The Tissue Protective Functions of Interleukin-22 can be Decoupled from Pro-Inflammatory Actions Through Structure-Based Design

Summary

Interleukin-22 (IL-22) acts on epithelial cells to promote tissue protection and regeneration, but can also elicit pro-inflammatory effects, contributing to disease pathology. Here, we engineered a high affinity IL-22 super-agonist that enabled the structure determination of the IL-22–IL-22Rα–IL-10Rβ ternary complex to a resolution of 2.6 Å. Using structure-based design, we systematically destabilized the IL-22–IL-10Rβ binding interface to create partial agonist analogs that decoupled downstream STAT1 and STAT3 signaling. The extent of STAT bias elicited by a single ligand varied across tissues, ranging from full STAT3-biased agonism to STAT1/3 antagonism, correlating with IL-10Rβ expression levels. In vivo, this tissue-selective signaling drove tissue protection in the pancreas and gastrointestinal tract without inducing local or systemic inflammation, thereby uncoupling these opposing effects of IL-22 signaling. Our findings provide insight into the mechanisms underlying the cytokine pleiotropy and illustrate how differential receptor expression levels and STAT response thresholds can be synthetically exploited to endow pleiotropic cytokines with enhanced functional specificity.

Graphical Abstract

eTOC blurb

Saxton et al engineer a high affinity interleukin (IL)-22 super-agonist that enables the structure determination of the IL-22–IL-22Rα–IL-10Rβ ternary complex. IL-22 receptor agonists designed based on these structural insights elicit activation of STAT3 but not STAT1, and promote epithelial protection and regeneration without inducing local or systemic inflammation.

Introduction

Cytokines play key roles in controlling immune responses and promoting tissue homeostasis (Akdis et al., 2016; Morris et al., 2018). A notable feature of most cytokines is their ability to elicit distinct and often opposing effects depending on the biological context (Lin and Leonard, 2019; Wang et al., 2009). This functional diversity is due in part to the broad expression profiles of many cytokine receptors, as well as their ability to simultaneously activate multiple intracellular signaling pathways and gene expression programs (Delgoffe et al., 2011; O’Shea and Plenge, 2012). Although the pleiotropic nature of cytokines is essential for instilling redundancy and feedback into immune responses, it also represents a significant barrier to their use as therapeutics.

The Interleukin (IL)-10 family cytokine IL-22 elicits diverse biological effects and plays a central role in maintaining tissue homeostasis during inflammation (Dudakov et al., 2015; Ouyang and O’Garra, 2019; Sabat et al., 2014). Multiple immune cell types including CD4⁺ T cells and Type 3 Innate Lymphoid Cells (ILC3s) produce IL-22 at barrier interfaces, where it acts on epithelial cells to enhance tissue integrity and promote regeneration (Basu et al., 2012; Cella et al., 2009; Sonnenberg et al., 2011a; Sonnenberg et al., 2011b). IL-22 exerts these protective effects on a variety of tissues including the pancreas, gastrointestinal (GI) tract, liver, and thymus, and treatment with exogenous IL-22 is protective in mouse models of pancreatitis, inflammatory bowel disease (IBD), acute liver damage, and graft-versus host disease (GvHD)-mediated intestinal damage (Dudakov et al., 2012; Feng et al., 2012; Hanash et al., 2012; Lindemans et al., 2015; Sugimoto et al., 2008; Zenewicz et al., 2007; Zenewicz et al., 2008). The unique ability of IL-22 signaling to promote recovery from inflammatory tissue damage without suppressing immune function provides an appealing therapeutic approach for autoimmune diseases and distinguishes it from current treatment modalities, such as corticosteroids or cytokine antagonists (Sabat et al., 2014).

Despite these beneficial functions, IL-22 can have pathogenic potential in some contexts as well, most notably in IBD and psoriasis (Bernshtein et al., 2019; Boniface et al., 2005; Fujita, 2013; Gunasekera et al., 2020; Zheng et al., 2007). The pro-inflammatory effects of IL-22 are due in large part to the induction of neutrophil recruiting chemokines such as CXCL1 in the skin, liver and GI tract, as well as production of acute-phase response proteins such as Serum-Amyloid A (SAA)-1/2, which promotes activation of inflammatory Th17 cells (Bernshtein et al., 2019; Lee et al., 2020; Liang et al., 2010; Sano et al., 2016). Indeed, administration of exogenous IL-22 significantly increases the serum levels acute-phase response proteins in both mice and humans (Liang et al., 2010; Rothenberg et al., 2019; Tang et al., 2019).

Mechanistically, IL-22 signals through a heterodimeric receptor complex consisting of a high affinity subunit, IL-22Rα, and a low affinity subunit, IL-10Rβ (Ouyang and O’Garra, 2019; Xie et al., 2000). Both IL-22Rα and IL-10Rβ are shared receptor subunits that engage additional cytokines in the IL-10 superfamily, and thus are capable of eliciting a wide range of biological effects (Ouyang and O’Garra, 2019). IL-10Rβ is expressed ubiquitously, whereas IL-22Rα is expressed primarily on epithelial cells, dictating the target cell specificity of IL-22 (Wolk et al., 2004). Upon binding to IL-22Rα, IL-22 facilitates the dimerization of IL-22Rα and IL-10Rβ, bringing together the intracellular receptor associated kinases JAK1 and TYK2, which phosphorylate both each other as well as tyrosine residues on the intracellular domain (ICD) of IL-22Rα. These phospho-tyrosines in turn recruit STAT transcription factors, primarily STAT1 and STAT3, enabling their activation. While expression of the tissue protective and regenerative genes downstream of IL-22 are mediated by STAT3, many of the pro-inflammatory effects are due to STAT1 (Bachmann et al., 2013; Muhl, 2013; Ouyang and O’Garra, 2019; Pickert et al., 2009). This paradigm of simultaneous activation of STAT1 and STAT3, which control opposing gene expression programs, is shared by many other cytokine receptors and contributes to their pleiotropic effects (Lin and Leonard, 2019; Regis et al., 2008).

Functionally selective, or “biased,” agonists that uncouple downstream signaling responses have been extensively characterized for G-protein coupled receptors (GPCRs) (Wootten et al., 2018), and such agonists are possible for cytokine receptors as well (Kim et al., 2017; Mohan et al., 2019; Moraga et al., 2015). However, these approaches rely on extensive structural information, which is currently lacking for the IL-22 receptor complex. This is due primarily to the extremely low affinity of IL-22 for its shared receptor subunit, IL-10Rβ, which hinders complex assembly in vitro (Jones et al., 2008).

Here, we employed a directed evolution approach to develop a high-affinity IL-22 variant, enabling us to solve the crystal structure of the IL-22–IL-22Rα–IL-10Rβ signaling complex to 2.6 Å resolution. Using this structure, we designed a series of IL-22 analogs that calibrated STAT1 and STAT3 signaling by exploiting distinct STAT response thresholds downstream of IL-22. One of these engineered ligands elicited tissue-selective STAT activation in vivo, thereby uncoupling the tissue protective and pro-inflammatory functions of IL-22. Together, these results reveal how natural mechanisms underlying cytokine pleiotropy can be exploited pharmacologically in order to tune cytokine function for therapeutic benefit.

Results

Stabilizing the IL-22 Receptor Complex Through Directed Evolution

In order to stabilize the ternary IL-22 receptor complex for structural studies, we used yeast surface display to affinity-mature the IL-22–IL-10Rβ interaction (Angelini et al., 2015). We designed a site directed library in which we randomized six amino acid positions in mouse IL-22, predicted to be in close proximity to the IL-10Rβ binding interface based on information from the partial complex structure of IL-22 bound to IL-22Rα as well as homology to interferon (IFN)-λ, which engages IL-10Rβ (Figure 1A) (Jones et al., 2008; Mendoza et al., 2017). This mutant IL-22 library was displayed on the surface of yeast, pre-bound with the unlabeled extracellular domain (ECD) of IL-22Rα and selected with the fluorescently labelled ECD of IL-10Rβ (Figure 1B). This process was repeated over five rounds of in vitro evolution (Figure S1A), yielding progressive improvements in IL-10Rβ binding (Figure 1C).

Figure 1: Stabilizing the IL-22 Receptor Complex Through Directed Evolution.

(A) Structure of the human IL-22–IL-22Rα partial complex (PDB ID: 3DLQ), with corresponding residues selected for randomization in mouse IL-22 yeast-display library shown in red.

(B) Schematic strategy for yeast display-based affinity maturation of IL-22 (MACS, magnetic activated cell sorting, FACS, fluorescence-activated cell sorting).

(C) Histograms of fluorescent intensity indicating Streptavidin-Alexa Fluor-647 (SA-647)-labelled IL-10Rβ binding to yeast-displayed wild-type IL-22, yeast populations at selected rounds of directed evolution, and the final high affinity IL-22 clone (Super-22a).

(D) Binding titration of SA-647-IL-10Rβ on yeast with surface-displayed WT IL-22 or affinity matured clones, Super-22a and Super-22b. Yeast were pre-bound with 1 μM unlabeled IL-22Rα.

(E). Dose response curve for phospho-STAT3 in mouse EC4 (hepatocyte) cells stimulated with WT IL-22 or indicated variants for 20 minutes and analyzed by flow cytometry. Data are mean +/− SD for two replicates, shown as a percent of maximal WT IL-22 mean fluorescent intensity (MFI).

We selected two high-affinity clones from the final yeast population, each containing five mutations relative to wild-type mouse IL-22, named ‘Super-22a’ and ‘Super-22b’ (Figure S1B). Both clones exhibited improved binding to the IL-10Rβ ECD when displayed on the surface of yeast (Figure 1D). In addition, both clones displayed enhanced activity in cultured cells, with approximately 5 to 10-fold reduced EC₅₀ on both mouse and human epithelial-derived cell lines, as determined by phosphorylation of STAT1 and STAT3 (Figures 1E and S1C). These engineered high-affinity IL-22 variants therefore showed improved binding to IL-10Rβ while retaining the ability to functionally dimerize both receptor subunits.

