**1. Introduction**

Interleukin-1 (IL-1) family cytokines play important roles in regulating and initiating inflammatory and immunological responses [1]. The IL-1 family includes eleven cytokines comprising seven agonist ligands, three receptor antagonists, and an anti-inflammatory cytokine [2]. Interleukin 33 (hereafter called as "IL-33") cytokine is identified as one of the IL-1 family agonist ligands [3]. It was first regarded as an alarmin that is released to signal immune system when a cell or tissue is damaged or stressed [4]. Recently, IL-33 has been considered as an important factor of the immune system involved in allergic inflammation and chronic diseases such as asthma, atopic dermatitis, and allergic rhinitis [5–7].

IL-33, expressed in endothelial cells, fibroblasts, epithelial cells, and other cells, binds to its receptor suppressor of tumorigenicity 2 (ST2)/interleukin 1 receptor-like 1 (IL1RL1), which formed heterodimer with co-receptor, IL-1 receptor accessory protein (IL1RAcP) [4,8]. There are two types of ST2 isoforms: the transmembrane form, ST2, and soluble form, sST2, covering residues 19 through 321 of the ectodomain (hereafter, we call all isoforms as simply "ST2"). The ST2 is expressed on various immune cells including innate lymphoid group 2 cells (ILC2s), mast cells, dendritic cells, macrophages, basophils, and type 2 helper T cells (Th2), and it is linked to Th2 effector functions [9,10]. IL-33 exerts its biological functions followed by binding to ST2 expressed in immune cells, and it is

mainly associated with Th2 responses through the production of inflammatory cytokines IL-5 and IL-13 [3,11]. The heterodimer complex formation activates downstream signaling complex formation. Myeloid differentiation primary response 88 (MyD88) first binds to heterodimeric receptor and leads to the recruitment of interleukin-1 receptor-associated kinase 1 (IRAK1), IRAK4 and tumor necrosis factor (TNF) receptor-associated factor 6 (TRAF6), and these subsequently activate mitogen-activated protein kinases (MAPKs) and nuclear factor κB (NF-κB) signaling pathways to promote inflammatory cytokine production [9,12,13]. It seems that IL-33 has the potential to activate Th2 cytokine-mediated allergic inflammation and related diseases, suggesting that blockade of the IL-33/ST2 signaling axis can be a new therapeutic strategy for allergic inflammation and chronic inflammatory diseases [14–16].

Several strategies have been developed to suppress the IL-33 mediated downstream signaling pathway to prevent chronic diseases and allergic inflammation: antagonists against IL-33, and antagonists against ST2 or sST2 binding to IL-33 [17]. Here, we describe the discovery and characterization of single-chain variable fragment (scFv) monoclonal antibodies (mAbs) directly targeting IL-33 to inhibit IL-33 binding to ST2. Although there are diverse antibodies against IL-33 for various purposes, they are mainly derived from immunizing living organisms with immunoglobulin G forms [16,18] or monoclonal antibodies for IL-33 detection [15,19]. Since immunoglobulin G (IgG) forms of antibodies are hard to handle and not suitable to further engineering, we used a human synthetic single-chain variable fragment (scFv) antibody library to screen IL-33 specific mAbs in vitro.

The discovery of mAbs using phage display library was performed with five rounds of biopanning, and enzyme-linked immunosorbent assay (ELISA) was used to determine the antibodies affinity. Using immunoblotting, we observed their cross-reactivity, and two types of mutant-based epitope mapping were implemented to identify the binding epitope domains. The inhibition effect of antibody was verified by glutathione *S*-transferase (GST) pull-down assay and human cell-expressing ST2 and IL1RAcP-based assay. The antibody seems to have therapeutic function by interfering with IL-33 binding to the ST2 receptor, heterodimeric receptor complex formation, and blocking the IL-33/ST2 signaling axis.

