery featuring five ion pairs (Lys-38α/Glu-61IL-2, Arg-36α/Glu-62IL-2, Glu-1α/Lys-35IL-2, Asp-6α/Arg-38IL-2, and Glu-29α/Lys-43IL-2; Fig. 2). Seven hydrogen bonds are formed between IL-2 and IL-2Rα (Table 3, which is published as supporting information on the PNAS web site). No significant differences in side- and main-chain conformations are observed at the core of the IL-2/IL-2Rα interface, but some variation in total buried surface area and shape correlation Sc in the two complexes (0.71 and 0.64, by using a 1.7-Å probe) is noted (Table 1). A slight rigid-body rotation (10°) of IL-2Rα with respect to IL-2 seems to result from different crystal packing environments that would give rise to the slight variations in shape complementarity and buried surface.

Fig. 2.

Fig. 2.

Interface between IL-2Rα and IL-2. Open-book representation of the IL-2/IL-2Rα interface. The electrostatic potential was mapped onto the molecular surface and contoured at ±35kT/eV (blue/red). The interface features a hydrophobic center, flanked by a large number of salt bridges and other polar contacts. The strong electrostatic component of this interaction serves to rapidly capture IL-2 and, thus, to dominate the kon rate of IL-2 binding to the IL-2R.

Architecture of IL-2Rβ and γc.

The ectodomains of IL-2Rβ and γc possess a cytokine-binding homology region, which is divided in two fibronectin type-III (FN-III) domains termed D1 and D2. Each domain contains seven β-strands that form a sandwich of two antiparallel β-sheets. The N-terminal D1 domains of IL-2Rβ and γc both conform to the hybrid type of the Ig fold (21) where the C′ strand hydrogen bonds with both faces of the β-sandwich (Fig. 1), similar to erythropoietin receptor (EPOR) (22). In contrast, the D2 domains belong to the switched type of Ig folds, where strand C′ hydrogen bonds only with strand C. Domain D1 (residues 6–99) and D2 (103–209) of IL-2Rβ are disposed at an elbow angle of 73°, characteristic of class I cytokine receptors (GHR, 64°; IL-4R, 82°; EPOR, 62° and 77°), whereas in γc the elbow angles between D1 (32–125) and D2 (129–226) increases to 92° and 98°, respectively, more typical of class II cytokine receptors (IL-10R1, 93°; IFN-αR2, 100°; tissue factor, 118°) (23). The distinct elbow angle in γc is also accompanied by a more extended interdomain region. Finally, the D1 and D2 linker regions of IL-2Rβ (residues 100–102) and γc (126–128) both adopt a 310 helical conformation.

A common feature of class I cytokine receptor domains includes two highly conserved disulfide bridges in D1 that link strands A to B, and strands C′ to E, as in IL-2Rβ and γc. IL-2Rβ D1 contains an additional disulfide between strands C and F (Cys-33β-Cys-84β). D2 of γc exhibits another unique disulfide that connects loops BC2 and FG2 (Cys-160γ-Cys-209γ) and significantly contributes to the interface with IL-2 (Fig. 4). In D2 of both IL-2Rβ and γc, a tryptophan-arginine ladder on sheet C′-C-F-G, that contains the highly conserved WSXWS motif, is present, as in other cytokine receptors (18, 24).

Fig. 4.

Fig. 4.

Interface of γc with IL-2 and IL-2Rβ. This γc interface (green/red) with IL-2 (yellow/red) is characterized by the smallest total buried surface area and formation of only two hydrogen bonds. Strikingly, the elbow region of γc engages only partially with IL-2. Furthermore, γc exhibits a large elbow angle, which is atypical of class-I cytokine receptors. These structural properties are likely to contribute to promiscuity of this receptor subunit. The large contact area with IL-2Rβ is colored brown and includes three residues (red) which also interact with IL-2 (red residues).

In IL-2Rβ, a particularly well defined glycan structure, consisting of four sugars N-linked to Asn-17, is wedged between D1 and D2 (Fig. 1A and Fig. 6, which is published as supporting information on the PNAS web site). The α1,6-linked core fucose forms four hydrogen bonds with Arg-105β and Leu-106β of D2. Moreover, the four sugars make numerous van der Waals contacts with D1 and D2, which may help stabilize a precise orientation of the respective fibronectin type-III domains to each other.

