Thymic stromal lymphopoietin receptor antagonist

The present invention relates to a Thymic Stromal Lymphopoietin Receptor antagonist. More specifically, it relates to a mutant form of Thymic Stromal Lymphopoietin, capable of binding the receptor, but unable to induce signaling. The invention relates further to the use of the antagonist to treat inflammatory disorders.

Allergic diseases of the airways, skin, and gut such as asthma (Holgate, 2012), chronic obstructive pulmonary disease (COPD) (Decramer et al., 2012), atopic dermatitis (Schmitt et al., 2013), and eosinophilic esophagitis (Abonia and Rothernberg, 2012) afflict hundreds of millions of individuals worldwide thereby imposing a daunting and multifaceted healthcare and socioeconomic burden. Paradoxically, the cellular context behind the onset of such chronic inflammatory disorders centers on type 2 helper T cell (T _H 2)-mediated inflammatory responses primed by activated dendritic cells (DC), which in normal physiology orchestrate essential immune responses via T _H 1 , T _H 2, or T _H 17 cells to fine-tune clearance of infection (Lambrecht et al., 2012; Licona-Limon et al., 2013). In addition, there is a clear pathophysiological connection among some of these conditions. For instance, a staggering 70% of patients with atopic dermatitis go on to develop asthma via what has been called "atopic march" (Spergel, 2010), while adults with active asthma are strongly predisposed for acquiring COPD when compared to non-asthmatic individuals (Guerra, 2009).

Over the last decade TSLP (Quentmeier et al., 2001 ; Reche et al., 2001 ; Sims et al., 2000), an IL-2 family cytokine expressed by lung- and gut epithelial cells, and epidermal keratinocytes in the skin, emerged as a critical initiator of STAT-5 mediated T _H 2 inflammatory pathways driving the activation of immature DC, mast cells, basophils, eosinophils, and lymphocytes into a type 2 polarizing phenotype (Bell et al., 2013; Ziegler and Artis, 2010; Ziegler, 2012; Ziegler et al., 2013). Furthermore, such TSLP-mediated initiation of T _H 2 responses appear to result from an orchestrated cellular cascade involving DC, CD4 ⁺ T cells and basophils (Leyva-Castillo et al., 2013). Signaling by TSLP proceeds via a complex with its unique cognate receptor, TSLPR (encoded by the CRLF2 gene) (Pandey et al., 2000; Park et al., 2000) and the interleukin 7 (IL- 7) receptor a-chain (IL-7Ra), a receptor that also serves as the cognate receptor for IL-7 to regulate T-cell development and homeostasis (Mackall et al., 201 1 ). TSLP is now widely considered as the master switch for most prevalent inflammatory allergic disorders in humans, such as the triad of atopic diseases (asthma, atopic dermatitis and atopic rhinitis), COPD, and eosinophilic esophagitis (Ziegler, 2012; Ziegler et al., 2013; Noti et al., 2013; Redhu and Gounni, 2012; Siracusa et al., 2013a). A fascinating dimension to TSLP-mediated signaling has been the recent discovery that TSLP acts as a potent molecular liaison between the skin epithelium and neuronal cells to trigger itch, an incurable discomfort associated with atopic dermatitis (Wilson et al., 2013). In addition, TSLP has been found to contribute to extramedullary hematopoiesis eliciting progenitors with differentiating potential into effector cells (Siracusa et al., 2013b). Nonetheless, TSLP-mediated signaling appears to impact a much broader pathology landscape within multiple organ systems, including the abrogation of T _H 1 /T _H 17 responses (Roan et al., 2012) and the development of non-allergen-induced conditions such as idiopathic pulmonary fibrosis (I PF) (Datta et al., 2013), and tumor progression in breast- and pancreatic cancer (De Monte et al., 201 1 ; Pedroza-Gonzales et al., 201 1 ). We note that the link of TSLP to cancer is gaining a genetic dimension as shown recently by the prevalence of rearrangements and mutations in the TSLPR gene (CRLF2) in pediatric acute lymphoblastic leukemia (Perez-Andreu et al., 2013). Such wide pathology profile coupled to the high levels of expression of TSLP in allergy (Ziegler and Artis, 2010; Ziegler et al., 2013), the identification of TSLP as a genetic risk factor for the development of asthma (Hunninghake et al., 2010; Liu et al., 2012; Torgerson et al., 201 1 ) and eosinophilic esophagitis (Rothenberg et al., 2010), have elevated therapeutic targeting of TSLP signaling to center stage (Borowski et al., 2013; Romeo et al., 2013; Zhang et al., 201 1 ). In this regard, a recent study employing a primate animal model has shown that blockade of TSLPR attenuates allergic inflammation (Cheng et al., 2013), and that TSLP is pivotal to the development of resistance to corticosteroid treatment during airway inflammation (Kabata et al., 2013).

Despite the established pathophysiologic pleiotropy of TSLP signaling, the structural and molecular principles underlying the assembly of the TSLP:TSLPR:I L-7Ra signaling complex have remained poorly understood, in what has grown to be the missing link between the diverse research disciplines investigating TSLP-mediated signaling. In addition, how IL-7Ra participates in two distinct signaling complexes with largely non-overlapping functionalities had remained intriguing (Walsh, 2012). Surprisingly we found that the binding of TSLP to the TSLPR is essential for complex formation with I L-7Ra, and for further signaling. Even more surprisingly, we identified a mutant TSLP that still can bind to the TSLP receptor, but is unable to induce signaling.

A first aspect of the invention is a mutant TSLP, capable of inhibiting TSLP induced signaling. Inhibition of TSLP induced signaling, as used here, means that said mutant TSLP induces TSLP induced signaling 5, 10, 25, 50, 100, 250, 500, 1000, 2500 or 5000 times less than wild type TSLP. Preferably, said mutant TSLP induces no detectable TSLP induced signaling. Preferably, said mutant TSLP is still capable of binding the TSLPR. Preferably, said mutant is mutated in an amino acid, selected from the group of T46, S45, K49, D50, K101 , A41 , A42, M97 and A104 (numbering based on the human sequence as represented in SEQ ID No.1 .) The mutant TSLP, as described here, may carry one or more mutations. In a preferred embodiment, the mutant TSLP is at least mutated at position T46. In one specific embodiment the invention provides TSLP having two specific mutation at positions T46 and S45. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions T46 and K49. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions T46 and D50. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions T46 and K101. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions T46 and A41. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions T46 and A42. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions T46 and M97. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions T46 and A104. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions S45 and K49. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions S45 and D50. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions S45 and K101. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions S45 and A41. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions S45 and A42. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions S45 and M97. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions S45 and A104. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions K49 and D50. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions K49 and K101. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions K49 and A41. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions K49 and A42. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions K49 and M97. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions K49 and A104. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions D50 and K101. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions D50 and A41. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions D50 and A42. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions D50 and M97. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions D50 and A104. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions K101 and A41. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions K101 and A42. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions K101 and M97. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions K101 and A104. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions A41 and A42. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions A41 and M97. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions A41 and A104. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions A42 and M97. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions A42 and A104. In yet another specific embodiment the invention provides TSLP having two specific mutations at positions M97 and A104.

Another aspect of the invention is the use of a mutant TSLP according to the invention as a TSLP receptor antagonist. Still another aspect of the invention is a method to block TSLP induced signaling in a person in need of it, said method comprising administration of a mutant TSLP, preferably a mutant TSLP according to the invention to said person.

Yet another aspect of the invention is a mutant TSLP, preferably a mutant TSLP according to the invention, as a medicament. The mutant TSLP, as described here, may be modified, or it may be fused to another protein. As a non-limiting example, the mutant TSLP may be pegylated to increase the clearing time in the body, or it may be fused to an antibody to facilitate the delivery to target cells.

Another aspect of the invention is a mutant TSLP, preferably a mutant TSLP according to the invention, in treatment of an allergic disease. Preferably, said allergic disease is selected from the group consisting of asthma, chronic obstructive pulmonary disease, atopic dermatitis and eosinophilic esophagitis.

Still another aspect of the invention is a mutant TSLP, preferably a mutant TSLP according to the invention, in treatment of cancer. Preferably, said cancer is pediatric acute lymphoblastic leukemia. Another aspect of the invention is a pharmaceutical composition comprising mutant TSLP, preferably a mutant TSLP wherein said mutant is mutated in an amino acid, selected from the group of T46, S45, K49, D50, K101 , A41 , A42, M97 and A104 (numbering based on the human sequence as represented in SEQ ID No.1 ). The pharmaceutical compositions may comprise other compounds, such as but not limited to a suitable excipient. The mutant TSLP, as described here, may be modified, or it may be fused to another protein. As a non-limiting example, the mutant TSLP may be pegylated to increase the clearing time in the body, or it may be fused to an antibody to facilitate the delivery to target cells. BRIEF DESCRIPTION OF THE FIGURES

Figure 1. Isolation of glycan-minimized ternary TSLP:TSLPR:IL-7Ra complexes. Binary _{TS L} pNi23Q. _{TS L p R} Ni22Q _comp | _ex expressed in HEK293S GnTI ^" ' ^" cells was isolated by size- exclusion chromatography (SEC) using a Superdex 200 column (red profile). Following treatment with endoglycosidase H (EndoH) and addition of a molar excess of refolded IL-7Ra ectodomain, the ternary TSLP ^N123Q :TSLPR ^N122Q :IL-7Ra complex was separated from EndoH and the excess of IL-7Ra by SEC (blue profile). Elution volumes of protein standards are indicated above the chromatograms. The Coomassie-stained SDS-PAGE gel shows the binary complex before and after EndoH-treatment, and the resulting ternary complex used to obtain crystal forms 1A and 1 B. The sequence-based molecular masses (i.e. without glycosylation) for recombinant mTSLP, mTSLPR and IL-7Ra are 15 kDa, 23 kDa and 27 kDa respectively. The TSLP ^N123Q :TSLPR ^N53Q :IL-7Ra complex (crystal form 2) was reconstituted and purified via an identical approach.