Structure of the IL-22 Receptor Complex

In contrast to wild-type IL-22, both high affinity IL-22 clones (Super-22a and Super-22b) formed a stable ternary complex with IL-22Rα and IL-10Rβ, as assessed by co-migration during gel filtration (Figures S2A and S2B). A complex containing a glyco-mutant variant of Super-22a bound to glyco-mutant variants of IL-22Rα and IL-10Rβ enabled crystallization of the complex (Figure S1C). The structure was determined by molecular replacement to 2.6 Å resolution using models derived from structures of the partial IL-22/IL-22Rα complex (PDB ID: 3DGC) (Jones et al., 2008) and monomeric IL-10Rβ (PDB ID: 3LQM) (Yoon et al., 2010).

The overall architecture of the IL-22–IL-22Rα–IL-10Rβ ternary complex is similar to that of other class 2 cytokine receptor complexes and consists of three distinct binding interfaces (sites 1–3, Figures 2A–C). At site 1, multiple loops from both Class 2 Homology Region (C2HR) domains of IL-22Rα (D1 and D2) engage IL-22 helices α1 and α5, as well as loop L2, burying 961 Ų of the IL-22 surface. The lower affinity site 2 contact is formed by both D1 and D2 of IL10Rβ, which engage IL-22 primarily through loops L2, L3 and L4 in D1, along with L5 and L6 in D2. L5 of IL-10Rβ forms extensive contacts with the bottom of IL-22 helix α1, while loops L2, L3 and L4 engage helix α3 (Figures 2A and 2B). Finally, site 3 consists of the receptor-receptor “stem contacts” between the D2 domains of IL-22Rα and IL-10Rβ, forming a smaller interface with only three hydrogen bond contacts and 421 Ų of buried surface area (Figure S2D).

Figure 2: Structure of the IL-22 Receptor Complex.

(A-C) Three views of the 2.6 Å resolution structure of the IL-22 receptor complex showing high-affinity IL-22 (super-22a) in yellow, IL-22Rα in blue, and IL-10Rβ in pink (PDB ID: 6WEO)

(D) Structural superposition of mouse IL-22–IL-10Rβ complex (PDB ID 6WEO) with human IFN-λ/IL-10Rβ complex (PDB ID: 5T5W), showing change in the relative orientation of IL-10Rβ between complexes.

(E) Surface representations of IL-22 (yellow, left) and IFN-λ (green, right) bound to IL-10Rβ, showing the different relative positions of three key aromatic residues in IL-10Rβ.

The IL-22 receptor complex structure presented here reveals several notable differences with the structure of IFN-λ in complex with IFN-λR1 and IL-10Rβ (Mendoza et al., 2017). In particular, the site 2 contact formed between IL-10Rβ and IL-22 differs from that observed in the IFN-λ complex. In the IFN-λ bound structure, IL-10Rβ engages primarily the N-terminal edge of helix α3 in IFN-λ, appearing to clasp the “front” of the cytokine (Mendoza et al., 2017), whereas IL-10Rβ engages IL-22 via the bottom and central portion of helix α3. This differential binding mode can be readily observed when the two complexes are aligned via superposition of the cytokine ligand, showing an approximately 40° relative rotation of the IL10Rβ D1 domains between the complexes (Figure 2D). The altered orientation of IL-10Rβ in these two complexes also results in a change in the relative positioning of three aromatic residues in IL-10Rβ, Tyr59, Tyr82, and Trp143, which contact the ligand in both structures (Figure 2E) (Mendoza et al., 2017).

The site 2 contact forms an extensive interface burying 723 Ų of the IL-22 surface. A close-up view of this interface reveals several key residues in IL-22 that mediate the interaction with IL10Rβ (Figures 3A–3C, Fig. S3A). On helix α1 of IL-22, 22-Arg55 forms a salt-bridge with Rβ-Glu141, which simultaneously contacts the adjacent N-linked glycan modification on 22-Asn54 (Figure 3B). In addition, 22-Tyr51 on helix α1 makes multiple contacts with IL-10Rβ, including a hydrogen bond with Rβ-Glu139 as well as a perpendicular pi-stacking interaction with Rβ-Trp143. Moreover, 22-Glu117 and 22-Lys124 in helix α3 form ionic contacts with Rβ-Lys81, Rβ-Asp84, and Rβ-Glu109 in loops L3 and L4 of IL-10Rβ (Figure 3B and 3C). Near the C-terminal edge of helix α3, 22-Gln48 in Loop L1 forms an additional hydrogen bond with Rβ-Asp150 in the D2 domain.

Figure 3: Design of STAT3-Biased IL-22 Receptor Agonists.

(A-C) Close-up views of the IL-22–IL-10Rβ binding interface. Hydrogen-bonds and salt-bridges are shown as black dashed-lines. Mutated residues in super-22a from affinity maturation are italicized.

(D) Dose response curves for phospho-STAT3 (left) and phospho-STAT1 (right) in HT-29 cells stimulated with WT IL-22 or indicated variants for 20 minutes and analyzed by flow cytometry. Data are mean +/− SD for three replicates, shown as a percent of maximal WT IL-22 mean fluorescent intensity (MFI).

(E) Normalized E_max values for phospho-STAT3 and phospho-STAT1 calculated from sigmoidal dose-response curves shown in (D). Data are mean +/− SD for three replicates.

(F) Immunoblot for the indicated proteins in lysates prepared from human colorectal-derived HT-29 cells stimulated with 100 nM WT IL-22 or the indicated variants for 20 minutes.

In addition to the contacts described above, the structure also reveals the basis for the affinity enhancement resulting from the yeast-display selected mutations (Fig. S1B). Most notably, the S45R mutation in Super-22a likely converts a hydrogen bond contact with Rβ-Gln198 in WT IL-22, into a salt-bridge with Rβ-Asp197 observed in our structure (Figure 3C). In addition, the Q49S mutation facilitates a hydrogen-bond contact with backbone of Rβ-Ile148, while the Q116W mutation is primarily forming a hydrogen-bond contact via the indole nitrogen with the carbonyl of Rβ-His78 (Figure 3B). Overall, the affinity enhancing mutations in Super-22a are primarily improving upon pre-existing polar contacts made between WT IL-22 and IL-10Rβ, rather than creating novel contact sites. This gives confidence that the role of affinity maturation was to enhance pre-existing contacts, and therefore not perturb the overall docking geometry of the wild-type cytokine with IL-10Rβ. Thus, the affinity-matured structure is a faithful representation of the “wild-type” IL-22 complex.

Design of STAT3-biased IL-22 Receptor Agonists

Having resolved the structure of the IL-22–IL-10Rβ interface, we turned our attention to designing IL-22 variants with the ability to calibrate signal strength through the IL-22 receptor. Our strategy was to systematically destabilize IL-22-driven IL-22Rα–IL-10Rβ dimerization in fine increments, in order to create a series of partial agonists. To implement this strategy, we designed a set of IL-22 mutants targeting the IL-10Rβ-binding site (Figure 3B) and assessed their ability to activate of STAT signaling in colonic epithelial-derived HT-29 cells. Mutation of IL-22 residue Tyr51 to alanine (Y51A) fully eliminated both STAT3 and STAT1 activation (Figure 3D and 3E), consistent with its central role in the IL-10Rβ-binding interface (Figure 3A). By contrast, mutation of key contact residue Glu117, either alone (E117A, “22-B1”) or in combination with Asn46 (N46A, E117A, “22-B2”), resulted in only a slight decrease in STAT3 activation, but an almost abrogation of both STAT1 and STAT5 activation (Figures 3D–3F, Fig. S3B and S3C). A complex mutant consisting of Q116A, K124A, Q128A, and S45E (“22-B3”) elicited greater biased agonism, inducing substantial activation of STAT3 with negligible activation of STAT1 and STAT5 (Figures 3D–3F, Fig. S3B and S3C). The related IL-10Rβ-binding mutants comprising Q116A, K124D, and Q128A (“22-B4”), and Q48A, Q116A, K124D, and Q128A (“22-B5”) also yielded biased agonism, albeit with accompanying reduced STAT3 activation (Fig. S3D). Moreover, human IL-22 variants containing the corresponding mutations as 22-B1, B2, and B3 also elicited STAT3-biased signaling in these cells (Fig. S3E–S3G). Thus, mutations in IL-22 that partially disrupt the IL-10Rβ-binding interface preferentially activate STAT3 over STAT1 and STAT5, resulting in STAT3-biased signaling.

Biased IL-22 Variants Exploit Non-Canonical Mechanism of STAT3-Activation by IL-22Rα

To gain further insight into how reducing the affinity of IL-22 for IL-10Rβ resulted in biased agonism, we first analyzed whether the STAT3-biased variants still elicited phosphorylation of receptor associated JAK1 and TYK2 kinases, the key initiating step in the activation of STAT1 and STAT3 (Morris et al., 2018). HT-29 cells treated with 22-B3 showed only partially reduced phosphorylation of JAK1 and TYK2 compared to cells treated with WT IL-22, despite a near absence of STAT1 phosphorylation, suggesting that reduced JAK1 and TYK2 activity does not fully explain the pronounced STAT3 bias in these cells (Figures 4A and S4A).

Figure 4: Biased IL-22 Variants Exploit Non-Canonical Mechanism of STAT3 activation by IL-22Rα.

(A)Immunoblot for the indicated proteins in lysates prepared from HT-29 cells stimulated with 100 nM WT IL-22 or the indicated variants for 20 minutes.

(B, C) Immunoblots for the indicated proteins in lysates and anti-HA immunoprecipitates prepared from HEK-293T cells transiently expressing the indicated HA-IL22Rα constructs and stimulated with 10 nM WT IL-22 or the indicated variants for 20 minutes.