Interface Between IL-2Rβ and IL-2.

IL-2 forms the second largest interface with IL-2Rβ (1,150 Å2, Table 1) with good shape complementarity (0.68) and high specificity via 10 hydrogen bonds (Table 3). Recognition of IL-2 by IL-2Rβ is dominated by the central, protruding Tyr-134β, which contributes 17% of the buried surface area on IL-2Rβ (Fig. 3). For IL-2, Asp-20IL2 and His-16IL2 seem to be the most critical residues; Asp-20IL2 hydrogen bonds to His-133β and Tyr-134β, whereas His-16IL2 is tucked into a slot created by Tyr-134β, Gln-188β, and the methyl groups of Thr-74β and Thr-73β. Major van der Waals contacts in the interface are also made with Arg-41β, Val-75β, His-133β, Leu-19IL2, Asp-84IL2, Asn-88IL2, and Val-91IL2.

Fig. 3.

Fig. 3.

Interface of IL-2Rβ with IL-2 and γc. The interface of IL-2Rβ (cyan/red) and IL-2 (yellow/red) is intermediate in buried surface area among the three IL-2/IL-2R interfaces. Asp-20 and His-16 of IL-2 are located at the center of the binding region and are involved in many van der Waals contacts and hydrogen bonds. The interface features the highest number of hydrogen bonds, consistent with its high specificity and low koff rate of IL-2 dissociation. The large contact area with γc is colored brown and includes two residues (red) that also interact with IL-2 (red residues).

Interface Between γc and IL-2.

The γc/IL-2 interaction (970 Å2) is the smallest of the three IL-2/IL-2R interfaces, although with good surface complementarity (0.72). Furthermore, this contact surface features only two hydrogen bonds (Table 3) and weak electrostatic interactions (Fig. 4), indicating low specificity. The contribution of main-chain atoms to the buried surface area is intermediate (17% on average) between those of IL-2Rα (21%) and IL-2Rβ (12%) (Table 1). Tyr-103γ, His-159γ, and Leu-208γ together with the disulfide (Cys-160γ-Cys-209γ) contribute 53% of the buried surface on γc and represent the major binding determinant. The importance of the disulfide bridge was previously appreciated from mutagenesis data (25, 26), but its direct involvement in ligand binding was unexpected. With a buried surface area of 72 Å2 and one hydrogen bond, Gln-126IL2 is the most critical IL-2 residue that contacts γc in accordance with biochemical data (27). Glu-15IL2, Thr-123IL2, and Ile-129IL2 contribute other significant interactions to the IL-2/γc interface.

Interface Between IL-2Rβ and γc.

In stark contrast to the relatively small IL-2/γc interface, the IL-2R structure reveals a surprisingly large and specific interaction between the D2 receptor domains of IL-2Rβ and γc (1,640 Å2). The predominantly polar contacts are evenly distributed among the 22 residues of IL-2Rβ and 31 residues of γc without any obvious hot-spot residues. The low percentage of aromatic residues among the side-chain contributions (12%), in combination with 14 hydrogen bonds, seems to enforce much higher specificity of γc with IL-2Rβ than of γc to IL-2. Glu-136β and His-138β form the hinge of the D1/D2 IL-2Rβ elbow and significantly interact with both IL-2 and γc. Similarly, Tyr-182γ, Pro-207γ, and Leu-208γ form the hinge of the D1/D2 γc elbow and interact with both IL-2 and IL-2Rβ.

Interface Between IL-2Rα and γc: Crystal Artifact or Physiologically Relevant.

A crystal contact of considerable size and specificity is formed between γc and a symmetry-related IL-2Rα molecule in the neighboring unit cell. Remarkably, this interaction is observed in both copies of the asymmetric unit, as well as in another crystal form (17). IL-2Rα D2 strands H, I, and J interact with the surface of the γc elbow region that is not involved in IL-2 binding. Structural features include high shape complementarity (0.68 and 0.78, respectively), a large buried surface area (1,150 Å2) and nine interdomain hydrogen bonds. In this arrangement, the D2 domains of IL-2Rβ/γc are parallel to each other, with their C termini toward the membrane (Fig. 7, which is published as supporting information on the PNAS web site). Such lateral interactions would lead to one-dimensional lattices of IL-2R complexes on the cell surface that could promote higher-order signaling assemblies, although no measurable affinity between soluble IL-2Rα and γc has been detected (28). However, this same set of experiments also did not detect any binding between IL-2Rβ and γc, which in fact form an extensive interface in the fully assembled IL-2R.

