Linker minimization and characterization of Fc-fused interleukin-17A for increased in vivo half-life

Abstract

Interleukin-17A (IL-17A) is a cytokine involved in pro-inflammatory responses and tissue regeneration, with potential therapeutic and research applications. However, its short serum half-life limits in vivo use. Here, we report the systematic design of fc-IL-17A fusion proteins for extended half-life. Through computational analysis of 25 design variants using AlphaFold, we found that IL-17A's native N-terminal unstructured region functions as a crucial natural linker that cannot be effectively replaced by artificial sequences. We therefore generated mouse and human fc-IL-17A variants using direct N-terminal fusion without additional linkers. The resulting proteins retain IL-17A's ability to stimulate IL-6 production and erythroid cell growth. Pharmacokinetic analysis confirms that the Fc fusion increases the serum half-life in mice from 1.5 to 13 hours post-subcutaneous injection. This enables tractable experimental use of IL-17A in vivo for studying its role in inflammation and tissue repair. We further perform pharmacokinetics and pharmacodynamics modeling and propose a dosing regimen with reduced frequency of injection for delivering comparable IL-17A activity. This work provides a valuable pharmacological tool for injectable delivery, enabling investigation of IL-17A's biological functions in homeostasis and disease and exploration of its therapeutic potential in tissue regeneration.

Introduction

Interleukin 17A (IL-17A) is a member of the IL-17 cytokine family. It is the signature cytokine of the T helper 17 (Th17) cell lineage, playing a central role in their differentiation and function (Harrington et al. 2005; Park et al. 2005). While IL-17A is principally secreted by Th17 cells, other cells such as T cells, T cells, natural killer cells, neutrophils, and epithelial cells are also sources of the cytokine (Cua and Tato, 2010). It is involved in host defenses against mucosal infections and contributes to the pathology of autoimmune diseases such as psoriasis (Lowes et al. 2008) and rheumatoid arthritis (Kotake et al. 1999). IL-17A's function is not limited to inflammatory responses; it also stimulates the proliferation of various cell types including T cells (Yao et al. 1995), mesenchymal stem cells, (Huang et al. 2009; Ono et al. 2016), erythroid progenitor cells (Tusi et al. 2018) and skin epithelial cells after injury (Konieczny et al. 2022). These findings suggest that IL-17A may have a broader role in tissue regeneration and homeostasis, highlighting the potential therapeutic applications of the cytokine in regenerative medicine.

The IL-17A receptor complex, IL17RA/IL17RC (Wright et al. 2008) is expressed in many cell types, indicating that IL-17A may elicit responses in a diverse range of cells. The direct effects of IL-17A signaling and the full extent of its biological functions are yet to be fully elucidated. As the response to IL-17A is highly pleiotropic, its precise transcriptional impact has been difficult to identify, both in vitro (Ruddy et al. 2004; Shen et al. 2005) and in vivo (Cui et al. 2023). Controlled in vivo systemic administration of IL-17A could help probe this pleiotropism, providing a clearer understanding of IL-17A's direct functional effects on a diverse set of cell types across tissues.

The systemic in vivo delivery of IL-17A is hindered by its poor pharmacokinetic profile, with a reported elimination half-life of approximately 20 minutes following intravenous injection in mice (Kumar et al. 2016). The modest molecular weight of IL-17A (~35 kDa) likely leads to rapid renal clearance, making it challenging to study the direct effects of IL-17A signaling in spatially distributed systems like the blood or in multi-tissue disease pathologies such as psoriatic arthritis (Wang et al. 2017). Previous studies have attempted to overcome this issue by introducing IL-17A in vivo using adenoviral vectors or episomal DNA to induce constitutive IL-17A production (Schwarzenberger et al. 2000; Adamopoulos et al. 2015). However, these methods lack fine dosage control and may elicit uncontrolled inflammatory responses (Liu and Muruve, 2003), emphasizing the need for alternative approaches to introduce IL-17A in vivo.

Several protein engineering strategies can be used to overcome the short half-life of small proteins, including cytokines. Here we employ one such approach: fusing the protein of interest to the Fc domain of a human IgG (Way et al. 2005; Yu et al. 2013). This fusion has multiple benefits: 1) it increases the molecular size, thus mitigating renal clearance; 2) it leverages the neonatal Fc receptor (FcRn)-Fc recycling pathway to prolong serum half-life (Roopenian and Akilesh, 2007), and 3) it allows for the use of affinity-based purification methods during production. Computational structure prediction tools now enable systematic optimization of fusion protein design to ensure preservation of native protein structure as a proxy for function.

