Conformational Changes Mediate Interleukin-10 Receptor 2 (IL-10R2) Binding to IL-10 and Assembly of the Signaling Complex*

Interleukin-10 receptor 2 (IL-10R2) is a critical component of the IL-10·IL-10R1·IL-10R2 complex which regulates IL-10-mediated immunomodulatory responses. The ternary IL-10 signaling complex is assembled in a sequential order with the IL-10·IL-10R1 interaction occurring first followed by engagement of the IL-10R2 chain. In this study we map the IL-10R2 binding site on IL-10 using surface plasmon resonance and cell-based assays. Critical IL-10R2 binding residues are located in helix A adjacent to the previously identified IL-10R1 recognition surface. Interestingly, IL-10R2 binding residues located in the N-terminal end of helix A exhibit large structural differences between unbound cIL-10 and cIL-10·IL-10R1 crystal structures. This suggests IL-10R1-induced conformational changes regulate IL-10R2 binding and assembly of the ternary IL-10·IL-10R1·IL-10R2 complex. The basic mechanistic features of the assembly process are likely shared by six additional class-2 cytokines (viral IL-10s, IL-22, IL-26, IL-28A, IL28B, and IL-29) to promote IL-10R2 binding to six additional receptor complexes. These studies highlight the importance of structure in regulating low affinity protein-protein interactions and IL-10 signal transduction.

3). However, IL-10 also exhibits immunostimulatory properties that include the proliferation and differentiation of thymocytes, mast cells, and B cells (4 -6). Preliminary clinical data suggest antagonists that block IL-10 immunostimulatory properties may have therapeutic potential in the treatment of lupus and other inflammatory diseases (7). cIL-10 biological activities are mediated by receptor engagement and assembly of the cIL-10⅐IL-10R1⅐IL-10R2 complex, which activates the JAK/STAT signaling pathway (8,9). Based on a variety of cell surface and solution binding studies, the ternary signaling complex is assembled in a sequential manner, with IL-10R1 binding to cIL-10 first with high affinity (ϳ1 ϫ 10 Ϫ10 M) followed by low affinity interactions between IL-10R2 and the cIL-10⅐IL-10R1 binary complex (Fig. 1). Surface plasmon resonance (SPR) studies show IL-10R2 exhibits at least a 10-fold increase in affinity for the cIL-10⅐IL-10R1 binary complex (K d ϭ 234 M) compared with IL-10 alone (K d ϳ 2 mM (10,11)). The cooperative binding mechanism by which IL-10R2 recognition is modulated by IL-10R1 has not been determined. Furthermore, the data suggest cIL-10 antagonists that bind tightly to IL-10R1 but do not engage the low affinity IL-10R2 chain could be designed if the mechanisms governing IL-10R1 and IL-10R2 binding are identified.
Several herpes viruses express cIL-10 mimics that aid viruses in establishing persistent infections in their hosts (12). In particular, cytomegalovirus and Epstein-Barr virus encode biologically functional IL-10 homologues cmvIL-10 and ebvIL-10, respectively (13,14). With sequence identities of 83% (ebvIL-10) and 27% (cmvIL-10), both viral IL-10s bind and signal through the IL-10 receptor complex inducing overlapping biological activities with cIL-10 (15). Interestingly, sequence identity is not correlated with IL-10R1 or IL-10R2 affinity, suggesting the mechanisms governing receptor recognition and assembly are more complex than simple amino acid changes on a common ligand scaffold.
In addition to the importance of IL-10R2 in modulating cIL-10 signaling, the IL-10R2 chain also participates in a broad array of protein-protein interactions with at least five additional class-2 cytokines (IL-22, IL-26, IL-28A, IL-28B, and IL-29) that share ϳ10 -20% sequence identity with cIL-10 and exhibit a diverse array of biological activities (16 -18). Recently, alanine scanning mutagenesis identified five hotspot residues on helices A and D of IL-22 that are important for IL-10R2 binding (19). However, because helices A and D in cIL-10 and * This work is supported by National Institutes of Health Grant AI47300. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental material including Tables I-III  IL-22 exhibit extensive structural and sequence diversity, it is unclear if each molecule shares a common structural or energetic IL-10R2 binding site (19). Analysis of IL-10R2 binding to a series of overlapping 15-mer peptides, derived from cIL-10 and IL-22, provides initial evidence the sites are quite different (20). In this report, we identify the IL-10R2 binding site on cIL-10 and show that cIL-10⅐IL-10R2 interactions are dependent on IL-10R1-mediated conformational changes in cIL-10. Residues important for IL-10R2 binding are located at the N-and C-terminal ends of helix A and form two distinct interaction surfaces (site 2a and 2b) located adjacent to the previously identified site 1a and site 1b IL-10R1 binding sites (21). The overall location of the IL-10R2 binding sites on cIL-10 and IL-22 are conserved. However, the most important residues for IL-10R2 binding are located in site 2b for cIL-10 and site 2a for IL-22. The importance of structure in IL-10R2 recognition is emphasized by the lack of sequence conservation across the IL-10R2 binding cytokines. However, each molecule conserves key structural features, including an N-terminal disulfide bond and a hydrophobic sequence after the AB loop, which allow the proteins to access significant conformational changes required for their biological function. These studies further define the molecular mechanisms governing IL-10R2 receptor sharing and provide insights into the design of cIL-10 antagonists that might be useful in a variety of inflammatory diseases.