Discussion

Despite only 10–20% sequence identity to other cytokine receptors, the IL-2Rβ and γc ectodomains of the IL-2 receptor have long been recognized to consist of two tandem fibronectin type-III domains that contain sequence signature motifs as in other cytokine receptors (13, 29). As expected, a DALI search (30) identified other cytokine receptors, such as IL-4R (Z-score of 17.3, rms deviation (rmsd) of 2.4 Å over 176 residues) and erythropoietin receptor (16.5, 3.4 Å over 182 residues) as the closest structural homologs of IL-2Rβ and γc. However, the crystal structure of the complete IL-2/IL-2R complex now enables analysis of the organization of the quaternary assembly and the nature and extent of each of the IL-2/IL-2R interfaces and receptor subunit interactions, as well as the role of the individual receptor subunits in the high-affinity complex assembly and signal transduction.

The precise role of IL-2Rα has so far remained elusive. This receptor subunit, together with IL-2Rβ, can form a pseudo-high-affinity site for IL-2 on the cell surface (Kd ≈ 300 pM), yet IL-2Rβ itself binds IL-2 only with modest affinity (Kd ≈ 450 nM) (31). In contrast, a soluble form of IL-2Rβ comprising solely its ectodomain exhibits comparable IL-2 affinity (Kd ≈ 144 nM) to its membrane-bound form, but only ≈2-fold increased affinity (Kd ≈ 64 nM) upon addition of soluble IL-2Rα (28). Minor changes observed in the IL-2 structure in response to IL-2Rα binding are probably sufficient to explain the 2-fold increase in IL-2Rβ affinity for IL-2 in solution. However, because IL-2Rα does not make any interactions with IL-2Rβ, the massive difference in the Kd of IL-2 with membrane-anchored IL-2Rβ in the presence or absence of IL-2Rα suggests that this subunit primarily serves as a ligand carrier. For potent cell signaling, secreted IL-2 must be captured at the cell surface to minimize its loss by diffusion away from the cell. The high affinity of IL-2Rα for IL-2 and the large excess of IL-2Rα over IL-2Rβ and IL-2Rγ on activated T cells facilitate this capture and delivery to IL-2Rβ or to any IL-2Rβ/γc complex through two-dimensional cell surface diffusion. Computational modeling suggests that a stepwise assembly mechanism leads to a greater number of high-affinity signaling complexes than signaling via a preformed, fully assembled IL-2R, considering the thermodynamics, kinetics, receptor density, and cell density of the IL-2/T cell system (32). Alternatively, cooperativity in formation of the ternary IL-2/IL-2Rα/IL-2Rβ complex may result from a preformed IL-2Rα/IL-2Rβ complex on the cell surface, which would then undergo large-scale conformational rearrangements upon assembly to higher-order IL-2 signaling complexes. Isothermal titration calorimetry (ITC) experiments (28) have, in fact, measured significant affinity for this heterodimer (Kd ≈ 280 nM), but the structure of the quaternary IL-2 complex is difficult to reconcile with this finding.

The preponderance of charge–charge interactions at the IL-2/IL-2Rα interface may serve to dominate the kinetics of IL-2 recruitment by its receptor. Long-range electrostatic interactions are known to accelerate the association rate kon, whereas short range interactions govern the dissociation rate koff (33). The rapid rate of IL-2 association with IL-2Rα (9) is then consistent with strong electrostatic interactions and high electrostatic complementarity, whereas the slow dissociation rate ascribed to IL-2Rβ (9) correlates well with the properties of the IL-2/IL-2Rβ interface (Table 1).