#### **2. Results**

#### *2.1. Selection of scFvs Specific to IL-33*

Human IL-33 is composed of three domains: N-terminal nuclear domain (residues 1–65), central domain (residues 66–111) and C-terminal IL-1-like cytokine domain (residues 112–270) [3,8,20]. The N-terminal and central domains of IL-33 are cleaved by caspase-1 to produce the mature form [3]. IL-33 is susceptible to oxidation by forming disulfide bonds among cysteine residues (C208, C227, C232, and C259). IL-33 oxidation reportedly drives a conformational change and inactivates its ST2-dependent cytokine activities [21,22]. To prevent from oxidation, we mutated C208 and C232 to serine and compared the activity of C208S/C232S mutant with that of IL-33 wild-type (WT). The purity of WT and C208S/C232S mutant was confirmed by SDS-PAGE (Figure S1A,B). The recognition of IL-33 WT and C208S/C232S mutant by a selected scFv (see next section) was comparable as corroborated by SDS-PAGE and immunoblot analyses (Figure S1C). We used GST-IL-33 WT for biopanning and characterizations and IL-33 C208S/C232S mutant for cell signaling analysis.

We performed five rounds of biopanning to select scFvs specific to IL-33 using a large synthetic human scFv library in two distinct conditions according to the number of negative selections (Table S1). GST-IL-33 and GST were used as antigens for positive and negative selections of biopanning, respectively. Ten scFv clones with high OD450 values in response to IL-33 compared to the negative selection were selected by ELISA screening (Figure S2A). Of the ten clones, clones with mutations in the backbone frame or duplicated sequences were excluded through multiple protein sequence alignment (Figure S2B). Finally, six clones (C1\_1E1, C2\_1D5, C2\_2A10, C2\_2E1, C2\_2E12, and C2\_2H5) were chosen based on their high binding signals at 450 nm without mutations in the amino acid sequences. Multiple sequence alignment revealed that these six clones have different amino acid residues mostly in the third

complementarity determining region in heavy chain (CDR-H3) and the second complementarity determining region in light chain (CDR-L2).

The *E. coli* cell lysates containing overexpressed His6-tagged scFvs were prepared to determine the binding affinity of selected scFvs with IL-33. The dissociation constants (*K*d) values of C1\_1E1, C2\_1D5, C2\_2A10, C2\_2E1, C2\_2E12, and C2\_2H5 by ELISA were estimated to be 48, 36, 57, 35, 28, and 31 nM, respectively (Figure 1A). The *K*<sup>d</sup> value of C2\_2E12 that showed the highest affinity using cell lysate was further measured using the purified proteins by ELISA (Figure S2C and Figure S1B). The *K*<sup>d</sup> value of the purified C2\_2E12 was 38 nM (Figure 1B), which is consistent with the value estimated using the cell lysate. We selected C2\_2E12 for further characterizations.

The cross-reactivity of C2\_2E12 was checked for two interleukins belonging to the same subfamily (GST-IL-1β and IL-6) and three unrelated proteins (GST, bovine serum albumin (BSA) and IlvC). Immunoblot assay results showed that C2\_2E12 only reacted with IL-33 (Figure 1C). It is interesting that C2\_2E12 did not react with IL-1β and IL-6, since IL-33, IL-1β, and IL-6 belong to the same subfamily. Human IL-33 shows low sequence identities to IL-1β and IL-6 despite all three belonging to the same subfamily: 13.5% with IL-1β and 12.9% with IL-6, respectively (Figure 1D). The structure of IL-1β (PDB ID: 1L2H) is similar to that of IL-33 (PDB ID: 4KC3) with a root mean square deviation (r.m.s.d.) of 1.93 Å, while the structure of IL-6 (PDB ID: 1ALU) is completely different from that of IL-33. Given the low sequence similarities and structural differences, no cross-reactivity of C2\_2E12 for IL-1β and IL-6 seems to be reasonable. The cross-reactivity results clearly demonstrate that C2\_2E12 specifically binds to IL-33.

