Engineered TYK2 dependence enables myeloid-selective IL-10 receptor agonism

IL-10 is a potent anti-inflammatory cytokine that maintains immune tolerance^1,6. Loss-of-function mutations in IL10 or its receptors IL10RA (encoding IL-10Rα) and IL10RB (encoding IL-10Rβ) cause early-onset inflammatory bowel disease (IBD) and autoinflammatory disorders^1,7. IL-10 signalling is initiated when IL-10 binds to IL-10Rα and IL-10Rβ to form a tetrameric receptor complex, which is constitutively associated with Janus kinase 1 (JAK1) and tyrosine kinase 2 (TYK2), respectively^1,8. This binding triggers JAK1-dependent phosphorylation of IL-10Rα and signal transducer and activator of transcription 3 (STAT3) to suppress pro-inflammatory gene expression and promote immune tolerance^8,9. IL-10 primarily exerts its tolerogenic effects in myeloid cells and antigen-presenting cells, such as dendritic cells (DCs) and macrophages, by inhibiting cytokine production, antigen presentation and T cell priming^{5,10,11,12,13}. Loss of IL10RA or STAT3 in these cells leads to colitis and systemic inflammation^10,12,13.

Despite its central role in immune tolerance, the dual functionality of IL-10 severely limits its therapeutic use^6,14. Whereas IL-10 potently suppresses inflammation in myeloid cells, higher concentrations activate T cells, promote IFNγ-dependent myelopoiesis and reduce platelet and erythrocyte counts^2,4,5,15. Early clinical trials in Crohn’s disease and psoriasis revealed a narrow therapeutic window, with dose-limiting anaemia and thrombocytopaenia from unintended T and B cell activation^5,14. Conversely, its ability to enhance cytotoxic T cell function and T helper type 1 (T_H1) cell differentiation has driven the development of pegylated IL-10 as a cancer immunotherapeutic for solid tumours¹⁶. However, IL-10’s limited therapeutic index, stemming from these opposing immunosuppressive and immunostimulatory effects plus haematological toxicities, remains a major obstacle^16,17,18. Previous attempts to engineer selective IL-10 variants have been constrained by short half-life, immunogenicity and incomplete separation of beneficial and harmful activities^14,19,20.

Activation of JAK family kinases requires receptor-mediated dimerization that enables transphosphorylation^21,22. Native IL-10 is a domain-swapped dimer^19,23 that assembles a (IL-10Rα)₂–(IL-10Rβ)₂ signalling complex, enabling JAK1 activation through both TYK2–JAK1 and JAK1–JAK1 transphosphorylation pathways^21,22,23 (Fig. 1a). JAK1 is required for IL-10Rα phosphorylation and downstream STAT3 activation⁹ (Extended Data Fig. 1a,b).

To dissect these signalling routes, we used a selective allosteric JAK1 inhibitor (VVD-118313), which targets mouse C816 and selectively blocks JAK1 transphosphorylation without inhibiting kinase domains²². In bone-marrow-derived macrophages (BMDMs), JAK1–JAK1 transphosphorylation inhibition by VVD-118313 partially reduced IL-10-induced STAT3 phosphorylation but did not abrogate functional responses, consistent with compensation through TYK2-mediated transphosphorylation (Fig. 1b and Extended Data Fig. 1c–e). In TYK2-deficient BMDMs, VVD-118313 completely abrogated STAT3 phosphorylation and IL-10 function, indicating that at least one transphosphorylation route, either JAK1–JAK1 or TYK2–JAK1, must remain intact for signalling competence. As a control, the ATP-competitive JAK1 kinase inhibitor (upadacitinib) fully blocked IL-10-mediated STAT3 phosphorylation and suppression of LPS-induced TNF and IL-6 (Fig. 1b and Extended Data Fig. 1c–e).

Consistent with these findings in mouse BMDMs, pharmacological inhibition of TYK2 with zasocitinib partially reduced IL-10-driven STAT3 phosphorylation in human monocytes (Fig.1c, Extended Data Fig. 1f and Supplementary Fig. 2) yet did not impair functional suppression of lipopolysaccharide (LPS)-induced TNF and IL-6 in primary blood mononuclear cells (PBMCs) (Fig. 1d). As a control, zasocitinib fully blocked IL-23- and IFNα-induced STAT3 phosphorylation and downstream signalling (Extended Data Fig. 2a). CRISPR-mediated TYK2 knockout in human macrophages produced similar results (Extended Data Fig. 2b,c), confirming that TYK2 is dispensable for IL-10-mediated TNF and IL-6 inhibition in macrophages²⁴.

By contrast, both TYK2 inhibitor and CRISPR knockout diminished IL-10-induced STAT3 phosphorylation in CD8⁺ T cells (Fig. 1e,f and Extended Data Figs. 1g and 2d) and completely abolished augmentation of granzyme B and IFNγ production (Fig. 1g,h), indicating a strict dependence on TYK2 in T cells.

We next tested whether tuning signalling amplitude could exploit this differential kinase dependence. An engineered IL-10 monomer¹⁹ that engages a single IL-10Rα–IL-10Rβ heterodimer (Fig. 1a) restricts signalling to a single TYK2–JAK1 transphosphorylation route (Extended Data Fig. 2e–h). The monomer induced around 50% maximal STAT3 phosphorylation in blood monocytes yet retained substantial TNF suppression and partial IL-6 suppression (Fig. 1i,j). In T cells, the monomer did not induce STAT3 phosphorylation (Fig. 1i, Extended Data Fig. 1g and Supplementary Fig. 2) or increase IFNγ and granzyme B production (Fig. 1k). These data show distinct thresholds across cell types: myeloid cells tolerate attenuated input, whereas T cells require maximal signalling, enabling natural myeloid selectivity with attenuated agonists.

Collectively, these observations indicate that the natural immune cell-type-specific dependence on TYK2, combined with tuning signalling strength, can be exploited to generate selective IL-10 agonists that uncouple myeloid immunosuppression from stimulation of T cells.

Although attenuated ligands such as IL-10 monomer and affinity-tuned IL-10 muteins provide an in vitro proof of concept for myeloid selectivity²⁰, their therapeutic potential is constrained by susceptibility to naturally occurring anti-IL-10 autoantibodies²⁵, thermal instability, reduced potency for TNF and IL-6 suppression (Fig. 1j), and residual T and B cell activation²⁰. To overcome these constraints and enable precise control of IL-10R signalling, we pursued a single-domain antibody (nanobody) approach. Nanobodies are well suited for receptor agonism owing to their small size (12–14 kDa), high affinity and favourable developability properties²⁶.

Alpacas were immunized with the human IL-10Rα extracellular domain (ECD), IL-10Rβ ECD or an IL-10Rα–IL-10Rβ heterodimeric Fc fusion protein (Fig. 2a). Variable heavy domain of heavy chain (VHH) repertoires derived from B cells were used to construct phage libraries, from which diverse anti-IL-10Rα and anti-IL-10Rβ VHH families were isolated (Supplementary Tables 1 and 2). VHH pairs were formatted as dimeric 6×His-tagged or tetrameric Fc fusion proteins, with either anti-IL-10Rα or anti-IL-10Rβ positioned at the N terminus and screened with an IL-10Rα–IL-10Rβ-expressing STAT3 reporter cell line (Fig. 2b).

Among 882 combinations tested, 182 constructs exhibited STAT3 reporter activities and 21 molecules achieved over 90% TNF and over 85% IL-6 inhibition in LPS-stimulated human PBMCs (Supplementary Fig. 3). However, most tetravalent IL-10R agonistic nanobodies recapitulated some degree of the T cell stimulatory profile of native IL-10, as evidenced by increased IFNγ and granzyme B secretion by CD8⁺ T cells and exemplified by a pan-IL-10R agonist (Supplementary Fig. 4), highlighting the challenge of achieving functional selectivity.

