Design of Glycoengineered IL-4 Antagonists Employing Chemical and Biosynthetic Glycosylation

Interleukin-4 (IL-4) plays a key role in atopic diseases. It coordinates T-helper cell differentiation to subtype 2, thereby directing defense toward humoral immunity. Together with Interleukin-13, IL-4 further induces immunoglobulin class switch to IgE. Antibodies of this type activate mast cells and basophilic and eosinophilic granulocytes, which release pro-inflammatory mediators accounting for the typical symptoms of atopic diseases. IL-4 and IL-13 are thus major targets for pharmaceutical intervention strategies to treat atopic diseases. Besides neutralizing antibodies against IL-4, IL-13, or its receptors, IL-4 antagonists can present valuable alternatives. Pitrakinra, an Escherichia coli-derived IL-4 antagonist, has been evaluated in clinical trials for asthma treatment in the past; however, deficits such as short serum lifetime and potential immunogenicity among others stopped further development. To overcome such deficits, PEGylation of therapeutically important proteins has been used to increase the lifetime and proteolytic stability. As an alternative, glycoengineering is an emerging strategy used to improve pharmacokinetics of protein therapeutics. In this study, we have established different strategies to attach glycan moieties to defined positions in IL-4. Different chemical attachment strategies employing thiol chemistry were used to attach a glucose molecule at amino acid position 121, thereby converting IL-4 into a highly effective antagonist. To enhance the proteolytic stability of this IL-4 antagonist, additional glycan structures were introduced by glycoengineering utilizing eucaryotic expression. IL-4 antagonists with a combination of chemical and biosynthetic glycoengineering could be useful as therapeutic alternatives to IL-4 neutralizing antibodies already used to treat atopic diseases.


Removal of non-conjugated IL-4 protein using iodoacetyl-activated agarose
Iodoacetyl-activated agarose beads (SulfoLink TM , Thermo Fisher Scientific) were used to remove non-reacted IL-4 protein harboring free sulfhydryl-groups derived from chemical glycosylation.
One mg of IL-4 glycoconjugate mixture was dissolved in 50 mM Tris-HCl, 1 mM EDTA pH 8.0 and subjected in batch mode to 1 mL of activated resin. The suspension was repeatedly mixed and incubated at 21°C for 45 min to ensure coupling of any IL-4 protein containing a free thiol group to the resin. The suspension was then filled into an empty column jacket and the flow through was collected. Thereafter, the resin was washed three times with one column volume coupling buffer and the flow through was again collected to completely elute IL-4 glyco-conjugates containing no free thiol group from the resin. Fractions containing IL-4 glycoconjugate were pooled, dialyzed against 1 mM HCl and the protein was freeze-dried for storage.

Recombinant expression of IL-4 in HEK293 cells
For production of IL-4 and variants carrying biosynthetic complex N-glycans the HEK293 cell lines Freestyle293 (Thermo Fisher Scientific) and Expi293 (Thermo Fisher Scientific) were used, which were maintained in suspension in the respective FreeStyle293 or Expi293 medium following the manufacturer`s recommendation. Additional N-glycosylation sites were introduced into the IL-4 coding sequence using two-step PCR mutagenesis (oligonucleotide sequences see Table S1).
The modified IL-4 cDNA were then cloned into a modified version of the expression vector pHLsec 1 . The 5' end of the IL-4 gene (encoding for residues His25 to Ser153) was fused to a DNA fragment encoding a hexahistidine sequence followed by a short linker (SSG) and an 8mer se-

Electrospray Ionization Mass Spectrometry Analysis (ESI-MS)
ESI-MS analyses of protein samples were performed on a Q Exactive™ Hybrid Quadrupole-Or-bitrap™ mass spectrometer (Thermo Fisher Scientific). Freeze-dried protein samples were dissolved in methanol/water/acetic acid (49.5/49.5/1) and adjusted to a concentration of 2 µM. Direct sample injection was performed and ions were detected in positive mode from 500 to 2500 m/z.
For evaluation, the FID signal (free induction decay) from 58 scans was deconvoluted to the single protonated ion mode using the software mMass (http://www.mmass.org).
ESI-MS analysis of low-molecular weight compounds, i.e. SMCC and SMCC-glucosamine conjugates, was done using ultra-performance liquid chromatography-tandem mass spectrometry

Surface Plasmon Resonance (SPR) Interaction Analysis.
All biosensor experiments were carried out employing a ProteOn TM XPR 36 system (BioRad) using (v/v) methanol, 6 % (v/v) glycerol and dried between cellophane film.    Figure S2). For comparison, enzymatically deglutathionylated IL-4 F82D R121C-SH containing a free thiol group is included. Mass spectrometry analysis of the IL-4 variant F82D

S8
R121C protein derived from HEK293 cell expression and hence carrying a complex N-glycan at Asn38 was performed after hydrolysis of the N-glycan moiety with endoglycosidase PNGaseF.
Abbreviations are SMCC-GlcN for SMCC-glucosamine conjugate, Glc for glucose and 4acGlc for glucose tetraacetate. S10  In order to show that the introduction of (an) additional cysteine residue(s) for chemical glycoconjugation does not alter the proteolytic stability of IL-4 an additional set of proteolysis experiments similar as described in materials and methods were conducted. As a different supplier/lot of trypsin was used wildtype IL-4 (derived from E. coli), IL-4 F82D (Q20N T28N K61N) (harboring four N-glycosylation sites) and IL4 F82D (derived from HEK293 cells thus harboring one N-glycosylation site) were used as internal references to allow comparison of the proteolytic half-lifes determined for these set of proteins with those in Table 3 After Coomassie staining the gels were scanned using a flatbed scanner (Mikrotec) and the scans were analyzed using the Analyze Gels tool in the software ImageJ to quantify the protein bands on the SDS gels. The volume data from the protein bands derived were then fitted with a one S12 phase decay routine using Prism version 9 to yield half times for proteolytic degradation. Two independent proteolysis experiments were performed. The half-life of wildtype IL-4 reived from E.
coli was set to 1 and the other half-life values were normalized to this value.

S24
materials and methods). Samples were taken at different time points and analyzed by SDS-PAGE analysis. Residual non-proteolyzed protein was quantified by gel densitometry using the software ImageLab (BioRad) and employing the volume tool. For relative quantification, staining intensities of the IL-4 protein bands were normalized to protein bands of the molecular weight standard.
"Proteolytic half-lives" were determined using the software Prism applying the "one phase decay" fitting model. Alle experiments were performed as independent triplicates. treated with PNGase F. The protein band observed at an apparent molecular weight of about 35 kDa in lanes 3, 5 and 7 is due to PNGase F added to the reaction mixture.