Thus, the IL-2Rα/IL-2 complex is likely to be preferentially delivered to IL-2Rβ, forming the pseudo-high-affinity complex, in agreement with isothermal titration calorimetry experiments (28). This complex then recruits γc, possibly from γc trimers, burying a large surface area between the receptor subunits (Table 1) and forming a tight three-way junction comprising IL-2Rβ, γc, and IL-2. Consistent with the lowest Kd value measured for a single step in IL-2R assembly (28), association of γc with IL-2/IL-2β constitutes the ultimate thermodynamic driving force toward the assembly of the IL-2 signaling complex. On activated T cells, IL-2/IL-2β would be part of the ternary IL-2/IL-2Rα/IL-2β complex whereas, on certain natural killer cells and monocytes, IL-2/IL-2β would not be associated with IL-2Rα due to the lack of this receptor subunit on these cells (9). Intriguingly, the buried surface areas of IL-2Rβ and of γc that are involved in receptor/receptor and receptor/IL-2 contacts nearly span the entire length of IL-2Rβ and of γc. Because the buried surface areas of IL-2Rβ and of γc that are involved in receptor/receptor and receptor/IL-2 contacts overlap somewhat (Figs. 3 and 4), we propose that IL-2 binding to IL-2Rβ may induce conformational changes in the contiguous γc contact area of IL-2Rβ, which primes the IL-2/IL-2Rα/IL-2Rβ and the IL-2/IL-2Rβ complex, respectively, for γc recruitment. The IL-2-dependent IL-2Rβ/γc association is further enhanced by three γc residues (Tyr-182γ, Pro-207γ, and L208γ), which interface with both IL-2 and IL-2Rβ, two IL-2Rβ residues (Glu-136β and His-138β) which are correspondingly buried in IL-2/γc interface, and by three IL-2 residues (LeuIL12, Glu-15IL2, and Leu-19IL2), which interface with both receptor subunits (Fig. 5). This cooperative mechanism could then account for the nondetectable affinities (<50 μM) of IL-2Rβ to γc in the absence of IL-2, and IL-2 to γc (28), and would also provide an essential safety mechanism against premature signaling via the IL-2Rβ/γc dimer in the absence of ligand. Collectively, the three-way junction between IL-2, IL-2Rβ, and γc provides a compelling structural basis for cooperativity in IL-2/IL-2R complex assembly initiated by, and therefore dependent on, IL-2 or IL-2/IL-2Rα.

Fig. 5.

Fig. 5.

Three-way junction between IL-2, IL-2Rβ, and γc. IL-2 (yellow ribbon representation), IL-2Rβ, and γc (the surfaces are colored as in Figs. 3 and 4) form a three-way junction at the heart of the high-affinity IL-2 signaling complex. The network of residues that mediate these contacts (colored red) provides a compelling structural basis for cooperativity in the IL-2/IL-2R complex assembly.

The structural features of the IL-2/γc interface provide a plausible mechanism of how the common γ chain γc is capable of binding to six different cytokines (34). First, the IL-2/γc interface is characterized by the smallest buried surface area, and only a portion of the D1/D2 junction of γc actually contacts IL-2 (Fig. 4). Second, the relatively large elbow angle between D1 and D2 of γc with respect to other cytokine class I receptors increases the prospective ligand-binding surface in the interdomain region. Most strikingly, little specificity is conferred upon the IL-2/γc interface due to the formation of only 2 hydrogen bonds, as opposed to 7 and 10 in the other two IL-2/IL-2R interfaces. Our structural results also agree well with extensive mutagenesis work on defining the epitopes on γc for various cytokines (25, 26, 35). Tyr-103γ, Cys-160γ, and Cys-209γ, identified as hot-spot residues in all of the cytokine/γc interfaces, form a hydrophobic patch in the center of the IL-2/γc interface. Because certain residues located around this hot spot were exclusively implicated in IL-4 binding (Ile-100γ and Leu-102γ) or IL-21 binding (Asn-44γ, Leu-161γ, Glu-162γ), a general model for degenerate cytokine recognition by γc was proposed in which the various binding sites were largely overlapping, but not identical (26, 35). The lack of effect on cytokine binding of mutations involving Lys-97γ, Phe-156γ, Leu-157γ, Asn-159γ, or His-184γ also revealed that this portion of the interdomain region, which in the x-ray structure does not contact IL-2 but IL-2Rα of an adjacent complex in the crystal (see Results), does not seem to be involved in binding to other cytokines. Given the relatively small nature and the low specificity of the IL-2/γc interface that must largely overlap with other cytokine/γc interfaces, the γc specificity in the IL-2R complex must then arise from specific interactions of γc with IL-2Rβ. Whether the combination of a relatively specific receptor/receptor interaction, but relatively nonspecific receptor/cytokine interaction, also holds for other signaling complexes involving γc remains an essential, but unanswered, question. In summary, the composite γc-binding site on the binary (IL-2/IL-2Rβ) or tertiary (IL-2/IL-2Rα/IL-2Rβ) complex serves both cooperativity and specificity purposes in the assembly of the high-affinity IL-2 signaling complex.