In this study, we detail the design and purification of human and mouse IL-17A Fc fusion proteins (fc-hIL-17A and fc-mIL-17A, respectively) aimed at enhancing the cytokine's in vivo half-life. We first performed systematic in silico linker optimization to understand how fusion design affects IL-17A dimerization. We then expressed, purified, assessed protein function by demonstrating that these proteins retain IL-17A’s specific pro-inflammatory activity and its non-canonical activity in promoting erythroid cell growth. Further, we measure the pharmacokinetics (PK) of recombinant mouse IL-17A (mIL-17A) and compare it with that of our fusion protein to assess conferral of PK benefits. We then use a pharmacokinetics and pharmacodynamics (PK/PD) model to examine dosing regimens that achieve comparable biological activity in vivo for mIL-17A and fc-mIL-17A. Our findings offer a pharmacological tool for researchers studying IL-17A in vivo and open new possibilities for the therapeutic use of IL-17A in regenerative medicine.

Materials and methods

Cell culture

The FreeStyle™ 293-F (HEK293-F) cell line was grown in serum-free and antibiotic-free FreeStyle™ 293 expression medium (Gibco). HEK293-F cell suspensions were maintained in a 37°C incubator containing 8% CO₂ on an orbital shaker platform rotating at 135 rpm. The cells were sub-cultured at cell densities between 0.1x10⁶-3x10⁶ cells/ml. The NIH/3 T3 cell line was purchased from ATCC and grown in Dulbecco's Modified Eagle's Medium (DMEM) containing 100 U/ml of penicillin, 100 μg/ml of streptomycin and 10% (v/v) heat-inactivated fetal bovine serum (all purchased from Gibco) in a 37°C incubator containing 5% CO₂. Cells were sub-cultured at cell densities between 0.5x10⁵ - 5x10⁵ cells/ml.

Expression and purification of fc-mIL-17A and fc-hIL-17A

The fc-IL-17A cDNA fragments (Supplementary Table 1) were inserted in silico into the multiple cloning site of the pSecTag2 DNA expression vector (Invitrogen) and the plasmids were synthesized de novo by Twist Biosciences. Plasmid DNA was transformed into DH5α competent Escherichia coli (New England Biolabs) and amplified. Endotoxin-free DNA expression vectors were purified using the Qiagen Plasmid Midi kit. HEK293-F cells were transfected with 1 μg of the expression vectors per million cells using 293 Fectin following the manufacturer’s protocol (Invitrogen). Culture supernatant was harvested five days post-transfection and filtered through 0.22 μm membrane filters to remove cellular debris. Filtered supernatant was subsequently concentrated using a 10 kDa molecular weight cut-off (Pierce) concentrator and applied to a Protein A affinity column (Pierce). To reduce IgG-mediated aggregate formation during the purification process, low-pH buffers were avoided in the treatment of the proteins (Latypov et al. 2012). Instead, the bound protein fractions were washed and eluted using the Gentle Ag/Ab Binding Buffer (pH 8.0) and Elution Buffer (pH 6.6) following the manufacturer’s instructions. Dimers were further purified via size exclusion fast protein liquid chromatography (SE-FPLC) on AKTA Pure FPLC using a Superdex 200 10/300 GL column (Cytiva) running at 0.4 ml/min. Phosphate buffered saline (pH 7.4) was used as the final buffer for storage of the Fc fusion proteins. Protein identity and purity were confirmed by SDS-PAGE under reducing (50 mM Dithiothreitol; Sigma) and non-reducing conditions. Gels were stained using SimplyBlue Safe Stain (Invitrogen) for visualization. Protein concentrations were estimated using Bradford Assay and bovine serum albumin protein standards (Pierce).

AlphaFold simulations of fc-mIL-17A and fc-hIL-17A

Structural predictions of fc-mIL-17A and fc-hIL-17A were conducted using AlphaFold3 (Abramson et al. 2024).The resulting models were assessed for structural confidence via predicted Local Distance Difference Test (pLDDT) scores and visualized with PyMol to examine the overall structure and dimerization domains. The hIL-17A fragment from the AlphaFold predicted structure was aligned with the crystal structure of hIL-17A (Liu et al. 2013) accessed from RCSB (PDB: 4HR9) using PyMol’s align command.

For in silico exploration of the fc-hIL-17A fusion protein design, we systematically explored 24 different linker types and linker lengths using AlphaFold modeling. To evaluate whether these constructs maintained the critical oligomeric interfaces, we compared them to reference crystal structures using DockQ scoring (Mirabello and Wallner, 2024). Specifically, we assessed preservation of the hIL-17A dimerization interface by comparison to the crystal structure of hIL-17A in complex with its receptors, IL-17RA and IL-17RC (PDB: 7ZAN), and the Fc dimerization interface by comparison to the IgG1-Fc crystal structure (PDB: 5JII). DockQ analysis was performed using default parameters with allowed mismatches of 2 and 3 residues for the IL-17A and Fc comparisons, respectively. This dual scoring approach enabled quantitative evaluation of how different linker designs impact the structural predictions of both fusion protein components.