### *2.2. Epitope Mapping*

To determine the epitope region in IL-33 for C2\_2E12, a series of GST-IL-33112-270 N-terminal deletion mutants were constructed by the insertion of a stop codon at the end of each α-helix or β-strand of IL-33 based on the crystal structure of human IL-33 (PDB ID: 4KC3) [23] (Figure 2A). Immunoblot analysis revealed that residues 149–158 of the IL-33 comprised the epitope region, which corresponded to its receptor ST2 binding site in the crystal structure of the IL-33:ST2 complex (PDB ID: 4KC3). We found that the other five scFv clones (C1\_1E1, C2\_1D5, C2\_2A10, C2\_2E1, C2\_2E12, and C2\_2H5) also recognized the same epitope region in the IL-33 (Figure 2B). Alanine scanning mutagenesis was performed to determine the critical residue(s) in the epitope region (Figure 2C). Each residue in GST-IL-33149-158 was substituted to alanine by site-directed mutagenesis PCR. The effects of the IL-33 mutants were analyzed by immunoblots with C2\_2E12 as the primary antibody. Alanine substitutions of L150 and K151 of IL-33 reduced the binding with C2\_2E12, rendering these the key residues in the epitope region. To obtain further insights on alanine scanning results at the molecular level, we performed molecular docking between IL-33 and C2\_2E12 using the HADDOCK server with restraints that only L150 and K151 residues of IL-33 and CDR residues of C2\_2E12 should participate in interactions. Although alanine scanning data showed that L150 of IL-33 is a key residue of IL-33 and C2\_2E12 binding, L150 seemed to not interact with any residue of C2\_2E12. Alternatively, L150 seemed to possibly interact with the surrounding hydrophobic residues of the 149–158 epitope region of IL-33, and it also seemed to play an important role in maintaining the shape of the loop (Figure 2D). It seems that the L150A mutant inhibits the interaction between IL-33 and C2\_2E12 by local conformational changes of the loop. The docked structural model of IL-33:C2\_2E12 suggested that K151 of IL-33 seemed to interact electrostatically with the acidic pocket of C2\_2E12 composed of D164, S166, Y168, A218, and Y230 (Figure 2E). This structural analysis with a docked model between IL-33 and C2\_2E12 supports that L150 and K151 residues of IL-33 are important for their binding to C2\_2E12.

**Figure 1.** Molecular characterizations of anti-IL-33 single-chain antibody variable fragments (scFvs). Enzyme-linked immunosorbent assay (ELISA)-based affinity determination of anti-IL-33 scFvs. (**A**) Affinity determination of the top six clones that exhibited high binding signals against IL-33. (**B**) Affinity determination of the purified C2\_2E12, which seems to have the highest binding signal and good protein condition among six clones. *K*<sup>d</sup> values of C1\_1E1, C2\_1D5, C2\_2A10, C2\_2E1, C2\_2E12, and C2\_2H5 were estimated by kinetic analysis. ELISA was repeated three times. (**C**) Immunoblot analysis of the recombinant proteins using C2\_2E12 for primary antibody (0.5 mg·mL<sup>−</sup>1, 1:100 dilution) and anti-hemagglutinin-horse radish peroxidase (anti-HA-HRP) for secondary antibody (0.2 mg·mL<sup>−</sup>1, 1:5000 dilution). Expression of recombinant interleukin cytokines (GST-IL-33, GST-IL1β, and GST-IL6) and unrelated proteins (GST, BSA, and IlvC) in *E. coli* BL21 (DE3) as revealed by SDS-PAGE analysis. IB, immunoblot. (**D**) Multiple protein sequence alignment of IL-33, IL1β, and IL6. BSA: bovine serum albumin, IL: interleukin.