Figure 2. Structure of the TSLP:TSLPR:IL-7Ra complex a. View of the ternary TSLP complex (structure determined from crystal form 1 B). TSLP is shown in cartoon mode with its four helices (aA - aD) colored green and loop regions colored grey. The position of the AB crossover loop is indicated. The extracellular regions of TSLPR and IL-7Ra, each comprising two Fnlll-like domains (D1 and D2), are shown in cartoon mode overlaid with a transparent surface and are colored magenta and cyan, respectively. TSLPR and IL-7Ra strand and loop regions contributing to site I, II and III are labeled. Disulfide- bridges are shown as yellow spheres. The insets show a detailed view of the cytokine-receptor interactions in site I and II (crystal form 1A), and the IL-7Ra receptor-receptor interactions in site III (crystal form 1 B). Interacting residues are shown as sticks. Polar interactions are indicated with dashed lines. Water molecules are shown as red spheres, b. Surface representation of TSLP with the TSLPR binding-epitope colored magenta, and the IL-7Ra epitope colored cyan. The total buried surface areas of site I and site II are indicated. Helix A, with its characteristic π-helical turn, contributes to both sites I and II and is superposed as a black outline on top of the TSLP surface, c. View of the four helical bundle core of TSLP. Trp132 is located at the center of the core and engages in stacking interactions with Tyr34 that stabilize the π-helix turn in helix A. Trp132, Thr71 and Thr90 coordinate the water molecule in the core of TSLP. The void volume observed in the core of TSLP is displayed as a blue mesh. The volume of the internal cavity (105 A ³ ) was calculated with the CASTp server, using a probe radius of 1.1 A.

Figure 3. Evolutionarily conserved features in TSLP a. Structure-based sequence alignments for sequences of mature TSLP from different mammalian species. The top secondary structure corresponds to mTSLP. Dashed lines indicate that the corresponding region was missing in the structural model. The four helices of the helical bundle fold are labeled A to D. Conserved residues are shown as white letters on a black background. Semi-conserved residues are shown in black bold and are boxed. mTSLP residues colored magenta interact with TSLPR, residues colored blue interact with IL-7Ra. Orange lines indicate the conserved disulfide pattern of TSLP. The stretch of basic residues in the CD-loop, present in a subset of TSLP sequences, is indicated in red. TSLP sequences were derived from the NCBI-database: M. musculus: NP_067342.1 , H. sapiens: NP_149024.1 , R. norvegicus: XP_001067649.1 , M. mulatta: XP_001 100503.1 , F. catus: XP_003981290.1 , O. garnettii: XP_003788993.1 , E. caballus: NP_001 157535.1 , A. melanoleuca: XP_002919224.1 b. View of the core of mTSLP centered around Trp132. Hydrophobic residues that constitute the core are shown as green sticks. Residues that constitute the hydrophobic core and coordinate the trapped water molecule (red sphere) are evolutionary conserved. The internal void volume in the core of TSLP is shown as a blue mesh. Figure 4. Molecular dynamics simulation to characterize unbound TSLP

A superposition of 10 representative snapshots along the 60-ns MD-simulation is shown in two orientations. In the left view, TSLP is orientated as in Figure 1 a, while the right orientation allows a clear view on the π-turn in the kinked helix A. The MD-simulation included a model of the highly flexible long overhand CD-loop (colored white) not visible in the electron density (residues 103-1 14). Residues are colored according to their average RMSD-value with respect to the reference crystallographic model determined to a resolution of 1.90 A. The position of residue Ile37, located in the π-turn in helix aA is indicated.

Figure 5. Cross-species sequence alignment for the TSLPR ectodomain

The secondary structure corresponds to mTSLPR. Dashed lines indicate that the corresponding region was missing in the structural model. Strands are labeled A1 -G1 for TSLPRD1 and A2-G2 for TSLPRD2. Conserved residues are shown as white letters on a black background. Semi-conserved residues are shown in black bold and are boxed. mTSLPR residues colored green interact with mTSLP. Orange lines indicate the disulfide pattern of mTSLPR. The atypical PSXW(S/T) motif in TSLPRD2 is indicated by a red box. The predicted transmembrane (TM)-region is shaded blue. Sequences were derived from the NCBI- database: M. musculus: NP_057924.3, H. sapiens: NP_071431.2, R. norvegicus: NP_604460.2, S. scrofa: XP_003361089.1 , S. simum: XP_004440836.1 , O. aries: XP_004022563.1 , O. rosmarus: XP_004400052.1 , C. lupus: XP_005641 109.1 .

Figure 6. Cross-species sequence alignment for the IL-7Ra ectodomain

The secondary structure elements shown correspond to those in mlL-7Ra. Dashed lines indicate that the corresponding region was missing in the structural model. Strands are labeled A1 -G1 for IL-7RaD1 and A2-G2 for IL-7RaD2. Conserved residues are shown as white letters on a black background. Semi-conserved residues are shown in black bold and are boxed. mlL- 7Ra residues colored green interact with mTSLP, hlL-7Ra residues that interact with hlL-7 are colored orange. Positions that interact with both TSLP and IL-7 are indicated by red dots. Orange lines indicate the disulfide pattern of IL-7Ra. The WSXWS motif in IL-7RaD2 is indicated by a red box. The predicted transmembrane (TM)-region is shaded blue. Sequences were derived from the NCBI-database: M. musculus: NP_032398.3, H. sapiens: NP_002176.2, F. catus: XP_003981510.1 , L. africana: XP_003408018.1 , C. lupus: XP_855408.1 , B. Taurus: NP_001 192816.1 , O. Aries: XP_004017102.1 , M. domestica: XP_001373020.1 and G. gallus: NP_001073575.1 .

Figure 7. Degeneracy of IL-7Ra in TSLP and IL-7 signaling complexes a. Structural alignment of the mTSLP:mlL-7Ra assembly in the ternary TSLP complex (Crystal form 1A, pH 4.5) with the structure for the binary hlL-7:hlL-7Ra complex (pdb code 3DI2, chain A and B) based on the superposition of mlL-7Ra and hlL-7Ra (the overall RMSD for the IL- 7Ra receptors is 0.93 A). The side-chains of residues mTSLP-137 and hlL-7-V40 are shown in sphere mode. b. The interaction of mTSLP and hlL-7 with IL-7Ra is predominantly hydrophobic and is centered around the insertion of a hydrophobic residue protruding from the π-helical turn in helix A into an evolutionary conserved platform at presented by the D1 -D2 junction of IL-7Ra. In both complexes, the hydrophobic residues constituting this platform contribute 60-70% to the interface area of IL-7Ra. c. Cartoon representation indicating the dual role of IL-7Ra in TSLP and IL-7 mediated signaling, d. A structural alignment of the TSLP ternary complex with the determined structures of other IL-2 family signaling complexes (pdb codes 2B5I, 3BPL, 3BPO, 4GS7) based on the superposition of the D2-D2 assemblies of each complex shows that the relative orientation and interaction mode between the membrane proximal Fnlll-like domains is well conserved while a larger degree of structural heterogeneity is observed for the cytokine-receptor interactions. The IL-2Ra and IL-15Ra receptors were omitted for clarity.

Figure 8. ITC and SPR binding studies for the assembly of the ternary TSLP:TSLPR:IL- 7Ra complex a. ITC thermograms for the titration of TSLP with TSLPR (left), and for the titration of IL-7RD with preformed TSLP:TSLPR complex (right) are shown. Data were fitted with a 'single-site binding model', giving the apparent molar reaction enthalpy (ΔΗ°), apparent entropy (AS°), dissociation constant (KD) and stoichiometry of binding (N) of complex formation, b. The association of IL-7Ra with immobilized TSLP was analyzed by SPR equilibrium binding analysis resulting in a KD value of 2.3 μΜ. c. Single-cycle kinetics data for the association of IL-7Ra with the TSLP:TSLPR complex was fitted with a 1 :1 Langmuir binding model. The determined dissociation constants (K _D ) and kinetic parameters (k _on , k _off ) are shown.

Figure 9. Binding and activity assays with mouse TSLP, TSLPR and IL-7Ra mutants

In a competitive binding assay TSLP-SEAP fusion protein was displaced from H EK293T cells expressing TSLPR with either wild type TSLP or TSLP point mutants (grey shaded area). TSLP-induced STAT5 activity was measured using a luciferase-based reporter system in H EK293T cells expressing wild type and/or mutant forms of either TSLPR or IL-7Ra upon incubation with wild type or mutant TSLP. The determined half-maximal inhibitory concentration (IC50)-values for the competition assays and the half-maximum effective concentration (EC50)-values for STAT5-activation assays are reported. Control experiments with wild-type TSLP and, TSLPR and IL-7Ra receptors are shown at the top. TSLP- displacement assays and STAT5-activation assays were carried out for TSLP-mutants TSLP i37E (a, b), TSLPF39A_YI33A (d, e) and TSLPDS6R (g, h). STAT5 activation assays were carried out for receptor mutants IL-7Ra _Y2 i3R (c), TSLPR _G I96R (f) and TSLPR _R110 E (0- Figure 10. Binding and activity assays with human TSLP mutants a. In a competitive binding assay TSLP-SEAP fusion protein was displaced from HEK293T cells expressing TSLPR with either wild type human TSLP (black squares) or the S45R+T46R human TSLP mutant (white spheres), b. Human TSLP-induced STAT5 activity was measured using a luciferase-based reporter system in HEK293T cells expressing wild type human TSLPR and human IL-7Ra upon incubation with wild type human TSLP (black squares) or the S45R+T45R human TSLP mutant (white spheres). EXAMPLES

Materials and methods to the examples

Recombinant protein production in mammalian cells cDNA fragments encoding full-length mouse TSLP (NP_067342.1 ; residue 1 - 140) and the extracellular fragments of mouse TSLPR (NP_057924.3; residue 1 - 222) and mouse IL-7Ra (NP_032398.3; residue 1 - 239) were chemically synthesized (Genscript). Each cDNA fragment was flanked by an EcoRI and Kpnl restriction site for cloning into the pHL-expression vector (Aricescu et al., 2006) in frame with a C-terminal hexahistidine-tag. For crystallization purposes, we also designed single-site N-glycosylation mutants for TSLP (N21 Q, N26Q and N123Q) and TSLPR (N53Q and N122Q) (Genscript). HEK293T or HEK293S GnTI ^" ' ^" cell lines (Reeves et al., 2002) were grown in high-glucose DMEM medium (Lonza) supplemented with 10% heat-inactivated fetal calf serum (Sigma), 10 ⁶ units. L ^"1 penicillin G and 1 gL ^"1 streptomycin in a 5% C02 atmosphere at 310 K. Small-scale transient expression experiments were conducted in 6-well tissue-culture plates. Prior to transfection, the medium of confluent cell cultures was replaced with serum-free medium. Subsequently, cells were transfected with purified plasmid DNA mixed with 25 kDa branched polyethylenimine (Sigma) in a 1 : 1 .5 ratio. For co-transfection experiments expression plasmids were used in a 1 : 1 ratio. Transfected cells were allowed to express recombinant proteins for 5 days before the medium was harvested. His-tagged proteins were detected in the conditioned medium by Western blot analysis using an anti-His(C-term)-HRP antibody (Invitrogen). Largescale expression experiments were conducted in 175 cm ² tissue culture flasks or 850 cm2 roller flasks. Conditioned medium (typically 1 -2 L) was clarified by centrifugation and filtered through a 0.22 μηη bottle-top filter and loaded onto a Talon SuperFlow column for purification via metalaffinity chromatography. Recombinant proteins and complexes were further purified by SEC using a Superdex 75 or Superdex 200 column with HBS pH 7.4 as running buffer.