NIH/3 T3 IL-6 stimulation assay

The IL-6 stimulation assay was adapted from (Ruddy et al. 2004). NIH/3 T3 cells were plated at 30% confluence in 24-well plates. 1 ng/ml of human TNF-α (Peprotech) was added along with different doses of murine IL-17A (R&D Systems) and human IL-17A (hIL-17A; Peprotech), or Fc fusion proteins. All doses were performed in biological triplicates. Cells were stimulated for 48 hours before the media were collected and flash frozen for storage at −80°C. The IL-6 concentration in media was measured using the Mouse IL-6 Uncoated ELISA Kit (Invitrogen) with technical triplicates per sample.

Erythroid colony forming assay

Erythroid colony forming units (CFU-E) colony forming assays were performed with bone marrow cells harvested from 8–12-week-old wild type and IL-17RA knockout C57B6J mice. In 6-well plates, 5x10⁵ bone marrow cells were plated in 1 ml of 1% (w/v) methylcellulose (Sigma) supplemented with 25 μM Beta-mercaptoethanol (Sigma), 17 mM NaHCO₃ (Sigma), 20% (v/v) fetal bovine serum (Gibco), 2 mM L-glutamine (Gibco), 100 U/ml penicillin (Gibco), 100 μg/ml streptomycin (Gibco), 1 mg/ml Human Transferrin (Sigma) and 0.2 U/ml Procrit (Epoetin alfa) (Janssen). Commercially available murine IL-17A (R&D Systems) or Fc fusion proteins were added in specified doses. Plates were incubated for three days in a 37°C incubator fed with 5% CO₂. Wells were subsequently stained with peroxide activated diaminobenzidine and imaged. Colonies were manually classified and counted per condition.

Pharmacokinetics measurements

8–16-week-old C57B6J mice (n = 2–4 per dose) from Jackson Laboratory were subcutaneously injected with two doses of mIL-17A and fc-mIL-17A. This dosing regimen complies with the mouse protocol issued to the Socolovsky laboratory by the Institutional Animal Care and Use Committee (IACUC) of the UMASS Chan Medical School. Over the course of 24 and 72 hours, < 20 μl volumes of whole blood were collected in heparin-coated capillary tubes per time point via tail-nick. Terminal blood collections were performed via cardiac puncture. Blood was immediately centrifuged at 3000 g for five minutes and plasma was collected, transferred into a protein lo-bind centrifuge tube (Eppendorf), and immediately flash frozen in liquid nitrogen. Samples were stored at −80°C prior to analysis. The small volumes of plasma were analyzed in technical triplicates using the mouse IL-17A ProQuantum Immunoassay Kit (Invitrogen) on a CFX96 Touch Real-Time PCR machine (BioRad).

Pharmacokinetics and pharmacodynamics modeling

For the PK modeling, blood plasma serum concentrations of mIL-17A or fc-mIL-17A were fit to a three-compartment depot model using the Scipy package (v1.11.4) in Python (v3.10.13). The functions optimize.differential_evolution and optimize.minimize were initialized with default parameters and used to minimize a weighted least squares error-based objective function. Model construction and parameters are given in the supplementary methods. Following curve fitting with all the data, bootstrapping was performed to evaluate parameter robustness: data were resampled with replacement 500 times and run through the same curve fitting algorithm to obtain a range of values for each parameter. This range was used to determine the standard error of each parameter. The elimination rate constant, k_e, was first estimated through data fitting, and this was then used to calculate the protein elimination half-life (t_1/2 = Ln(2)/k_e) for each molecule. For PK/PD modeling, the fitted PK model was utilized to obtain predicted concentrations for theoretical injection dose sizes and dose frequencies. These concentrations were then converted to activity using the PD E_max model that was fitted to the IL-6 stimulation data. Detailed description of the model is found in supplementary methods. The scripts for model fit and predictions are available on GitHub: https://github.com/Qiucwu/Wu_et_al_2024/

Results

Systematic in silico linker optimization of N-terminal fc-IL-17A fusion design

When designing Fc-fusion proteins, two key functional considerations must be preserved: the biological activity of the target protein and the mechanism by which it engages its receptor. IL-17A is active as a homodimer, and this homodimer engages with the IL-17RA receptor through the C-terminus of the protein (Liu et al. 2013; Goepfert et al. 2022). Preservation of homodimer formation and ensuring the receptor binding region remains accessible are therefore primary design requirements. To avoid interference with IL-17RA binding, Fc attachment must be positioned at the N-terminus rather than the more commonly used C-terminal fusion approach.

The next design consideration is how to link the Fc domain: what type of linker to choose, whether rigid or flexible, and what length. IL-17A’s structure presents several interesting aspects for linker optimization. The protein contains a native N-terminal unstructured region that is typically absent in crystal structures, raising questions about whether this region is necessary for dimerization and whether its specific sequence is functionally important for dimerization. Further, if extension of the fusion domain is required, what optimal linker length is required to preserve dimerization.