Selectivity and potency were highly clone specific. One tetravalent construct (Fig. 2c) exhibited picomolar potency together with near-complete myeloid selectivity: half-maximum inhibitory concentration (IC₅₀) of 2.1 pM for LPS-induced TNF and 6.9 pM for IL-6 in PBMCs, with no significant induction of IFNγ or granzyme B in T cells or B cell proliferation (Fig. 2d,e and Extended Data Fig. 3a–d). We designated this molecule as the myeloid-selective IL-10R agonist (hereafter, msIL-10R agonist). Surface plasmon resonance (SPR) analysis confirmed high-affinity binding to human IL-10Rα (K_D = 2.8 nM) and IL-10Rβ (K_D = 1.1 nM) with substantially higher affinity for IL-10Rβ relative to native IL-10 (Supplementary Fig. 5 and Supplementary Table 3).

The msIL-10R agonist demonstrated improved lymphocyte selectivity compared with affinity-detuned, myeloid-biased IL-10 muteins²⁰ (Extended Data Fig. 3a–d). Whereas the IL-10 muteins 10DE and D25K retained approximately 30% maximal STAT3 phosphorylation in T cells and induced IFNγ production²⁰, the msIL-10R agonist elicited minimal signalling (<10% E_max) in CD4⁺ T cells and CD8⁺ T cells (Fig. 2f, Extended Data Fig. 3a–c and Supplementary Fig. 6) and did not induce the cytotoxic effector program (Fig. 2e and Extended Data Fig. 3c). Furthermore, in contrast to IL-10 and the IL-10 muteins, the msIL-10R agonist exhibited minimal phosphorylation of STAT3 (pSTAT3) in B cells (Extended Data Fig. 3a,b) and did not promote B cell expansion or plasmablast differentiation (Extended Data Fig. 3d), demonstrating improved selectivity across lymphocyte populations.

Transcriptional profiling by RNA-sequencing (RNA-seq) showed highly correlated transcriptional responses to IL-10, IL-10–Fc and the msIL-10R agonist in resting and LPS-stimulated PBMCs, including comparable suppression of inflammatory genes such as CSF3, IL6 and PTGES (Fig. 2g and Supplementary Fig. 7). By contrast, only IL-10 and IL-10–Fc induced cytotoxic T cell-associated genes (such as GZMB and BCL6) in activated T cell cultures, whereas the msIL-10R agonist induced no significantly differentially expressed genes in T cells (Fig. 2h and Supplementary Fig. 8).

Together, through empirical high-throughput screening of approximately 1,000 tetravalent nanobody assemblies, we identified an msIL-10R agonist that recapitulates IL-10’s anti-inflammatory program in myeloid cells while uncoupling its immunostimulatory effects in lymphocytes.

The msIL-10R agonist comprises tandem VHH domains fused to an Fc scaffold, with an anti-IL-10Rα VHH (clone E07) linked to an anti-IL-10Rβ VHH (clone A05) (hereafter, E07-A05) in each arm (Fig. 2c). To define its mode of receptor engagement, we determined the cryo-electron microscopy (cryo-EM) structure of IL-10Rα ECD and IL-10Rβ ECD in a complex with E07-A05 tandem VHH.

Owing to the small size of the receptor–VHH complex (around 78 kDa) for cryo-EM, we used a recently developed anti-VHH fiducial Fab (NabFab; Fab 1) to facilitate particle alignment and reconstruction²⁷. NabFab recognizes an engineered binding epitope within VHH A05 distal to the E07-A05 linker and all CDRs and does not affect biological activities (Extended Data Fig. 4 and Supplementary Fig. 9). The cryo-EM map of the IL-10Rα–IL-10Rβ–E07-A05 complex was resolved to an overall resolution of 3.14 Å (Fig. 3a and Supplementary Fig. 10). The local resolution ranges from 2.8 Å at the IL-10Rβ–VHH A05 interface to about 4 Å at the IL-10Rα–VHH E07 interface, indicating conformational flexibility between the receptor-binding modules. To further define the E07 epitope, we determined a second cryo-EM structure of deglycosylated IL-10Rα ECD in a complex with E07 and an anti-VHH Fab (Fab 2) at a resolution of 2.40 Å (Fig. 3b and Supplementary Fig. 11). Integration of these datasets yielded a composite molecular model of the complete receptor-bound assembly (Fig. 3c).

Comparison with receptor complexes formed by IL-10, IL-22 and IFNλ reveals a distinct receptor organization induced by the msIL-10R agonist^23,28,29 (Fig. 3c,d and Extended Data Fig. 5). VHH A05 binds to an epitope on IL-10Rβ that partially overlaps the cytokine-binding surface used by IL-10 family cytokines. However, A05 binding induces a local conformational rearrangement of the Asn138–Trp143 loop of IL-10Rβ, positioning Trp143 and Tyr140 into a cleft formed by CDR2 and CDR3 of VHH A05 (Fig. 3e). Additional contacts involving CDR2 and CDR3 generate an extensive interaction network that stabilizes receptor engagement (Fig. 3e).

By contrast, VHH E07 recognizes a large surface spanning the junction between the tandem fibronectin type III domains of IL-10Rα (Fig. 3f). This epitope is situated on the opposite side of the IL-10-binding site and is not in contact with known disease-associated IL10RA mutations linked to IBD³⁰ (Fig. 3g). Consistent with this observation, disease-associated IL-10-binding interface mutations in IL-10Rα (R117C, R117H and Y64C) markedly impaired TNF suppression by native IL-10 and, to an even greater extent, by affinity-detuned IL-10 muteins (Fig. 3h). By contrast, the msIL-10R agonist retained full activity in cells expressing these disease-associated IL10RA variants (Fig. 3h).

To define the structural determinants for agonism, we selectively disrupted receptor engagement by introducing binding-disrupting mutations into either E07 or A05. Loss of IL-10Rα or IL-10Rβ binding, whether in cis or trans, markedly reduced STAT3 phosphorylation and impaired suppression of proinflammatory cytokines, including TNF, IL-6 and IL-1β (Extended Data Fig. 6a–c). Maximal potency and efficacy required simultaneous cross-linking of two IL-10Rα with either one or two IL-10Rβ molecules (Extended Data Fig. 6a–c), demonstrating that the tetravalent structure is essential for full agonist activity.

Importantly, agonism and myeloid selectivity were highly sensitive to molecular geometry. Altering the linker length between E07 and A05, either shortening or extending the GS linker, substantially reduced activity (Extended Data Fig. 6d). Likewise, reversing the topology of the tandem VHHs completely abolished signalling. Whereas the lead E07-A05/E07-A05 construct functioned as a potent myeloid-selective IL-10R agonist, the topologically inverted A05-E07/A05-E07 construct displayed no detectable activity in the STAT3 reporter assay (Extended Data Fig. 6e). Notably, constructs with symmetric architectures (A05-A05/E07-E07) increased pSTAT3 activity and enhanced myeloid activity, but also restored T cell activation, inducing approximately 20% of the IFNγ response elicited by native IL-10 (Extended Data Fig. 6e–g). These results indicate that signalling output and cellular selectivity are governed not simply by receptor occupancy, but by the precise topology and geometry of receptor cross-linking.

Taken together, these structural and functional studies demonstrate that the msIL-10R agonist assembles IL-10Rα and IL-10Rβ into a receptor geometry distinct from that induced by native IL-10 family cytokines. This unique tetravalent assembly provides the structural foundation for the potent myeloid-selective signalling properties of the msIL-10R agonist.

We next investigated the molecular basis for myeloid selectivity of the msIL-10R agonist. In human macrophages, the msIL-10R agonist induced TYK2, JAK1 and STAT3 phosphorylation, with more sustained TYK2 phosphorylation than IL-10 at the 2 h timepoint (Fig. 4a and Supplementary Fig. 12). By contrast, in T cells, the msIL-10R agonist elicited negligible JAK1 or TYK2 phosphorylation and substantially reduced STAT3 activation, as assessed by both immunoblotting (Fig. 4b and Supplementary Fig. 13) and flow cytometry (Fig. 2g and Supplementary Fig. 8).