EXPERIMENTAL PROCEDURES
Mutagenesis-Mutations to generate sIL-10R1 T213C , and all variants of cIL-10 were carried out using the QuikChange sitedirected mutagenesis kit (Stratagene) and confirmed by DNA sequencing.
Expression and Purification of cIL-10 Variants and IL-10 Receptors-Wild-type and variant cIL-10 proteins were overexpressed in Escherichia coli BL21 (DE3) strain, refolded, and purified as previously described (22). The extracellular domains of IL-10R1 (sIL-10R1), sIL-10R1 T213C , and IL-10R2 (sIL-10R2) were expressed in Drosophila S2 cells and purified by using affinity chromatography as previously described (10,21,23). Purified sIL-10R1 T213C was reduced with 2 mM ␤-mercaptoethanol in 50 mM phosphate buffer (pH 7.0) containing 200 mM NaCl and 5 mM EDTA. ␤-Mercaptoethanol was removed by a desalting column (GE Healthcare) and biotinylated (sIL-10R1 T213C-Bt ) with 1 M PEO-maleimide activated biotin (Pierce) followed by exhaustive dialysis against 10 mM Hepes (pH 7.0). SPR Experiments-All SPR experiments were performed on a BIAcore 2000 system (BIAcore) at 20°C using HBS running buffer (10 mM Hepes (pH 7.4), 0.15 M NaCl, and 0.005% Surfactant P20). Immobilization of all proteins on the chip surfaces was performed at a flow rate of 5 l/ml. Surfaces were regenerated by injecting 2 M MgCl 2 (pH 7.4) for 1 min followed by HBS running buffer containing 20 mM EDTA for 1 min at the flow rate of 50 l/min. Streptavidin (Pierce) in 10 mM sodium acetate (pH 5.0) was immobilized at the level of 1900 -3600RU by aminecoupling chemistry on CM5 chips (BIAcore) that had been activated with a 1:1 mixture of 0.1 M N-hydroxysuccinimide and 0.1 M 3-(N,N-dimethylamino)propyl-N-ethylcarbondiim-ide. Surfaces were blocked by the injection of 1 M ethanolamine for 7 min. sIL-10R1 T213C-Bt in 10 mM Hepes (pH 7.0) was captured on streptavidin surfaces at final densities of 80 -240 RU for CM5 chips. Control surfaces of free streptavidin or streptavidin containing the biotinylated extracellular domain of IL-22R1 were used and gave identical results.
Kinetic interaction experiments were performed by injecting serially diluted monomeric cIL-10s (cIL-10M1s) over sIL-10R1 T213C-Bt surfaces in random order at 50 l/min. The data were prepared by the method of double referencing as described by Rich and Myszka (24). Using BIAevaluation software version 3.2 (BIAcore), the response curves were globally fit to a 1:1 Langmuir binding model. For cIL-10⅐sIL-10R1 interactions, cIL-10 was amine coupled to CM5 chips. Amine coupled IFN-4 was used as the control surface. sIL-10R1 was serially diluted and injected over the amine-coupled cIL-10 surfaces in random order at 50 l/min. IL-10R2 binding was evaluated using cIL-10 or cIL-10 alanine mutants amine coupled to CM-5 chips at high levels (1020 -1230 RU). The binary cIL-10⅐sIL-10R1 complex was formed by injecting 500 nM of sIL-10R1 to saturate the cIL-10 surface. This was followed by a second injection of 700 M sIL-10R2 and 500 nM sIL-10R1. The contribution of sIL-10R2 to the binding response was calculated by normalizing sensorgrams with the responses of the first injection and then by subtracting a base-line response obtained by the injection of 500 nM sIL-10R1 only. Equilibrium responses were obtained by averaging the responses obtained in the final 10 s of the latter injection.