Production of recombinant mlL-7Ra in E. coli

A synthesized cDNA fragment corresponding to the mature extracellular domain of mlL-7Ra (residues 21 -239) (Genscript) was cloned into pET15b vector between the Ndel and BamHI sites and in frame with a cleavable N-terminal His-tag. mlL-7Ra was expressed in the E. coli Rosettagami B (DE3) strain (Novagen). Inclusion bodies were isolated and refolded as previously described (Verstraete et al., 2009), followed by cleavage of the N-terminal His-tag with biotinylated thrombin followed by removal of thrombin on streptavidin agarose beads. As a polishing step, mlL-7Ra was injected on a Superdex 75 column with HBS pH 7.4 as running buffer. Preparation of crystallization-grade TSLP:TSLPR:IL-7Ra

Binary complexes TSLP ^N123Q :TSLPR ^N53Q and TSLP ^N123Q :TSLPR ^N122Q were produced by cotransfection in HEK293S GnTT' ^" cells. Following purification by metal-affinity chromatography and SEC, and trimming of N-linked glycans by EndoH-treatment (incubation overnight with 7.5 kU of EndoH (NEB) per mg of complex in a volume of 1 - 2 mL HBS-buffer), binary TSLP:TSLPR complexes were mixed with a molar excess of in vitro refolded IL-7Ra produced in E. coli. To isolate the ternary TSLP : TSLPR : IL-7Ra complex from the excess of IL-7Ra and EndoH the protein mixture was injected onto a Superdex 200 column with HBS as running buffer. Fractions corresponding to the ternary TSLP : TSLPR : IL-7Ra complex were pooled and concentrated by centrifugal ultrafiltration to a concentration of 7 mg/mL.

Crystallization of the TSLP : TSLPR : IL-7Ra ternary complex

Nanolitre crystallization experiments were set up at room temperature using a Mosquito crystallization robot (TTP Labtech) and sparse-matrix screens (Molecular Dimensions, Hampton Research). The TSLP:TSLPR:IL-7Ra ternary complex crystallized in condition D2 of the MIDAS-HT screen (Molecular Dimensions). Subsequent crystal optimization and diffraction experiments only led to weakly diffracting crystals. The TSLP ^N123Q :TSLPR ^N122Q :IL-7Ra complex crystallized in condition B2 of the ProPlex screen (crystal form 1 ). Optimized crystals were grown in 300 mM CaAc2, 300 mM NaAc pH 4.5 - 5.0 and 14 % PEG 4000 and cryoprotected with Paraton-N oil. A subset of the crystals derived from this condition was soaked into 300 mM NaCI, 100 mM Tris-HCI pH 7.5, 18 % PEG 4000. The TSLP ^N123Q :TSLPR ^N53Q :IL-7Ra complex (crystal form 2) crystallized in a condition derived from the previously identified D2 condition of the MIDAS screen HT screen: 25 % pentaerythritol ethoxylate, 100 mM Tris-HCI pH 8.0. These crystals were flash frozen directly into liquid nitrogen for data collection.

Crystallographic data collection, structure determination and refinement X-ray diffraction measurements were conducted from single crystals of the TSLP ^N123Q : TSLPR ^N122Q :IL-7Ra and TSLP ^N123Q :TSLPR ^N53Q :IL-7Ra complexes under cryogenic conditions (100 K) at synchrotron beam lines P13 (Petralll, DESY) and Proxima 2A (SOLEIL). All data were integrated and scaled using the XDS suite (Kabsch, 2010). Each crystal form contained one copy of the ternary complex in the crystal asymmetric unit. The structure of the JSLP ^N123Q : TSLPR ^N122Q : IL-7Ra complex was determined by maximum-likelihood molecular replacement as implemented in the program suite PHASER (McCoy, 2007), using search models derived from the structure of the human IL-7:IL-7Ra complex (PDB entry 3DI2) (McElroy et al., 2009). After initial rounds of model refinement in PHENIX and model (re)building in Coot (Emsley et al., 2010), a model for TSLPR derived from the IL-13Ra1 structure (PDB entry 3BPO) (LaPorte et al., 2008) was modeled manually into the electron density. The crystallographic model was further completed by additional rounds of refinement and model rebuilding. The crystal structure of the TSLP ^N123Q :TSLPR ^N53Q :IL-7Ra complex (crystal form 2) was determined by molecular replacement based on the structure of crystal form 1 and was refined with PHENIX. Comparison of the two structures showed that the membrane-proximal Fnlll-like domains (TSLPR _D2 and IL-7Ra _D 2) remain separated in crystal form 1 , while lateral receptor interactions are observed in crystal form 2. Since the crystallization conditions of crystal form 2 are closer to physiological conditions (pH 8 vs. pH 5.1 ) and given the prevalence of receptor receptor interactions in the IL-2 family, we hypothesized that crystal form 2 represents the biologically relevant assembly. When crystals of crystal form I were incubated in a stabilizing solution buffered at neutral pH, the crystals retained their X-ray diffraction capability and gave rise to a 2.55 A dataset (Crystal form 1 , pH 7.5). Importantly, the resulting structure revealed that TSLPR-D2 had adopted a conformation relative to IL-7Ra-D2 as seen in crystal form 2. We can thus conclude that TSLPR and IL-7Ra engage in lateral contacts upon complex formation with TSLP. All refined crystallographic models were validated using Molprobity as implemented in the PHENIX suite. Coordinates and structure factors for the TSLP:TSLPR:IL-7Ra complexes have been deposited in the Protein Data Bank (www.rcsb.org). The crystal structures of the _{TS L} pNi23Q. _{TS L p R} Ni22Q. | _L _ _{7 Ra comp} | _{ex at ac} j _d j _{c and neu} tral pH have accession codes 4NN5 and 4NN6, respectively. The structure of the TSLP ^N123Q :TSLPR ^N53Q :IL-7Ra complex (Crystal form 2) is available via code 4NN7.

Structural superpositions, structure-based sequence alignments, and structural analyses

The structures for mTSLP and hlL-7 were superimposed with Chimera (Yang et al., 2012). Sequence alignments of mature TSLP, TSLPR and IL-7Ra from different mammalian species were aligned using ClustalW (Larkin et al., 2007) and ESPript (Gouet et al., 1999). Protein- protein interaction interfaces were analyzed using EBI-PISA (Krissinel and Henrick, 2004) and the program Probe (Word et al., 1999) as implemented in Phenix. The volume of the internal cavity in TSLP was calculated with the CASTp server (Dundas et al., 2006) , using a probe radius of 1 .1 A. Figures containing structural models were prepared in PyMOL (DeLano and Lam, 2005).

SEC-MALS

Protein samples at 3-4 μΜ were injected through a Anotop-10 0,02 μηη syringe filter (Whatmann) onto a WTC-030S5 silica SEC column (Wyatt), with HBS pH 7.4 as running buffer, coupled to a online UV-detector (Shimadzu), a multi-angle light scattering-angle laser Wyatt miniDAWN TREOS instrument and a Wyatt Optilab T-rEX refractometer at 25° C. A refractive index increment (dn/dc) value of 0.185 ml/g was used for protein concentration and molecular mass determination. Data were analyzed using the ASTRA6 software.

Isothermal titration calorimetry Experiments were carried out using a MicroCal iTC200 instrument (GE Healthcare) at 37° C, and data were analyzed using the Origin ITC analysis software package. TSLP and the TSLPR and IL-7Ra ectodomains were produced in HEK293T without the addition of kifunensine. All proteins were exchanged to the same buffer, 15 mM HEPES, 150 mM NaCI, pH 7.4 by size- exclusion chromatography. Protein concentrations were measured spectrophotometrically at 280 nm using calculated theoretical extinction coefficients and all solutions were extensively degassed prior to use. Titrations were always preceded by an initial injection of 0.5 μΙ_, and were carried out using 2 μΙ_ injections applied 150 s apart. The sample was stirred at a speed of 1000 rpm throughout. The thermal titration data were fit to the "one binding site model", and apparent molar reaction enthalpy (ΔΗ°), apparent entropy (AS°), dissociation constant (Kd) and stoichiometry of binding (N) were determined.

Surface Plasmon Resonance

Full-length TSLP and the TSLPR-ectodomain were cloned between the EcoRI and Kpnl sites of the pHL-AVITAG vector (Aricescu, 2006). Prior the transfection in HEK293T cells, the culture medium was changed to serum-free DMEM medium to which 100 μΜ D-biotin was added. To allow specific C-terminal in vivo biotinylation the pHL-TSLP-AVITAG or pHL- TSLPRAVITAG constructs were co-transfected with the pDisplay-BirA-ER plasmid (Howarth, Nat.Prot., 2008) in a 3 : 2 ratio. Five days post-transfection the medium (100 mL) was harvested and loaded onto a Ni Sepharose column. Recombinant proteins were eluted with imidazole and loaded onto a Superdex 75 column with HBS as running buffer. SPR experiments were performed using a Biacore X-100 instrument at 25°C with HBS-EP+ (GE Healthcare) as running buffer. 50 - 150 RU of TSLP-AVITAG or TSLPR-AVITAG were coupled to an SA sensor chip to which streptavidin was pre-immobilized (GE Healthcare). Single-cycle kinetics and equilibrium SPR experiments were conducted with fully glycosylated TSLP and, TSLPR and IL- 7Ra receptor ectodomains. No aspecific binding was observed to the reference flow path. After double-reference subtraction, sensorgrams were analyzed using the Biacore data evaluation software (version 2.0.1 ). Kinetic parameters were fitted to a 1 :1 Langmuir or heterogeneous ligand model. To measure the interaction of IL-7R! with the TSLP-TSLPR complex, an SAChip on which TSLP was immobilized was used and a nearly-saturating concentration (250 nM) of TSLPR was included in the running buffer and all samples. Molecular Dynamics

All Molecular Dynamics (MD) simulations were performed using GROMACS (Pronk et al., 2013) and the Amber99SB-ILDN force field (Lindorff-Larsen et al.,2010). The initial geometry for the protein was taken from our crystallographic refinement. A structure for the missing residues 84-95 was generated by using MODELLER (Eswar et al., 2008), missing atoms were added and the protonation states of titratable residues were optimally assigned using GROMACS (Pronk et al., 2013). The protein was then solvated in a cubic box of (56.6 A) with 5177 TIP3P water molecules and sodium ions were added to ensure total charge neutrality. Electrostatics were treated with particle-mesh Ewald (PME) using a short-range cutoff of 1.4 nm, and van der Waals interactions were switched off between 1 .0 to 1 .1 nm. Neighbor lists were updated every 5 fs. The entire molecular system was first minimized for 2.000 steps with a steepest descent algorithm. Then, the system was equilibrated at 300K and 1 bar, holding all non-hydrogen protein atoms fixed (with a force constant of 1000 kJ/mol/nm ² ) and allowing the surrounding water to relax for 100ps. The pressure was controlled with an isotropic Berendsen barostat applied to the entire system with a time constant of LODps and a compressibility of 4.5x10 ⁵ bar ^"1 . The temperature was controlled with two Nose-Hoover thermostats applied to the protein and solvent respectively with a time constant of 0.2ps. A 1 fs timestep was used in all MD simulations. Equilibrated cell dimensions were adopted (55.82 A) and all further MD simulations were conducted in the NVT ensemble. Over the course of 700 ps, restraints on protein atoms were gradually reduced. This was followed by production simulations totaling 60 ns. No restraints were imposed during these simulations.