Notably, signalling induced by the msIL-10R agonist was strictly TYK2 dependent. Genetic deletion or pharmacological inhibition of TYK2 completely abolished STAT3 phosphorylation and anti-inflammatory activity in macrophages (Fig. 4a,c–e). This dependency was further validated in a STAT3–luciferase reporter system, whereby msIL-10R agonist activity required the presence of TYK2 (Fig. 4f). Notably, non-myeloid-selective tetravalent IL-10R agonists, including the pan-IL-10R agonist and A05-A05/E07-E07 construct, retained partial STAT3 phosphorylation in TYK2-knockout cells (Fig. 4f), highlighting the unique TYK2 reliance by the msIL-10R agonist. Similarly, in BMDMs from human IL10RA/IL10RB-knock-in mice, TYK2 inhibition abrogated suppression of TNF and IL-6, whereas JAK1 transphosphorylation inhibitor VVD-118313 had minimal effects (Extended Data Fig. 7a–d). Together, these results indicate that the distinct receptor cross-linking geometry imposed by the msIL-10R agonist impairs productive JAK1–JAK1 transphosphorylation, rendering signalling propagation strictly dependent on TYK2 kinase activity. This mechanism provides a molecular explanation for both its reduced signalling amplitude and its myeloid selectivity.

Myeloid cells (monocytes, macrophages, DCs) express 2- to 18-fold higher surface copies of both IL-10Rα and IL-10Rβ than do T, B or natural killer (NK) cells (Fig. 4g). To determine whether receptor abundance governs msIL-10R agonist selectivity, we ectopically expressed IL-10Rα and/or IL-10Rβ in activated CD8⁺ T cells. Increasing IL-10Rα expression to levels comparable to those observed on human monocytes (Extended Data Fig. 8a) substantially restored STAT3 phosphorylation induced by msIL-10R agonist or IL-10 monomer (Fig. 4h and Extended Data Fig. 8b–e). Co-expression of both IL-10Rα and IL-10Rβ further enhanced granzyme B production by msIL-10R agonist relative to non-overexpressed control (Fig. 4h and Extended Data Fig. 8b,c). Overexpression of IL-10Rβ, TYK2 or JAK1 individually was insufficient (Extended Data Fig. 8b–i). Thus, differential IL-10Rα and IL-10Rβ copy number is the critical determinant of myeloid-selective IL-10R agonism.

We reasoned that, if receptor abundance dictates selectivity, then pairing IL-10Rα with a receptor enriched on lymphocytes should redirect STAT3 agonist signalling toward T cells. IL-2Rγ (common γ chain) is expressed around tenfold higher on T and NK cells than on myeloid cells³¹. We therefore generated a panel of bispecific nanobodies targeting IL-10Rα and IL-2Rγ. Two lead constructs, termed T/NK IL-10R agonist, preferentially induced STAT3 and STAT1, but not STAT5, phosphorylation in CD8⁺ T and NK cells³², promoted granzyme B production in T cells and exhibited minimal activity in monocytes (Fig. 4i,j and Extended Data Fig. 9a–c). Similar to the msIL-10R agonist, activity was highly dependent on receptor geometry, as a symmetric tandem VHH topology completely abolished signalling despite preserving receptor-binding affinity (Extended Data Fig. 9d–f). Taken together, these data demonstrate that cell-type-specific IL-10R agonism is dictated by the relative abundance and pairing of target receptors, enabling selectivity through an engineered bispecific nanobody.

IL-10 limits tissue inflammation primarily by inhibiting cytokine production and DC maturation^11,13,33. The msIL-10R agonist fully recapitulated these activities, potently blocking GM-CSF–driven monocyte into CD1a⁺ DCs (Fig. 5a) and suppressing zymosan-stimulated TNF and IL-23 release from DCs (Fig. 5b). In antigen-presentation assays, pretreatment of DCs with the msIL-10R agonist or IL-10–Fc strongly impaired expansion of influenza HA–tetramer⁺ T cells after peptide pulsing (Fig. 5c). The agonist also mirrored native IL-10 in attenuating immune-complex-triggered TNF production through FcγR signalling in macrophages (Fig. 5d). Thus, these data show that the msIL-10R agonist potently suppresses DC differentiation, cytokine secretion and T cell priming, and macrophage proinflammatory cytokine production.

As the msIL-10R agonist lacks cross-reactivity with mouse IL-10 receptors, we evaluated its in vivo efficacy in human IL10RA/IL10RB-knock-in mice. In an LPS challenge model, msIL-10R agonist and IL-10–Fc equivalently suppressed systemic TNF, IL-6 and IL-12 production³⁴ (Fig. 5e). In a keyhole limpet haemocyanin (KLH)/alum-induced delayed-type hypersensitivity model, the msIL-10R agonist reduced ear swelling (Fig. 5f), reduced neutrophil infiltration in draining lymph nodes (Fig. 5g) and attenuated KLH-specific IFNγ and TNF recall responses ex vivo (Fig. 5h). Similarly, in dextran sulfate sodium (DSS)-induced colitis (Fig. 5i), the msIL-10R agonist improved the colon weight/length ratio (Fig. 5j), attenuated systemic neutrophil expansion (Fig. 5k) and reduced tissue expression of proinflammatory genes including Tnf, Il1b and Csf3r (Fig. 5l). Collectively, these results demonstrate that the msIL-10R agonist delivers myeloid-restricted anti-inflammatory activity across multiple humanized in vivo models, faithfully recapitulating the immunoregulatory function of native IL-10.

Native IL-10 has a short serum half-life (2.7–4.5 h in humans), and long-acting variants such as PEGylated IL-10 or IL-10 Fc fusions extend exposure but drive undesirable T cell activation and haematological toxicities including anaemia/thrombocytopenia in preclinical and clinical studies^2,4,6. In cynomolgus monkeys, a fully humanized version of the msIL-10R agonist exhibited dose-proportional exposure, sustained receptor occupancy on blood monocytes and a terminal half-life of approximately 8 days (Extended Data Fig. 10a,b). Ex vivo LPS challenge of whole blood from dosed animals confirmed durable suppression of TNF for more than 3 weeks after a single dose (Extended Data Fig. 10c). Importantly, whereas IL-10–Fc induced IL-18 (a reported biomarker for T cell activation and haematological toxicity in cynomolgus monkeys^4,35) at 0.25 mg kg⁻¹, and thrombocytopaenia and anaemia, the msIL-10R agonist showed no significant dose-dependent IL-18 elevation nor haematological abnormalities at doses up to 100 mg kg⁻¹ (Extended Data Fig. 10d,e).

Thus, the msIL-10R agonist achieves prolonged myeloid-selective IL-10R agonism with an excellent safety profile, eliminating the T cell stimulation and haematotoxicity observed with IL-10–Fc.

Here we describe a fully synthetic, myeloid-selective IL-10 receptor agonist that recapitulates the potent immunosuppressive activity of IL-10 in monocytes, macrophages and DCs while uncoupling the T cell stimulatory effects that have historically limited IL-10 therapeutics.

Our studies reveal a cell-type-specific requirement for TYK2 downstream of IL-10R signalling. Whereas myeloid cells retain anti-inflammatory responses despite TYK2 deficiency and partial attenuation of STAT3 signalling, T cells exhibit a much greater dependence on TYK2 activity and maximal pathway activation. We further demonstrate that pSTAT3 signalling strength can be finely tuned through nanobody-mediated engineering of receptor geometry, thereby enabling the development of the msIL-10R agonist. The tetravalent architecture of the msIL-10R agonist imposes a distinct mode of receptor cross-linking that renders signalling strictly TYK2 dependent and highly sensitive to receptor density. Myeloid cells express substantially higher surface levels of both IL-10Rα and IL-10Rβ, enabling efficient receptor clustering and robust downstream signalling. By contrast, lower receptor abundance on T and B cells appears insufficient to achieve the avidity threshold required for productive signalling. Although receptor abundance emerged as the determinant of selectivity, additional mechanisms may contribute. T cells may possess higher signalling thresholds or enhanced negative regulatory mechanisms, including phosphatases that attenuate IL-10R signalling³⁶, thereby further restricting responses to attenuated IL-10R agonists. Together, these findings suggest that receptor copy number and cell-intrinsic signalling thresholds cooperate to establish the natural myeloid bias observed with attenuated IL-10R signalling and provide a mechanistic framework for engineering cell-type-selective cytokine agonists³⁷.