To address these questions, we performed a systematic in silico analysis of 25 design variants using AlphaFold structural modeling (Supplementary Table 1, Fig. 1) to explore three key parameters: (1) truncation of IL-17A’s N-terminal unstructured region, (2) replacement of artificial linkers of varying flexibility and length, and (3) extension of additional linker sequences beyond the native N-terminal region. We used DockQ scoring to quantify the preservation of both the IL-17A and Fc dimerization interfaces, a measure of structural similarity on a scale of 0 to 1. For IL-17A, we compared predicted structures to the crystal structure of hIL-17A in complex with its receptors (PDB: 7ZAN). For the Fc domain, we assessed preservation of the IgG1-Fc dimerization interface by comparison to the native IgG1-Fc crystal structure (PDB: 5JII). This dual scoring approach enabled quantitative evaluation of how different linker designs impact both fusion protein components.

Figure 1.

In silico modeling of linker and trunctation designs using AlphaFold. (A,B) AlphaFold-predicted protein structures of dimeric direct fusion (A) fc-mIL-17A (pTM/ipTM scores: 0.53/0.49) and (B) fc-hIL-17A (pTM/ipTM scores: 0.59/0.59) are shown. N-terminus (N) and C-terminus (C) are labeled on the protein. The unstructured region on the N-terminus of mIL-17A and hIL-17A is shown in yellow. (B inset) AlphaFold predicted hIL-17A fragment of fc-hIL-17A aligned with the crystal structure of hIL-17A (magenta) obtained from RCSB PDB (accession 4HR9). (C) DockQ scores comparing predicted fc-hIL-17A structures to the crystal structure of IL-17A in complex with its receptor (PDB: 7ZAN). Designs tested include truncation of the N-terminal unstructured region (yellow), truncation with replacement by flexible (GGGGS)xn (red) or rigid A(EAAAK)xnA (blue) linkers, and addition of these linkers to the full-length protein (pink and light blue, respectively). The direct fusion design is shown in black. Error bars represent standard error of the mean from the top three AlphaFold models. (D) DockQ scores comparing the same designs to the crystal structure of human IgG1 Fc (PDB: 5JII). Color scheme and error bars as in (A).

We first evaluated a direct fusion design, attaching the human IgG1 Fc domain to the N-terminus of IL-17A without additional linkers. AlphaFold structural analysis shows that the IL-17A domain in this design maintains dimerization interface similar to the native crystal structure (Fig. 1A and B) with a DockQ score of 0.66. This is nearly identical to the score obtained for unfused hIL-17A alone (0.68), indicating that direct Fc fusion does not compromise the IL-17A interface. The Fc domain also maintained high interface preservation with a DockQ score of 0.92.

The N-terminal region of IL-17A is typically unresolved in crystal structures due to its inherent flexibility, so we assessed the effects of truncating this region. Surprisingly, this region proved crucial for maintaining proper dimerization – its removal reduced the IL-17A interface DockQ score from 0.66 to 0.53 (Fig.1C). This effect could not be rescued by either flexible (GGGGS)_xN or helical and more rigid A(EAAAK)_xNA linkers of similar sizes. While additional linkers on top of the unstructured region slightly improved DockQ scores (up to 0.70), no clear length dependence emerged.

For the Fc interface (Fig.1D), flexible (GGGGS)_xN linkers showed length-dependent improvement only in constructs lacking the IL-17A N-terminal unstructured region, while rigid A(EAAAK)_xNA linkers maintained consistently lower scores regardless of length. These findings show that the direct fusion design of Fc and IL-17A without additional linkers preserves both IL-17A and Fc interfaces (DockQ scores 0.66 and 0.92, respectively) and provides the most straightforward strategy for building a fusion protein for IL-17A specifically.