Site-directed mutagenesis for cellular activity assays and binding studies with mouse TSLP

A codon optimized DNA sequence coding for mTSLP with a C-terminal GGSGGS linker was ligated into the BsrGI/Bglll opened pMET7-CRH2-SEAP-Flag vector (Zabeau et al., 2005). The resulting pMET7-mTSLP-SEAP-Flag allows the expression of a mTSLP-secreted alkaline phosphatase fusion protein (mTSLP-SEAP). A codon optimized mTSLPR DNA sequence was ligated into the Clal/Xbal opened pMet7-SigK-flag-human leptin_receptor-HA vector. The resulting pMET7-Flag-mTSLPR allows the expression of a Flag tagged mTSLPR. A codon optimized mlL-7Ra DNA sequence was ligated into the BspEI/Xbal opened pMet7-SigK-HA- mousejeptin vector (Peelman et al., 2004). The resulting pMET7-HA-IL-7Ralpha allows the expression of an HA tagged mlL-7 receptor. Site directed mutations in these vectors were introduced via the Quickchange protocol (Stratagene). Site directed mutations of pHL-mTSLP- His were first introduced in the pUC57-mTSLP vector, followed by ligation of the EcoRI/Kpnl mutant mTSLP DNA fragment into the EcoRI/Kpnl opened pHL-mTSLP-His vector. Site-directed mutagenesis for cellular activity assays and binding studies with human TSLP

A codon optimized DNA sequence coding for hTSLP with a C-terminal GGSGGS linker was ligated into the BsrGI/Bglll opened pMET7-CRH2-SEAP-Flag vector (Zabeau et al., 2005). The resulting pMET7-hTSLP-SEAP-Flag allows the expression of a hTSLP-secreted alkaline phosphatase fusion protein (hTSLP-SEAP). A codon optimized hTSLPR DNA sequence was ligated into the Clal/Xbal opened pMET7-Flag-mTSLPR vector. The resulting pMET7-Flag- hTSLPR allows the expression of a Flag tagged hTSLPR. A codon optimized human IL-7Ra DNA sequence was ligated into the BspEI/Xbal opened pMET7-HA-IL-7Ralpha. The resulting pMET7-HA-IL-h7Ralpha allows the expression of an HA tagged hlL-7 receptor. A S45R+T46R site directed mutation of pHL-hTSLP-His were first introduced in the pUC57-hTSLP-R127A- R130S vector via the Quickchange protocol (Stratagene), followed by ligation of the EcoRI/Kpnl mutant hTSLP DNA fragment into the EcoRI/Kpnl opened pHL-mTSLP-His vector.

Competitive mouse TSLP-SEAP/mouse TSLP receptor cell binding assay

HEK293T cells were transfected with pMET7-mTSLP-SEAP using linear PEI (Polysciences). The day after transfection, medium was replaced with Optimem medium (Life technologies). Three days after transfection, the medium containing the secreted TSLP-SEAP fusion protein was harvested. HEK293T cells were transfected with pMet7-FLAG-mTSLPR in 6-well plates using linear PEI (Polysciences). Two days post transfection, the cells were detached with 5mM EDTA in phosphate buffered saline (Life Technologies) and washed in FACS buffer (1 % foetal bovine serum, 0.5 mM EDTA in phosphate buffered saline). 130.000 cells were incubated for 1 h at 6°C with 15-fold diluted mTSLP-SEAP containing supernatant and different concentrations of unlabeled TSLP or TSLP mutant in FACS buffer. The cells were washed three times with FACS buffer, and the bound alkaline phosphatase activity was quantified using the PhosphaLight kit (Tropix) in a TopCount chemiluminescence counter (Packard). The data were fitted to a log inhibitor versus response curve in Graphpad Prism.

Competitive human TSLP-SEAP/human TSLP receptor cell binding assay

HEK293T cells were transfected with pMET7-hTSLP-SEAP using linear PEI (Polysciences). The day after transfection, medium was replaced with Optimem medium (Life technologies). Three days after transfection, the medium containing the secreted hTSLP-SEAP fusion protein was harvested. HEK293T cells were transfected with pMet7-FLAG-hTSLPR in a 75 cm ² flask using linear PEI (Polysciences). Two days post transfection, the cells were detached with 5mM EDTA in phosphate buffered saline (Life Technologies) and washed in FACS buffer (1 % foetal bovine serum, 0.5 mM EDTA in phosphate buffered saline). 130.000 cells were incubated for 1 h30min at 6°C with 8-fold diluted hTSLP-SEAP containing supernatant and different concentrations of unlabeled hTSLP or hTSLP mutant in FACS buffer. The cells were washed three times with FACS buffer, and the bound alkaline phosphatase activity was quantified using the PhosphaLight kit (Tropix) in an Envision chemiluminescence counter (Hewlett- Packard). The data were displayed as a log inhibitor versus response curve. STAT5 reporter activation studies with mouse TSLP

HEK293T cells were co-transfected with 1 12.5 ng pMET7-Flag-mTSLPR, 1 12.5 ng pMET7- HA-IL-7Ralpha, 675 ng empty pMET7 vector and 100 ng pGL3-p-casein-luci reporter plasmid per well of a 6-well plate. The pGL3-p-casein-luci luciferase reporter contains 5 repeated STAT5-responsive motifs of the β-casein promoter. The day after transfection, the cells were detached with cell dissociation buffer (Life Technologies), and resuspended in DMEM + 10% foetal bovine serum. 50% of the cells were seeded in a new six-well plate for FACS analysis. 2% of the cells were seeded per well in 96 well plates and stimulated with increasing concentrations of mTSLP. On day two after transfection, the luciferase activity in the 96 well plates was determined on a TopCount chemiluminescence counter as described previously63. Fold induction of luciferase activity was calculated by dividing the luminescence signal (cps) of the mTSLP stimulated cells by the luminescence signal of the unstimulated cells. The data were fitted to a log agonist versus response curve in Graphpad Prism. The Surface expression of the flag-tagged mTSLPR was determined on a FACSCalibur (BD Biosciences) using mouse monoclonal anti-FLAG M2 antibody (Sigma) plus Alexfluor488 labeled goat anti-mouse antibodies (Molecular Probes). HA-tagged mlL-7R! expression was determined via an FITC- labeled mouse monoclonal anti-HA antibody (Sigma). Relative receptor expression was determined by calculating the mean, geometrical mean and median Alexafluor488 fluorescence from the FACS histograms, and comparing these values to the corresponding values of mock (pMet7) transfected cells. STAT5 reporter activation studies with human TSLP

HEK293T cells were co-transfected with 1 .3 ng pMET7-Flag-hTSLPR, 1 .3 ng pMET7-HA-hlL- 7Ralpha, 890 ng empty pMET7 vector and 89 ng pGL3-p-casein-luci reporter plasmid per well of a 6-well plate. The pGL3-p-casein-luci luciferase reporter contains 5 repeated STAT5- responsive motifs of the β-casein promoter. The day after transfection, the cells were detached with cell dissociation buffer (Life Technologies), and resuspended in DMEM + 10% foetal bovine serum. 2% of the cells were seeded per well in 96 well plates and stimulated with increasing concentrations of hTSLP or hTSLP mutant. On day two after transfection, the luciferase activity in the 96 well plates was determined on a Envision chemiluminescence counter (Hewlett-Packard). The induction of luminescence by luciferase activity, expressed as counts per second, was plotted versus the hTSLP/hTSLP mutant concentration used for stimulation.

Example 1 : Production of recombinant TSLP complexes suitable for crystallographic studies To enable structural studies of the TSLP:TSLPR:IL-7Ra ternary complex by X-ray crystallography we developed a combined strategy entailing co-expression of mouse TSLP:TSLPR binary complexes with homogeneous N-linked GlcNAc ₂ Man ₅ glycan trees in HEK293S GnTT' ^" cells and production of mouse IL-7Ra in E. coli coupled to refolding in vitro (verstraete et al., 2009). Following purification of TSLP:TSLPR binary complexes by metal- affinity chromatography and size-exclusion chromatography (SEC), and shaving off accessible N-linked glycans by Endoglycosidase H (EndoH) treatment, we were able to assemble the TSLP:TSLPR:IL-7Ra ternary complex by mixing the purified TSLP:TSLPR complex with a molar excess of purified recombinant IL-7Ra. However, initial crystallization trials based on this recombinant material failed to yield well-diffracting crystals, prompting us to resort to variants of the TSLP:TSLPR binary complex with minimized N-linked glycosylation based on targeted permutation of putative asparagine N-linked glycosylation sites to glutamine. Based on our original protein production and purification strategy, we could prepare highly monodisperse preparations of two ternary complexes, namely TSLP _N1 23Q:TSLPR _N53 Q: I L- 7Ra and TSLP _N1 23Q:TSLPR _N12 2Q:IL-7Ra (Fig. 1 ). Example 2: Overall structure of the TSLP:TSLPR:IL-7Ra complex

Purified TSLP _N1 23Q:TSLPR _N5 3 _Q :IL-7Ra and TSLP _N12 3Q:TSLPR _N12 2Q:IL-7Ra ternary complexes proved readily amenable to crystallization trials and led to two crystal forms, allowing us to determine three crystal structures of the ternary complex including one at 1 .9 A resolution allowing detailed delineation of the cytokine-receptor interfaces (Table 1 ).