Previous efforts to achieve selective IL-10 signalling primarily relied on affinity tuning through partial disruption of IL-10Rβ engagement²⁰. Although these approaches demonstrated a proof of concept for in vitro functional selectivity, residual STAT3 signalling and T cell activation were consistently observed²⁰ (Extended Data Fig. 3). By contrast, the msIL-10R agonist achieves improved cell type selectivity between myeloid and lymphoid populations through engineered receptor geometry rather than cytokine mutagenesis, which is constrained by the native cytokine interaction interfaces. The structural basis of receptor engagement also offers advantages in disease settings in which native IL-10 signalling is compromised. A substantial fraction of patients with very-early-onset IBD have IL10RA mutations that impair ligand–receptor interactions. As the msIL-10R agonist engages epitopes spatially distinct from the native cytokine-binding interface, signalling remained preserved across multiple pathogenic IL10RA variants examined in this study (Fig. 3h). Moreover, the fully synthetic nanobody scaffold lacks sequence homology to IL-10 and would be expected to resist neutralization by endogenous anti-IL-10 autoantibodies reported in IBD²⁵.

In summary, we identify a cell-type-specific requirement for TYK2 in IL-10R signalling that arises from differential receptor abundance and signalling thresholds across immune cell populations. By exploiting these natural differences, we engineered a precision IL-10 therapeutic that preserves full anti-inflammatory potency while substantially reducing historical on-target liabilities. The msIL-10R agonist therefore represents a differentiated therapeutic candidate for IBD and other myeloid-driven inflammatory diseases, while the underlying engineering principles provide a generalizable blueprint for cell-type-specific cytokine agonists and antagonists across immunology and oncology.

Alpaca immunization and VHH immune phage library biopanning

Six alpacas were immunized subcutaneously with the ECD of hIL-10Rα, hIL-10Rβ or a hIL-10Rα–hIL-10Rβ heterodimer–hIgG1 Fc fusion (1 mg, five times at 2 week intervals; Lampire Biological Laboratories). All procedures were performed by Lampire under IACUC-approved protocols and applicable animal welfare regulations. No statistical methods were used to predetermine the sample size because the study was intended to isolate the antibody rather than compare groups. Animals were not randomized, and investigators were not blinded. Serum titre values were evaluated by IL-10Rα and IL-10Rβ enzyme-linked immunosorbent assay (ELISA) on days 35 and 49. Peripheral blood lymphocytes were isolated on days 49 and 63, RNA was extracted (Qiagen, 74104) and converted to cDNA, and VHH regions (~400 bp) were amplified by two-step PCR. The VHH product and pMES4 phagemid vector (Creative Biogene, OVT3038) were digested, ligated and transformed into Escherichia coli TG1 cells by electroporation; the library size was estimated by colony counting. The TG1 library was infected with M13K07 helper phage (Antibody Design lab, PH010L) to generate VHH-presenting phages, and rescue with polyethylene glycol precipitation was performed as described previously³⁸. Phage libraries were panned against hIL-10Rα or hIL-10Rβ immobilized on MaxiSorp plates (Thermo Fisher Scientific, 439454); non-specific phage was washed away and bound phage was eluted at low pH, neutralized and used to infect TG1 cells for subsequent rounds. Enrichment was monitored by comparing eluted phage titre from antigen-coated versus uncoated wells, and stringency was increased each round by adding washes and alternating blocking reagents. Enriched phages were plated for single-colony phage ELISA.

Generation of humanized and knockout mouse lines

Double-humanized B-hIL10RA/hIL10RB mice (Biocytogen, 111871) were obtained by intercrossing B-hIL10RA and B-hIL10RB mice, both generated on a C57BL/6 background by homologous recombination. B-hIL10RA mice express a chimeric receptor in which mouse Il10ra exons 1–6 (signal peptide, ECD, transmembrane domain) were replaced with the human sequences while retaining the mouse cytoplasmic domain; B-hIL10RB mice similarly carry humanized Il10rb exons 2–6 (Supplementary Figs. 14 and 15). Tyk2-deficient mice were generated by Taconic Biosciences using CRISPR–Cas9 editing (deletion of exons 4–23) on the C57BL/6NTac background.

Protein expression and purification

Designed VHH molecules were codon-optimized, synthesized and cloned as bivalent VHH-hIgG1 with a short Gly-Ser linker. Plasmids were transiently transfected using the Expi293 or ExpiCHO systems (Thermo Fisher Scientific, A14635, A29133), and VHH-Fc constructs were purified by protein A affinity chromatography (5 ml HiTrap MabSelect-SuRe column, Cytiva, 11003495) followed by size-exclusion chromatography (SEC) (HiLoad 16/600 Superdex 200 pg column, Cytiva, 28989335); purity was assessed by analytical SEC, SDS–PAGE and mass spectrometry. The tandem VHH E07-A05, individual VHHs, anti-VHH tool Fab (Fab 1; PDB: 7PIJ) and framework-mutant variants were expressed and purified similarly. Null-binding variants were generated by introducing CDR mutations into VHH E07 or VHH A05 to abolish binding to IL-10Rα and IL-10Rβ, respectively. Human IL-10Rα and IL-10Rβ ECDs were expressed with C-terminal 8×His or hIgG1 Fc tags and purified by Ni-NTA or protein A chromatography followed by SEC. To improve cryo-EM map quality, IL-10Rα ECD was partially deglycosylated with PNGase F (37 °C, 1 h) and repurified by SEC. Complexes containing IL-10Rα ECD, IL-10Rβ ECD, tandem VHH E07-A05 and tool Fab were assembled, analysed by SEC-HPLC in 25 mM HEPES pH 7.5/150 mM NaCl. IL-10 muteins (10DE, D25K, E96K) were cloned into the PB510B mammalian expression vector (SBI System biosciences) containing an N-terminal HSA signal peptide and a C-terminal 6×His tag, and expressed in TurboCHO cells (GenScript) and purified by AmMag Ni magnetic beads (GenScript, L00776-4) followed by SEC (HiLoad 16/600 Superdex 75 pg column, Cytiva) in 25 mM HEPES pH 7.4/150 mM NaCl. Detailed protein and antibody sequence information is provided in the Supplementary Table 4.

Human PBMC stimulation

PBMCs from healthy donors were stimulated with 10 ng ml⁻¹ LPS in the presence of PBS, recombinant human IL-10 (Peprotech, 200-10) or msIL-10R agonist (0.005–1,000 nM). For TYK2 inhibitor experiments, cells were pretreated with DMSO or zasocitinib (MedChemExpress, HY-150096, 1 μM) before stimulation. After 24 h, supernatant TNF, IL-6 and IL-1β were measured using the Meso Scale Discovery (MSD, K15049D) assay and analysed in GraphPad Prism.

MDM differentiation and CRISPR editing

Monocytes were isolated from PBMCs by negative selection (StemCell Technologies, 19359) and differentiated into MDMs in M-CSF–supplemented RPMI for at least 6 days. For TYK2 knockout, freshly isolated monocytes were nucleofected with Cas9–guide RNA ribonucleoprotein (RNP) complexes (Lonza 4D Nucleofector) and differentiated for 6 days before downstream phosphoflow or cytokine assays. RNPs were generated by combining 2 μl gRNA (100 μM) and 2 μl of Cas9 (20 μM) (EditCo) in supplemented P3 nucleofection buffer (Lonza, V4SP-3096). The solution was incubated at room temperature for 20 min. In total, 10⁶ monocytes were resuspended in 20 μl of pre-formed RNPs in P3 buffer and nucleofected using Lonza 4D Nucleofector using pulse code CM-137.