Upon activation, the high-affinity receptor undergoes rapid endocytosis and subsequent dissociation in the endosome (36). The prevalence of strong electrostatic interactions within the IL-2/IL-2Rα interface (Fig. 2) also provides an excellent structural solution for promoting the dissociation of IL-2Rα from the quaternary IL-2 signaling complex at the lower pH within endosomes. IL-2Rα can recycle back to the plasma membrane whereas IL-2, IL-2Rβ, and γc undergo degradation. It is noteworthy that the x-ray structures presented here as well as elsewhere (17) were both determined at acidic pH (5.1 and 6.1, respectively). The determined assemblies may, therefore, be more representative of the structure as it occurs in the endosome, where it is poised for IL-2Rα disengagement. Engineered IL-2 mutants with significantly increased affinity for IL-2Rα increase the potency and persistence of IL-2, possibly by preventing dissociation of IL-2 from IL-2Rα in the endosome and causing an IL-2/IL-2Rα complex to be recycled to the cell surface (37). This mechanism could account for the differential signaling responses for IL-2 versus IL-15, despite their shared receptor (IL-2Rβ/γc), because each ligand binds a unique α chain with very different affinities [Kd ≈ 10 nM and Kd ≈ 10 pM (membrane-bound), respectively] (38), and each receptor complex exhibits a distinct pattern of persistence on the cell surface (39), further supporting the role of IL-2Rα as a regulatory ligand carrier.

In conclusion, the crystal structure of the human IL-2 signaling complex has given us the opportunity to address key residual issues on IL-2R assembly, the specific role of the IL-2Rα chain in modulating the signaling response, and the ability of γc to act as a component of a variety of additional receptors, as well as to suggest lateral assembly of receptor complexes on the cell surface that could enhance signal propagation.

Materials and Methods

For details on expression, purification, complex formation, crystallization, and structure determination, see Supporting Materials and Methods. In brief, the ectodomains of the human IL-2 receptor chains IL-2Rα, IL-2Rβ, and γc were individually cloned and expressed in insect cells. Recombinant human IL-2 was kindly provided by Amgen as a gift to K.A.S. The high-affinity IL-2 receptor ligand complex was reconstituted from the individual protein chains by incubation at 37°C and purified by gel filtration. The IL-2/IL-2R complex crystallized in triclinic space group P1. The crystal structure was determined to 3.0 Å resolution by molecular replacement by using the coordinates from an independently assembled IL-2/IL-2R complex (17). Data processing and final refinement statistics are shown in Table 2.

Supplementary Material

Supporting Information

Acknowledgments

We thank the Advanced Light Source staff at the beamline 8.2.1 for their assistance, K. C. Garcia for generously providing coordinates before publication, and M. Carson for providing software. This research was supported by a Damon Runyon Cancer Research Foundation postdoctoral fellowship (to D.J.S.) and a Skaggs predoctoral fellowship (to E.W.D.). This is publication 17665-MB from The Scripps Research Institute.

Abbreviations

IL-2R

IL-2 receptor

Dn

domain n.

Footnotes

Conflict of interest statement: No conflicts declared.

Data deposition: The atomic coordinates and structure factors have been deposited in the Protein Data Bank, www.pdb.org (PDB ID code 2ERJ).