Table 1. Crystallographic data collection and refinement statistics

Crystal form 1

Acidic pH Neutral pH Crystal form 2

Crystal form 1A Crystal form 1 B

Data collection

Space group 9 2 2 2 P 2 _λ 2 _λ 2 P 2 _\ 2 _\ 2 _\

Cell dimensions

a, b, c (A) 150.06, 79.76, 52.14 147.95, 75.16, 51.44 35.77, 50.1 1 , 249.71 α, β, γ{°) 90, 90, 90 90, 90, 90 90, 90, 90

Resolution (A) 50.00-1 .90 (2.01 -1.90) 50.00-2.54 (2.70-2.55) 50-3.77 (4.00-3.78)

Rmeas (%) 8.6 (67.9) 10.8 (84.6) 26.7 (77.8)

Mean Ι / σ(Ι) 19.9 (3.9) 1 1 .2 (1.9) 9.2 (3.0)

Completeness (%) 99.7 (98.4) 99.5 (97.8) 99.5 (97.7)

Redundancy 13.2 (13.5) 6.5 (6.4) 1 1.9 (10.7)

Refinement

Resolution (A) 50.00-1.90 50.00-2.55 50.00-3.78

No. reflections 50285 19427 4962

Rwork / Rfree 0.1698 / 0.2009 0.2587 / 0.2966 0.2765 / 0.2865

No. atoms

Protein 3804 3362 3297

Water 330 3 -

ADP (A ² )

Protein 38.7 87.7 101 .0

Water 40.5 50.7 - r.m.s. deviations

Bond lengths (A) 0.01 1 0.003 0.004

Bond angles (°) 1.22 0.86 0.87

Values in parentheses refer to the highest-resolution shell. Each of the reported datasets was obtained from a single crystal.

The structure of the TSLP _N12 3Q:TSLPR _N12 2Q:IL-7Ra complex (crystal form 1 A) was determined by maximum-likelihood molecular replacement as implemented in the program suite PHASER (McCoy, 2007), using search models derived from the structure of the human IL-7:I L-7Ra complex (PDB entry 3DI2) ((McElroy et al., 2009). We subsequently used this structure to obtain the crystal structure of the TSLP _N1 23Q:TSLPR _N5 3 _Q :IL-7Ra complex (crystal form 2). Both structures show that the ternary TSLP complex resembles the canonical architecture seen in other I L-2 family complexes, whereby the cytokine bridges the two receptors by interacting with cytokine-binding epitopes presented at the D1 -D2 junctions of each receptor. However, comparison of the two structures showed that while lateral receptor-receptor between the membrane-proximal domains (D2) of TSLPR and IL-7Ra could be observed in crystal form 2, the two receptors remain separated in crystal form 1 A. Since the crystallization conditions of crystal form 2 are closer to physiological conditions (pH 8 vs. pH 5.1 for crystal form 1A) and given the prevalence of receptor-receptor interactions in the IL-2 family, we hypothesized that crystal form 2 represents the biologically relevant assembly. Moreover, when crystals of crystal form 1A were incubated in a crystal stabilizing solution buffered at neutral pH, the crystals retained their X-ray diffraction quality yielding a dataset to 2.55 A dataset (crystal form 1 B, pH 7.5). Importantly, the resulting structure revealed that TSLPR-D2 had adopted a conformation relative to IL-7Ra-D2 as seen in crystal form 2, allowing us to conclude that TSLPR and IL- 7Ra are indeed poised to engage in lateral contacts upon complex formation with TSLP at the cell surface. Thus our structural studies establish that in the extracellular complex TSLP wedges between TSLPR and IL-7Ra via two equally extensive interaction interfaces, site I (burying 1350 A2) and site II (burying 1250 A2), to establish a T-shaped ternary assembly that enables interactions between the membrane-proximal domains of the two receptors via site III (900 A2) (Fig. 2a-b and Table 2).

Table 2. Interactions at the TSLP:TSLPR, TSLP:IL-7Ra and TSLPR:IL-7Ra interfaces

Site I: TSLP:TSLPR interactions Site II: TSLP:IL-7Ra interactions

Hydrogen bonds and salt bridges Hydrogen bonds and salt bridges

TSLP TSLPR Distance TSLP IL-7Ra Distance

Tyr34 OH Ala193 0 2.73 Asn36 Οδ1 Asn21 1 Νδ2 2.98

Phe51 O Gln109 Νε2 3.08 Asn36 Νδ2 Asp212 Οδ1 3.24

Gln53 N Arg108 O 2.86 Asp41 Οδ1 Tyr159 OH 2.54

Gln53 O Arg108 Νη1 2.73 Asp41 Οδ1 Lys158 Νζ 3.09

Ile54 O Arg108 Νη1 2.84 Arg88 Νη1 Leu78 O 3.45

Asp56 Οδ Arg89 Νε Leu100 O 2.89

Arg1 10 Νε 3.00

1

Asp56 Οδ Arg89 Νη1 Tyr159 OH 3.03

Arg1 10 Νη2 3.39

1

Asp56 Οδ Arg89 Νη2 Phe99 O 3.14

Arg1 10 Νη2 3.23

2

Asp56 Οδ

Gly90 N 2.81 Bridging water molecules

2

Cys57 O Arg1 10 Νη2 3.04 TSLP IL-7Ra Water

Ser134 Ογ Arg1 10 Νε 3.01 Arg89 Νη2 His53 Ns2 A143

Gln137 Νε

His194 O 2.88 Asp45 Οδ1 Lys158 Νζ A144 2

Gln137 Οε

Arg1 10 Νη1 3.24 Asn36 O Leu 160 N A145 1

Gln137 Οε

Ser1 12 Ογ 2.67

1

van der Waals contacts

Bridging water molecules TSLP IL-7Ra

TSLP TSLPR Water Lys32 Asp212

Gln137 Tyr195 0 A142 Tyr213 Νε2

Ile33 Ile102

van der Waals contacts Tyr213

TSLP TSLPR Asn36 Tyr213

Tyr34 Ala193 Phe214

Cys35 Ala193 Ile37 Tyr213

Phe39 Ala 192 Leu101

Ala193 Ile102

Glu52 Arg108 Phe214

Gln53 Leu91 Tyr159

Ile54 Leu91 His40 Lys158

Glu55 Leu91 Lys161

Asp56 Leu91 Thr85 Leu78

Tyr133 Gly196 Arg88 Leu78

Met136 His194 Arg89 Leu 100

Gln137 His194 Ile102

Tyr195 Glu92 Leu 100

Site III: TSLPR:IL-7Ra interactions

Potential hydrogen bonds and salt

TSLPR IL-7Ra Distance

Glu153 Οε

Arg140 Νη2 3.79

1

Glu153

Arg140 Νη2 3.13

Οε2

Gln162

Arg187 N 3.57

Οε1

Ser165

Lys162 Νζ 2.83

0

Gly174 0 His185 Νε2 3.56

Gly175 0 Thr188 Ογ1 3.27

Arg180 Νη 2.44

Ala143 0

2

van der Waals contacts

TSLPR IL-7Ra

His151 His185

Gln162 Arg187

Thr164 Phe184

Val173 His185

Gly174 Ser182

Gly175 Val181

Asp177 Pro191

⁵ The TSLP SLPR and TSLP:IL-7Ra cytokine-receptor interactions (Site I and Site II) are described based on the 1.90 A structure of the ternary complex (Crystal form 1A, pH 4.5). The IL-7Ra:TSLPR receptor- receptor interactions (Site III) are described based on the 2.55 A resolution structure determined from crystal form 1 at neutral pH (Crystal form IB, pH 7.5) . Protein-protein interactions were analyzed using the PISA server at the European Bioinformatics Institute and the program Probe as implemented in Phenix.

Example 3: TSLP and TSLPR carry unique structural features Our studies have revealed for the first time structural information for TSLP and TSLPR. TSLP adopts a short-chain four-helix bundle fold resembling IL-742 (r.m.s.d. 1 .23 A for 56 Cot atoms) despite the abysmally low sequence identity between the two cytokines, and has an extended helix A at the N-terminus and markedly shorter B and C helices (Fig. 2a). The strong conservation of core residues and cysteines participating in disulfide bonds (C25-C98, C57- C63, C78-C121 ) indicates that the observed fold can be extrapolated to all mammalian TSLPs (Fig. 3). Surprisingly, an ordered water molecule in the core of TSLP is coordinated by a conserved triad of residues (T71 , T90, W132), adjacent to a large internal void volume (105 A ³ ) that runs from the BC- to the AC-face of the fold (Fig. 2c). Given such unusual cytokine core packing we carried out full-atom molecular dynamics simulations on unbound TSLP (Fig. 4) and found that unlike other IL-2 family cytokines possessing conformational plasticity (Levin et al., 2012; Bondensgaard et al., 2007), the TSLP core remains largely invariable including the buried water molecule. However, our simulations uncovered significant structural heterogeneity at the atypical ττ-helical turn localized midway into the kinked helix A (Fig. 4).

TSLPR is evolutionarily and structurally related to the common cytokine receptor γ chain (y _c ) and IL-13Ra1 (Reche et al., 2001 ; Wang et al., 2009) and is organized into tandem fibronectin type III (Fnlll)-like domains (D1 and D2) with unique characteristics (Fig. 2a). For instance it carries a conserved PSxW(S/T) sequence cassette instead of the WSxWS sequence fingerprint in other type I cytokine receptors, and displays an atypical disulfide-bridge network (Fig. 2a, Fig. 5 & 6). The latter is hallmarked by a cis peptide bond between C168 and C169 facilitating a rare type of surface-exposed disulfide-bridge (Fass, 2012) in the C2-E2 turn in TSLPR _D2 , and a C181 -C219 disulfide anchoring the atypical C-terminal tail against TSLPR _D2 . Finally, in the structure resulting from crystal form 1 a, a single GlcNAc carbohydrate moiety could be modeled at N53 in TSLPR, which is a well conserved N-linked glycosylation site in mammalian TSLPR sequences (Fig. 5). This GlcNAc residue interacts with residues located in strands D and E of TSLPR-D1 and may help stabilize the structure in this region.