T cell isolation and in vitro T cell assay

Human bulk or CD8⁺ T cells were isolated from PBMCs by negative selection (>99% purity, StemCell Technologies, 17951, 17953) and activated with plate-bound 10 μg ml⁻¹ anti-CD3 and soluble 5 μg ml⁻¹ anti-CD28 for 3 days. T cells were then cultured with or without IL-10, msIL-10R agonist, IL-10 monomer or IL-10 muteins for 3 additional days, restimulated with anti-CD3/anti-CD28 (5 μg ml⁻¹ each) overnight and assessed for secreted IFNγ and granzyme B by ELISA and MSD (R&D, DIF50C, DGZB00; MSD, K251TTK) or by intracellular flow cytometry.

CRISPR–Cas9 gene editing in T cells and HEK293T cells

Activated human T or CD8⁺ T cells were nucleofected with Cas9–gRNA RNP complexes and poly-l-glutamic acid using a 4D Nucleofector (Lonza) and expanded for 3 days before use. RNP complexes were generated by incubating guide RNA (EditCo Bio) with recombinant Cas9 protein (IDT, 10007803) for 20 min at room temperature. Cells were electroporated using the EO115 pulse program. To generate TYK2-knockout HEK293T IL-10 pSTAT3 reporter cells, Cas9–RNP complexes targeting TYK2 were assembled using a predesigned human TYK2 gene knockout kit (EditCo) and delivered by electroporation using the Neon Transfection System (Thermo Fisher Scientific). After recovery, single-cell clones were isolated by flow cytometry single-cell sorting and expanded. TYK2 knockout was confirmed by DNA-seq.

Lentiviral transduction

Human CD8⁺ T cells were activated with anti-CD3/anti-CD28 Dynabeads overnight, debeaded and transduced with concentrated lentivirus encoding IL10RA-mCherry, IL10RB-GFP, JAK1-GFP or TYK2-mCherry (cloned into the GLV3 backbone under the EF1A promoter with C-terminal P2A-mCherry/EGFP reporters) by spin-infection on retronectin-coated plates (100 μg ml⁻¹, Takara, T100B). Cells were rested 2–3 days before functional assays. The gating strategy for transduced T cells is shown in Supplementary Fig. 16.

pSTAT3 and pSTAT1 staining

After stimulation, PBMCs, MDMs or T cells were surface stained, fixed, permeabilized with ice-cold methanol and stained with anti-pSTAT3 (pY705) and anti-pSTAT1 (pY701). Cells were analysed by flow cytometry (LSR Fortessa) and quantified in FlowJo. The gating strategy is shown in Supplementary Fig. 17.

Copy numbers of IL-10Rα and IL-10Rβ

PBMCs, macrophages (detached from UpCell plates with cold PBS) and monocyte-derived DCs (CellXVivo, CDK004) were stained with lineage markers and titrated anti-IL-10Rα and anti-IL-10Rβ antibodies with matched isotype controls. The delta geometric mean fluorescence intensity (gMFI) was calculated as the sample gMFI minus the isotype gMFI. Absolute receptor copy numbers were determined using Quantum Simply Cellular anti-rat and anti-mouse IgG bead standards (Bangs Laboratories, 817B, 815B). Antibody details are provided in the Reporting Summary.

Immunoblot analyses

Macrophages or T cells were serum-starved, pretreated with DMSO, JAK1 inhibitor upadacitinib (10 µM) or TYK2 inhibitor zasocitinib (1 µM), and stimulated with or without IL-10 or msIL-10R agonist (10 nM). The lysates were resolved on NuPAGE 4–12% Bis-Tris gels, transferred to nitrocellulose and probed for phosphorylated and total TYK2, JAK1, STAT3, STAT1 and GAPDH. Chemiluminescence was detected on the ChemiDoc MP system and quantified in Image Lab, with brightness/contrast adjustments applied uniformly across all groups. pTYK2, pJAK1 and pSTAT3 were normalized to total TYK2 and expressed as the fold change relative to the unstimulated control. Antibody details are provided in the Reporting Summary.

IL-10Rα–IL-10Rβ pSTAT3 reporter assay

An HEK293T reporter line stably expressing human IL10RA, IL10RB and a STAT3-responsive firefly luciferase reporter was used to measure IL-10 receptor agonist activity. Reporter cells were incubated with eight-point dose–responses of test molecules or recombinant IL-10 (positive control) for 3 h, and luminescence was measured using an EnVision reader (Revvity) after Bio-Glo addition (Promega, G7941). The half-maximum effective concentration and E_max were calculated by four-parameter logistic regression on the IL-10-normalized signal.

Inhibition of LPS-mediated PBMC cytokine secretion

Thawed PBMCs were pre-incubated with eight-point dose–responses of IL-10 or antibodies for 30 min, stimulated with LPS (1 ng ml⁻¹) for 20 h, and supernatant IL-6 and TNF were measured by time-resolved Förster resonance energy transfer (Revvity, TRF1223C) and MSD (K151UCK), respectively. Data were normalized to positive/negative controls and fitted by four-parameter logistic regression.

Evaluation of the IL-10R agonist activity on primary T cells

CD8⁺ T cells from three donors were expanded for 3 days with anti-CD3/anti-CD28, incubated with eight-point antibody dose–responses for 3 days, restimulated with anti-CD3/anti-CD28 for 6 h and the supernatant IFNγ was measured by MSD (K151TTK). Data were normalized and fitted by four-parameter logistic regression.

RAW264.7 knockout and IL10RA-mutant cell line generation

Endogenous Il10ra and Il10rb were disrupted in RAW264.7 cells by CRISPR–Cas9 nucleofection using a Lonza 4D Nucleofector with pulse code DS-136, and single-cell knockout clones were validated by genomic sequencing and flow cytometry. IL-10R-deficient cells were then reconstituted by lentiviral expression of chimeric receptors (human IL-10Rα/IL-10Rβ ECDs fused to the corresponding mouse intracellular domains, linked by P2A), including IBD-associated IL-10Rα extracellular mutants. To assess anti-inflammatory activity, cells were stimulated with LPS (10 ng ml⁻¹) with or without IL-10, IL-10 muteins (10DE, D25K) or msIL-10R agonist (0.1–10 nM) for 6 h with monensin, and intracellular TNF was measured by flow cytometry. For immune complex assays, RAW264.7 cells expressing human IL-10Rα and IL-10Rβ were cultured with mouse IgG2a antibody (10 μg ml⁻¹) and 10 nM isotype control, IL-10 Fc, or msIL-10R agonist for 1 h before addition of 10 μl Dynabeads Protein A to generate immune complexes. After 24 h, culture supernatants were harvested and cytokines quantified by MSD.

SPR measurement of binding kinetics

Antibody binding to human and cynomolgus macaque (cyno) IL-10Rα and IL-10Rβ was measured on the Biacore 8K system. The C1 chip (Cytiva VR100535) was preconditioned with glycine-NaOH/Triton X-100, activated with s-NHS/EDC chemistry, and anti-human Fc (Southern Biotech, 2014-1) was amine-coupled in sodium acetate (pH 5.0) to a final density of around 500 RU for antibody capture through the Fc tag; excess activated carboxyls were quenched with ethanolamine. Antibodies were captured at 50 nM to around 50–60 RU per channel. After ten blank buffer injections for stabilization, human and cyno IL-10Rα were injected as a threefold concentration series (150 nM top) and IL-10Rβ as a threefold series (1,000 nM top), using 180 s association and 30 min dissociation at 25 °C in 10 mM HEPES pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.005% Tween-20 and 0.2 mg ml⁻¹ BSA, with 10 mM glycine (pH 1.5) regeneration between cycles. Data were fit to a simple 1:1 kinetic model to derive k_on and k_off, from which K_D was calculated (K_D = k_off/k_on).