References

  • 1.Morgan D. A., Ruscetti F. W., Gallo R. Science. 1976;193:1007–1008. doi: 10.1126/science.181845. [DOI] [PubMed] [Google Scholar]
  • 2.Robb R. J., Smith K. A. Mol. Immunol. 1981;18:1087–1094. doi: 10.1016/0161-5890(81)90024-9. [DOI] [PubMed] [Google Scholar]
  • 3.Smith K. A., Favata M. F., Oroszlan S. J. Immunol. 1983;131:1808–1815. [PubMed] [Google Scholar]
  • 4.Taniguchi T., Matsui H., Fujita T., Takaoka C., Kashima N., Yoshimoto R., Hamuro J. Nature. 1983;302:305–310. doi: 10.1038/302305a0. [DOI] [PubMed] [Google Scholar]
  • 5.Baker P. E., Gillis S., Smith K. A. J. Exp. Med. 1979;149:273–278. doi: 10.1084/jem.149.1.273. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Smith K. A. Science. 1988;240:1169–1176. doi: 10.1126/science.3131876. [DOI] [PubMed] [Google Scholar]
  • 7.Klebb G., Autenrieth I. B., Haber H., Gillert E., Sadlack B., Smith K. A., Horak I. Clin. Immunol. Immunopathol. 1996;81:282–286. doi: 10.1006/clin.1996.0190. [DOI] [PubMed] [Google Scholar]
  • 8.Robb R. J., Munck A., Smith K. A. J. Exp. Med. 1981;154:1455–1474. doi: 10.1084/jem.154.5.1455. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Wang H. M., Smith K. A. J. Exp. Med. 1987;166:1055–1069. doi: 10.1084/jem.166.4.1055. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Leonard W. J., Depper J. M., Uchiyama T., Smith K. A., Waldmann T. A., Greene W. C. Nature. 1982;300:267–269. doi: 10.1038/300267a0. [DOI] [PubMed] [Google Scholar]
  • 11.Sharon M., Klausner R. D., Cullen B. R., Chizzonite R., Leonard W. J. Science. 1986;234:859–863. doi: 10.1126/science.3095922. [DOI] [PubMed] [Google Scholar]
  • 12.Teshigawara K., Wang H. M., Kato K., Smith K. A. J. Exp. Med. 1987;165:223–238. doi: 10.1084/jem.165.1.223. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Takeshita T., Asao H., Ohtani K., Ishii N., Kumaki S., Tanaka N., Munakata H., Nakamura M., Sugamura K. Science. 1992;257:379–382. doi: 10.1126/science.1631559. [DOI] [PubMed] [Google Scholar]
  • 14.He Y. W., Malek T. R. Crit. Rev. Immunol. 1998;18:503–524. doi: 10.1615/critrevimmunol.v18.i6.20. [DOI] [PubMed] [Google Scholar]
  • 15.Smith K. A. Ann. N.Y. Acad. Sci. 1995;766:263–271. doi: 10.1111/j.1749-6632.1995.tb26674.x. [DOI] [PubMed] [Google Scholar]
  • 16.Smith K. A. Med. Immunol. 2004;3:3–22. doi: 10.1186/1476-9433-3-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Wang X., Rickert M., Garcia K. C. Science. 2005;310:1159–1163. doi: 10.1126/science.1117893. [DOI] [PubMed] [Google Scholar]
  • 18.de Vos A. M., Ultsch M., Kossiakoff A. A. Science. 1992;255:306–312. doi: 10.1126/science.1549776. [DOI] [PubMed] [Google Scholar]
  • 19.Rickert M., Wang X., Boulanger M. J., Goriatcheva N., Garcia K. C. Science. 2005;308:1477–1480. doi: 10.1126/science.1109745. [DOI] [PubMed] [Google Scholar]
  • 20.Sugamura K., Asao H., Kondo M., Tanaka N., Ishii N., Ohbo K., Nakamura M., Takeshita T. Annu. Rev. Immunol. 1996;14:179–205. doi: 10.1146/annurev.immunol.14.1.179. [DOI] [PubMed] [Google Scholar]
  • 21.Bork P., Holm L., Sander C. J. Mol. Biol. 1994;242:309–320. doi: 10.1006/jmbi.1994.1582. [DOI] [PubMed] [Google Scholar]