Example 4: Binding interfaces in the TSLP:TSLPR:IL-7Ra complex

Arguably the centerfold of the TSLP:TSLPR:IL-7Ra complex is the kinked helix A in TSLP, which contributes to both the TSLP:TSLPR interface (site I) and the TSLP:IL-7Ra interface (site II) (Fig. 2a, b). In site I, helix A, helix D and the long overhand AB loop contact a TSLPR epitope defined by the EF1 -loop and the C-terminus of β-strand G1 in TSLPR _D1 , the 310-helix in the D1 -D2 linker region and the a-helical turn in the FG2-loop of TSLPR _D2 (Fig. 2a). Overall, the TSLP:TSLPR interface is characterized by a large number of hydrogen-bonds and salt- bridge interactions (Table 2), indicative of a highly specific interaction. In contrast to site I, site II is dominated by hydrophobic interactions and sees TSLP helix A pairing up with helix C to engage IL-7Ra via the CC'1 and EF1 loops in IL-7Ra _D i and the BC2 and FG2 loops in IL-7Ra _D2 (Fig. 2a and Table 2). A key interaction in site II appears to be mediated by I37 which protrudes from the ττ-helical turn on helix A of TSLP into a conserved hydrophobic pocket in the D1 -D2 junction of IL-7Ra defined by residues L101 , 1102, Y159, Y213 and F214 (Fig. 2a). The pronounced hydrophobic character of site II when compared with site I is consistent with the role of IL-7Ra as a shared receptor in the TSLP and IL-7 signaling complexes. The non-specific nature of site II is further supported by the observation that human IL-7Ra can engage mouse TSLP:TSLPR to form a high affinity complex (Park et al., 2000).

Helix B and the long CD overhand loop, which is largely disordered in all three structures, remain as the only segments in TSLP that do not contribute to any binding interfaces. However, the C-terminal half of the CD loop in many mammalian TSLP sequences including human TSLP, carries a highly basic amino acid cassette with as many as 7 consecutive basic residues (Fig. 3a). We propose that such a local concentration of positive charges may mediate interactions with the extracellular matrix to enhance availability in local tissues (Fry and Mackall, 2002; Vaday and Lider, 2000). Also, the CD loop may serve as an entry point for engineering new properties into antagonistic variants of TSLP for enhancing bioavailability and stability in a therapeutic context.

Contrary to the rather extensive interfaces observed between cognate receptors and shared receptors in ternary complexes of other IL-2 family members (typically 1300-1700 A2) (Wang et al., 2009), the extent of site III between TSLPR and IL-7Ra is much more limited (900 A2). Thus, site III is primarily constructed via contacts between residues in the EF2 loop of TSLPR _D2 , most notably TSLPR ^Gly175 , with the C'2 and £2 strands of IL-7Ra _D2 (Fig. 2a and Table 2). Example 5: Structural basis of the functional duality of IL-7Ra

Interestingly, the TSLP-bound conformation of mouse IL-7Ra is very similar to that of human I L-7Ra observed in the human IL-7:IL-7Ra complex (McElroy et al., 2009), providing possible insights into the duality and degeneracy of signaling via I L-7Ra (overall r.m.s.d. 0.93 A) (Fig. 7a). The cornerstone of the two interaction interfaces appears to be the accommodation of a hydrophobic β-branched amino acid projecting from the ττ-helical turn of helix A in TSLP and IL-7 (TSLP ¹³⁷ and IL-7 ^V40 ) by an evolutionarily conserved hydrophobic pocket at the D1 -D2 junction of IL-7Ra (Fig. 7b). The residues constituting this platform account for 60-70% of the IL-7Ra interface area in the respective binary interactions. In both cases, the cytokine:! L-7Ra interface is further strengthened by additional van der Waals contacts mediated by branched hydrophobic residues protruding from the CC'1 and EF1 loops of IL-7RD1 (Fig. 7b) and specific hydrogen-bonding interactions between main-chain carbonyl oxygen atoms in the CC'1 and EF1 loops in IL-7Ra _D i with long side-chains reaching out from helix C. Thus, IL-7Ra establishes its binding degeneracy to serve both as cognate and shared receptor in two functionally distinct signaling assemblies (Fig. 7c) by using an evolutionarily conserved hydrophobic platform coupled to the hydrogen-bonding potential of main-chain carbonyl oxygen atoms.

We further note that in the IL-7Ra binding epitopes of TSLP and IL-7 very few of the IL- 7Ra interacting residues are conserved between TSLP and IL-7. However, the majority of the IL-7Ra interacting residues do occupy equivalent positions in the A and C helices of TSLP and IL-7. Structural comparison between the mTSLP:IL-7Ra and hlL-7:IL-7Ra complexes suggests that IL-7 and TSLP establish IL-7Ra binding mainly by nonspecific van der Waals interactions rather than the structural mimicry of specific polar interactions. Along this line, the π-helical turn in the kinked A-helix of TSLP and IL-7 is a key structural element employed for their interaction with IL-7Ra and provides excellent surface complementarity with the hydrophobic platform presented by IL-7Ra. This π-helical turn was likely acquired by a single amino-acid insertion in the ancestral a-helix in IL-2 family cytokines and illustrates how the introduction of a π-helical turn can steer functional diversification (Cooley et al., 2010). Intriguingly, the relative orientation of TSLPR _D2 and IL-7Ra _D 2 resembles that observed for yc _D 2 and IL-13R _D 2 and their co-receptors in all other structurally characterized ternary complexes of IL-2 family cytokines (IL-2Rp:yc, IL-4Ra:yc and IL-4Ra:IL-13Ra1 ) (LaPorte et al., 2008;Ring et al., 2012) (Fig. 7d). This points to possible evolutionary pressure to maintain the relative orientation of the membrane proximal Fnlll-like domains, which raises confidence that the IL- 7:IL-7Ra:yc ternary complex will likely follow suit.

Example 6: TSLPR primes TSLP for high affinity binding to IL-7Ra

To dissect the thermodynamic and kinetic profile underlying the assembly of the ternary TSLP complex we employed isothermal titration calorimetry (ITC) and surface plasmon resonance (SPR) based on recombinant TSLP, TSLPR and I L-7Ra receptor ectodomains expressed in mammalian cells with native-like glycosylation. In the first instance we used ITC to characterize binary complexes of TSLP with its two receptors. Whereas the TSLP:TSLPR binary complex could be readily formed with an equilibrium dissociation constant (KD) of 58 nM coupled to a strongly enthalpic thermodynamic fingerprint consistent with the polar character of site I (Fig. 8a), no binding could be detected for the TSLP:IL-7Ra binary interaction at the low μΜ concentrations used in the ITC experiment. However, titration of IL-7Ra with TSLP:TSLPR complex isolated by SEC resulted in the formation of a high affinity complex with a KD-value of 1 .5 nM (Fig. 8a). This demonstrates that prior binding of TSLP to TSLPR is critical for recruiting IL-7Ra to a high-affinity ternary complex. To confirm the oligomeric state of TSLP complexes, we have performed multi-angle light scattering (MALS) measurement combined with SEC on recovered samples following ITC measurements and found that the assembly stoichiometry in solution agrees with our crystal structures. We subsequently conducted SPR experiments employing an immobilized biotinylated version of TSLP to further characterize the cooperative assembly of the TSLP ternary complex. This led to a reliable estimate for the inherently weak affinity of the TSLP:IL-7Ra binary complex (KD=2.3 μΜ) (Fig. 8b). We were also able to determine that binding of IL-7Ra to preassembled TSLP:TSLPR complex in the presence of saturating concentrations of TSLPR (250 nM) occurs with a fast kon of 3.3 x 106 M ^" V ¹ coupled to a moderately slow koff of 8.7 x 10 ^"3 s ^"1 to yield the high affinity complex (KD=2.6 nM) (Fig. 8c).

Example 7: Identification of putative interface hot spots in the TSLP ternary complex

To validate our mechanistic insights and to identify possible hotspot regions in the cytokine- receptor interfaces defined by sites I and II, we screened a set of TSLP, TSLPR and IL-7Ra point mutants by cellular assays. We found that an I37E point-mutant in TSLP retains high affinity binding to TSLPR (Fig. 9a), but is unable to establish a signaling complex in a STAT5- based activity assay (Fig. 9b). Moreover, the structurally reciprocal point-mutant Y213R in IL- 7Ra (Fig. 2a) also abolished signaling (Fig. 9c). Together, these results identify I37 in TSLP and Y213 in IL-7Ra as complementary functional hot-spots, and suggest that the I37E TSLP variant could serve as a starting point for antagonist development. Two mutant variants in site I of TSLP, F39A/Y133A and D56R showed drastically reduced affinity for TSLPR (Fig. 9d,g ), and an increased EC50 value in the STAT5 activation assay (Fig. 9e,h), in line with their role in TSLPR binding. TSLP-D56 makes a salt-bridge with TSLPR-R1 10 and TSLP-Y133 contacts TSLPR-G196 (Fig. 2a). A TSLPR R1 10E mutation increases the EC50 value of STAT5 activation, while a G196R mutant is completely inactive, hinting the importance of these interactions in site I (Fig. 9f,i). Our functional probing of the receptor-receptor interface defined by site III was based on four point-mutations. Three of them concern conserved residues in TSLPR (Q162A, D177S, and R180S) (Fig. 4) and one probes the importance of position 185 in IL-7Ra (H185A). Residue H185 corresponds to residue S185 in human IL-7Ra which is mutated to cysteine at a high frequency in T-ALL and B-ALL patients (Sochat et al., 201 1 ; Zenatti et al., 201 1 ). We found that all four site III mutants were indistinguishable from wild-type receptors in our cellular activity assays. The overwhelming conservation of such receptor-receptor orientation in all other structurally characterized ternary complexes of IL-2 family cytokines (Fig. 7d) and the fact that we observed two nearly identical site III contacts in two distinct crystal forms provides confidence about the biological relevance of our structural observations and mechanistic deductions.

On the base of the mouse model, the TSLP amino acids that are essential for the receptor interaction have been identified in human TSLP. The equivalent of mouse TSLP I37 is human TSLP T46 (numbering according to SEQ ID No.1 ). Other residues that are essential in the receptor interaction are S45, K49, D50 and K101. To a lesser extent, the residues A41 , A42, M97 and A104 do play a role. To validate the identification of the human TSLP residues, we screened a human TSLP double mutant by cellular assay. We found that introduction of the S45R+T46R substitutions in human TSLP reveals a mutant TSLP that binds with high affinity to the human TSLPR (Fig. 10a), but induces TSLP induced signaling at least 10 times less than human TSLP which does not contain said mutations of residues S45 and T46. (Fig. 10b).