(D) Model of how IL-22 variants induce biased STAT signaling by exploiting the distinct mechanisms of STAT3 and STAT1 activation downstream of IL-22Rα.

We next asked whether biased IL-22 variants induced different levels of receptor ICD phosphorylation by reconstituting IL-22 receptor signaling using HEK-293T cells transiently expressing HA-tagged IL-22Rα. Despite triggering robust STAT3 activation, the biased IL-22 variants B1, B2, and B3 all induced negligible levels of IL-22Rα phosphorylation, correlating with the loss of STAT1 signaling. (Figures 4B and S4B). Furthermore, in cells expressing an IL-22Rα mutant lacking all tyrosine residues in the ICD (IL22Rα-YtoF), IL-22 still induced weak phosphorylation of STAT3 but not STAT1 (Figures 4C and S4C). This is consistent with the finding that unlike STAT1, STAT3 can pre-associate with IL-22Rα through a non-canonical mechanism that is independent of receptor phosphorylation, via binding to the C-terminal domain (CTD) of the receptor (Figure 4C) (Dumoutier et al., 2009). Our data further expands on this finding by revealing that these alternative modes of receptor binding also result in differential sensitivities of STAT1 and STAT3 signaling to IL-22 receptor dimerization, which can be exploited by designed variants with reduced affinity for IL-10Rβ, eliciting STAT3-biased agonism (Figure 4C).

Together, these data support a model in which the IL-22 receptor has different activation thresholds for STAT1 and STAT3, which can be “tuned” by modulating the affinity of IL-22 for IL-10Rβ (Figure 4D). Whereas STAT3 is robust to changes in IL-22Rα ICD phosphorylation due to its two distinct modes of receptor binding, STAT1 strictly requires receptor phosphorylation and is therefore highly sensitive to these changes (Figure 4D) (Dumoutier et al., 2009). These differential STAT response thresholds likely enhance the functional specificity of natural IL-22 signaling, and our data suggests they can also be exploited pharmacologically to generate STAT3-biased agonists.

Biased IL-22 Variant 22-B3 Elicits Tissue Selective STAT1/3 Activation In Vivo

In order to assess the effects of biasing STAT signaling downstream of the IL-22 receptor in vivo, we treated mice with a single intraperitoneal injection of WT IL-22 or 22-B3 and then analyzed STAT3 and STAT1 phosphorylation in several IL-22 responsive tissues after 30 minutes (Figure 5A). In the pancreas, 22-B3 induced full activation of STAT3 relative to WT IL-22, and a partially reduced activation of STAT1, thereby exhibiting weak STAT3-biased agonism in this tissue (Figures 5B and S5A). In the colon, however, 22-B3 retained full activation of STAT3, but induced no detectable phosphorylation of STAT1, whereas WT IL-22 robustly activated both pathways (Figures 5C and S5A). 22-B3 therefore functioned as a strong STAT3-biased-agonist in the colon, consistent with the results obtained in the colorectal HT-29 cells in vitro (Figure 3F). However, 22-B3 elicited no detectable signaling in the liver or skin, as assessed by phosphorylation of STAT3 (Figures 5D, 5E, and S5A). 22-B3 also failed to activate STAT1 or STAT3 signaling in human liver-derived HepG2 cells in vitro, indicating that the lack of activity in the liver is due to a cell-intrinsic effect, not differences in bioavailability or sensitivity to the natural IL-22 antagonist IL-22 binding protein (IL-22BP) (Figures 5F and S5B, S5C, and S5D). 22-B3 was also able to inhibit WT IL-22 signaling in HepG2 cells, demonstrating that 22-B3 retains the ability to bind IL-22Rα on these cells despite failing to elicit STAT activation (Figure 5G).

Figure 5: Biased IL-22 Variant 22-B3 Elicits Tissue Selective STAT Responses In Vivo.

(A)Schematic for characterization of 22-B3 activity in vivo. Mice were administered PBS or 200 μg/mouse recombinant WT IL-22 or 22-B3, and organs were isolated after 30 minutes (protein analysis), 6 hours (RNA analysis) or 24 hours (serum analysis).

(B-E) Immunoblots for the indicated proteins in tissue lysates prepared from the indicated organs. Phospho-STAT1 signal in the skin and liver were below the limit of detection.

(F) Relative expression of IL-22Rα and IL-10Rβ analyzed by qPCR (normalized to Gapdh). Data are mean +/− SEM (n=3).

(G, H) Immunoblot for the indicated proteins in lysates prepared from human hepatocyte-derived HT-29 cells stimulated with 100 nM WT IL-22 and/or 22-B3 for 20 minutes.

(I) Immunoblot for the indicated proteins in lysates prepared from HEK-293T cells transiently expressing HA-IL-22Rα alone or in combination with HA-IL10Rβ, stimulated with 10 nM WT IL-22 or 22-B3 for 20 minutes.

(J) Schematic illustrating different agonist profiles of 22-B3 across tissues, as a function of IL-10Rβ expression.

Given that the mutations in 22-B3 were designed to weaken the interaction with IL-10Rβ (Figure 3), we hypothesized that differences in IL-10Rβ expression could disproportionately affect 22-B3 signaling relative to that of WT IL-22, explaining the observed differences in signaling across tissues. Indeed, qPCR analysis of IL-22Rα and IL-10Rβ expression in these four mouse tissues revealed that the strength of 22-B3 signaling across tissues correlated well with IL-10Rβ expression, which was highest in the pancreas, followed by the colon, skin, and then liver (Figure 5H). Consistent with this, increasing the levels of IL-10Rβ, but not IL-22Rα, in HEK-293T cells was sufficient to reduce the extent of STAT3 bias elicited by 22-B3 (Figures 5I, S5E, S5F, and S5G). Together, these results demonstrate that the level of IL-10Rβ expression is a primary determinant of the type of STAT signaling induced by IL-22 partial agonists.

Thus, the IL-22 variant 22-B3 elicited distinct STAT1 and STAT3 agonist profiles across tissue types in vitro and in vivo, behaving as a weak STAT3-biased agonist in the pancreas, a strong STAT3-biased agonist in the colon, and a neutral antagonist of both STAT1 and STAT3 in the liver (Figure 5J). These results suggest that the distinct STAT1/3 activation thresholds downstream of IL-22 can be modulated by both IL-10Rβ affinity as well as IL-10Rβ expression, enabling tissue-selective agonism for a single IL-22 receptor ligand.

22-B3 Uncouples Expression of STAT1 and STAT3 Target Genes In Vivo

We next asked how biasing STAT activation downstream of IL-22 would alter the expression of IL-22 target genes in vivo. As 22-B3 showed the greatest STAT3-bias in the colon, which is also a major site of both the protective and pro-inflammatory functions of IL-22 (Bernshtein et al., 2019; Sugimoto et al., 2008; Zenewicz et al., 2008), we performed RNA sequencing (RNA-Seq) along with qPCR analysis of colonic tissue from mice treated with a single injection of either WT IL-22 or 22-B3 (Figure 6A). 22-B3 retained the ability to induce expression of key IL-22 target genes involved in promoting mucosal barrier integrity, including the antimicrobial peptides regenerating islet-derived protein (Reg)3β and Reg3γ, as well as the mucins Muc1 and Muc13, to the same extent as WT IL-22 (Figures 6B and 6C). 22-B3 also retained the ability to induce expression of a large set of DNA-damage response genes, including BRCA1 and RAD51 (Gronke et al., 2019), as well as key anti-inflammatory genes such as Il18BP and Socs3 (Figure 6B). One notable exception however was the anti-inflammatory gene Socs1, which mediates a negative feedback pathway downstream of STAT1 and was induced by WT IL-22 but not 22-B3 (Figure 6B) (O’Shea and Murray, 2008).

Figure 6: Uncouples Expression of STAT1 and STAT3 Target Genes In Vivo.

22-B3 (A) Correlation analysis of gene expression changes induced by 22-B3 compared to WT IL-22 in colon tissues analyzed by RNA-seq. Mice were treated via I.P. injection with 200 μg WT IL-22 or 22-B3 for 6 hours.

(B) Heatmaps showing gene expression changes relative to PBS-treated controls (log₂FC) in colon tissue of mice administered 22-B3 or WT IL-22, analyzed by RNA-seq. Mice were treated via I.P. injection with 200 μg WT IL-22 or 22-B3 for 6 hours.

(C) Relative expression of IL-22 target genes analyzed by qPCR. RNA was isolated from the colon of mice treated via I.P. injection with PBS or 200 μg WT IL-22 or 22-B3 for 6 hours. Data are mean +/− SD for two independent biological replicates, each analyzed in triplicate.

(D) Heatmaps showing relative expression changes (log₂FC) of genes in mouse colons following treatment with 22-B3 or WT IL-22, analyzed by RNA-seq. Mice were treated via I.P. injection with 200 μg WT IL-22 or 22-B3 for 6 hours.

(E) Relative expression of IL-22 target genes analyzed by qPCR. RNA was isolated from the colon of mice treated via I.P. injection with PBS or 200 μg WT IL-22 or 22-B3 for 6 hours. Data are mean +/− SD for two independent biological replicates, each analyzed in triplicate. *=p<0.05, **=p<0.01.

In epithelial cells, STAT1 is canonically activated by type 1 and type 3 interferons and primarily drives expression of genes involved in viral response and inflammation (O’Shea and Murray, 2008; Morris et al., 2018). Consistent with this, we found that WT IL-22 induced a large set of pro-inflammatory interferon stimulated genes (ISGs), including Isg15, Apobec, and Tmem173 (STING). By contrast, 22-B3 failed to induce any expression of these ISGs, indicating a lack of STAT1 activity in these cells (Figures 6D and 6E). Moreover, treatment with WT IL-22 but not 22-B3 increased expression of the Tumor Necrosis Factor (TNF) family members including Tnfsfr1b (Tnfr2) and Tnfsf10 (Trail), potentially reflecting altered immune cell recruitment or activation in this tissue (Figures 6D and 6E). In addition, unlike WT IL-22, 22-B3 also failed to induce colonic expression of the pro-inflammatory, chemokine Cxcl1, which is reported to play a key role in the pathologic effects of IL-22 in the GI tract (Figure 6E) (Bernshtein et al., 2019). Together, these results show that, by biasing IL-22R signaling toward STAT3, 22-B3 induced expression of tissue protective, stress-responsive, and anti-inflammatory gene sets in the colon without the accompanying inflammatory response associated with WT IL-22 signaling.