Cryo-EM sample preparation and data acquisition

To facilitate cryo-EM study of the IL-10Rα–IL-10Rβ–tandem VHH E07-A05 complex, framework mutations (Lys116/Pro119/Lys244/Pro247) were introduced into the tandem VHH so that a tool anti-VHH Fab (Fab 1; PDB: 7PIJ) could bind to and enlarge the particle. When IL-10Rα ECD, IL-10Rβ ECD, tandem VHH and tool Fab 1 were mixed at a 1:1:1:2 molar ratio (0.8 mg ml⁻¹, in 25 mM HEPES pH 7.5 and 150 mM NaCl), the Fab bound preferentially to the engineered site on A05 but heterogeneously to E07, of which the engineered Fab site is not solvent exposed in the complex (Extended Data Fig. 4). To improve homogeneity, IL-10Rα ECD, IL-10Rβ ECD and tandem VHH were pre-incubated on ice for 1 h before adding tool Fab 1 to a 1.2:1:1:1 final molar ratio, which markedly improved particle homogeneity in 2D and 3D classification (Supplementary Fig. 10). For the IL-10Rα–VHH E07 complex, partially deglycosylated IL-10Rα ECD, VHH E07 and a second anti-VHH Fab (Fab 2, GenScript, A02165) were mixed at a 1.2:1:1 molar ratio (1.3 mg ml⁻¹, in 25 mM HEPES pH 7.5 and 150 mM NaCl) and incubated on ice for 1 h. Deglycosylation significantly improved IL-10Rα density in the map. Complexes were cross-linked with 0.05% glutaraldehyde (15 min on ice in the dark), and a 3.5 µl aliquot was applied to a glow-discharged UltrAuFoil R1.2/1.3 grid (R1.2/1.3, Quantifoil), blotted in Vitrobot Mark IV (Thermo Fisher Scientific) (2 s, 4 °C, 100% humidity), and plunge-frozen in liquid ethane. Both untilted and stage-tilted (15–30°) data were acquired to improve angular sampling. Videos were collected in EPU (Thermo Fisher Scientific) on the Titan Krios system (300 kV) (Thermo Fisher Scientific) with a Falcon 4i direct electron detector (Thermo Fisher Scientific) at ×165,000 nominal magnification (0.729 Å per pixel), a defocus range of −0.6 to −2.2 µm, and a Selectris X energy filter (10 eV slit), over a 5.85 s exposure with 50 frames and a total dose of around 50 e⁻ Å⁻².

Cryo-EM image processing and model building

All data processing was performed in CryoSPARC (v.4)³⁹ with patch motion correction and patch contrast transfer function (CTF) estimation; images with CTF fit to 5.5 Å or better were retained. For the IL-10Rα–IL-10Rβ–tandem VHH E07-A05–anti-VHH Fab complex (Supplementary Fig. 10), 15,755 untilted videos (two sessions of 6,460 and 9,295) and 7,508 tilted (15–30°) videos were collected, processed separately and merged after 2D classification, yielding 2,730,139 particles used for iterative heterogeneous refinement, 3D classification and non-uniform refinement with local and global CTF refinement to give a 3.14 Å map (FSC = 0.143; 101,358 particles). For the IL-10Rα–VHH E07–anti-VHH Fab complex (Supplementary Fig. 11), 19,416 untilted videos (two sessions of 9,680 and 9,736) and 7,461 tilted (15–30°) videos yielded 1,215,819 particles used for ab initio reconstruction, heterogeneous refinement and non-uniform refinement with local and global CTF refinement, giving a 2.40 Å map (FSC = 0.143; 458,458 particles). The IL-10Rα structure (PDB: 1Y6K) and an AlphaFold⁴⁰ model of VHH E07 were docked into the 2.40 Å map by rigid-body fitting in ChimeraX⁴¹, manually adjusted in Coot⁴² and refined with phenix.real_space_refine⁴³. The refined IL-10Rα–VHH E07 model, together with IL-10Rβ (PDB 3LQM) and the anti-VHH tool Fab (PDB 7PIJ), served as initial models to refine against the 3.14 Å complex map. The Gly-Gly-Gly-Ser inter-VHH linker was not resolved and was omitted from model building. Structural figures were generated and validated using standard cryo-EM refinement metrics, with full processing pipelines, particle counts and map–model statistics provided in Supplementary Figs. 10 and 11 and Supplementary Table 5.

DSS-induced colitis model

Ten-week-old female humanized IL10RA/IL10RB mice received 2% DSS in drinking water for 7 days and were dosed three times weekly with isotype control, IL-10 Fc or msIL-10R agonist (1 nmol per mouse). Colon length and weight were measured on day 10 and colonic gene expression was quantified by TaqMan qPCR. Neutrophil gating is shown in Supplementary Fig. 18.

LPS-induced sepsis model

Six-week-old female humanized IL10RA/IL10RB mice received intraperitoneal LPS (E. coli O111:B4, 4 mg kg⁻¹) with a single dose of isotype control, human IL-10 Fc or msIL-10R agonist. Serum TNF, IL-6 and IL-12p70 were measured 4 h later by MSD (N05048A-1).

KLH-induced DTH sensitization recall model

Humanized IL10RA/IL10RB mice were sensitized subcutaneously with KLH/complete Freund’s adjuvant (Sigma, H7017; Chondrex, 7009), dosed intraperitoneally on day 4 with isotype, IL-10 Fc or msIL-10R agonist (2 mg kg⁻¹), rechallenged intradermally with KLH in the ear on day 5, and taken down on day 8. Draining lymph node cells were restimulated ex vivo with KLH, and supernatant cytokines were measured by MSD (N05048A-1).

RNA extraction, library generation and sequencing for RNA-seq analysis

Total RNA was extracted (Qiagen, 74104), quality-checked and used (500 ng) for TruSeq stranded mRNA library preparation and paired-end sequencing on an Illumina NextSeq2000. A Nextflow pipeline performed quality control (FastQC, RNA-SeQC2, MultiQC), alignment to GRCh38 with STAR and gene quantification with featureCounts^44,45. Differential expression used TMM normalization and edgeR (glmQLFit/glmQLFTest) with Benjamini–Hochberg FDR < 0.05.

DC differentiation and cytokine secretion

CD14⁺ monocytes were differentiated into DCs with IL-4 and GM-CSF for 6 days, with IL-10 Fc or msIL-10R agonist (10 nM) present throughout differentiation, and CD1a/CD14 expression was assessed by flow cytometry. The gating strategy is shown in Supplementary Fig. 17. For activation, DCs were stimulated with zymosan with or without IL-10–Fc or msIL-10R agonist molecules for 24 h, and TNF and IL-23 were measured by MSD.

DC antigen presentation to HA-specific CD4⁺ T cells

Monocyte-derived DCs were differentiated (R&D Systems, CDK004) and matured (IL-4, GM-CSF, TNF and LPS) in the presence of isotype control, IL-10 Fc or msIL-10R agonist (1 or 10 nM), then pulsed with influenza haemagglutinin HA307 peptide (Influenza A HA 307–319, DRA1*01:01/DRB1*04:01, PKYVKQNTLKLAT). HA307-reactive CD4⁺ T cells from an HLA-DRB1*04:01 donor were expanded, tetramer-sorted and co-cultured with peptide-pulsed DCs. Expansion of tetramer⁺ T cells was assessed by flow cytometry using the gating strategy shown in Supplementary Fig. 18.

Pharmacokinetic assessment in cynomolgus monkey

Plasma concentrations of msIL-10R agonist and IL-10 Fc were measured by an MSD electrochemiluminescence assay using anti-human Fc capture (Southern Biotech, 9048-08) and SULFO-TAG anti–IL-10Rα detection, with a lower limit of quantification of 15 ng ml⁻¹ and concentrations back-calculated from four-parameter logistic standard curves.

In vitro human B cell activation and plasmablast differentiation

Isolated human B cells (StemCell Technologies, 17954) were cultured with recombinant CD40 ligand (R&D, 6420-CLB, 1 μg ml⁻¹) and IL-10 (10 nM), fed on day 3, and collected on day 6. Plasmablasts (CD19⁺CD27⁺CD38⁺) were quantified by flow cytometry, and the gating strategy is shown in Supplementary Fig. 19.