Structure Comparisons of Unbound cIL-10 and cIL-10⅐sIL-10R1 Complexes-To ensure the structures used to interpret the mutagenesis data were of the highest quality possible, we completed the refinement of unbound cIL-10, 1INR, to give a final structure with PDB code 2H24. In addition, crystals of the cIL-10⅐sIL-10R1 complex that diffract to 2.5Å resolution, compared with 2.9 Å for the previous 1J7V structure, were used to refine the cIL-10⅐sIL-10R1 model (1Y6K) for the analysis. No significant changes were observed in the new structures, and the interpretation of the data is the same regardless of the structure used. Details of the structure refinements are described in the supplemental material. Structures used in the analyses were cIL-10 (2H24, 2ILK), ebvIL-10 (1VLK), cIL-10⅐sIL-10R1 (1Y6K), ebvIL-10⅐sIL-10R1 (1Y6M), cmvIL-10⅐sIL-10R1 (1LQS), and IL-22 (1M4R, 1YKB). Figs. 4 and 5 were generated using Ribbons (25).
STAT3 Activation Assays-Human peripheral blood monocytes were isolated by elutriation in a Beckman JE-6B centrifugal elutriator as described previously (26). Monocytes were incubated with wild-type cIL-10 or cIL-10 mutants for 30 min at 37°C. After cytokine treatment, whole cell lysates were prepared as described previously (27). Immunoblot analysis was performed using rabbit anti-phospho-STAT3 (Tyr 705 , Cell Signaling Technology) or anti-STAT3 (Santa Cruz Biotechnologies) antibodies.
Statistical Methods-The Kolmogorov-Smirnov method was used to confirm the receptor binding, and biological activity datasets were normally distributed. Analysis of variance was performed to confirm each dataset contained significant overall differences (p Ͻ 0.05). A 2-sided Dunnett's test, which controls the type I experiment-wise error, was utilized to determine significant differences (p Ͻ 0.05) between cIL-10M1 or cIL-10 and their respective mutants.

Experimental Strategy to Identify the IL-10R2 Binding Site-
The dissociation constant of the soluble IL-10R2 chain for the soluble cIL-10⅐sIL-10R1 binary complex is approximately ϳ230 M (11). In this study we sought to define specific cIL-10 residues that mediate IL-10R2 binding and signal transduction. Regions of cIL-10 targeted for mutagenesis were identified based on previous studies that defined the IL-10R2 binding site on IL-22, a structural homolog of cIL-10 (19). To map the IL-10R2 binding site, cIL-10 mutants were initially screened for IL-10R1 affinity using SPR, and their ability to induce the proliferation of human TF-1 cells transfected with the IL-10R1 chain (Fig. 1). Mutation of cIL-10 residues that bind IL-10R2 are expected to display essentially wild-type affinity for sIL-10R1 but exhibit a reduced capacity to induce TF-1/cIL-10R1 cell proliferation.
IL-10⅐IL-10R1 SPR Assay-A new SPR assay was developed to accurately evaluate the binding kinetics of cIL-10 alanine mutants for IL-10R1 (Fig. 1B). Threonine-213 in sIL-10R1 was mutated to a cysteine (sIL-10R1 T213C ) and biotinylated (sIL-10R1 T213C-Bt ) with maleimide biotin for oriented and homogeneous capture on streptavidin CM5 chips. The T213C mutation was made in the stalk region of sIL-10R1 just before where IL-10R1 enters the membrane to simulate its orientation on the cell surface (Fig. 1B). Attempts to accurately fit sensorgram data collected by injecting dimeric cIL-10 over the sIL-10R1 T213C-Bt surface were not successful because cIL-10 dimers formed mixtures of 1:1 and 1:2 cIL-10 dimer-sIL-10R1 T213C-Bt complexes (data not shown). This mixture occurs because streptavidin-labeled sIL-10R1 T213C-Bt molecules do not form 2-fold-related receptor pairs on the sensorgram surfaces as required for bivalent IL-10R1 binding (11,21).