Experimental validation of direct Fc and IL-17A fusion design

Based on the structural predictions, we generated direct Fc and mouse and human IL-17A fusion proteins by fusing the human IgG1 Fc domain to the N-terminus of IL-17A (Fig. 1A and B and Supplementary Table 2). The native N-terminal signal peptide of IL-17A was removed, and the fusion proteins were placed downstream of an Igκc signal peptide to facilitate cellular export of the secreted proteins (Fig. 2A and B). We produced the recombinant proteins in HEK293-F cells and then purified them using Protein A affinity chromatography. We used size exclusion chromatography (SE-FPLC) to characterize the resulting protein products and specifically purify dimeric proteins, as biologically active IL-17A is dimeric and Fc fusion proteins are known to form aggregates (Latypov et al. 2012). Both fc-mIL-17A and fc-hIL-17A yielded approximately 8 mg/L of purified dimeric protein. The SE-FPLC chromatogram of fc-mIL-17A (Fig. 2C) and fc-hIL-17A (Fig. 2D) displayed multiple protein peaks that likely corresponded to different oligomeric states. To confirm the oligomeric states and maximize collection of the dimeric proteins, the eluents from SE-FPLC were collected in two fractions (F1 and F2) and their size and identity were evaluated by SDS-PAGE under reducing (Fig. 2E) and non-reducing (Fig. 2F) conditions. Under reducing conditions, both eluent fractions contained bands of roughly 40 kDa, consistent with all protein fractions being composed of monomeric sub-units of fc-[m/h]IL-17A (43 kDa). The appearance of double bands likely reflects the heterogenous glycosylation at IL-17A sites, including N45 and HEK293-specific O-glycosylation at T28 (Liu et al. 2013; Geoghegan et al. 2013), as well as glycosylation at the conserved Fc N297 site characteristic of HEK293 expression (Dekkers et al. 2016). Under non-reducing conditions, the first eluent fraction (F1) contained multiple bands corresponding to dimeric (~ 80 kDa), tetrameric (~ 160 kDa), and aggregate (> 260 kDa) forms of fc-[m/h]IL-17A. We did not resolve, however, whether the oligomerization is due to proper IL-17A dimerization, Fc dimerization in the hinge region, or both. Eluent fraction F2 contained only two bands corresponding to the dimeric and tetrameric forms of fc-[m/h]IL-17A, with >90% purity in the dimeric form. Thus, eluent fraction F2 is comprised of highly enriched protein and is used for the remainder of this work.

Figure 2.

Design and purification of fc-mIL-17A and fc-hIL-17A. (A,B) Schematics of the fc-mIL-17A (A) and fc-hIL-17A (B) constructs. (C,D) Representative size exclusion fast protein liquid chromatography (SE-FPLC) chromatograms for fc-mIL-17A (C) and fc-hIL-17A (D). Small peaks were identified as oligomers, tetramers, and dimers of fc-[m/h]IL-17A. The black boxes labeled 1 and the grey boxes labeled 2 indicate SE-FPLC collection fractions obtained during the intervals of 20–30 min and 30–40 minutes for fraction 1 and 2, respectively. (E) Reducing SDS-PAGE of SE-FPLC elution fractions for both fc-mIL-17A and fc-hIL-17A shows both elution fractions contain monomeric fc-[m/h]IL-17A (~ 40 kDa). (F) Non-reducing SDS-PAGE of SE-FPLC elution fraction for fc-mIL-17A and fc-hIL-17A shows the oligomeric state of protein in each elution fraction. Most of the oligomers in elution fraction 1 are tetramers or larger aggregates. Most of the oligomers in elution fraction 2 are dimers.

Human and mouse fc-IL-17A fusion proteins synergize with TNF-α to induce IL-6 production in NIH3T3 cells

To measure the pro-inflammatory activity of IL-17A, we assessed the synergy of IL-17A and TNF-α to induce IL-6 production in NIH3T3 fibroblast cells (Ruddy et al. 2004; Supplementary Fig. 1). We measured IL-6 protein levels in the cell supernatant after co-stimulation of NIH3T3 cells with TNF-α and wild-type human or mouse IL-17A as well as in our fusion constructs (Fig. 3A). Cells were stimulated with 1 ng/ml of TNF-α and either fc-[m/h]IL-17A or commercially available mIL-17A and hIL-17A for 48 hours at a range of concentrations (0.1 pM - 57 nM). Both fc-mIL-17A and fc-hIL-17A synergized with TNF-α to induce a dose-dependent increase of IL-6 production in NIH3T3s, comparable to mIL-17A and hIL-17A, respectively (Fig. 3B and C). fc-mIL-17A and fc-hIL-17A fusion proteins showed 2.9-fold and 1.7-fold increase, respectively, in EC₅₀ values relative to their wild-type counterparts (Table 1). This modest decrease in potency of the fusion proteins compared to the native proteins may be the result of reduced rotational diffusion or steric effects of the Fc fragment (Chen et al. 2013). Nonetheless, the proteins are functional at concentrations that are compatible with physiological administration.

Figure 3.

Fc-mIL-17A and fc-hIL-17A retain IL-6-inducing activity comparable to unfused cytokine counterparts. (A) Schematic of IL-6 induction assay for measuring fc-[h/m]IL-17A and [h/m]IL-17A activity. (B,C) IL-6 production in NIH/3 T3 as a function of dose of fc-mIL-17A (black circle) (B) and mIL-17A (grey square) or fc-hIL-17A and hIL-17A (C). Values show min-max normalized IL-6 protein abundance as measured by ELISA. The data points show mean and error bars show standard error of the mean (SEM) with n > 3 replicates per dose. The lines represent the fit of the data to an E_max pharmacodynamics model with the light shaded region highlighting the 95% confidence interval of the fit. The E_max fit parameters for each curve are shown in Table 1.

