Optimized synthesis, polymer conjugation, and proof‐of‐concept studies of the gd‐IgA1 epitope for antibody‐scavenging therapies in IgA nephropathy

IgA nephropathy (IgAN) is the most common glomerular autoimmune disease and has severe long‐term consequences for patients, with 40% of the patients eventually progressing to end‐stage renal disease. Despite the severity, no causal treatment is currently available. While the pathogenesis of IgAN is complex, disease severity is linked to autoantibodies against the gd‐IgA1 epitope, a stretch in the hinge region of IgA1 that lacks O‐glycans and is found in the characteristic immune complexes deposited in the kidneys of IgAN patients. One elegant, causal approach would be to remove the anti‐gd‐IgA1 autoantibodies and consequently reduce the immune complex burden on the kidneys. The administration of synthetic polymers that present autoantigens in a multivalent manner have been established as promising therapeutic strategies in other autoimmune diseases and may be applied to IgAN. We here present an improved protocol for the synthesis of the gd‐IgA1 epitope, its successful coupling to a poly‐L‐lysine polymer and proof‐of‐concept experiments that the polymer‐bound synthetic glycopeptide is able to capture the IgAN autoantibodies, making this approach a promising way forward for developing a targeted treatment option for IgAN patients.


| INTRODUCTION
IgA nephropathy (IgAN) is a highly prevalent and severe glomerular autoimmune disease that leads to a gradual reduction of kidney function up to end-stage renal disease.
Although the progression is often slow and shows large variability between patients, 40% of the patients will eventually require hemodialysis or a kidney transplant Floege & Barratt, 2021;Suzuki & Novak, 2021). The pathology is complex and heterogenous, but all patients show IgA1-IgG or IgA1-IgM immune complex (IC) deposition in the glomeruli of the kidney (Floege & Barratt, 2021;Wyatt & Julian, 2013). IgA1, the more common of the two IgA subclasses, has a unique hinge region with a high number of serine, threonine and proline residues. Typically, three to six Ser and Thr within this stretch are O-glycosylated with βgalactosyl-(1➝3)-N-acetylgalactosamine, with further addition of sialic acids being possible. In IgAN patients, however, some IgA1 immunoglobulins lack proper galactosylation of the N-acetylgalactosamine moieties, thereby giving rise to galactose-deficient IgA1 (gd-IgA1) (Figure 1a) (Suzuki & Novak, 2021). These gd-IgA1 are prone to aggregation and act as immunogens, which leads to the formation of autoantibodies, mostly of the IgG class but also IgA or IgM. Both aspects favor the formation of IC deposits in the kidney. The presence of gd-IgA1 IC is thought to drive the disease, although IgAs without erroneous glycosylation are also found in the deposits. Furthermore, the complement protein C3 is colocalized with these IC and complement activation can be typically observed; several complement inhibitors are therefore in clinical development as potential treatment options for IgAN Suzuki & Novak, 2021). The complexity of the disease contributes to the lack of disease-specific interventions, with current treatments being largely restricted to efforts preserving kidney function (e.g., by antihypertensive drugs) and, in specific cases, immunosuppression Floege & Barratt, 2021). Importantly, it was demonstrated that circulating IgA and IgA-IC levels correlate with the clinical severity of the condition, thereby rendering the removal of those complexes an attractive therapeutic strategy (Suzuki & Novak, 2021).
We present here a first proof-of-concept study, in which we were able to synthesize the gd-IgA1 epitope as a glycopeptide on a larger scale and with less equivalents of the difficult-to-obtain buildings blocks when compared to previous reports (Bolscher et al., 2010). When conjugated to a poly-L-lysine polymer, the epitope showed binding both to a commercially available antibody and to patient-derived autoantibodies. Thereby, such polymers may provide suitable candidates for treating IgAN using the previously established Antibody-Catch™ strategy (Aliu et al., 2020). This technology, in which disease-causing antibodies are sequestered by an epitope-bearing polymer, has already been successfully pursued for other conditions. Upon autoantibody-binding, the polymer complexes are rapidly degraded by the mononuclear phagocyte system, thus reducing the antibody burden (Aliu et al., 2020).

| Materials
Non-glycosylated amino acids were from Carbolution (St. Ingbert), solvents were from VWR (Radnor) or Carl Roth (Karlsruhe), resins were from CEM (Matthews) and other reagents were from Sigma-Aldrich (St. Louis). The glycosylated building blocks were custom synthesized by Samuel Pharma (Shandong). Poly-L-lysine (400er) was from Polypeptide Therapeutic Solutions (Paterna). All reagents and solvents were used without further purification. Sera from patients with positive anti-gd-IgA1 titers were obtained from the University Hospital of Basel. The use of patient sera was approved by the Ethics Committee of Northwestern and Central Switzerland (EKNZ). Informed consent was obtained from all non-anonymized participants. Healthy donor control serum was obtained from Sigma (H4522, lot SLB6544).

| Synthesis and purification of 1c
SPPS was performed on a Liberty Blue automated peptide synthesizer (CEM) using a microwave-assisted Fmoc/t-Bu SPPS strategy on Rink Amide ProTide LL (0.19 mmol/g) resin. The final conditions were 4 min double coupling at 95°C for the coupling with 6 eq. of the non-glycosylated amino acids, and 10 min single coupling at 95°C for the coupling with 2 eq. of the glycosylated amino acids 2 or 3, followed by a capping step after coupling 2 or 3. For coupling any of the amino acids, 6 eq. each of HOAt and HATU and 12 eq. of 2,4,6-trimethylypridine were used. Capping was performed with 10% Ac 2 O in DMF, and Fmoc-deprotection was achieved with 10% piperazine F I G U R E 1 (a) Hinge region sequence of IgA1 containing the gd-IgA1 epitope. IgAN patients lack the galactose (yellow circles), only carrying the N-acetylglucosamine (yellow squares) glycans. (b) Used glycosylated building blocks in the synthesis of the gd-IgA1 epitope.

| Synthesis of derivatized glycopeptide 4
Triethylamine (5 μL, 35.7 μmol) and γthiobutyrolactone (3.1 μL, 35.7 μmol) were added to a solution of 1c (3.92 mg, 1.19 μmol) in dry methanol (0.3 mL). The reaction mixture was stirred at RT under argon atmosphere overnight, the solvent removed under reduced pressure, the residue dissolved in H 2 O (0.5 mL) and washed with EtOAc (0.5 mL, three times). The aqueous phase was lyophilized and 3.34 mg of crude 4 was obtained as white solid.

KM55-coated plate
Maxisorp 96 well microtiter plates (Nunc) were coated with 75 μL/well KM55 rat anti-human gd-IgA1 IgG (IBL 30117066, lot 1F-701) at 1 μg/mL in Dulbecco's phosphatebuffered saline (DPBS) overnight at 4°C. Plates were washed four times with DPBS containing 0.05% Tween 20 (PBST) and unspecific binding sites were blocked with incubation buffer (IB; 0.3% non-fat dry milk in PBST; 100 μL/well) for 2 h at RT. A 2-fold serial dilution of polymer 5 (PN-251) was prepared in IB starting from 5400 ng/ mL. A quantity of 50 μL/well was added to the respective wells and incubated for 2 h at RT. Plates were washed four times and polymer 5 was detected with anti-poly-l-lysine F(ab) 2 -HRP (AbD27389, Bio-Rad). After 1 h incubation, plates were washed four times and 50 μL TMB substrate (Thermo Fisher) was added. The colour reaction was stopped before 30 min with 0.16 M sulfuric acid (Thermo Fisher) and absorbance (OD) was read at 450 nm using a microtiter plate reader (Synergy H1, Biotek).

| Detection of anti-gdIgA1-IgG and -IgA
Maxisorp plates were coated with 75 μL/well of polymer 5 (PN-251) at 1 μg/mL in DPBS overnight at 4°C. Plates were washed four times with PBST and nonspecific binding sites were blocked with IB (100 μL/well) for 2 h at RT. Patient and healthy donor control serum was diluted 1:50 in IB. A quantity of 50 μL/well was added to the respective wells and incubated for 2 h at RT. Plates were washed four times and IgG or IgA autoantibodies bound to glycopeptide 5 were detected with biotinylated anti-human IgG (SAB3701268, Sigma-Aldrich) or biotinylated anti-human IgA (SAB3701227, Sigma), respectively. Detection antibodies were diluted 1:2500 in IB, 50 μL/well was added and incubated for 2 h at RT. Following another wash (4×), 50 μL of 1:2000-diluted ExtrAvidin-HRP (E2886, Sigma) was added and incubated for 1 h at RT. Plates were washed four times and the colorimetric readout was performed as described above.

| RESULTS AND DISCUSSION
We aimed to synthesize the gd-IgA1-epitope glycopeptide 1 in a linear fashion using solid-phase peptide synthesis (SPPS), introducing the glycoamino acids as acetyl-protected building blocks that is Fmoc-O-β-(2-acetamido-3,4,6-tri-O-acetyl-2-deoxy-αd-galactopyranosyl)-L-Ser-OH 2 and Fmoc-O-β-(2-acetamido-3,4,6-tri-O-acetyl-2-deoxy-αdgalactopyranosyl)-L-Thr-OH 3 (Figure 1b) during SPPS. This was followed by cleavage from the solid phase, concomitant with removal of the acid-labile protecting groups, and base-catalyzed deacetylation (Figure 2a), all based on previously described conditions to synthesize the gd-IgA1 epitope fragment 1a (Figure 1a, Bolscher et al., 2010). Due to the low scale (12 nmol) used by Bolscher et al. (2010), the synthesis required further optimization efforts for our purpose to obtain sufficient amounts of product and to lower the equivalents of the expensive glycosylated building blocks 2 and 3. The initial strategy consisted of coupling the C-terminal carboxylic acid directly to a poly-L-lysine polymer. Thus, we started the synthesis on a Cl-MPA ProTide resin (0.17 mmol/g) using standard DIC/Oxyma coupling agents on a 25 μmol scale. However, neither coupling at room temperature (RT) with 1 eq. or at 50°C with 2 eq. of 2 and 3 yielded any product (Table 1, entries 1, 2), despite the successful use of these reactions by others for related peptides (Matsushita et al., 2005;Ohyabu et al., 2016). Increasing the eq. of 2 and 3, while also switching the coupling agents to HATU/HOBt/DIPEA and capping after each coupling step of 2 and 3, allowed us for the first time to detect product 1a, although only at a 2% crude yield (Table 1, entry 3). We therefore decided to switch to an alternative polymer coupling procedure, requiring a free amino group on the glycopeptide that would allow us to use Rink Amide resins that often result in higher yields. Indeed, the crude yield for 1b increased to 7% by switching to a Rink Amide ProTide resin (0.19 mmol/g) on a 50 μmol scale (Table 1, entry 4). By increasing the reaction temperature, switching from HOBt to HOAt for higher reactivity as shown for other particularly challenging couplings steps (Palitzsch et al., 2016), and substituting DIPEA with 2,4,6-trimethylpyridine (TMP), which has been demonstrated to improve yields and reduce epimerization in the coupling during SPPS of glycosylated amino acids (Zhang et al., 2012), we could substantially improve the crude yield of 1b to 55% by only using 2 eq. of 2 and 3. The observed degradation of the product was probably T A B L E 1 overview over the different reaction conditions employed in the synthesis of the gd-IgA1 epitope. 6 eq. of the coupling reagents and non-glycosylated amino acid were used. In the case of a base (DIPEA, TMP), 12 eq. were used. due to pH changes for work-up after the deacetylation step. Controlling the pH during work-up prevented degradation (Table 1, entry 5). Finally, we introduced an additional Tyr at the N-terminus for improved monitoring by UV and NMR, and a Lys for coupling to the activated polymer backbone. Thereby, we could isolate 1c in 16% yield after LCMS purification, again by using only 2 eq. of 2 and 3 (Table 1, entry 6). The production of sufficient amounts of the glycopeptide epitope allowed us to conjugate 1c to the poly-L-lysine (400-mer). In a first step, the free amine of 1c reacted with γthiobutyrolactone to give glycopeptide 4 with a free thiol (Figure 2b), which was subsequently coupled, as previously described (Thoma et al., 1999), to 2-chloroacetyl-derivatized poly-L-lysine to yield polymer 5 (PN-251) with an epitope loading degree of 10% (as determined by 1 H-NMR) (Figure 2c).

Entry number Product
For the functional evaluation and proof-of-concept study of the epitope-presenting polymer, we tested 5 in an ELISA with the commercially available KM55 antibody, which recognizes the gd-IgA1 epitope and has previously been used to determine gd-IgA1 levels in patient samples (Yasutake et al., 2015). Plates were coated with KM55, incubated with 5, and the poly-L-lysine polymer was detected with a horseradish peroxidase (HRP)-conjugated anti-poly-L-lysine antibody. KM55 was able to recognize 5 (PN-251), suggesting that the epitope is presented on the polymer in a similar fashion as it is in vivo (Figure 3a). To assess whether the reactivity and specificity is maintained in the case of patient antibodies, we coated plates with 5 and incubated them with IgAN patient sera or sera from healthy donors. Anti-gd-IgA1-IgG and IgA was detected using biotinylated anti-human IgG/IgA and HRP-streptavidin. From 10 patient sera, 8 showed an increased signal compared to the healthy control, thereby confirming the reactivity, and suggesting that patient antibodies might indeed be amenable to sequestration by polymer 5 (Figure 3b). The two non-responders in this study might be attributed to the expected heterogeneity of glycosylation patterns within the IgAN patient population, which affects the reactivity of individual autoantibodies toward a specific glycopeptide epitope. In addition, variations in autoantibody titers may affect the detection signal.

| CONCLUSION
Herein, we presented an improved synthetic protocol for the chemical synthesis of the gd-IgA1 epitope and its successful conjugation to a poly-L-lysine polymer, and demonstrated that the polymer-bound epitope was able to capture autoimmune IgGs in patient sera. To optimize the synthetic protocol for this challenging glycopeptide, which not only contains five glycosylated residues but also 10 prolines (out a total of 20 residues), the switch to the more reactive coupling reagents HATU/HOAt under addition of TMP as base proved to be critical as did higher reaction temperatures. This allowed us to isolate the gd-IgA1 epitope with a good overall yield of 16% while only using 2 eq. of the expensive glycosylated building blocks and on a relatively large scale of 50 μmol. Importantly, the resulting glycopeptide could be effortlessly conjugated to a polymer using our standard protocols. Finally, the synthetic glycopeptide-containing polymer was recognized by both commercial and patient-derived autoantibodies F I G U R E 3 (a) Dose-response plot of detected glycopeptide polymer 5 (PN-251) using an anti-poly-L-lysine antibody, after capturing 5 (PN-251) with the anti-gd-IgA1 antibody KM55 immobilized on a microtiter well plate. (b) Anti-gd-IgA1 IgG and IgA bound to immobilized glycopeptide polymer 5 from IgAN patients and healthy donors at 2% serum concentration. against gd-IgA1. These findings provide an important base for the future development of the approach, as it suggests that such polymers could be used, either in vivo or ex vivo, to capture ICs and, consequently, reduce disease burden in IgAN. At the same time, this approach could also be employed for diagnostic purposes in combination with an extended panel of glycopeptide epitopes; thereby, the molecular profile of pathogenic IgA could be determined in serum and patients could be subjected to a personalized therapeutic option based on the corresponding glycopeptide polymer. Future studies should therefore focus on the synthesis and evaluation of distinct glycopeptide epitopes.

ACKNOWLEDGMENTS
We thank Dr. Rachel Hevey (University of Basel) for the valuable discussions during the revision of the manuscript. This study was partially supported by funding from the Swiss National Science Foundation (grant 31003A_176104 to DR). Open access funding provided by Universitat Basel.

CONFLICT OF INTEREST STATEMENT
DR, CB, and KFK have no conflict of interest to declare. LP, KM, PH and NP worked for Polyneuron Pharmaceuticals, which had commercial activities and interests in this disease area.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.