Tissue-Selective IL-22 Signaling Promotes Tissue Protection Without Inducing Inflammation

A therapeutically relevant function of IL-22 is the ability to protect intestinal epithelium and promote recovery of the tissue after damage. (Hanash et al., 2012; Lindemans et al., 2015). In order to test whether the STAT3-biased activity of 22-B3 is sufficient to maintain this beneficial property, we analyzed the ability of 22-B3 to promote intestinal recovery following radiation exposure. In the context of intestinal organoid cultures, we found that 22-B3 increased crypt cell viability following ex vivo radiation damage, and also increased expression of the intestinal stem cell (ISC) markers Lgr5 and Olfm4 (Figures 7A and 7B). 22-B3 also enhanced recovery of the intestinal epithelium following in vivo radiation damage, as measured by increased crypt depth and crypt circumference in the terminal ileum, demonstrating that 22-B3 retained the ability to promote epithelial regeneration (Figure 7C).

Figure 7: Tissue-Selective IL-22 Signaling Promotes Tissue Protection Without Inducing Inflammation.

(A) Viability of SI organoids treated with WT IL-22 (10 nM) or 22-B3 (10 nM) following irradiation, measured by MTT assay (n=3 mice per group).

(B) Relative expression of ISC markers Lgr5 and Olfm4 determined by qPCR in SI organoids cultured with either WT IL-22 (10 nM) or 22-B3 (10 nM) for 5 days. (n=3 mice per group).

(C) Quantitative analysis in the terminal ileum of total crypts number per circumference (left, n=3 mice per group), and average crypt depth (right, n=8 mice per group) at baseline and 5 days after total body irradiation, in mice treated via I.P. injection with PBS, WT IL-22 (20 μg/mouse) or 22-B3 (20 μg/mouse).

(D) Mice (5 per group) were pre-treated via I.P. injection with PBS or WT IL-22 (50 μg/mouse) or 22-B3 (50 μg/mouse) at 20 hours and 2 hours before initiating pancreatitis via 6 hourly I.P. injections of Cerulein (50 μg/kg). Serum was isolated 1 hour after final Cerulein injection for amylase and lipase measurements.

(E) Relative expression of IL-22 target genes analyzed by qPCR. RNA was isolated from the pancreas of mice treated via I.P. injection with PBS or 200 μg WT IL-22 or 22-B3 for 6 hours. Data are mean +/− SD for two independent biological replicates, each analyzed in triplicate.

(F) Mice (3 per group) were treated with PBS or the indicated amount of WT IL-22 or 22-B3. 24 hours later serum was isolated and levels of SAA-1/2 and Haptoglobin were analyzed by ELISA. Data are mean +/− SEM for three independent biological replicates.

(G, H) Relative expression of IL-22 target genes analyzed by qPCR. RNA was isolated from the skin (G) or liver (H) of mice treated via I.P. injection with PBS or 200 μg WT IL-22 or 22-B3 for 6 hours. Data are mean +/− SD for two independent biological replicates, each analyzed in triplicate. *=p<0.05, **=p<0.01.

To assess whether the 22-B3 also retains the protective effects of IL-22 across tissues, we next assessed its therapeutic efficacy in a mouse model of cerulein-induced acute pancreatitis. 22-B3 mitigated pancreatic damage in this model, to the same extent as WT IL-22, as measured by reduced serum levels of pancreatic lipase and amylase relative to PBS-treated controls shortly after disease onset (Figure 7D). Consistent with this, 22-B3 administered to healthy mice promoted pancreatic expression of Reg3β and Reg3γ (Figure 7E), which mediate the protective effects of IL-22 in pancreatitis (Xue et. al., 2012). 22-B3 also showed weaker induction of the inflammatory chemokine Cxcl1 in the pancreas, consistent with modest STAT3-biased agonism observed in this tissue (Figures 5B and 7E).

A potential limitation for the therapeutic use of IL-22 is its ability to induce pro-inflammatory effects both in the skin, leading to dermal inflammation and acanthosis (Zheng et al., 2007), and systemically, through the induction of hepatic acute phase-response proteins (APPs) (Liang et al., 2010; Rothenberg et al., 2019; Tang et al., 2019). In addition to serving as markers of systemic inflammation, APPs such as Serum Amyloid A (SAA)-1/2 can also stimulate Th17-mediated inflammation and contribute to disease pathology (Lee et al., 2020; Sano et al., 2016). Consistent with previous reports, we found that a single injection of WT IL-22 in healthy mice was sufficient to increase serum levels of the APPs SAA-1/2 and Haptoglobin (Figure 7F) (Liang et al., 2010). However, administration of 22-B3 had no detectable effect on SAA-1/2 or Haptoglobin levels, even when administered as high as 200 μg/mouse (Figure 7F). Furthermore, unlike WT IL-22, 22-B3 did not induce hepatic expression of Cxcl1 or the pro-inflammatory cytokine Il1b (Figure 7G). Similarly, in the skin, WT IL-22 but not 22-B3 promoted expression of Cxcl1 and Il6, potential drivers of inflammation in this tissue (Figure 7H). Together, these data show that the tissue-selective and STAT3-biased signaling elicited by 22-B3 was sufficient to retain the tissue protective and regenerative functions of IL-22 in vivo, without inducing the major pro-inflammatory effects associated with IL-22 signaling.

Discussion

Here, we have used structure-based design of IL-22 analogs to show that: (1) STAT1 and STAT3 activation by IL-22 can be uncoupled by exploiting distinct STAT response thresholds, (2) differential receptor expression across tissues correlates with shifts in these thresholds, enabling tissue-selective agonism, and (3) tissue-selective modulation of STAT1/3 signaling can uncouple the opposing tissue protective and pro-inflammatory functions of IL-22. Thus, these findings suggest that a combination of IL-22 affinity and receptor expression level modulate the strength of IL-22 signaling, and therefore the extent of STAT activation. Although these mass action-based mechanisms naturally endow IL-22 with some degree of tissue and functional selectivity, the cytokine remains highly pleiotropic. We have exploited these mechanisms to engineer IL-22 analogs with enhanced selectivity and the ability to uncouple the distinct physiological effects of IL-22 signaling.

Co-activation of the functionally opposed STAT1 and STAT3 transcription factors is a common feature of many pleiotropic cytokines in addition to IL-22, including IL-2, IL-6, and IL-21 (Lin and Leonard, 2019; O’Shea and Murray, 2008; Regis et al., 2008). The data presented here provide a mechanistic framework for understanding how these two pathways can be selectively uncoupled for therapeutic benefit. Specifically, our experiments in vitro revealed natural differences in STAT1 and STAT3 response thresholds downstream of IL-22, which could be pharmacologically manipulated by controlling the stability of the IL-22 receptor signaling complex. The signaling induced by these engineered IL-22 variants exists on a spectrum, ranging from full STAT1/3 agonism (WT IL-22), to inactive (IL-22-Y51A), with intermediate signal strength resulting in STAT3-biased activity (22-B1-B3). A relevant analogy to our studies is represented by the Type I Interferons, which exhibit the ability to differentially signal through a common IFNAR1/IFNAR2 heterodimer. However, in that case, the human genome encodes 16 different IFN sub-types, each of which has naturally evolved to impart distinct stabilities to the IFNAR1/IFNAR2 heterodimer, resulting in diverse signaling profiles (Thomas et al., 2011). Our results show that for cytokines like IL-22, where only a single ligand exists, a similar level of functional diversification is possible by engineering synthetic cytokine analogs based on these principles.

Mechanistically, these biased mutants induced substantially less IL-22Rα phosphorylation relative to WT IL-22, likely due to the reduced recruitment of IL-10Rβ and its associated intracellular kinase TYK2. Whereas STAT1 can only be recruited to the IL-22 receptor by binding to phosphorylated tyrosines on the ICD of IL-22Rα, STAT3 can associate with IL-22Rα independently of phospho-tyrosine binding, and therefore be activated even in the absence of receptor phosphorylation (Dumoutier et al., 2009). As a result of these two distinct mechanisms of activation, STAT3 signaling is robust across a wide range of ligand affinities, whereas STAT1 can be more readily “tuned out” by partial agonists.

Functional characterization of the engineered IL-22 variant 22-B3 in vivo revealed that a single ligand can also elicit distinct STAT1/3 agonist profiles across tissues, depending on the level of IL-10Rβ expression. Specifically, 22-B3 elicited STAT1/3 antagonism in the skin and liver, strong-STAT3 bias in the colon, and weak STAT3-biased agonism in the pancreas, correlating with IL-10Rβ expression in these tissues. This suggests that along with changes in IL-10Rβ binding affinity, changes in IL-10Rβ receptor expression can also shift apparent STAT activation windows, thereby modulating IL-22 signaling in a tissue-specific manner. This concept of functional tissue-selectivity, whereby a single ligand can elicit distinct agonist/antagonist profiles through the same receptor in different tissue contexts, is well established in nuclear hormone receptor pharmacology (Riggs and Hartmann, 2003), and our data extends this concept to cytokine receptors as well.

Previous studies have shown that exposure to type 1 and type 3 interferons can prime epithelial cells to elicit increased STAT1 activation in response to IL-22, enhancing the pro-inflammatory effects of IL-22 in these cells (Bachmann et al., 2013; Hernandez et al., 2015). Although an understanding of the mechanisms underlying the crosstalk between these cytokines requires further study, our data suggests that the regulation of IL-10Rβ expression may be one way for cells to shift the balance of IL-22-dependent STAT1/3 activation. The regulation of IL-10Rβ expression by other cytokines may therefore provide a natural mechanism for tuning the tissue protective versus pro-inflammatory functions of IL-22 depending on the specific the tissue context or disease state.

The ability of 22-B3 to uncouple the tissue protective and pro-inflammatory functions of IL-22 in vivo results from a combination of both its biased STAT activation and altered tissue specificity. For example, by activating STAT3 but not STAT1 in the colon, 22-B3 retains the ability to induce tissue protective genes such as Reg3β, Reg3γ, and Muc1, but no longer induces pro-inflammatory STAT1-regulated ISGs. However, acute phase response proteins such as SAA-1/2 are STAT3-target genes (Alonzi et al., 2001); therefore the lack of a systemic increase in these proteins is instead likely due to the antagonist activity of 22-B3 in the liver. Due to its inability to activate STAT3 in the skin and liver, 22-B3 likely does not retain the tissue protective effects of WT IL-22 in these tissues, and future studies should address whether it is possible to uncouple IL-22 functions in those contexts as well. Nonetheless, these data suggest that 22-B3 or similar IL-22 variants could have therapeutic utility in the treatment of inflammatory diseases in the pancreas and GI tract, with significantly reduced risk of side effects associated with inflammation in the skin and liver.

While our studies have focused on IL-22, many cytokines play important roles in controlling host immune responses and promoting tissue homeostasis, but often also exert pleiotropic or counterproductive effects that can limit their use as therapeutics (Lin and Leonard, 2019; Wang et al., 2009). As a result, while cytokine receptor antagonists have achieved significant clinical success, there are far fewer such examples for cytokine receptor agonists, and new approaches are needed to control signaling by cytokines in order to capture their therapeutic benefits. Here, we used a pharmacological approach, conceptually analogous to GPCR medicinal chemistry, to develop functionally selective variants that uncouple the tissue protective and pro-inflammatory activity of IL-22. These molecules not only open up new therapeutic opportunities, but also provide insights into the signaling mechanisms through which pleiotropic cytokines exert their diverse functions.

Limitations of the study

The mouse studies reported here were performed in the context of acute inflammation and short-term treatment with recombinant 22-B3. We therefore did not assess the efficacy of 22-B3 in the treatment of chronic inflammatory disease requiring a long-term treatment regimen. In addition, the in vivo signaling and RNA-sequencing experiments in this study were performed using bulk tissue samples. However, it is possible that 22-B3 also elicits differential signaling activity on individual cell types within these tissues, due to cell-type differences in IL-10Rβ expression. Future studies using single-cell transcriptomics and proteomics approaches are needed to shed light on this and provide further insight into the mechanisms underlying the therapeutic effects of both wild-type IL-22 and the engineered variants described here.

STAR METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, K. Christopher Garcia (kcgarcia@stanford.edu).

Materials availability

All unique/stable reagents generated in this study are available from the Lead Contact with a completed Materials Transfer Agreement.

Data and code availability

Structure factors and coordinates have been deposited in the Protein Data Bank with identification number PDB: 6WEO (IL-22 receptor complex). Diffraction images have been deposited in the SBGrid Data Bank. Next-generation sequencing data files from the mouse transcriptome study were deposited to the NCBI Gene Expression Omnibus (GEO) data repository with accession number GSE160217.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mammalian cell lines and culture conditions

HEK-293T (ATCC CRL-3216), HT-29 (ATCC HTB-38), HepG2 (ATCC HB-8065), and EC4 (gift from D. Felsher) cells were all grown in DMEM (Dulbecco’s Modified Eagle Medium) supplemented with 10% v/v fetal bovine serum, penicillin-streptomycin, 1 mM sodium pyruvate, 10 nM HEPES and 2 mM GlutaMAX^™ (Gibco). The cells were maintained at 37°C with 5% CO2. Expi293F cells were grown in serum free Expi293 expression media (Thermo) and maintained at 37°C with 5% CO₂ with gentle agitation.

Insect cell lines and culture conditions

For recombinant protein expression, baculovirus was produced in Spodoptera frugiperda (Sf9) ovarian cells (ATCC) maintained in Sf-900 III medium (GIBCO) with 10% FBS (Fisher Scientific) and GlutaMAX (GIBCO). Protein was expressed by baculoviral infection of Trichoplusia ni (Hi5) ovarian cells (Expression Systems) maintained in ESF 921 Insect Cell Culture Medium (Expression Systems). Insect cells were grown at 27°C with ambient CO₂ and gentle agitation.

Yeast strains and culture conditions

EBY100 yeast (ATCC) were cultured at 30°C with shaking in YPD media (Sigma) and selected in media containing glucose and casamino acids (SDCAA): 1 L deionize water with 20 g dextrose (Sigma), 6.7 g yeast nitrogen base (RPI), 5 g Bacto casamino acids (GIBCO), 10.4 g sodium citrate (Sigma), and 6.4 g citric acid monohydrate (Sigma), pH 4.5. Expression was induced at 20°C with shaking in galactose containing media (SGCAA) which resembles SDCAA but includes 20 g galactose (Sigma) instead of glucose.

Mouse organoid culture

To dissociate the crypts, the small intestine was incubated at 4 °C in EDTA (5 mM) for 30 min, and for mouse intestinal organoids 100 crypts per well were suspended in Matrigel composed of 50% advanced DMEM/ F12 medium (Gibco) and 50% growth-factor-reduced Matrigel (Corning). After the Matrigel polymerized, complete ENR medium containing advanced DMEM/ F12 (Sigma), 2 mM Glutamax (Invitrogen), 10 mM HEPES (Sigma), 100 U/ml penicillin, 100 μg/ml streptomycin (Sigma), 1 mM N-acetyl cysteine (Sigma), N2 supplement (Invitrogen), 50 ng ml−1 mouse EGF (Peprotech), 100 ng/ml mouse Noggin (Peprotech) and 10% human R-spondin-1-conditioned medium from R-spondin-1-transfected HEK 293T cells were added to the cultures. The media was replaced every 2–3 days. Along with medium changes, treatment wells received WT IL-22 (10 nM) or 22-B3 (10 nM).

In vivo animal studies

Female C57BL/6J mice (Cat# 000664) age 8–12 weeks were obtained from Jackson Labs. Mice were maintained on a 12-hour light-dark cycle. All mouse experiments were conducted according to protocols approved by the Institutional Animal Care and Use Committee, Administrative Panel on Laboratory Animal Care (APLAC) protocol.

METHOD DETAILS

Protein production and purification

For yeast-binding studies and affinity maturation, the extracellular domains (ECDs) of mouse IL-22Rα (16–228) and IL-10Rβ (20–220) were cloned into the pAcGP67a baculoviral vector with an N terminal GP64 signal peptide, C-terminal 3C cleavage site followed by a biotin-acceptor peptide tag (BAP tag, GLNDIFEAQKIEW) and 6xHis tag. The baculovirus stocks were prepared by co-transfection of the BestBac^™ DNA (Expression Systems) and the pAcGP67-A DNA into Spodoptera frugiperda (Sf9). Next, the viruses were used to infect the Trichoplusia ni (Hi5) cells. The proteins were purified from the supernatant of baculovirus infected Hi5 cells 48 hours after infection and purified with Ni-NTA resin (Qiagen) followed by size-exclusion chromatography (SEC) on a Superdex 200 column (GE). The proteins were maintained in HEPES buffered saline (HBS, 20 mM HEPES pH 7.4, 150 mM sodium chloride). IL-10Rβ ECD was site-specifically biotinylated at the C-terminal BAP tag using BirA ligase and re-purified by size exclusion chromatography.

For crystallographic studies, the fully glycosylated Super-22b and glyco-mutant Super-22a (N68Q, N97Q), as well as glyco-mutant versions the IL-22Rα (N80D, N87D, T89Q) and IL-10Rβ (N49Q, N102Q, N161Q, N199Q) ECDs (Fig. S2A) (Jones et al., 2008; Yoon et al., 2010) were cloned into the pAcGP67a baculoviral vector with an N terminal GP64 signal peptide and C-terminal 6xHis tag, expressed and purified as described above, followed by size-exclusion chromatography (SEC) on a Superdex 75 or 200 column (GE). Following SEC, the individual proteins were incubated overnight at a 1:1:1.2 molar ratio of IL-22:IL-22Rα:IL-10Rβ in the presence of carboxypeptidase A and B before purification by SEC on an S200 column (GE).

For signaling experiments, mouse IL-22 variants were cloned into the pD649 mammalian expression vector containing an N-terminal HA signal peptide and C-terminal 6xHis-tag. DNA was transiently transfected into Expi-293F cells (Thermo) using Expifectamine transfection reagent (Thermo). 96 hours after transfection, cell supernatant was harvested and proteins were purified with Ni-NTA resin (Qiagen) followed by size-exclusion chromatography (SEC) on a Superdex 75 column (GE) in HEPES buffered saline (HBS, 30 mM HEPES pH 7.4, 150 mM sodium chloride). For in vivo studies, endotoxin was removed using the NoEndoHC Column Kit (Proteus), and endotoxin removal was confirmed using the Pierce^™ Chromogenic Endotoxin Quant Kit (ThermoFisher). Mouse IL-22BP was cloned into the pAcGP67a baculoviral vector with an N terminal GP64 signal peptide, C-terminal 3C cleavage site followed by a biotin-acceptor peptide tag (BAP tag, GLNDIFEAQKIEW) and 6xHis tag and purified from Hi5 cells as described above.

Yeast display, library assembly, and affinity maturations

Mouse IL-22 (residues 34–179) containing a C-terminal Myc-tag was displayed on the surface of S. cerevisiae strain EBY100 as a C-terminal fusion to Aga2 using the pCT302 vector. A mutant mIL-22 library containing six randomized residues at the predicted IL-10Rβ contact site was generated by primer assembly PCR using degenerate codons (NNK). Electroporation, rescue and expansion of the yeast library were performed as described previously (Chao et al., 2006). The final library contained approximately 6.5 × 10⁸ yeast transformants.

Library selection was conducted as described previously with some modifications. Briefly, the initial selections (rounds 1–3) were conducted using magnetic activated cell sorting (MACS). For rounds 1 and 2, 5 × 10⁹ and 1 × 10⁸ cells, respectively, were pre-incubated in 1 μM IL-22Rα and selected with paramagnetic streptavidin microbeads (Miltenyi) that were pre-coated with 400 nM biotinylated IL-10Rβ. IL-10Rβ-binding clones were isolated using MACS LS columns (Miltenyi). For round 3, 1.0 × 10⁸ yeast were pre-incubated in 1 μM mIL-22Rα and stained and selected with 500 nM monomeric biotinylated IL-10Rβ labeled with streptavidin conjugated to Alexa Fluor 647. IL-10Rβ-binding clones were isolated using MACS LS columns (Miltenyi) in combination with Anti-Cy5/Anti-Alexa Fluor 647 MicroBeads (Miltenyi). In rounds 4 and 5, library selection was performed using two-color fluorescence activated cell sorting (FACS) to normalize apparent affinity by protein expression on the cell surface. The yeast library was pre-incubated with 1 μM mIL-22Rα followed by 100 nM and 10 nM of biotinylated mIL-10Rβ, for round 4 and 5, respectively. The cells were washed twice with PBE (PBS, pH 7.4% + 0.5% (w/v) BSA + 2 mM EDTA, pH 8.0) and co-labeled with streptavidin-Alexa Fluor 647 (mIL-10Rβ binding) and anti-c-Myc-Alexa Fluor 488 (mIL-22 variant cell surface expression) for 15 min at 4°C. Alexa647⁺Alexa488⁺ yeast were purified using a SH800S Cell Sorter (Sony Biotechnology). At the conclusion of the selections, post-selection library at each round was simultaneously expanded and incubated with varying concentrations of biotinylated mIL-10Rβ for 1 hr at 4°C, washed twice with PBE, then stained with fluorescently labeled streptavidin for 10 min at 4°C to assess the enrichment of high-affinity clones via flow cytometry. In addition, 100 μL of post-round 5 library was used to extract library DNA using the Zymoprep Yeast Plasmid Miniprep II Kit (Zymo Research), according to the manufacturer’s instructions. The extracted DNA was transformed into DH5α E. coli and plated to sequence the individual clones.

To measure relative binding affinities, individual clones were displayed on the yeast surface and incubated with 1 μM mIL-22Rα and increasing concentrations of biotinylated mIL-10Rβ. The cells were then stained with streptavidin-Alexa Fluor 647 for 15 minutes and analyzed on a CytoFLEX Flow Cytometer (Beckman Coulter).

Crystallization, data collection, and refinement

Following purification, the IL-22/IL-22Rα/IL-10Rβ complex was concentrated to 10 mg/ml. Crystals of the fully glycosylated “Super-22b” complex were grown by hanging drop vapor diffusion with 1 μl of protein mixed with an equal volume of reservoir solution containing 0.2 M sodium potassium tartrate, 15% PEG 3350, and 0.1 M HEPES pH 7.8 at 20°. These crystals were subsequently microseeded into hanging drops containing 1 μl of the glyco-mutant “Super-22a” complex, mixed with an equal volume of 0.2 M sodium potassium tartrate, 12.5% PEG 3350, and 0.1 M HEPES pH 8.0 at 20°. Crystals were harvested and cryoprotected in mother liquor containing 25% Glycerol.

Diffraction data was collected at the Advanced Photon Source (APS) beamline 23 ID-B. The 2.6 Å resolution dataset was integrated and scaled using XDS (Kabsch, 2010) before merging symmetry-related reflections with aimless (Evans, 2011; Winn et al., 2011). Phases were solved by molecular replacement with Phaser (Adams et al., 2010; McCoy et al., 2007) using the crystal structures of human IL-22, IL-22Rα (PDB ID: 3DGC) and IL10-Rβ (PDB ID: 3LQM) as search models, identifying 36 molecules in the ASU (12 trimeric complexes). A native Patterson peak at (0.35, 0.31, 0) indicated the presence of translational non-crystallographic symmetry (tNCS). Although the presence of tNCS degraded the quality of the electron density maps, three of the complexes in the asymmetric unit had unambiguous electron density and were used for further structural analysis.

The initial high-resolution cutoff of was chosen such that CC_1/2 for the high-resolution shell was approximately 0.3. The high-resolution limit of the final data set was 2.6 Å, chosen by performing paired refinement tests of the final cycles of refinement using high resolution cutoffs of 2.6, 2.7, 2.8, 2.9, and 3.0 Å (Karplus and Diederichs, 2015). Model building was carried out using iterative rounds of reciprocal space refinement in Phenix.refine (Adams et al., 2010; Terwilliger et al., 2008) or Buster (Bricogne et al., 2016) and manual model building in Coot (Emsley et al., 2010). All data-processing steps were carried out with programs provided through SBgrid (Morin et al., 2013). Statistics for data collection and refinement are provided in Table S1. Protein-protein and protein-ligand interfaces were analyzed using PDBePISA (Krissinel and Henrick, 2007). All structure figures were made using ChimeraX (Goddard et al., 2018).

Phospho-flow signaling assays

HT-29 cells or EC4 cells were plated in 96-well plates and stimulated with WT or mutant IL-22 for 20 min at 37°C, followed by fixation with paraformaldehyde (Electron Microscopy Sciences) for 10 min at room temperature. The cells were permeabilized for intracellular staining by treatment with ice cold methanol (Fisher) for 30 min at −20°C. The cells were then incubated with the desired antibodies at a 1:50 dilution for 1 hour at room temperature in autoMACS buffer (Miltenyi). The background fluorescence of the unstimulated samples was subtracted from all samples. Data was acquired using CytoFlex, flow cytometer instrument (Beckman Coulter). The MFI values were background subtracted and represented as a percent of the maximal WT IL-22 value within each experiment and plotted in Prism 8 (GraphPad). The dose-response curves were generated using the “sigmoidal dose-response” analysis. Alexa Fluor 488 Mouse Anti-Stat1 (pY701) and Alexa Fluor 647 Mouse Anti-Stat3 (pY705) antibodies were purchased from BD Biosciences.

Immunoprecipitation and western blot

For signaling and immunoprecipitation experiments in HEK-293T cells, 0.7 × 10⁶ cells were plated in 6-well culture dishes coated with fibronectin (Millipore). 24 hours later, cells were transfected using FuGene 6 (Promega) with pLV-EF1a-IRES-Puro vector containing HA-tagged IL-10Rβ or IL-22Rα (WT, YtoF, or ΔCTD mutants, as described previously) (Dumoutier et al., 2009). 24 hours after transfection, cells were treated with IL-22 (WT or mutant) for 20 minutes at 37°C. Cells were then rinsed one time with ice-cold PBS and immediately lysed with Triton lysis buffer (1% Triton, 20 mM HEPES pH 7.4, 150 mM NaCl, 1 tablet of PhosSTOP phosphatase inhibitor cocktail (Roche), and 1 tablet of EDTA-free protease inhibitor (Roche) (per 10 ml buffer). The cell lysates were cleared by centrifugation at 13,000 rpm at 4°C in a microcentrifuge for 10 minutes. For anti-HA-immunoprecipitations, the magnetic anti-HA beads (Pierce) were washed 3 times with ice-cold lysis buffer with 300 mM NaCl. 20 μl of anti-HA beads was then added to clarified cell lysates and incubated with rotation for 1 hour at 4°C. Following immunoprecipitation, the beads were washed 4 times with ice-cold lysis buffer with 300 mM NaCl. Immunoprecipitated proteins and cell lysates were denatured by the addition of SDS sample buffer and boiling for 5 minutes, resolved by SDS-PAGE, and analyzed by immunoblotting. Western blot image densities were quantified using ImageJ.

For western blot based signaling assays, HT-29 (1 × 10⁶ cells/well) or HepG2 (0.6 × 10⁶ cells/well) cells were seeded in 6-well culture dishes. 24 hours later cells were treated with IL-22 (WT or mutant) for 20 minutes at 37°C, and lysates were prepared and analyzed as described above.

In vivo signaling, gene expression, and serum analysis

For in vivo signaling, experiments, C57BL/6J mice were administered 200 μl of PBS or IL-22 variants (200 μg/mouse) via I.P. injection. After 30 minutes, mice were euthanized and the pancreas, colon, liver, and tail skin tissues were isolated and flash frozen in liquid nitrogen. Tissues were disrupted with mortar and pestle in liquid nitrogen and 50 mg tissue was lysed in ice-cold RIPA lysis buffer (Thermo) supplemented with 1 tablet PhosSTOP (Roche) and one tablet of protease inhibitor cocktail (Roche) per 10 ml. Lysates were further homogenized via centrifugation through a QiaShredder column (Qiagen). Protein concentrations for each sample of a given tissue were normalized by Bradford assay (Bio-Rad) before being denatured in SDS sample buffer and resolved by SDS-PAGE and analyzed by immunoblot.

For gene expression analysis in mouse tissues by qPCR, C57BL/6J mice were administered 200 μL of PBS or IL-22 variants (200 ug/mouse) via I.P. injection. After 6 hours, mice were euthanized, and tissues were processed and homogenized as described above. RNA from each tissue was isolated using RNAeasy plus mini kit (Qiagen) per the manufacturer’s instructions. 1 μg RNA for each sample was used for cDNA generation with iScript Reverse Transcription Supermix (BioRad). Relative gene expression was measured by SYBR-green based qPCR using the comparative ΔC_t method and normalized to Gapdh expression. All samples were run in triplicate. The following mouse qPCR primers were obtained from IDT: Gapdh (5’GTG GAG TCA TAC TGG AAC ATG TAG3’, 5’AAT GGT GAA GGT CGG TGT G3’), Il22ra1 (5’CTC GTA TTC TCT CTG TTT GCC T3’, 5’CAT GAC CTG TTC TAC CGC TTA G3’), Il10r2 (5’CAG GAC GGA GAC TAT GAG GAT3’, 5’ACA GAA CAG GAG AGT GGA GT3’), Reg3b (5’TGT TAC TCC ATT CCC ATC CAC3’, 5’CTG AGG CTT CAT TCT TGT CCT3’), Reg3g (5’GAT TCG TCT CCC AGT TGA TGT3’, 5’CTC CAT GAC CCG ACA CTG3’), Muc1 (5’GAC TGC TAC TGC CAT TAC CTG3’, 5’CCT ACC ATC CTA TGA GTG AAT ACC3’), Cxcl1 (5’GTG CCA TCA GAG CAG TCT3’, 5’CCA AAC CGA AGT CAT AGC CA3’), Il6 (5’TCC TTA GCC ACT CCT TCT GT3’, 5’AGC CAG AGT CCT TCA GAG A3’), Il1b (5’CTC TTG TTG ATG TGC TGC TG3’, 5’GAC CTG TTC TTT GAA GTT GAC G3’), Isg15 (5’CCC CCA TCA TCT TTT ATA ACC AAC3’, 5’CAC AGT GAT CAA GCA TTT GCG3’), Tnfrsf1b (5’GCC TTC CTG TCA TAG TAT TCC TG3’, 5’TCT TCG AAC TGC AGC TGT G3’), Tnfsf10 (5’GTT GCT TCT CCG AGT GAT CC3’, 5’TCA GGA CAC CAT TTC TAC AGT TC3’).

For qPCR from mouse intestinal organoids, RNA was isolated from organoids after 5 days of in vitro culture. Reverse transcriptase was performed with a High-Capacity RNA-to-cDNA Kit (Applied Biosystems). Primers were obtained from PrimerBank: Gapdh (ID 6679937a1) Olfm4 (ID 71892419c1) and Lgr5 (ID 110624771c2). qPCR was performed on QuantStudio 7 Flex System (Applied Biosystems). cDNAs were amplified with SYBR master mix (Applied Biosystems) in QuantStudio 7 Flex System (Applied Biosystems). Relative amounts of mRNA were calculated by the comparative ΔC_t method with Gapdh as house-keeping genes.

For RNA-sequencing analysis, C57BL/6J mice were administered 200 μl of PBS or IL-22 variants (200 μg/mouse) via I.P. injection and RNA was isolated after 6 hours as described above. RNA-quality control was performed using Qubit and Bioanalyzer 2100 (Agilent). cDNA libraries were constructed using NEBnext® Ultra^™ II RNA Library Prep Kit for Illumina®, using 0.2 μg RNA input, and the mRNA enrichment method. Library QC was performed using Qubit, LabChip, and qPCR (KPA library quantification kit). cDNA libraries were loaded onto an Illumina NovaSeq 6000 sequencer, PE150 platform. RNA samples from two independent biological replicates were used for sequencing. Reference genome and gene model annotation files were downloaded from genome website browser (NCBI/UCSC/Ensembl) directly. Indexes of the reference genome was built using STAR and paired-end clean reads were aligned to the reference genome using STAR (v2.5) (Dobin et. al., 2013). The FPKM of each gene was calculated based on the length of the gene and reads count mapped to the gene and used to quantify relative gene expression levels (Mortazavi et. al., 2008).

For serum analysis of acute phase response proteins, C57BL/6J mice were administered 200 μl of PBS or IL-22 variants (2 μg or 200 μg per mouse) via I.P. injection. After 24 hours, blood was obtained by cardiac puncture, and serum was isolated by centrifugation at 2000 g for 10 minutes at 4°. Levels of SAA-1/2 (Invitrogen) and Haptoglobin (R&D) were measured by ELISA following the manufacturer’s instructions. Each sample was run in triplicate.

In vivo radiation model

C57BL/6J mice received 10 Gy of single-dose lethal irradiation. Mice were treated immediately after radiation with a single dose of PBS, WT IL-22 (20 μg/mouse), or 22-B3 (20 μg/mouse). For histopathology analysis, mice were euthanized for small intestine analysis 5 days after radiation using CO₂ asphyxiation. The small intestine was fixed in PFA 4%, sectioned, and stained with hematoxylin and eosin.