Table 1.

NIH3T3 dose response to IL-17A.

Protein	EC50 (nM) (95% CI)	EC50 (ng/ml) (95% CI)
mIL-17A	0.88 (0.68–1.2)	31 (24–40)
Fc-mIL-17A	2.6 (2.0–3.2)	220 (180–280)
hIL-17A	1.6 (1.1–2.3)	56 (40–80)
Fc-hIL-17A	2.8 (2.0–3.9)	240 (180–340)

EC₅₀, half-maximum effective concentration. EC₅₀ was obtained using a fit of experimental data from IL-6 production assay in NIH/3 T3 to a one-parameter E_max model. See supplementary methods for fit details.

Fc-mIL-17A induces murine erythroid colony expansion

We evaluated the ability of fc-mIL-17A to synergize with Epoetin alfa (Epo) in promoting bone marrow erythroid progenitor survival and growth, a function that was recently demonstrated by our group for mIL-17A (Tusi et al. 2018). We measured the number of erythroid colony-forming units (CFU-E) from primary mouse bone marrow cells in the presence of a constant dose of Epo (0.2 U/ml) and either increasing doses of fc-mIL-17A or one saturating dose of mIL-17A (Fig. 4A).

Figure 4.

Fc-mIL-17A induces growth in primary murine erythroid cells. (A) Schematic of erythroid colony forming unit (CFU-E) assay. (B) Number of CFU-Es per 5x10⁵ cells as a function of Fc-mIL-17A concentration from IL17RA^+/+ (black circle) and IL17RA^−/− (grey square) mouse bone marrow cells. (C) CFU-E counts after treatment with 20 ng/ml (0.6 pM) of mIL-17A for the same animals. Data points show mean (n = 3 per condition); error bars show SEM.

Similar to the observations made by Tusi et al. for mIL-17A, we observed a dose-dependent increase in CFU-E number in response to fc-mIL-17A (Fig. 4B). While the reported saturating dose of mIL-17A is 0.57 nM, or 20 ng/ml (Fig. 4C), a comparable response was observed at 5.81 nM, or 500 ng/ml for fc-mIL-17A. We confirmed that the response occurred specifically through the stimulation of IL17RA receptor. Cells from IL17RA knockout mice (IL17RA^−/−) show no increase in CFU-E count in response to mIL-17A, as seen previously by Tusi et al., nor in response to fc-mIL-17A. This demonstrates that fc-mIL-17A retains the erythroid growth-promoting activity of mIL-17A and that the growth signal is dependent on IL17RA binding in erythroid cells.

Prolonged in vivo half-life of fc-mIL-17A

To determine if the Fc domain prolongs the in vivo half-life of fc-mIL-17A compared to mIL-17A, we administered two molar equivalent doses of each protein subcutaneously to wild-type C57B6/J mice (n = 2–4 per dose). The doses were as follows: 50 ng/g (1.42 nmol/g) and 200 ng/g (5.71 nmol/g) of mIL-17A, and 125 ng/g (1.45 nmol/g) and 500 ng/g (5.81 nmol/g) of fc-mIL-17A. Serum concentrations were measured over 24 or 72 hours (Fig. 5A). Using high sensitivity immunoPCR, we reliably detected the proteins in small plasma volumes, enabling longitudinal analysis of the same mice. We confirmed that the Fc fragment did not interfere with fc-mIL-17A detection by the mIL-17A-specific antibody used in the immunoPCR (Supplementary Fig. 2).

Figure 5.

The increased half-life of fc-mIL-17A permits experimentally tractable dosing designs. (A) Schematic of tail nick blood collection schedule used to measure the pharmacokinetics (PK) of mIL-17A and fc-mIL-17A. (B) PK of subcutaneous injections for two different doses of fc-mIL-17A, 125 ng/g (black circle) and 500 ng/g (black downwards triangle), compared to two different doses of mIL-17A, 50 ng/g (grey square) and 200 ng/g (grey upwards triangle). Lines represent the PK model fits. Data points represents the mean concentration per timepoint; error bars represent SEM. (C) Schematic of two comparable dosing regimens that deliver a similar integrated activity of IL-17A by administering mIL-17A (grey) or fc-mIL-17A (black). (D) Model predictions for the serum concentration of mIL-17A (grey) and fc-mIL-17A (black) during a dosing experiment with doses and timing as shown in (C). The area under the curve (AUC) is reported in pM·hour (hr) and represents the time integrated concentration experienced in the blood. The AUC is 5.06x10⁴ pM·hr for fc-mIL-17A and 1.95x10⁴ pM·hr for mIL-17A. (E) The corresponding model predictions for the biological activity during the same dosing experiment. The area under the effect curve (AUEC) is reported as IL-6 production activity units (AU)·hr. The AUEC is 13.57 AU·hr for both fc-mIL-17A and mIL-17A.

We used a three-compartment depot model (Zierhut et al. 2008) to analyze the serum concentration measurements and determine the proteins’ elimination and serum half-lives. In pharmacokinetics (PK), the elimination half-life represents the time required for half of the drug to be eliminated from the body, while the serum half-life is the time needed for the drug concentration in the blood to be reduced by half (Buxton, 2006). The depot model conceptualizes the injection site as a ‘depot’ from which the drug is slowly absorbed into the systemic circulation, making it well-suited for fitting PK curves after subcutaneous administration (Dubbelboer and Sjögren 2022). This slow release affects the serum half-life but not the elimination half-life, derived from the elimination rate constant (k_e). Estimating confidence intervals using bootstrapping revealed that k_e is well-constrained by the data, allowing for reliable estimation of the elimination half-life (Supplementary Fig. 3  B and C). For mIL-17A, the model estimated an in vivo elimination half-life of 26 minutes (Supplementary Table 3), consistent with the reported ~20-minute half-life (Kumar et al. 2016). The experimentally measured serum half-life of unmodified mIL-17A is 1.5 hours (Fig. 5B, Table 2). In contrast, the fusion protein's elimination half-life and serum half-life increased to 9.2 hours and 13 hours, respectively.

Table 2.

Summary of measured pharmacokinetics of mIL-17A and fc-mIL-17A

Protein	Route	Dose (ug/kg)	Cmax (ng/ml)	Tmax (hr)	Serum Half-life (hr)	Elimination Half-Life (hr)	AUC (ng·hr/ml)	AUMC (ng·hr2/ml)	MRT (hr)
mIL-17A	s.c.	50	8.1	1.0	1.5	0.4	5.1	14	2.8
mIL-17A	s.c.	200	37	1.0	1.5	0.4	23	66	2.8
Fc-mIL-17A	s.c.	125	46	4.0	13	9.2	610	15 000	24
Fc-mIL-17A	s.c.	500	490	8.0	13	9.2	6500	160 000	24

s.c. Subcutaneous; C_max, maximum serum concentration; T_max, time to maximum serum concentration; Serum half-life, the calculated concentration half-life in serum based on the fit curve; Elimination half-life, the in vivo protein half-life calculated from elimination rate constant k_e; AUC, area under the serum concentration versus time curve; AUMC, area under the first moment of the serum concentration over time curve; MRT; mean residence time, a ratio of AUMC and AUC. T_max and C_max were obtained directly from experimental observations. AUC, AUMC, and MRT were derived using integration of the PK fit curve.

An experimentally tractable in vivo dosing design

The fusion protein's increased serum half-life offers the potential to achieve comparable in vivo activity to the native protein with fewer doses. However, the optimal dosing regimen depends on the chosen pharmacological goal: maintaining a minimum blood serum level, achieving a maximal blood serum level, or delivering the same time-integrated induced activity, known as the area under the effect curve (AUEC) (Chen et al. 2022). To evaluate the benefit of the fusion protein, we focused on identifying a dosing regimen (dose size and frequency) that produces similar AUEC as we are interested in comparing functional activity between the native and fusion proteins. To this end, we employed a pharmacokinetics and pharmacodynamics (PK/PD) model that combines the parameters from the pharmacodynamics (PD) model in Fig. 3B and PK model from Fig. 5B (parameters given in Supplementary Table 3). We evaluate the AUEC in units of activity units (AU) · hours (hr), where one AU corresponds to a maximal cellular response as measured in Fig. 2.

We modelled in silico injection experiments to estimate the effective exposure, as measured by area under the concentration curve (AUC), and the effective activity, as measured by AUEC. The model shows that a single dose of 4 nmol/g (334 ng/g) fc-mIL-17A has a 2.6-fold higher effective exposure than six doses of 5.2 nmol/g (182 ng/g) of mIL-17A injected every 12 hours for 72 hours (Fig. 5C and D; AUC of 5.06x10⁴ pM and 1.95x10⁴ pM · hr for fc-mIL-17A and mIL-17A, respectively). This dosing regimen also has comparable predicted activity between the molecules (Fig. 5E, AUEC of 13.57 AU · hr for both fc-mIL-17A and mIL-17A). This six-fold reduction in dosing frequency also decreases total amount of protein required (3.2-fold reduction in weight, 7.8-fold in molar quantity) due to the prolonged activity of fc-mIL-17A within the system.

Discussion

In this study, we designed, synthesized, and characterized fc-IL-17A fusion proteins with considerably longer half-lives than native IL-17A. We determined that they have slightly lower activity than the native protein in two different in vitro functional assays. Nevertheless, a PK/PD model demonstrates that the total amount of fusion protein that would need to be delivered is lower over time, as compared to native IL-17A protein.