Example 8: Mechanistic assembly of the TSLP-mediated signaling complex

Together our data show that the formation of the TSLP signaling complex occurs sequentially and cooperatively, with the TSLP:TSLPR binary complex priming the binding affinity of TSLP to IL-7Ra by three orders of magnitude to establish a highly stable ternary complex. Thus, engagement of IL-7Ra in TSLP-mediated signaling can only become relevant once the TSLP:TSLPR concentrations have reached a high enough level, thereby allowing IL-7Ra to serve its dual signaling role via its cognate IL-7:IL-7Ra:YC complex (Fig. 7c). Our binding studies are consistent with cellular binding studies showing that mouse TSLP binds with moderate affinity to TSLPR alone (KD = ~ 8 nM) and with high affinity (KD = ~ 56 pM) in the presence of both TSLPR and IL-7Ra (Park et al., 2000). Importantly, our structural studies point to two possible scenarios that could underlie cooperative assembly of the

TSLP:TSLPR:IL-7Ra complex. In the first instance, recruitment of IL-7Ra could be attributed to multipoint attachment mediated by two relatively weak binding sites (site II and site III) as recently suggested for other cytokine-receptor complexes (LaPorte et al., 2008; Verstraete and Savvides, 2012). Secondly, the TSLP:TSLPR binary complex might induce conformational changes in TSLP to enable high affinity binding to IL-7Ra by analogy to other IL-2 family members Levin et al., 2012; Hage et al., 1999). We propose that the structural heterogeneity and flexibility of the ττ-helical turn at the elbow of helix A in TSLP (Fig. 2 and 4) could serve as a switch that can be activated upon TSLPR binding to render I37 amenable to high affinity binding to IL-7Ra (Fig. 8). Finally, the structural work we presented here provides a starting point for rationalizing clinically relevant point-mutations occurring in TSLPR and IL-7Ra. For instance, a F232C mutation in TSLPR has been identified in patients with B-cell acute lymphoblastic leukemia (B-ALL) leading to disulfide-linked TSLPR and constitutive JAK activation (Yoda et al, 2010). This mutation is located at the very end of the juxtamembrane region that extends from the construct of TSLPR in the crystal structure of the ternary complex leading to the transmembrane (TM) helix of TSLPR. Nonetheless, the location of position 232 had remained unclear. After analysis of the membrane insertion potential (Senes et al., 2007) of sequences of different lengths that could correspond to the JM and TM in TSLPR we propose that position 232 is likely inserted in the membrane as an interfacial residue. This means that its mutation to cysteine could render TSLPR readily amenable to disulfide linkage to another TSLPR, a process that can readily take place within the lipidic environment of the membrane (White and Wimley1999). A large number of mutations in the juxtamembrane and TM segments of IL-7Ra have been identified in B-ALL and T-ALL patients and have been proposed to serve as gain-of-function mutations (Sochat et al., 201 1 ; Zenatti et al., 201 1 ). Several of these mutations introduce cysteine residues which may covalently dimerize IL-7Ra (Walsh, 2012; Sochat et al., 201 1 ; Zenatti et al., 201 1 ) in a manner that resembles our proposal for the F232C somatic mutation in TSLPR. Nonetheless, a somatic mutation at position 185 in human IL-7Ra (S185C) (Sochat et al., 201 1 ), would map to H185 in site III at the interface of the mouse TSLP:TSLPR:IL-7Ra complex (Fig. 2a) far away from the juxtamembrane and TM regions. While the recently proposed assembly of a twofold-symmetric dimer of IL-7Ra linked by a disulfide bond between opposing C185 residues is an interesting working hypothesis (Walsh, 2012; McElroy et al., 2012) it also appears to be incompatible with the recent crystal structure of dimeric wild-type IL-7Ra (McElroy et al., 2012) that features a very different dimer interface. However, what we can be confident about based on our structure of TSLP:TSLPR:IL-7Ra complex is that position 185 is embedded in a part of IL-7Ra that has evolved to participate in receptor-receptor interactions. This strengthens the notion that a disulfide-bond at this position could be well poised to lock two membrane proximal domains of IL-7Ra in a signaling-competent orientation. REFERENCES

Abonia, J. P. & Rothenberg, M. E. Eosinophilic esophagitis: rapidly advancing insights. Annu Rev Med 63, 421 -434, doi:10.1 146/annurev-med-041610-134138 (2012).

Aricescu, A. R., Lu, W. & Jones, E. Y. A time- and cost-efficient system for high-level protein production in mammalian cells. Acta Crystallogr D Biol Crystallogr 62, 1243-1250, doi:S0907444906029799 [pii] 0.1 107/S0907444906029799 (2006).

Bell, B. D. et al. The transcription factor STAT5 is critical in dendritic cells for the development of TH2 but not TH1 responses. Nat Immunol 14, 364-371 , doi:10.1038/ni.2541 (2013).

- Bernat, B., Pal, G., Sun, M. & Kossiakoff, A. A. Determination of the energetics governing the regulatory step in growth hormone-induced receptor homodimerization. Proc Natl Acad Sci U S A 100, 952-957, doi:10.1073/pnas.0235023100 (2003).

Bondensgaard, K. et al. The existence of multiple conformers of interleukin-21 directs engineering of a superpotent analogue. J Biol Chem 282, 23326-23336, doi:10.1074/jbc.M701313200 (2007).

Borowski, A. et al. Expression analysis and specific blockade of the receptor for human thymic stromal lymphopoietin (TSLP) by novel antibodies to the human TSLPRalpha receptor chain. Cytokine 61 , 546-555, doi:10.1016/j.cyto.2012.10.025 (2013).

Cheng, D. T. et al. Thymic stromal lymphopoietin receptor blockade reduces allergic inflammation in a cynomolgus monkey model of asthma. J Allergy Clin Immunol 132, 455-

462, doi:10.1016/j.jaci.2013.05.01 1 (2013).

Cooley, R. B., Arp, D. J. & Karplus, P. A. Evolutionary origin of a secondary structure: pi- helices as cryptic but widespread insertional variations of alpha-helices that enhance protein functionality. J Mol Biol 404, 232-246, doi:10.1016/j.jmb.2010.09.034 (2010).

- Datta, A. et al. Evidence for a Functional Thymic Stromal Lymphopoietin Signaling Axis in Fibrotic Lung Disease. J Immunol, doi:10.4049/jimmunol.1300588 (2013).

Decramer, M., Janssens, W. & Miravitlles, M. Chronic obstructive pulmonary disease. The Lancet 379, 1341 -1351 (2012).

DeLano, W. L. & Lam, J. W. PyMOL: A communications tool for computational models. Abstracts of Papers of the American Chemical Society 230, U1371 -U1372 (2005).

De Monte, L. et al. Intratumor T helper type 2 cell infiltrate correlates with cancer- associated fibroblast thymic stromal lymphopoietin production and reduced survival in pancreatic cancer. J Exp Med 208, 469-478, doi:10.1084/jem.20101876 (201 1 ).

Dundas, J. et al. CASTp: computed atlas of surface topography of proteins with structural and topographical mapping of functionally annotated residues. Nucleic Acids Res 34, W1 16-1 18, doi:10.1093/nar/gkl282 (2006). Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta Crystallogr D Biol Crystallogr 66, 486-501 , doi:S0907444910007493 [pii]10.1 107/S0907444910007493 [doi] (2010).

Eswar, N., Eramian, D., Webb, B., Shen, M. Y. & Sali, A. Protein structure modeling with MODELLER. Methods Mol Biol 426, 145-159, doi:10.1007/978-1 -60327-058-8_8 (2008).

Fass, D. Disulfide bonding in protein biophysics. Annual review of biophysics 41 , 63-79, doi: 10.1 146/annurev-biophys-05051 1 -102321 (2012).

- Fry, T. J. & Mackall, C. L. lnterleukin-7: from bench to clinic. Blood 99, 3892-3904 (2002).

Gouet, P., Courcelle, E., Stuart, D. I. & Metoz, F. ESPript: analysis of multiple sequence alignments in PostScript. Bioinformatics 15, 305-308, doi:btc028 [pii] (1999).

Guerra, S. Asthma and chronic obstructive pulmonary disease. Current opinion in allergy and clinical immunology 9, 409-416, doi:10.1097/ACI.0b013e3283300baf (2009).

Hage, T., Sebald, W. & Reinemer, P. Crystal structure of the interleukin-4/receptor alpha chain complex reveals a mosaic binding interface. Cell 97, 271 -281 , doi:S0092- 8674(00)80736-9 [pii] (1999).

Holgate, S. T. Innate and adaptive immune responses in asthma. Nature medicine 18, 673- 683, doi:10.1038/nm.2731 (2012).

Hunninghake, G. M. et al. TSLP polymorphisms are associated with asthma in a sex- specific fashion. Allergy 65, 1566-1575, doi:10.1 1 1 1/j.1398-9995.2010.02415.x (2010). - Kabata, H. et al. Thymic stromal lymphopoietin induces corticosteroid resistance in natural helper cells during airway inflammation. Nature communications 4, 2675, doi:10.1038/ncomms3675 (2013).

- Kabsch, W. XDS. Acta Crystallogr D Biol Crystallogr 66, 125-132, doi:S0907444909047337 [pii] 0.1 107/S0907444909047337 [doi] (2010).

- Krissinel, E. & Henrick, K. Secondary-structure matching (SSM), a new tool for fast protein structure alignment in three dimensions. Acta Crystallogr D Biol Crystallogr 60, 2256-2268, doi:10.1 107/s0907444904026460 (2004).

Lambrecht, B. N. & Hammad, H. Lung dendritic cells in respiratory viral infection and asthma: from protection to immunopathology. Annu Rev Immunol 30, 243-270, doi:10.1 146/annurev-immunol-02071 1 -075021 (2012).

LaPorte, S. L. et al. Molecular and structural basis of cytokine receptor pleiotropy in the interleukin-4/13 system. Cell 132, 259-272, doi:S0092-8674(08)00047-0 [pii] 10.1016/j.cell.2007.12.030 [doi] (2008).

- Larkin, M. A. et al. Clustal W and Clustal X version 2.0. Bioinformatics 23, 2947-2948, doi:btm404 [pii]10.1093/bioinformatics/btm404 [doi] (2007).

Levin, A. M. et al. Exploiting a natural conformational switch to engineer an interleukin-2 'superkine'. Nature 484, 529-533, doi:10.1038/nature10975 (2012). Leyva-Castillo, J. M. et al. Skin thymic stromal lymphopoietin initiates Th2 responses through an orchestrated immune cascade. Nature communications 4, 2847, doi:10.1038/ncomms3847 (2013).