**Figure 2.** Epitope mapping of C2\_2E12. (**A**) Immunoblot analysis of 14 GST-IL-33 deletion mutants to map the IL-33 epitope at secondary structural element level for scFv clone C2\_2E12. Residue numbers of the mutants are shown. (**B**) Immunoblot analysis of four GST-IL-33 deletion mutants to map the IL-33 epitopes for scFv clones C1\_1E1, C2\_1D5, C2\_2A10, C2\_2E1, and C2\_2H5. (**C**) Immunoblot analysis of alanine scanning for GST-IL-33149-158 for recognition by the scFv clone C2\_2E12. Residue numbers and identities are shown. In panels (**A**) through (**C**), Coomassie Blue stained gel is shown at the bottom. Amount of the loaded protein per lane was 1 μg. The scFv clones were used as the primary antibody (0.5 mg·mL<sup>−</sup>1, 1:100 dilution), and anti-HA-HRP was used as the secondary antibody (0.2 mg·mL<sup>−</sup>1, 1:5000 dilution). (**D**,**E**) The HADDOCK-derived molecular docking of C2\_2E12 (green) to IL-33 (PDB: 4KC3, blue) complex. A homology structural model for C2\_2E12 was generated using SwissModel [24]. Both proteins are depicted as cartoon diagrams. (**D**) Residues of IL-33 in the epitope region recognized by C2\_2E12 are represented as stick models. Dash lines represent van der Waals atomic distances in Å. (**E**) Electrostatic interactions between C2\_2E12 and IL-33. The acidic pocket in C2\_2E12 consists of N163, D164, S166, Y168, A218, and Y230. L150 and basic K151, the two key residues in the epitope region of IL-33 are depicted as stick models. The figures in the panels (**D**) and (**E**) were generated using PyMOL (Schrödinger).

#### *2.3. Competitive Binding of C2\_2E12 to IL-33:ST2 Complex*

Residues 149–152 and 156 in the epitope of IL-33 for C2\_2E12 are reportedly involved in the interaction with the ectodomain (residues 19–321) of ST2, which is present in both the transmembrane and soluble isoforms [23]. Since the epitope of IL-33 for C2\_2E12 overlaps with the ST2 binding site, we hypothesized that the C2\_2E12 could function as a blocking antibody in the IL-33/ST2 signaling axis. To test the hypothesis, in vitro GST pull-down assay was performed. The antibody fragment

crystallizable (Fc) fusion of ST2, ST2-Fc, interacted with immobilized GST-IL-33 as expected. By contrast, ST2-Fc did not bind to the immobilized GST-IL-33 in the presence of C2\_2E11 (Figure 3A). To investigate whether C2\_2E12 inhibits IL-33:ST2 interaction in a dose-dependent manner, a series of concentrations of C2\_2E12 were used for competitive binding assay GST pull-down assay. The IL-33:ST2 interaction was reduced at 100-fold molar excess of C2\_2E12 (Figure 3B). Quantification of the IL-33:ST2 interaction in the presence of increasing concentrations of C2\_2E12 showed that the IL-33:ST2 interaction decreased in a concentration-dependent manner, leading to about 40% level at 100-fold molar excess of C2\_2E12 (Figure 3C). These results suggest that C2\_2E12 can act as a neutralizing antibody in the IL-33/ST2 signaling axis in vitro.

**Figure 3.** Interfering with IL-33 and suppressor of tumorigenicity 2 (ST2) complex formation by C2\_2E12. GST pull-down assay was used to observe the interaction of C2\_2E12 with IL-33 competitively with ST2 receptor in vitro. Pull-down assay was performed step by step. Immobilize GST-IL-33 in Glutathione Sepharose 4B and add ST2-Fc fusion proteins. Then, put C2\_2E12 in a dose-dependent manner. Binding was performed for 30 min every step. Molar concentration of the proteins for binding were GST-IL-33: ST2-Fc: C2\_2E12 = 1: 1: 0.1, 1:1:1, 1:1:10, and 1:1:100 (M). (**A**) Input loading (GST-IL-33 in lane 1, ST2-Fc in lane 2, and C2\_2E12 in lane 3) and proteins binding test (lane 4, 5, and 6) with pull-down assay were visualized by immunoblot and SDS-PAGE. (**B**) Inhibition of IL-33 and ST2 binding by anti-IL-33 antibody in a dose-dependent manner. After 4 μM of GST-IL-33 was immobilized in resin and 4 μM of ST2-Fc was added to the resin, C2\_2E12 (0.4, 4, 40, and 400 μM in lane 1, 2, 3, and 4) was added in a dose-dependent manner and visualized by immunoblot analysis and SDS-PAGE. (**C**) Quantification of the inhibitory effects of C2\_2E12 for the IL-33:ST2 interaction. Band intensities of ST2-Fc in panel (**B**) were quantified using ImageJ and normalized by dividing them by those of GST-IL-33. Relative intensities in reference to that of ST2-Fc with 0.4 μM C2\_2E12 are shown as a bar graph.