  • 22.Livnah O., Stura E. A., Johnson D. L., Middleton S. A., Mulcahy L. S., Wrighton N. C., Dower W. J., Jolliffe L. K., Wilson I. A. Science. 1996;273:464–471. doi: 10.1126/science.273.5274.464. [DOI] [PubMed] [Google Scholar]
  • 23.Walter M. R. Adv. Protein Chem. 2004;68:171–223. doi: 10.1016/S0065-3233(04)68006-5. [DOI] [PubMed] [Google Scholar]
  • 24.Syed R. S., Reid S. W., Li C., Cheetham J. C., Aoki K. H., Liu B., Zhan H., Osslund T. D., Chirino A. J., Zhang J., et al. Nature. 1998;395:511–516. doi: 10.1038/26773. [DOI] [PubMed] [Google Scholar]
  • 25.Olosz F., Malek T. R. J. Biol. Chem. 2000;275:30100–30105. doi: 10.1074/jbc.M004976200. [DOI] [PubMed] [Google Scholar]
  • 26.Zhang J. L., Buehner M., Sebald W. Eur. J. Biochem. 2002;269:1490–1499. doi: 10.1046/j.1432-1033.2002.02796.x. [DOI] [PubMed] [Google Scholar]
  • 27.Buchli P., Ciardelli T. Arch. Biochem. Biophys. 1993;307:411–415. doi: 10.1006/abbi.1993.1608. [DOI] [PubMed] [Google Scholar]
  • 28.Rickert M., Boulanger M. J., Goriatcheva N., Garcia K. C. J. Mol. Biol. 2004;339:1115–1128. doi: 10.1016/j.jmb.2004.04.038. [DOI] [PubMed] [Google Scholar]
  • 29.Bazan J. F. Biochem. Biophys. Res. Commun. 1989;164:788–795. doi: 10.1016/0006-291x(89)91528-3. [DOI] [PubMed] [Google Scholar]
  • 30.Holm L., Sander C. J. Mol. Biol. 1993;233:123–138. doi: 10.1006/jmbi.1993.1489. [DOI] [PubMed] [Google Scholar]
  • 31.Wu Z., Goldstein B., Laue T. M., Liparoto S. F., Nemeth M. J., Ciardelli T. L. Protein Sci. 1999;8:482–489. doi: 10.1110/ps.8.3.482. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Forsten K. E., Lauffenburger D. A. Mol. Immunol. 1994;31:739–751. doi: 10.1016/0161-5890(94)90148-1. [DOI] [PubMed] [Google Scholar]
  • 33.Selzer T., Albeck S., Schreiber G. Nat. Struct. Biol. 2000;7:537–541. doi: 10.1038/76744. [DOI] [PubMed] [Google Scholar]
  • 34.Sugamura K., Asao H., Kondo M., Tanaka N., Ishii N., Nakamura M., Takeshita T. Adv. Immunol. 1995;59:225–277. doi: 10.1016/s0065-2776(08)60632-x. [DOI] [PubMed] [Google Scholar]
  • 35.Zhang J. L., Foster D., Sebald W. Biochem. Biophys. Res. Commun. 2003;300:291–296. doi: 10.1016/s0006-291x(02)02836-x. [DOI] [PubMed] [Google Scholar]
  • 36.Gesbert F., Sauvonnet N., Dautry-Varsat A. Curr. Top. Microbiol. Immunol. 2004;286:119–148. [PubMed] [Google Scholar]
  • 37.Rao B. M., Driver I., Lauffenburger D. A., Wittrup K. D. Biochemistry. 2005;44:10696–10701. doi: 10.1021/bi050436x. [DOI] [PubMed] [Google Scholar]
  • 38.Fehniger T. A., Caligiuri M. A. Blood. 2001;97:14–32. doi: 10.1182/blood.v97.1.14. [DOI] [PubMed] [Google Scholar]
  • 39.Dubois S., Mariner J., Waldmann T. A., Tagaya Y. Immunity. 2002;17:537–547. doi: 10.1016/s1074-7613(02)00429-6. [DOI] [PubMed] [Google Scholar]
  • 40.Voss S. D., Leary T. P., Sondel P. M., Robb R. J. Proc. Natl. Acad. Sci. USA. 1993;90:2428–2432. doi: 10.1073/pnas.90.6.2428. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supporting Information

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences

RESOURCES