Licona-Limon, P., Kim, L. K., Palm, N. W. & Flavell, R. A. TH2, allergy and group 2 innate lymphoid cells. Nat Immunol 14, 536-542, doi:10.1038/ni.2617 (2013).

Lindorff-Larsen, K. et al. Improved side-chain torsion potentials for the Amber ff99SB protein force field. Proteins 78, 1950-1958, doi:10.1002/prot.2271 1 (2010).

Liu, W. et al. Two single nucleotide polymorphisms in TSLP gene are associated with asthma susceptibility in Chinese Han population. Experimental lung research 38, 375-382, doi:10.3109/01902148.2012.714840 (2012).

Mackall, C. L, Fry, T. J. & Gress, R. E. Harnessing the biology of IL-7 for therapeutic application. Nat Rev Immunol 1 1 , 330-342, doi:10.1038/nri2970 (201 1 ).

McCoy, A. J. Solving structures of protein complexes by molecular replacement with Phaser. Acta Crystallogr D Biol Crystallogr 63, 32-41 , doi:S0907444906045975 [pii] 10.1 107/S0907444906045975 (2007).

McElroy, C. A., Dohm, J. A. & Walsh, S. T. Structural and biophysical studies of the human IL-7/IL-7Ralpha complex. Structure 17, 54-65, doi:10.1016/j.str.2008.10.019 (2009).

McElroy, C. A. et al. Structural reorganization of the interleukin-7 signaling complex. Proc Natl Acad Sci U S A 109, 2503-2508, doi:10.1073/pnas.1 1 16582109 (2012).

- Noti, M. et al. Thymic stromal lymphopoietin-elicited basophil responses promote eosinophilic esophagitis. Nature medicine 19, 1005-1013, doi:10.1038/nm.3281 (2013). Pandey, A. et al. Cloning of a receptor subunit required for signaling by thymic stromal lymphopoietin. Nat Immunol 1 , 59-64, doi:10.1038/76923 (2000).

Park, L. S. et al. Cloning of the murine thymic stromal lymphopoietin (TSLP) receptor: Formation of a functional heteromeric complex requires interleukin 7 receptor. J Exp Med

192, 659-670 (2000).

Pedroza-Gonzalez, A. et al. Thymic stromal lymphopoietin fosters human breast tumor growth by promoting type 2 inflammation. J Exp Med 208, 479-490, doi:10.1084/jem.20102131 (201 1 ).

- Peelman, F. et al. Mapping of the leptin binding sites and design of a leptin antagonist. J Biol Chem 279, 41038-41046, doi:10.1074/jbc.M404962200 [doi]M404962200 [pii] (2004). Perez-Andreu, V. et al. Inherited GATA3 variants are associated with Ph-like childhood acute lymphoblastic leukemia and risk of relapse. Nature genetics, doi:10.1038/ng.2803 (2013).

- Pronk, S. et al. GROMACS 4.5: a high-throughput and highly parallel open source molecular simulation toolkit. Bioinformatics 29, 845-854, doi:10.1093/bioinformatics/btt055 (2013). Quentmeier, H. et al. Cloning of human thymic stromal lymphopoietin (TSLP) and signaling mechanisms leading to proliferation. Leukemia 15, 1286-1292 (2001 ).

Reche, P. A. et al. Human thymic stromal lymphopoietin preferentially stimulates myeloid cells. J Immunol 167, 336-343 (2001 ).

Redhu, N. S. & Gounni, A. S. Function and mechanisms of TSLP/TSLPR complex in asthma and COPD. Clinical and experimental allergy : journal of the British Society for Allergy and Clinical Immunology 42, 994-1005, doi:10.1 1 1 1/j.1365-2222.201 1.03919.x (2012).

Reeves, P. J., Callewaert, N., Contreras, R. & Khorana, H. G. Structure and function in rhodopsin: high-level expression of rhodopsin with restricted and homogeneous N- glycosylation by a tetracycline-inducible N-acetylglucosaminyltransferase l-negative HEK293S stable mammalian cell line. Proc Natl Acad Sci U S A 99, 13419-13424, doi:10.1073/pnas.212519299 [doi]212519299 [pii] (2002).

Ring, A. M. et al. Mechanistic and structural insight into the functional dichotomy between IL-2 and IL-15. Nat Immunol 13, 1 187-1 195, doi:10.1038/ni.2449 (2012).

Roan, F. et al. The multiple facets of thymic stromal lymphopoietin (TSLP) during allergic inflammation and beyond. Journal of leukocyte biology 91 , 877-886, doi:10.1 189/jlb.121 1622 (2012).

Romeo, M. J., Agrawal, R., Pomes, A. & Woodfolk, J. A. A molecular perspective on T2- promoting cytokine receptors in patients with allergic disease. J Allergy Clin Immunol, doi:10.1016/j.jaci.2013.08.006 (2013).

Rothenberg, M. E. et al. Common variants at 5q22 associate with pediatric eosinophilic esophagitis. Nature genetics 42, 289-291 , doi:10.1038/ng.547 (2010).

Schmitt, J. et al. Assessment of clinical signs of atopic dermatitis: A systematic review and recommendation. Journal of Allergy and Clinical Immunology, doi:http://dx.doi.org/10.1016/j.jaci.2013.07.008 (2013).

Senes, A. et al. E(z), a depth-dependent potential for assessing the energies of insertion of amino acid side-chains into membranes: derivation and applications to determining the orientation of transmembrane and interfacial helices. J Mol Biol 366, 436-448, doi:10.1016/j.jmb.2006.09.020 (2007).

Sims, J. E. et al. Molecular cloning and biological characterization of a novel murine lymphoid growth factor. J Exp Med 192, 671 -680 (2000).

Siracusa, M. C, Kim, B. S., Spergel, J. M. & Artis, D. Basophils and allergic inflammation. J

Allergy Clin Immunol 132, 789-801 , doi:10.1016/j.jaci.2013.07.046 (2013a).

Siracusa, M. C. et al. Thymic stromal lymphopoietin-mediated extramedullary hematopoiesis promotes allergic inflammation. Immunity 39, 1 158-1 170, doi:10.1016/j.immuni.2013.09.016 (2013b). Shochat, C. et al. Gain-of-function mutations in interleukin-7 receptor-alpha (IL7R) in childhood acute lymphoblastic leukemias. J Exp Med 208, 901 -908, doi:10.1084/jem.201 10580 (201 1 ).

Spergel, J. M. From atopic dermatitis to asthma: the atopic march. Annals of allergy, asthma & immunology : official publication of the American College of Allergy, Asthma, & Immunology 105, 99-106; quiz 107-109, 1 17, doi:10.1016/j.anai.2009.10.002 (2010).

Torgerson, D. G. et al. Meta-analysis of genome-wide association studies of asthma in ethnically diverse North American populations. Nature genetics 43, 887-892, doi:10.1038/ng.888 (201 1 ).

- Vaday, G. G. & Lider, O. Extracellular matrix moieties, cytokines, and enzymes: dynamic effects on immune cell behavior and inflammation. Journal of leukocyte biology 67, 149- 159 (2000).

Verstraete, K. et al. Efficient production of bioactive recombinant human Flt3 ligand in E. coli. Protein J 28, 57-65, doi:10.1007/s10930-009-9164-5 (2009).

- Verstraete, K. & Savvides, S. N. Extracellular assembly and activation principles of oncogenic class III receptor tyrosine kinases. Nat Rev Cancer 12, 753-766, doi:10.1038/nrc3371 (2012).

Walsh, S. T. Structural insights into the common gamma-chain family of cytokines and receptors from the interleukin-7 pathway. Immunological reviews 250, 303-316, doi:10.1 1 1 1/j.1600-065X.2012.01 160.x (2012).

- Wang, X., Lupardus, P., Laporte, S. L. & Garcia, K. C. Structural biology of shared cytokine receptors. Annu Rev Immunol 27, 29-60, doi:10.1 146/annurev.immunol.24.021605.090616 [doi] (2009).

White, S. H. & Wimley, W. C. Membrane protein folding and stability: physical principles. Annual review of biophysics and biomolecular structure 28, 319-365, doi:10.1 146/annurev.biophys.28.1.319 (1999).

Wilson, S. R. et al. The Epithelial Cell-Derived Atopic Dermatitis Cytokine TSLP Activates Neurons to Induce Itch. Cell 155, 285-295, doi:10.1016/j.cell.2013.08.057 (2013).

Word, J. M. et al. Visualizing and quantifying molecular goodness-of-fit: small-probe contact dots with explicit hydrogen atoms. J Mol Biol 285, 171 1 -1733, doi:10.1006/jmbi.1998.2400 (1999).

- Yang, Z. et al. UCSF Chimera, MODELLER, and IMP: an integrated modeling system. J Struct Biol 179, 269-278, doi:10.1016/j.jsb.201 1 .09.006 (2012).

Yoda, A. et al. Functional screening identifies CRLF2 in precursor B-cell acute lymphoblastic leukemia. Proc Natl Acad Sci U S A 107, 252-257, doi: 10.1073/pnas.091 1726107 (2010). Zabeau, L. et al. Leptin receptor activation depends on critical cysteine residues in its fibronectin type III subdomains. J Biol Chem 280, 22632-22640, doi: 10.1074/j be. M413308200 (2005).

Zenatti, P. P. et al. Oncogenic IL7R gain-of-function mutations in childhood T-cell acute lymphoblastic leukemia. Nature genetics 43, 932-939, doi:10.1038/ng.924 (201 1 ).

Zhang, F., Huang, G., Hu, B., Song, Y. & Shi, Y. A soluble thymic stromal lymphopoietin (TSLP) antagonist, TSLPR-immunoglobulin, reduces the severity of allergic disease by regulating pulmonary dendritic cells. Clin Exp Immunol 164, 256-264, doi:10.1 1 1 1/j.1365- 2249.201 1 .04328.x (201 1 ).

Ziegler, S. F. Thymic stromal lymphopoietin and allergic disease. J Allergy Clin Immunol 130, 845-852, doi:10.1016/j.jaci.2012.07.010 (2012).

Ziegler, S. F. & Artis, D. Sensing the outside world: TSLP regulates barrier immunity. Nat Immunol 1 1 , 289-293, doi:10.1038/ni.1852 (2010).

Ziegler, S. F. et al. The biology of thymic stromal lymphopoietin (TSLP). Advances in pharmacology 66, 129-155, doi:10.1016/B978-0-12-404717-4.00004-4 (2013).