In contrast to SPR studies with cIL-10, injecting a monomeric form of cIL-10 (cIL-10M1) over the sIL-10R1 T213C-Bt surface resulted in sensorgrams that could be globally fit to a simple 1:1 Langmuir interaction model (Fig. 1C). As previously described, cIL-10M1 is biologically active, and its structure is identical to one subunit of the cIL-10 dimer (10,22,29). Thus, cIL-10M1 provides an important tool for determining accurate binding parameters for the cIL-10⅐IL-10R1 site 1 interface and for mapping the IL-10R2 binding site. Based on the global fit of the data, the association (k 1 ) and dissociation rates (k Ϫ1 ) are 5.7 ϫ 10 5 M Ϫ1 s Ϫ1 and 2.6 ϫ 10 Ϫ4 s Ϫ1 , respectively. Using these kinetic constants, the equilibrium dissociation constant (K d ϭ k Ϫ1 /k 1 ) for the interaction between cIL-10M1 and sIL-10R1 T213C-Bt is 0.47 nM. The kinetic and equilibrium parameters are in good agreement with those from previous SPR analysis using cIL-10M1-immobilized surfaces and isothermal titration calorimetry data (10,11).
Identification of the IL-10R2 Binding Site on cIL-10M1-A total of 13 cIL-10M1 mutants, located on ␣-helices A and D, were evaluated for IL-10R1 binding on sIL-10R1 T213C-Bt surfaces, biological activity in TF-1/cIL-10R1 cell proliferation assays, and STAT3 activation in human monocytes ( Fig. 2 and supplemental Table I). Twelve of the 13 cIL-10M1 mutants exhibited sIL-10R1 binding affinities similar to cIL-10M1 ( Fig.   2 and supplemental Table I). In contrast, cIL-10M1 R24A displayed a 10-fold lower affinity for sIL-10R1 T213C-Bt (K d ϭ 4.7 nM). Thus, with the exception of cIL-10M1 R24A , mutants that exhibit reduced biological activity in the proliferation and STAT3 assays are likely involved in IL-10R2 binding.
To quantify the potency of each mutant in the proliferation assay, dose-response curves (Fig. 1D) were generated for each cIL-10M1 mutant, and the concentration required to induce 50% maximal proliferation (EC 50 ) was plotted in Fig.  2A along with the IL-10R1 affinity of each mutant (supplemental Table 1). These experiments identified two cIL-10M1 mutants (M22A and R24A) that displayed substantially higher EC 50 values than cIL-10M1. cIL-10M1 M22A exhibited the largest increase in EC 50 (EC 50 ϭ 23 ng/ml) and could only reach 80% of the maximal cIL-10M1 activity at concentrations as high as 100 g/ml (Fig. 1D). This was the only mutant that could not achieve maximal proliferation by increasing its concentration in the assay. cIL-10M1 R24A exhibited a similar increase in EC 50 value as cIL-10M1 M22A (EC 50 ϭ 14 ng/ml). After cIL-10M1 M22A and cIL-10M1 R24A , cIL-10M1 N21A and cIL-10M1 R32A exhibited the next highest EC 50 values (EC 50 ϳ 1 ng/ml), although they were not statistically different from cIL-10M1 (p Ͻ 0.05).
Because the TF-1/cIL-10R1 cell line used in the proliferation assay is an artificial cell line transfected with the IL-10R1 chain (15), we also evaluated the ability of each cIL-10M1 mutant to activate STAT3 in primary human monocytes, which express endogenous IL-10 receptor levels (Fig. 2B). Consistent with the results obtained using the TF-1/cIL-10R1 proliferation assay, cIL-10M1 M22A and cIL-10M1 R24A were completely inactive in the STAT3 activation assay. In addition, cIL-10M1 N21A and cIL-10M1 R32A also exhibited essentially no detectable STAT3 phosphorylation activity on monocytes. Furthermore, the S31A and S93A mutants showed reductions in STAT3 activation. In summary, STAT3 activation studies using human monocytes indicated cIL-10M1 residues Asn-21, Met-22, and Arg-32 are most critical for the cIL-10M1⅐IL-10R2 interaction, and Ser-31 and Ser-93 may play a minor role in IL-10R2 engagement. The Arg-24 mutant is clearly important for biological activity, but its role in IL-10R2 binding, rather than IL-10R1 binding, remains to be determined.
Characterization of cIL-10 Dimer Mutants-Because the initial alanine scan was performed with monomeric cIL-10 rather than dimeric cIL-10, putative IL-10R2 binding residues (Asn-21, Met-22, Arg-24, and Arg-32) as well as His-90 and Ser-93 were mutated in the cIL-10 dimer and characterized for IL-10R1 and IL-10R2 binding and the ability to activate STAT3 in primary human monocytes. To characterize the cIL-10⅐sIL-10R1 interactions, cIL-10 and cIL-10 mutants were amine-coupled to CM-5 chips, and multiple concentrations of sIL-10R1 were injected over the cIL-10 surfaces (Fig.  3A). The K d values obtained between amine-coupled cIL-10 or cIL-10 mutants and sIL-10R1 were slightly higher than when specific coupling of sIL-10R1 T213C-Bt to the chip surface was used (supplemental Table I). Nonetheless, the results of the cIL-10⅐sIL-10R1 binding studies were consistent with the cIL-10M1⅐sIL-10R1 T213C-Bt SPR analysis. Specifically, cIL-10M1 R24A exhibited the greatest decrease in sIL-10R1 affinity IL-10R2 Binding Assay-cIL-10 residues implicated in IL-10R2 binding were further evaluated in a direct IL-10R2 binding assay (Fig. 3B). For this assay, cIL-10 mutants (N21A, M22A, R24A, R32A, and H90A) were amine-coupled at high surface densities (ϳ1000 RU) onto CM-5 sensor chips. The ability of sIL-10R2 to bind amine-coupled cIL-10 mutant-sIL-10R1 binary complexes was evaluated because of the extremely weak affinity of sIL-10R2 for cIL-10 alone (K d ϳ2 mM (10)). Because of the large quantity of protein required to characterize IL-10R2 binding, only selected mutants were evaluated at a single sIL-10R2 concentration of 700 M. In this way, a relative sIL-10R2 binding strength could be assigned for each cIL-10 mutant. Consistent with its low biological activity, the cIL-10 M22A mutant exhibited the poorest binding to IL-10R2 (35% of cIL-10), whereas cIL-10 R24A and cIL-10 R32A exhibited intermediate IL-10R2 binding levels (65% of cIL-10). Thus, cIL-10 Arg-24 contributes to IL-10R2 as well as IL-10R1 binding. Alanine mutations of Asn-21 and His-90 marginally impaired IL-10R2 binding (80 -85% of cIL-10).
Next, we evaluated the ability of the cIL-10 dimer mutants to induce STAT3 in primary human monocytes (Fig. 3C). Once again, these studies confirm the critical importance of cIL-10 residues Met-22, Arg-24, and Arg-32 in mediating cIL-10-induced signaling. Additional concentration-dependent STAT3 activation studies showed mutation of Met-22, followed closely by Arg-32, disrupt IL-10R2 signaling the greatest on monocytes (supplemental Fig. S1). They also show cIL-10 mutants N21A, H90A, and S93A have a minimal impact on the induction of cIL-10 biological activity.
cIL-10M1 N21A/M22A/D25K/R32K Is a cIL-10 Antagonist-Three additional cIL-10M1 mutants were evaluated to determine whether it is possible to design a cIL-10 antagonist that binds with high affinity to IL-10R1 but cannot form a productive interaction with IL-10R2 to initiate cIL-10-mediated biological activities. To accomplish this, the cIL-10M1 N21A/R32E double mutant was constructed to disrupt possible Arg-32 chargecharge interactions with IL-10R2. Consistent with the importance of Arg-32, the EC 50 of cIL-10M1 N21A/R32E increased by about 9-fold (8.5 ng/ml) over the EC 50 values reported for the individual alanine mutants (0.95 and 1.0 ng/ml).
Structure of the IL-10R2 Binding Site-Residues corresponding to the IL-10R2 binding site were mapped onto the cIL-10⅐sIL-10R1 binary complex (Fig. 4). Consistent with IL-10R2 low affinity for cIL-10, the IL-10R2 binding surface is composed A, dissociation constants (K d ) determined by SPR for cIL-10 mutant-sIL-10R1 interactions are plotted as black bars. Schematic diagram depicting the kinetic SPR experiment used to obtain the values is shown to the right of the figure. B, relative IL-10R2 binding by cIL-10 mutants. IL-10R2 binding to cIL-10 or cIL-10 mutant-sIL-10R1 complexes were evaluated in a single point assay normalized to the wild-type cIL-10⅐IL-10R2 response. A schematic diagram depicting the SPR experimental setup used to obtain the data is shown to the right of the figure. The results for both assays are expressed as the mean of multiple experiments Ϯ S.D. Statistically significant differences (p Ͻ 0.05) between cIL-10 and cIL-10 mutants in the assays are denoted by asterisks (*). C, Western blot showing the ability of cIL-10 dimer mutants to induce phosphorylation of STAT3 in primary human monocytes. pY, phosphotyrosine. of no more than seven residues located on helices A and D. In contrast, the high affinity IL-10R1 binding site consists of 25 residues and buries ϳ1100 Å 2 of surface area (21). Residues critical for IL-10R2 binding separate into two distinct regions located adjacent to the IL-10R1 contact surfaces (site 1a and 1b). Based on this similarity, we designate Arg-32 and surrounding residues as the IL-10R2 site 2a and residues near Met-22 as the IL-10R2 site 2b contact surface. Our results generally agree with a theoretical model of the cIL-10⅐IL-10R1⅐IL-10R2 complex, built using crystal structures of cIL-10⅐IL-10R1 and IL-6⅐IL-6R␣⅐gp130 complexes (21,30,31). Biological and receptor binding assays identified site 2b, containing Met-22 and Arg-24, as the most energetically important region of the IL-10R2 binding surface.
IL-10R1 Induced Conformational Changes in the IL-10R2 Binding Site-Structural comparison of site 2b in unbound cIL-10 (PDB code 2H24) and the cIL-10⅐IL-10R1 complex (PDB code, 1Y6K) revealed cIL-10 residues important for IL-10R2 binding (Met-22 and Arg-24) undergo an IL-10R1-mediated conformational change (Fig. 4). The unbound cIL-10 structure (PDB code 2H24) was used for these comparisons because, unlike the cIL-10 structure PDB code 2ILK (32), the N-terminal and AB loop regions do not participate in extensive crystal contacts, suggesting 2H24 more accurately reflects the solution structure of cIL-10.
In unbound cIL-10 (2H24), the side chain of Leu-19 packs into a cleft between Met-154 and Thr-155 located on helix F (Fig. 4). As a result, the N-terminal end of helix A is bent by ϳ25°. However, upon IL-10R1 binding, IL-10R1 residue Phe-143 inserts itself into the pocket occupied by Leu-19 and Pro-20 in unbound cIL-10. This causes a 26°rigid body rotation of helix A residues 18 -22, which is accomplished almost exclusively by rotating the / torsion angles of Leu-23. This rotation straightens and lengthens helix A and results in main chain movements of up to 5.5 Å. The conformational change in helix A is accompanied by a significant reorganization and ordering of cIL-10 residues 12-18, which connects helix A via the disulfide bond between Cys-12 and Cys-108 to helix D. The term "ordering" is used to describe the observation that residues 12-17 are not observed in the final electron density maps of cIL-10 (2H24) but are visible in the cIL-10⅐sIL-10R1 complex (1Y6K). Comparison of ebvIL-10 and the ebvIL-10⅐sIL-10R1 structures (1VLK versus 1Y6M) reveals ebvIL-10 undergoes essentially the same rotation of helix A despite having only one residue compared with four in cIL-10, linking helix A to helix D.
Structural analysis sheds considerable light on the critical nature of Met-22 for efficient IL-10R2 binding. In unbound cIL-10, the side chain of Met-22 is located near the position of Leu-19 in the cIL-10⅐sIL-10R1 complex (Fig. 4). Upon IL-10R1 binding, the Met-22 side chain moves ϳ6 Å to pack against Leu-101 and Arg-104. Mutation of Met-22 to alanine is expected to significantly disrupt this packing interaction, which would subsequently alter the conformation of the entire N-terminal region. Thus, the role of Met-22 in the IL-10R2 interaction is likely 2-fold. First, it plays an important role in organizing/rigidifying the N-terminal region of cIL-10 around site 2b and possibly additional residues. Second, it can directly participate in interactions with the IL-10R2 chain.
Comparison of the IL-10R2 Binding Epitopes-A structural role for Met-22 is strengthened by comparing the putative IL-10R2 binding site of cmvIL-10 with cIL-10 (Fig. 5A). Met-22 in cIL-10 is replaced by an arginine in cmvIL-10 (Arg-22 cmv ). As described for Met-22 in cIL-10, Arg-22 cmv participates in numerous interactions between helices A and D. In particular, the guanido group of Arg-22 cmv forms four specific hydrogen bond interactions with cmvIL-10 residues Asp-18 cmv , Ser-100 cmv , and Asp-104 cmv (Fig. 5A). In addition to their structural roles, the aliphatic segments of Met-22 and Arg-22 cmv can both participate directly in van der Waals/hydrophobic interactions with IL-10R2.
Arg-24 and Arg-32 also exhibited significant reductions in IL-10R2 binding levels when mutated to alanine. Interestingly, the Arg-24 to alanine mutant was the only molecule tested that exhibited reduced affinity for IL-10R1. The 10-fold increase in sIL-10R1 off-rate observed for the cIL-10 R24A mutant can be rationalized by the loss of the interaction between the N⑀ atom The site 1a and site 1b IL-10R1 contact surfaces are shown in blue and magenta, respectively. cIL-10 residues mutated to alanine that have no effect on biological activity are shown in silver. Residues that exhibit reduced IL-10R2 binding and biological activity are colored according to their relative importance, with red being most important, followed by yellow, and then cyan. IL-10R1-induced conformational changes observed between free and IL-10R1-bound cIL-10 are shown in the inset. The backbone and side chains of unbound cIL-10 are cyan and green, respectively. IL-10R1-bound cIL-10 backbone and side chains are magenta and yellow, respectively. sIL-10R1 side chains are silver, and important residues are labeled and distinguished from cIL-10 residues by underlining. Oxygen atoms are red, and nitrogens are blue.
The IL-10R2 Binding Site on IL-10 NOVEMBER 17, 2006 • VOLUME 281 • NUMBER 46 of Arg-24 and the carbonyl oxygen of IL-10R1 residue Arg-191 (Fig. 4). In addition to IL-10R1 binding, SPR studies and bioassays showed Arg-24 is important for sIL-10R2 binding and biological activity. Thus, Arg-24 appears to be the only residue in the cIL-10 ternary complex that is involved in interactions with both IL-10R1 and IL-10R2. As observed for Met-22, IL-10R1 binding changes the side-chain conformation of Arg-24. These changes presumably influence both IL-10R1 and IL-10R2 binding through the hydrogen bond with the L5 receptor loop.
In contrast to Met-22 and Arg-24, Arg-32 is located at the other end of helix A in site 2a. As shown by the structural superpositions in Fig. 5, the side-chain position of cIL-10 Arg-32 is conserved in the IL-10R2 binding epitopes of cmvIL-10 and IL-22, which is consistent with its critical importance in cIL-10⅐IL-10R2 binding (Fig. 5) and STAT3 activation (supplemental Fig. S1). Despite a general structural correspondence, the main-chain and side-chain positions of Arg-32 differ consid-erably among the three cytokines. In fact, the C␣ atom positions of Arg-32 from cIL-10 and Arg-55 in IL-22 are separated by one turn of helix. Thus, structure-based primary sequence alignments of IL-10s and IL-22 do not identify the IL-10R2 binding residues (Fig.  6). Thus, the location of Arg-32 is only "loosely" conserved, which may reflect the inherent properties of the low affinity binding surface. In contrast, the main-chain and side-chains positions of 8 residues in the high affinity IL-10⅐IL-10R1 interfaces of cIL-10, ebvIL-10, and cmvIL-10 are highly conserved (33).
Previous studies have shown IL-22 exhibits a higher affinity for IL-10R2 (K d ϭ 14 M) than cIL-10 (K d ϭ 234 M). For IL-22, the energetic hotspot consists of Asn-54 and Arg-55 located in site 2a. As shown in Fig. 5B, Arg-55 in IL-22 forms an extensive hydrogen bond/salt bridge network with Tyr-51 and Glu-117, which might be responsible for the IL-22 higher affinity interaction with IL-10R2. In contrast to IL-22, the three energetically important IL-10R2 binding residues in cIL-10 (Met-22, Arg-24, and Arg-32) are spread out over site 2a (Arg-32) and site 2b (Met-22 and Arg-24). Interestingly, the most important residue in the cIL-10⅐IL-10R2 binding surface, Met-22, is located in site 2b rather than in site 2a, as observed for IL-22. Residues important in the cmvIL-10⅐IL-10R2 binding interface have not yet been identified. However, it is interesting that the N-terminal region of cmvIL-10 is very well ordered and contains two additional turns of ␣-helix. This suggests structural differences in site 2b may contribute to the increased IL-10R2 binding affinity observed for cmvIL-10 (11).
Mechanism of IL-10 Ternary Complex Assembly-The combination of SPR, crystallographic, and bioactivity data demonstrate the importance of cIL-10 conformational changes in promoting the assembly of the cIL-10⅐IL-10R1⅐IL-10R2 signaling complex. Previous studies have shown cIL-10⅐IL-10R1 affinity is regulated by the AB loop, which adopts alternate structures in the unbound and IL-10R1 bound states (11,21). Data from the current study now suggest productive engagement of the IL-10R2 chain is dependent on IL-10R1-induced conformational changes in the N terminus of helix A. A more detailed structural com-  . Sequence alignment and conformational changes required for complex assembly. Structurebased sequence alignment of cellular IL-10 (cIL10) residues 12-48 to Epstein-Barr virus (ebvIL10) and cytomegalovirus (cmvIL10) IL-10s and to IL-22 (IL22). Arrows denote key structural features (Cys-12 and the Leu-Leu-Leu repeat) of IL-10 that attach this flexible peptide segment to the helix bundle. Black boxes above the sequence correspond to ␣-helix A in unbound cIL-10 (2H24) and cIL-10 bound to sIL-10R1 (1Y6K). Residues that undergo conformational changes (at least 1Å C␣-C␣ difference) between the unbound and IL-10R1 bound structures are shown by dashed lines. The location of IL-10R1 site 1a and 1b and IL-10R2 site 2a and 2b are shown. Amino acids that participate in IL-10R1 binding are boxed. Residues important for IL-10R2 binding are colored according to Fig. 4. parison reveals 24 of 34 (71%) cIL-10 residues, between Cys-12 and Leu-46, change their positions by at least 1 Å (C␣-C␣ distance) upon IL-10R1 binding. All 24 residues map to the N terminus (residues Cys-12-Met-22) and AB loop (residues Val-33-Asn-44) segments of cIL-10.
For cIL-10 to accommodate these large conformational changes it must contain structural features that stabilize the ends of the flexible regions. For cIL-10, these regions are the Cys-12-Cys-108 disulfide bond at the N terminus and the Leu-Leu-Leu motif (residues 46 -48) found at the C-terminal end of the AB loop (Fig. 6). The N-terminal and AB loop regions are separated by residues 23-32, which form the structurally invariant core of helix A. Consistent with our studies, the Cys-12-Cys-108 disulfide bond has previously been shown to be critical for the stability and biological activity of cIL-10 (34). Interestingly, these structural features are conserved in all IL-10R2 binding cytokines, suggesting manipulation of the conformations of the N-terminal region may be a common mechanism to regulate receptor binding and complex assembly for other class-2 cytokines.
Structural studies on glycosylated IL-22, which contained 6 molecules in the asymmetric unit, have identified considerable conformational variability in the IL-10R2 binding site on IL-22 supporting this hypothesis (35,36). Furthermore, an anti-IL-22 antibody has been shown to increase the high affinity IL-22⅐IL-22R1 interaction, apparently by inducing conformational changes in the IL-10R2 binding site that are subsequently transmitted to the high affinity IL-22R1 binding site (37). These studies together with our current results emphasize the importance of conformational changes in promoting receptor assembly and biological activity of cIL-10, IL-22, and most likely, other members of the class-2 cytokine family. Further mechanistic studies on these molecules should aid in the design of novel methods to antagonize or enhance their broad biological activities.
The requirement of an IL-10R1-mediated conformational change in cIL-10 for productive IL-10R2 binding provides a mechanistic explanation for why cIL-10 ternary complex formation is sequential. For example, complex assembly depends on an initial interaction between cIL-10 and IL-10R1 followed by engagement of the IL-10R2 chain. The IL-10R1-induced conformational change in cIL-10 is thought to be at least partially responsible for the ϳ10-fold increase in cIL-10⅐IL-10R2 affinity in the presence of IL-10R1 (10). However, even the cIL-10 M22A mutant retains some IL-10R2 binding activity (Fig. 5), suggesting additional interactions between IL-10R1 and IL-10R2 chains may occur and contribute to the overall stability of the cIL-10⅐IL-10R1⅐IL-10R2 ternary complex. Such receptor-receptor interactions have been observed by crystallography in the growth hormone, IL-6, and IL-2 receptor systems and are predicted for cIL-10 (30, 31, 38 -40).
Recently, the IL-10R1 and IL-10R2 chains were shown to associate with one another on the cell surface in the absence of cIL-10 (28). Our data suggests one function of the IL-10R2 chain in the preassembled IL-10R1⅐IL-10R2 complex is to monitor IL-10R1-induced conformational changes in cIL-10, which promotes assembly of the "active" ternary complex. The low affinity and fast binding kinetics of the IL-10R2 chain are pre-dicted to make it exquisitely sensitive to the formation or dissociation of the cIL-10⅐IL-10R1 binary complex, which would facilitate the rapid activation or inactivation of cIL-10 transmembrane signaling.