BMDM differentiation, pSTAT3 signalling and cytokine-secretion assays

Bone marrow from wild-type or Tyk2-deficient mice was differentiated into BMDMs with M-CSF (Gibco 31502250UG) for 5–6 days. For pSTAT3 induction, BMDMs were pretreated with upadacitinib (Tocris, 7783, 10 µM) or the JAK1-selective inhibitor VVD-118313 (MedChemExpress, HY-151385, 100 nM) and stimulated with IL-10–Fc (10 nM); intracellular pSTAT3 was measured by flow cytometry. For cytokine secretion, BMDMs were pretreated with upadacitinib, VVD-118313 or zasocitinib (1 µM), stimulated with IL-10 Fc and LPS, and supernatant cytokines were measured by MSD.

Statistical analysis

All analyses were performed using two-tailed Student’s t-tests, log-normalized Welch’s t-tests or one-way/two-way ANOVA with correction for multiple comparisons (GraphPad Prism 8). All bar graphs show means with overlaid scatter and s.d. as indicated in each figure legend.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

1. Ouyang, W. & O’Garra, A. IL-10 family cytokines IL-10 and IL-22: from basic science to clinical translation. Immunity 50, 871–891 https://doi.org/10.1016/j.immuni.2019.03.020 (2019).

   Article  PubMed  CAS  Google Scholar 

2. Cardoso, A. et al. Interleukin-10 induces interferon-γ-dependent emergency myelopoiesis. Cell Rep. 37, 109887 https://doi.org/10.1016/j.celrep.2021.109887 (2021).

   Article  PubMed  CAS  Google Scholar 

3. Huhn, R. D. et al. Effects of single intravenous doses of recombinant human interleukin-10 on subsets of circulating leukocytes in humans. Immunopharmacology 41, 109–117 https://doi.org/10.1016/s0162-3109(98)00058-7 (1999).

   Article  PubMed  CAS  Google Scholar 

4. Yang, Z. et al. Application of pharmacokinetic/pharmacodynamic modeling to bridge mouse antitumor efficacy and monkey toxicology data for determining the therapeutic index of an interleukin-10 Fc fusion protein. Front. Pharmacol. 13, 829063 https://doi.org/10.3389/fphar.2022.829063 (2022).

   Article  PubMed  PubMed Central  CAS  Google Scholar 

5. Tilg, H. et al. Treatment of Crohn’s disease with recombinant human interleukin 10 induces the proinflammatory cytokine interferon gamma. Gut 50, 191–195 https://doi.org/10.1136/gut.50.2.191 (2002).

   Article  PubMed  PubMed Central  CAS  Google Scholar 

6. Saraiva, M., Vieira, P. & O’Garra, A. Biology and therapeutic potential of interleukin-10. J. Exp. Med. 217, e20190418 https://doi.org/10.1084/jem.20190418 (2020).

7. Glocker, E. O. et al. Inflammatory bowel disease and mutations affecting the interleukin-10 receptor. N. Engl. J. Med. 361, 2033–2045 https://doi.org/10.1056/NEJMoa0907206 (2009).

   Article  PubMed  PubMed Central  CAS  Google Scholar 

8. Weber-Nordt, R. M. et al. Stat3 recruitment by two distinct ligand-induced, tyrosine-phosphorylated docking sites in the interleukin-10 receptor intracellular domain. J. Biol. Chem. 271, 27954–27961 https://doi.org/10.1074/jbc.271.44.27954 (1996).

   Article  PubMed  CAS  Google Scholar 

9. Riley, J. K., Takeda, K., Akira, S. & Schreiber, R. D. Interleukin-10 receptor signaling through the JAK-STAT pathway. Requirement for two distinct receptor-derived signals for anti-inflammatory action. J. Biol. Chem. 274, 16513–16521 https://doi.org/10.1074/jbc.274.23.16513 (1999).

   Article  PubMed  CAS  Google Scholar 

10. Takeda, K. et al. Enhanced Th1 activity and development of chronic enterocolitis in mice devoid of Stat3 in macrophages and neutrophils. Immunity 10, 39–49 https://doi.org/10.1016/s1074-7613(00)80005-9 (1999).

    Article  PubMed  CAS  Google Scholar 

11. Fiorentino, D. F. et al. IL-10 acts on the antigen-presenting cell to inhibit cytokine production by Th1 cells. J. Immunol. 146, 3444–3451 (1991).

    Article  PubMed  CAS  Google Scholar 

12. Li, B. et al. IL-10 engages macrophages to shift Th17 cytokine dependency and pathogenicity during T-cell-mediated colitis. Nat. Commun. 6, 6131 https://doi.org/10.1038/ncomms7131 (2015).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

13. Zigmond, E. et al. Macrophage-restricted interleukin-10 receptor deficiency, but not IL-10 deficiency, causes severe spontaneous colitis. Immunity 40, 720–733 https://doi.org/10.1016/j.immuni.2014.03.012 (2014).

    Article  PubMed  CAS  Google Scholar 

14. Saxena, A. et al. Interleukin-10 paradox: a potent immunoregulatory cytokine that has been difficult to harness for immunotherapy. Cytokine 74, 27–34 https://doi.org/10.1016/j.cyto.2014.10.031 (2015).

    Article  PubMed  CAS  Google Scholar 

15. Tilg, H., Ulmer, H., Kaser, A. & Weiss, G. Role of IL-10 for induction of anemia during inflammation. J. Immunol. 169, 2204–2209 https://doi.org/10.4049/jimmunol.169.4.2204 (2002).

    Article  PubMed  CAS  Google Scholar 

16. Naing, A. et al. PEGylated IL-10 (pegilodecakin) induces systemic immune activation, CD8⁺T cell invigoration and polyclonal T cell expansion in cancer patients. Cancer Cell 34, 775–791 https://doi.org/10.1016/j.ccell.2018.10.007 (2018).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

17. Naing, A. et al. Pegilodecakin combined with pembrolizumab or nivolumab for patients with advanced solid tumours (IVY): a multicentre, multicohort, open-label, phase 1b trial. Lancet Oncol. 20, 1544–1555 https://doi.org/10.1016/S1470-2045(19)30514-5 (2019).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

18. Salkeni, M. A. & Naing, A. Interleukin-10 in cancer immunotherapy: from bench to bedside. Trends Cancer 9, 716–725 https://doi.org/10.1016/j.trecan.2023.05.003 (2023).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

19. Josephson, K. et al. Design and analysis of an engineered human interleukin-10 monomer. J. Biol. Chem. 275, 13552–13557 https://doi.org/10.1074/jbc.275.18.13552 (2000).

    Article  PubMed  CAS  Google Scholar 

20. Saxton, R. A. et al. Structure-based decoupling of the pro- and anti-inflammatory functions of interleukin-10. Science 371, eabc8433 https://doi.org/10.1126/science.abc8433 (2021).

21. Caveney, N. A. et al. Structural basis of Janus kinase trans-activation. Cell Rep. 42, 112201 https://doi.org/10.1016/j.celrep.2023.112201 (2023).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

22. Kavanagh, M. E. et al. Selective inhibitors of JAK1 targeting an isoform-restricted allosteric cysteine. Nat. Chem. Biol. 18, 1388–1398 https://doi.org/10.1038/s41589-022-01098-0 (2022).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

23. Josephson, K., Logsdon, N. J. & Walter, M. R. Crystal structure of the IL-10/IL-10R1 complex reveals a shared receptor binding site. Immunity 15, 35–46 https://doi.org/10.1016/s1074-7613(01)00169-8 (2001).

    Article  PubMed  CAS  Google Scholar 

24. Shimoda, K. et al. Tyk2 plays a restricted role in IFN alpha signaling, although it is required for IL-12-mediated T cell function. Immunity 13, 561–571 https://doi.org/10.1016/s1074-7613(00)00055-8 (2000).