QUANTIFICATION AND STATISTICAL ANALYSIS

Flow cytometry data was analyzed using FlowJo software (BD). qPCR data was analyzed using StepOne Software (Applied Biosystems). Statistical analyses were performed using Prism v8.4 (GraphPad Software). For signaling and gene expression analysis, data shown are mean ± standard deviation. For serum protein analysis and intestinal crypt measurements, data shown are mean ± standard error of the mean. Groups were compared using Student’s T test. All data are representative of two or more independent experiments. Additional information, including n values and number of replicates, can be found in figure legends.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-c-Myc-Alexa Fluor 488 (9B11)	Cell signaling technologies	Cat#2279S
Anti-phospho-STAT3 Y705-Alexa Flour 647 (4/P-STAT3)	BD Biosciences	Cat#557815
Anti-phospho-STAT1 Y701-Alexa Flour 488 (4a)	BD Biosciences	Cat#612596
Anti-phospho-STAT3 Y705	Cell signaling technologies	Cat#9131
Anti-phospho-STAT3 S727	Cell signaling technologies	Cat#9134
Anti-phospho-STAT1 Y701 (58D6)	Cell signaling technologies	Cat#7649
Anti-phospho-STAT1 S727	Cell signaling technologies	Cat#9177
Anti-phospho-STAT5 Y694 (C11C5)	Cell signaling technologies	Cat#9359
Anti-STAT3 (79D7)	Cell signaling technologies	Cat#4904
Anti-STAT1	Cell signaling technologies	Cat#9172
Anti-STAT5 (D3N2B)	Cell signaling technologies	Cat#25656
Anti-phospho-Tyrosine (P-Tyr-1000)	Cell signaling technologies	Cat#8954
Anti-HA (C29F4)	Cell signaling technologies	Cat#3724
Anti-rabbit Ig HRP	Dako	Cat#P0399
Anti-GAPDH	Cell signaling technologies	Cat#2118
Bacterial and Virus Strains
Mix & Go Competent Cells - DH5α	Zymo Research	Cat#T3007
Biological Samples
Chemicals, Peptides, and Recombinant Proteins
mIL-22	This paper	N/A
mIL-22 (“Super-22a”, E43H, S45R, Q49S, Q116W, Q128K)	This paper	N/A
mIL-22 (“Super-22a” glyco-mutant, E43H, S45R, Q49S, N68Q, N97Q, Q116W, Q128K)		N/A
mIL-22 (“Super-22b”, E43R, S45G, Q49G, Q116K, K124Y)	This paper	N/A
mIL-22 (22-B1, E117A)	This paper	N/A
mIL-22 (22-B2, N46A, E117A)	This paper	N/A
mIL-22 (22-B3, S45E, Q116A, K124A, Q128A)	This paper	N/A
mIL-22 (22-B4, Q116A, K124D, Q128A)	This paper	N/A
mIL-22 (22-B5, Q48A, Q116A, K124A, Q128A)	This paper	N/A
hIL-22	This paper	N/A
hIL-22 (h22-B1, E117A)	This paper	N/A
hIL-22 (h22-B2, N46A, E117A)	This paper	N/A
hIL-22 (h22-B3, S45E, Q116A, R124A, R128A)	This paper	N/A
mIL22Rα	This paper	N/A
mIL22Rα glyco-mutant (N80D, N87D, T89Q)	This paper	N/A
mIL10Rβ	This paper	N/A
mIL10Rβ glyco-mutant (N49Q, N102Q, N161Q, N199Q)	This paper	N/A
mIL-22BP	This Paper	N/A
EndoH	Produced in house	N/A
Kifunensine	Toronto Research Chemicals	Cat#K450000
Streptavidin	Sigma	Cat#189730
Alexa Flour 647 C2 Maleimide	Invitrogen	Cat#A20347
Cellfectin II	Gibco	Cat#10362100
cOmplete protease inhibitor cocktail	Roche	Cat#5056489001
PhosSTOP	Sigma	Cat#4906845001
1M HEPES	Gibco	Cat#15630-080
Penicillin-streptomycin	Gibco	Cat#15140163
2-mercaptoethanol	Sigma	Cat#M3148
TrypLE	Gibco	Cat#12604013
Carboxypeptidase A	Sigma	Cat#C9268
Carboxypeptidase B	Sigma	Cat#217356
Expi293 Expression Medium	Gibco	Cat#A1435101
Dulbecco’s Modified Eagle’s Medium (DMEM)	Gibco	Cat#11960-069
Phosphate Buffered Saline (PBS)	Gibco	Cat#20012-050
Fetal Bovine Serum	Sigma	Cat#F4135-500
Sf-900 III Media	Invitrogen	Cat#12658019
ESF 921 Insect Cell Culture Medium	Expression Systems	Cat#96-001-01
D-glucose (dextrose)	Sigma	Cat#G8270-5KG
YPD broth	Sigma	Cat#Y1375
Bacto Casamino Acids	Gibco	Cat#223050
Yeast Nitrogen Base (without amino acids)	RPI	Cat#Y20040
Sodium Citrate Tribasic Dihydrate	Sigma	Cat#S4641
Citric Acid Monohydrate	Sigma	Cat#C1909
D-galactose	Sigma	Cat#G0625
Biotin	Sigma	Cat#B4501
16% paraformaldehyde	Fisher Scientific	Cat#50-980-487
Cerulein	Sigma	Cat#C9026
Streptavidin/Alexa Fluor™ 647 conjugate	Produced in house	N/A
BirA	Produced in house (Fairhead and Howarth, 2015)	N/A
Recombinant murine EGF	Peprotech	315–09
Recombinant murine Noggin	Peprotech	250–38
Human R-spondin	Produced in house	N/A
N-acetyl cysteine	Sigma	A7250
Fibronectin	EMD Millipore	Cat#341635
Critical Commercial Assays
Mouse SAA-1/2 ELISA kit	Invitrogen	Cat#KMA0021
Mouse Haptoglobin ELISA kit	R&D	Cat#DY4409-05
iScript™ Reverse Transcription Supermix	BioRad	Cat#1708840
RNAEasy plus mini kit	Qiagen	Cat#1708840
Streptavidin microbeads	Miltenyi	Cat#130-048-101
Anti-Cy5/Anti-Alexa Fluor 647 Microbeads	Milteny	Cat#130-091-395
LS magnetic selection column	Miltenyi	Cat#130-042-401
EasySep Magnet	StemCell Technologies	Cat#18000
High Capacity NoEndo Columns	Protein Ark	Cat#Gen-NoE48HC
Chromogenic Endotoxin Quant Kit	Pierce	Cat#A39552
12% Mini-PROTEAN TGX Precast Protein Gels	BioRad	Cat#4561046
Trans-Blot Turbo RTA Mini 0.2μm PVDF Transfer Kit	BioRad	Cat#1704272
Blotting-Grade Blocker	BioRad	Cat#1706404
ECL Prime Western Blotting Detection Reagent	Cytiva	Cat#45002401
Hyperfilm ECL	Cytiva	Cat#28906839
Zymoprep Yeast Plasmid Miniprep II Kit	Zymo Research	Cat#D2004
Anti-HA Magnetic Beads	Thermo	Cat#88836
FuGene HD Transfection Reagent	Fisher	Cat#E2312
High-Capacity RNA-to-cDNA Kit	Applied Biosystems	Cat#4387406
NEBnext® Ultra II RNA Library Prep Kit for Illumina	NEB	Cat# E7770
ExpiFectamin 293 Transfection Kit	Thermo	Cat#A14525
Deposited Data
IL22-IL22Rα-IL10Rβ complex	PDB	6WE0
RNA-Seq data (raw and analyzed)	NCBI GEO	GSE160217
Experimental Models: Cell Lines
Human: HEK-293T	ATCC	CRL-1575
Human: HT-29	ATCC	HTB-38
Human: Panc-1	ATCC	CRL-1469
Human: HepG2	ATCC	HB-8065
Human: Expi293F	ThermoFisher	Cat#A14527
Mouse: EC4	(Cao et al.,2011)
Yeast: S. cerevisiae EBY100	ATCC	MYA-4941
Insect: Spodotera frugiperda (Sf9)	ATCC	CRL-1711
Insect: Trichoplusia ni (Hi5)	Expression Systems	Cat#94-002F
Experimental Models: Organisms/Strains
Mouse: C57BL/6J	The Jackson Laboratory	JAX:000664
Oligonucleotides
Gapdh Forward GTG GAG TCA TAC TGG AAC ATG TAG	IDT	N/A
Gapdh Reverse AAT GGT GAA GGT CGG TGT G	IDT	N/A
Il22ra1 Forward CTC GTA TTC TCT CTG TTT GCC T	IDT	N/A
Il22ra1 Reverse CAT GAC CTG TTC TAC CGC TTA G	IDT	N/A
Il10r2 Forward CAG GAC GGA GAC TAT GAG GAT	IDT	N/A
Il10r2 Reverse ACA GAA CAG GAG AGT GGA GT	IDT	N/A
Reg3b Forward TGT TAC TCC ATT CCC ATC CAC	IDT	N/A
Reg3b Reverse CTG AGG CTT CAT TCT TGT CCT	IDT	N/A
Reg3g Forward GAT TCG TCT CCC AGT TGA TGT	IDT	N/A
Reg3g Forward CTC CAT GAC CCG ACA CTG	IDT	N/A
Muc1 Forward GAC TGC TAC TGC CAT TAC CTG	IDT	N/A
Muc1 Reverse CCT ACC ATC CTA TGA GTG AAT ACC	IDT	N/A
Cxcl1 Forward GTG CCA TCA GAG CAG TCT	IDT	N/A
Cxcl1 Reverse CCA AAC CGA AGT CAT AGC CA	IDT	N/A
Il6 Forward TCC TTA GCC ACT CCT TCT GT	IDT	N/A
Il6 Reverse AGC CAG AGT CCT TCA GAG A	IDT	N/A
Il1b Forward CTC TTG TTG ATG TGC TGC TG	IDT	N/A
Il1b Reverse GAC CTG TTC TTT GAA GTT GAC G	IDT	N/A
Isg15 Forward CCC CCA TCA TCT TTT ATA ACC AAC	IDT	N/A
Isg15 Reverse CAC AGT GAT CAA GCA TTT GCG	IDT	N/A
Tnfrsf1 Forward GCC TTC CTG TCA TAG TAT TCC TG	IDT	N/A
Tnfrsf1 Reverse TCT TCG AAC TGC AGC TGT G	IDT	N/A
Tnfsf10 Forward GTT GCT TCT CCG AGT GAT CC	IDT	N/A
Tnfsf10 Reverse TCA GGA CAC CAT TTC TAC AGT TC	IDT	N/A
Olfm4 Forward CAGCCACTTTCCAATTTCACTG	PrimerBank	71892419c1
Olfm4 Reverse GCTGGACATACTCCTTCACCTTA	PrimerBank	71892419c1
Lgr5 Forward ACATTCCCAAGGGAGCGTTC	PrimerBank	110624771c2
Lgr5 Reverse ATGTGGTTGGCATCTAGGCG	PrimerBank	110624771c2
Recombinant DNA
pAcGP67a	BD	Cat# 554756
pCT_302	(Midelfort et al., 2004)	RRID:Addgene_41845
pD649	ATUM	Cat#PD649
pLV-EF1a-IRES-Puro	(Hayer et al., 2016)	RRID:Addgene_85132
pD649 HASP-mIL-22 (34–179)-6xHIS	This paper	N/A
pAcGP67a GP64SP-mIL-22 (34–179)-6xHIS	This paper	N/A
pAcGP67a GP64SP-mIL-22 (“Super-22a”, 34–179, E43H, S45R, Q49S, Q116W, Q128K)-6xHIS	This paper	N/A
pAcGP67a GP64SP-mIL-22 (“Super-22a” glyco-mutant, 34–179, E43H, S45R, Q49S, N68Q, N97Q, Q116W, Q128K)-6xHIS	This paper	N/A
pAcGP67a GP64SP-mIL-22 (“Super-22b”, 34–179, E43R, S45G, Q49G, Q116K, K124Y)-6xHIS	This paper	N/A
pD649 HASP-mIL-22 (22-B1, 34–179, E117A)-6xHIS	This paper	N/A
pD649 HASP-mIL-22 (22-B2, 34–179, N46A, E117A)-6xHIS	This paper	N/A
pD649 HASP-mIL-22 (22-B3, 34–179, S45E, Q116A, K124A, Q128A)-6xHIS	This paper	N/A
pD649 HASP-mIL-22 (22-B4, 34–179, Q116A, K124D, Q128A)-6xHIS	This paper	N/A
pD649 HASP-mIL-22 (22-B5, 34–179, Q48A, Q116A, K124A, Q128A)-6xHIS	This paper	N/A
pD649 HASP-hIL-22 (34–179) -6xHIS	This paper	N/A
pD649 HASP-hIL-22 (h22-B1, 34–179, E117A)-6xHIS	This paper	N/A
pD649 HASP-hIL-22 (h22-B2, 34–179, N46A, E117A)-6xHIS	This paper	N/A
pD649 HASP-hIL-22 (h22-B3, 34–179, S45E, Q116A, R124A, R128A)-6xHIS	This paper	N/A
pAcGP67a GP64SP-mIL22Rα (16–228)-6xHIS	This paper	N/A
pAcGP67a GP64SP-mIL22Rα glyco-mutant (16–228, N80D, N87D, T89Q)-6xHIS	This paper	N/A
pAcGP67a GP64SP-mIL10Rβ (20–220)-6xHIS	This paper	N/A
pAcGP67a GP64SP-mIL10Rβ glyco-mutant (20–220, N49Q, N102Q, N161Q, N199Q)-6xHIS	This paper	N/A
pCT302 Aga2-mIL-22-myc	This paper	N/A
pLV-EF1a-IRES-Puro-HA-hIL22Rα (16–574)	This paper	N/A
pLV-EF1a-IRES-Puro-HA-hIL22Rα YtoF (16–574, Y326F, Y351F, Y364F, Y377F, Y390F, Y403F, Y491F)	This paper	N/A
pLV-EF1a-IRES-Puro-HA-hIL22Rα ΔCTD (16–490)	This paper	N/A
pD649 HASP-HA-hIL-10Rβ (20–325)	This paper	N/A
pAcGP67a GP64SP-mIL-22BP (21–230)-6xHIS	This paper	N/A
Software and Algorithms
FlowJo v10.5	Tree Star	RRID:SCR_008520
GraphPad Prism 8.3.0	GraphPad Software	RRID:SCR_002798
Phenix	(Liebschner et al., 2019)	RRID:SCR_014224
Coot	(Emsley et al., 2010)	RRID:SCR_014222
UCSF ChimeraX	(Goddard et al., 2018)	RRID:SCR_015872
PISA	(Krissinel and Henrick, 2007)	RRID:SCR_015749
SBGrid	(Morin et al., 2013)	RRID:SCR_003511

Highlights.

• 2.6-Å resolution structure of a stabilized IL-22 receptor ternary complex

• Structure-based design of STAT3-biased IL-22 receptor agonists

• Biased IL-22 variant 22-B3 elicits tissue selective STAT3 activation in vivo

• 22-B3 uncouples the tissue protective and pro-inflammatory functions of IL-22