Here, we outline the design considerations essential for IL-17A fusion proteins and discuss possible approaches to further increase in vivo protein half-life.

Protein design considerations

Through systematic evaluation of 25 design variants, we discovered that IL-17A's native N-terminal unstructured region serves as a crucial natural linker for Fc fusion. Our analysis revealed that truncation of this region significantly compromised dimerization (DockQ score reduction from 0.66 to 0.53), while artificial linkers could not rescue this defect. The direct fusion design without additional linkers preserved both IL-17A and Fc interfaces well (DockQ scores 0.66 and 0.92, respectively), demonstrating that a simple design emerges from careful consideration of structural constraints.

Another fc-IL-17A fusion protein has previously been constructed in the literature, featuring an Fc fusion to the N-terminus connected by a ten amino acid glycine-serine (GGGGS)_x2 linker (Yao et al. 1995). Our in silico modeling included this design and showed that adding such a linker to IL-17A's unstructured region may marginally improve interface preservation (DockQ score increase from 0.66 to 0.70). However, our approach demonstrates that IL-17A's native N-terminal region alone provides sufficient flexibility for proper domain organization. Both designs maintain proper dimerization interfaces not only in silico but also experimentally - Yao et al. demonstrated biological activity in their cell-based assay, while we show functionality in both a similar cell-based assay and a novel primary cell erythroid assay. Given the similarity in structural predictions and experimental validation of both designs, the addition of the (GGGGS)_x2 linker, while potentially beneficial, appears not to be essential for producing functional fc-IL-17A fusion proteins with proper domain organization.

Possible further extensions of protein half-life

The extended half-life of our fc-mIL-17A fusion protein is consistent with half-lives of other Fc fusion constructs (Suzuki et al. 2010; Unverdorben et al. 2012; Yu et al. 2013), but all such constructs fall short of the intact human IgG1 half-life (Mankarious et al. 1988). The reasons for this discrepancy are not yet clear, but two factors may specifically contribute to the shorter half-life of our IL-17A fusion proteins compared to intact IgG1. First, the molecular weight of fc-IL-17A (~86 kDa) is smaller than IgG1 (~150 kDa) and may result in faster plasma clearance. Second, Fc fusion protein may have weakened FcRn-Fc interactions compared to whole IgG1 (Suzuki et al. 2010), and surface charge distribution can influence FcRn binding and protein half-life (Schoch et al. 2015), suggesting that the overall charge properties of fc-IL-17A might affect its FcRn interactions. As the FcRn recycling pathway plays a role in longevity of IgG, directly evaluating FcRn-Fc interactions of fusion proteins may reveal strategies for further prolonging protein half-life.

For extending protein half-life in murine studies, it may be useful to produce a fusion protein using murine IgG1 Fc as a fusion partner instead of the human Fc fragment as we have done here. Using the human Fc would likely lead to the development of anti-Fc antibodies in mice, compromising long-term studies (Gupta et al. 2022). While murine IgG1 corresponds structurally to human IgG4 rather than IgG1 (Bruhns 2012), it provides an optimal balance of Fc properties, specifically leveraging the high FcRn affinity to enhance protein recycling combined with low Fcγ receptor to minimize immunogenic responses. These characteristics are particularly important for studies requiring repeated dosing over the 1–2 weeks.

In conclusion, our findings indicate that Fc fusion with mIL-17A and hIL-17A enhances the in vivo half-life of the cytokine. These proteins effectively address the challenges associated with the experimental and therapeutic uses of IL-17A by offering a pharmacologically tractable delivery of IL-17A that requires less frequent doses and lower dose sizes.

Edited By: Dr. Feng Ni.

Author contributions

Qiu Chang Wu (Conceptualization [equal], Data curation [lead], Formal analysis [lead], Investigation [equal], Methodology [equal], Software [Lead], Visualization [lead], Writing - original draft [lead], Writing - review & editing [lead]), Jungmin Lee (Conceptualization [equal], Methodology [equal], Writing - review & editing [equal]), Aishwarya Swaminathan (Investigation [equal], Methodology [equal]), Ashley Winward (Investigation [equal], Methodology [equal]), Yung Hwang (Investigation [equal], Methodology [equal]), Merav Socolovsky (Conceptualization [equal], Funding acquisition [equal], Investigation [equal], Methodology [equal], Project administration [equal], Resources [equal], Supervision [lead], Writing - review & editing [equal]), Jeffrey C Way (Conceptualization [equal], Funding acquisition [equal], Methodology [equal], Project administration [equal], Resources [equal], Supervision [equal], Writing - review & editing [supporting]), Allon Klein (Conceptualization [lead], Funding acquisition [equal], Methodology [equal], Project administration [lead], Resources [lead], Supervision [lead], Writing - review & editing [lead]).

Conflict of interest

AMK is a cofounder of Somite Therapeutics, Ltd.

Funding

None declared.