    Article  PubMed  CAS  Google Scholar 

25. Griffin, H. et al. Neutralizing autoantibodies against interleukin-10 in inflammatory bowel disease. N. Engl. J. Med. 391, 434–441 https://doi.org/10.1056/NEJMoa2312302 (2024).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

26. Chan, A. C., Martyn, G. D. & Carter, P. J. Fifty years of monoclonals: the past, present and future of antibody therapeutics. Nat. Rev. Immunol. 25, 745–765 https://doi.org/10.1038/s41577-025-01207-9 (2025).

    Article  PubMed  CAS  Google Scholar 

27. Bloch, J. S. et al. Development of a universal nanobody-binding Fab module for fiducial-assisted cryo-EM studies of membrane proteins. Proc. Natl Acad. Sci. USA 118, e2115435118 https://doi.org/10.1073/pnas.2115435118 (2021).

28. Mendoza, J. L. et al. The IFN-λ-IFN-λR1-IL-10Rβ complex reveals structural features underlying type III IFN functional plasticity. Immunity 46, 379–392 https://doi.org/10.1016/j.immuni.2017.02.017 (2017).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

29. Saxton, R. A. et al. The tissue protective functions of interleukin-22 can be decoupled from pro-inflammatory actions through structure-based design. Immunity 54, 660–672 https://doi.org/10.1016/j.immuni.2021.03.008 (2021).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

30. Aschenbrenner, D. et al. Pathogenic interleukin-10 receptor alpha variants in humans—balancing natural selection and clinical implications. J. Clin. Immunol. 43, 495–511 https://doi.org/10.1007/s10875-022-01366-7 (2023).

    Article  PubMed  CAS  Google Scholar 

31. Leonard, W. J., Lin, J. X. & O’Shea, J. J. The γ_c family of cytokines: basic biology to therapeutic ramifications. Immunity 50, 832–850 https://doi.org/10.1016/j.immuni.2019.03.028 (2019).

    Article  PubMed  CAS  Google Scholar 

32. Cai, G., Kastelein, R. A. & Hunter, C. A. IL-10 enhances NK cell proliferation, cytotoxicity and production of IFN-γ when combined with IL-18. Eur. J. Immunol. 29, 2658–2665 (1999).

    Article  PubMed  CAS  Google Scholar 

33. Fiorentino, D. F., Zlotnik, A., Mosmann, T. R., Howard, M. & O’Garra, A. IL-10 inhibits cytokine production by activated macrophages. J. Immunol. 147, 3815–3822 (1991).

    Article  PubMed  CAS  Google Scholar 

34. Berg, D. J. et al. Interleukin-10 is a central regulator of the response to LPS in murine models of endotoxic shock and the Shwartzman reaction but not endotoxin tolerance. J. Clin. Invest. 96, 2339–2347 https://doi.org/10.1172/JCI118290 (1995).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

35. Iwai, Y. et al. An IFN-γ-IL-18 signaling loop accelerates memory CD8⁺ T cell proliferation. PLoS ONE 3, e2404 https://doi.org/10.1371/journal.pone.0002404 (2008).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

36. Zhao, Y. et al. Expanding the cytokine receptor alphabet reprograms T cells into diverse states. Nature 645, 1039–1050 https://doi.org/10.1038/s41586-025-09393-1 (2025).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

37. Yen, M. et al. Facile discovery of surrogate cytokine agonists. Cell 185, 1414–1430 https://doi.org/10.1016/j.cell.2022.02.025 (2022).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

38. Hust, M. & Lim, T. S. Phage Display: Methods and Protocols 2nd edn (Humana Press, 2023).

39. Punjani, A., Rubinstein, J. L., Fleet, D. J. & Brubaker, M. A. cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nat. Methods 14, 290–296 https://doi.org/10.1038/nmeth.4169 (2017).

    Article  PubMed  CAS  Google Scholar 

40. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 https://doi.org/10.1038/s41586-021-03819-2 (2021).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

41. Pettersen, E. F. et al. UCSF Chimera—a visualization system for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612 https://doi.org/10.1002/jcc.20084 (2004).

    Article  ADS  PubMed  CAS  Google Scholar 

42. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta Crystallogr. D 66, 486–501 https://doi.org/10.1107/S0907444910007493 (2010).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

43. Afonine, P. V. et al. Real-space refinement in PHENIX for cryo-EM and crystallography. Acta Crystallogr. D 74, 531–544 https://doi.org/10.1107/S2059798318006551 (2018).

    Article  ADS  CAS  Google Scholar 

44. Di Tommaso, P. et al. Nextflow enables reproducible computational workflows. Nat. Biotechnol. 35, 316–319 https://doi.org/10.1038/nbt.3820 (2017).

    Article  PubMed  CAS  Google Scholar 

45. Graubert, A., Aguet, F., Ravi, A., Ardlie, K. G. & Getz, G. RNA-SeQC 2: efficient RNA-seq quality control and quantification for large cohorts. Bioinformatics 37, 3048–3050 https://doi.org/10.1093/bioinformatics/btab135 (2021).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

Download references

Author notes

1. These authors contributed equally: D. Alejandro Canaria, Trang T. T. Nguyen, Huang Huang, Hui Xu, Lili Yue

Authors and Affiliations

1. Postdoctoral Program, Gilead Sciences, Foster City, CA, USA

   D. Alejandro Canaria, Cong Xing & Tangsheng Yi

2. Inflammation Research, Gilead Sciences, Foster City, CA, USA

   D. Alejandro Canaria, Trang T. T. Nguyen, Huang Huang, Lili Yue, Yongyao Fu, Cong Xing, Aron L. Martin, Zhiyu Huang, Meagan C. Main, Amanali Rodriguez, Xiaoli Lin, Maria L. Mouchess & Tangsheng Yi

3. Structural Biology & Chemistry, Gilead Sciences, Foster City, CA, USA

   Hui Xu

4. Discovery Sciences & Technologies, Gilead Sciences, Foster City, CA, USA

   Hoa Truong, Kerul Patel & Chen Chen

5. Protein Therapeutics, Gilead Sciences, Foster City, CA, USA

   Weimei Xing, George Delidakis, Sylvia Wong, Andrew Pae, Yu Yu, Ramsay Macdonald, Magdeleine Hung, Lei Chen, Nicole Schirle Oakdale & Chingwei V. Lee

6. Research Data Science, Gilead Sciences, Foster City, CA, USA

   Jayamary Divya Ravichandar

7. Non-Clinical Safety, Gilead Sciences, Foster City, CA, USA

   Ellen M. McGlinchey

8. Drug Metabolism, Gilead Sciences, Foster City, CA, USA

   Brian Carr & Megan L. Koleske

Contributions

H.H., C.V.L., C.C. and T.Y. conceptualized the idea and initiated the project. D.A.C., T.T.T.N., H.H., L.Y., C.X., A.L.M., Y.F., H.T., K.P., C.C. and T.Y. designed, performed and analysed the in vitro biology and biochemical experiments. H.T., K.P. and C.C. performed the screening and identification of antibodies. G.D., S.W., A.P., L.C., N.S.O. and C.V.L. performed antibody engineering and humanization. Y.Y., R.M., M.H. and W.X. performed protein purification. H.X. performed the cryo-EM experiment and structural analysis. Z.H., M.C.M., A.R., X.L., M.L.M. and T.Y. designed and performed in vivo experiments. E.M.M., B.C., L.Y., T.Y. and M.L.K. designed and performed analysis of non-human primate pharmacokinetic- and clinical-pharmacology-related experiments. H.H. and J.D.R. performed and analysed RNA-seq data. All of the authors approved and contributed to the editing and revision of the manuscript.

Corresponding authors

Correspondence to Chingwei V. Lee, Chen Chen or Tangsheng Yi.

Competing interests

All of the authors are current or former employees of Gilead Sciences and may hold stock or stock options in the company. C.C., L.C., G.D., H.H., M.L.K., C.V.L., N.S.O., H.X., T.Y. and L.Y. are listed as inventors on published patent WO2026112287 (Single immunoglobulin domain bispecific antibodies against IL-10R alpha and beta subunits and methods of use).

Peer review information

Nature thanks the anonymous reviewers for their contribution to the peer review of this work. Peer reviewer reports are available.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions