Cellular effects and orientation of immobilized immunoglobulin are correlated to the charge-mediated influence of the antibody variable region

The binding of a ligand to a cell surface receptor initiates a transmembrane signal, triggering a cascade of cellular events. However, in many cell systems, particularly in immune cells, simple ligand binding is insufficient to initiate signal transduction. Instead, receptor clustering is a necessary initial process, as a ligand must bind simultaneously to two or more receptors to enhance the cellular response significantly.¹ The extent of receptor clustering, rather than the number of ligands bound, determines the magnitude of the cellular response.

Since the advent of monoclonal antibodies (mAbs) in clinical therapy, receptor clustering has become crucial for activating immune effector mechanisms or functioning as molecular agonists to activate therapeutic signaling pathways.² For instance, the level of immune effector signaling through the fragment crystallizable (Fc) region of IgG1 binding to Fcγ receptors (FcγR) is driven by high-avidity cross-linking on immune complexes, inducing FcγR clustering.³ 

Consequently, immobilized immunoglobulin immunoassays have been established and widely used in various fields, including drug discovery, disease diagnosis, and clinical pharmacokinetics.⁴ These assays rely on the successful immobilization of antibodies, requiring sufficient density and correct orientation on the immobilized surface to maximize antigen interaction and subsequent signal production. However, it is unclear whether different epitope motifs of the same target mAbs influence the immobilized density and orientation in the solid-phase state.

In addition to FcγR signaling, we present a dendritic cell immunoreceptor (DCIR) as an example. DCIR, a C-type lectin receptor (CLR), is primarily expressed on the surface of most antigen-presenting cells, such as monocytes, dendritic cells (DCs), neutrophils, and B cells.⁵ DCIR contains an intracellular immunoreceptor tyrosine-based inhibitory motif (ITIM) and functions mainly as a pattern-recognition receptor, modulating inhibitory responses to pathogens to maintain immune system homeostasis.⁶ Literature has shown that anti-DCIR mAbs can modulate cytokine production in human dendritic cells through immobilized immunoglobulin immunoassays, inducing DCIR clustering.^7,8

We have generated a series of human anti-DCIR antibodies with different epitope motifs but the same IgG1 Fc region. Immobilized anti-DCIR mAbs not only trigger the effector response from FcγR through the Fc region but also induce inhibitory pathways from the DCIR intracellular ITIM motif through the fragment variable (Fv) region in the immobilized immunoglobulin immunoassay. The modulation of cytokine production and signal transduction in human monocytes, under different anti-DCIR antibody treatments, is influenced by the binding ability of these immobilized antibodies to the plate. Additionally, the isoelectric point (pI) of the antibody variable region, rather than the total antibody pI, shows a significant correlation with the surface density and orientation of the immobilized antibody on negatively charged plates. Modification methods, such as increasing hydrophobicity or ionic interactions, have improved the surface density and consistent orientation of immobilized anti-DCIR mAbs.

Overall, our study highlights the importance of the net charge of the antibody Fv region in affecting the surface density and orientation of antibodies in the immobilized immunoglobulin immunoassay.

The detailed protocol is described in our previous publication.⁹ Briefly, Sprague Dawley JR14 rats were immunized with a vector expressing full-length huDCIR cDNA. Supernatants from splenic B cell hybridomas were screened based on binding activity. The heavy and light chain variable regions (VH and VL) of the hybridomas were cloned into a human IgG1 backbone as chimeric mAbs and subsequently fully humanized. Antibodies were produced from Expi293 cells and assessed for thermal stability and monomer content (>96% monomer). Selected antibodies were evaluated for species cross-reactivity to huDCIR and cyno DCIR, and no binding was observed with other CLRs and pattern-recognition receptors, including L-SIGN, DC-SIGN, DNGR-1, Dectin-1, CD123, CD205, CD206, CD207, and CD301.⁹ LALA antibody mutants were generated as described by the Winter group,¹⁰ involving the replacement of leucine (L) with alanine (A) at positions 234 and 235 of humanized antibodies.

Human monocytes were isolated from peripheral blood mononuclear cells (PBMCs) of healthy donors using positive selection with anti-CD14 labeled magnetic beads (Cat:130-050-201, Miltenyi Biotec) according to the manufacturer’s protocol. The purity of the monocytes was assessed by CD14 expression, achieving ≥90% purity as confirmed by flow cytometry analysis. Antibodies were coated on plates overnight at 4 °C, followed by two phosphate-buffered saline (PBS) washes. Human monocytes (1 × 10⁵ cells/well) were cultured on the plate in the presence or absence of R848 (0.4 μM) (Cat:tlrl-r848-1, InvivoGen) for 24 h. Cell-free supernatants were harvested, and TNF-α, IL-6, and IL-10 levels were determined by sandwich ELISA.

The detailed protocol is described in our previous publication.¹¹ Briefly, antibodies were coated on plates overnight at 4 °C, followed by two PBS washes. Human monocytes (5 × 10⁶ cells/well) were cultured on the plate in the presence or absence of R848 for 30 min. Total cell lysates were prepared with IP lysis buffer (Cat:87787, ThermoFisher) containing protease and phosphatase inhibitors. Western blot analysis was performed using specific antibodies against SHP2 (Cat:3397S, Cell Signaling Technology), Syk (Cat:644302, BioLegend), DCIR (Cat:MAB1748-SP, R&D systems or clone 1G3), and GAPDH (Cat:2275-PC-100, R&D systems) as a control, as previously described.⁹ 

Experimental pI of DCIR antibodies was determined using the Maurice icIEF system (ProteinSimple, Santa Clara, CA, USA). An ampholyte master mix was prepared according to the vendor’s recommendations. 34 μl of a 1 mg/ml anti-DCIR antibody was added to a 166 μl ampholyte master mix to prepare samples consisting of 0.2 mg/ml protein, 4% pharmalyte, 0.35% methyl cellulose, 4 M urea, 10 mM arginine, and pI markers (5.85 and 9.5). Samples were injected using the default pressure setting for 55 s and prefocused for 1 min at 1500 V followed by 8 min at 3000 V. Electropherograms were detected with UV absorbance at 280 nm, and data were analyzed using the vendor software, Compass for iCE.

Samples were prepared in 15 mM histidine, pH 6.0, to a concentration of 1 mg/ml. A heparin column (POROS^™ Heparin 50 μm Column, 4.6 × 100 mm, 1.7 ml, Applied Biosystems) was equilibrated with buffer A (20 mM Tris, pH 7.5) at room temperature with a flow rate of 1.5 ml/min. A total of 15 μg protein was injected into the column. A linear gradient from 0% to 100% buffer B (20 mM Tris, pH 7.5, 1 M NaCl) over 11 min was applied. Buffer B concentration was maintained for 2 min before the column was re-equilibrated with buffer A for 4 min. Detection was performed using a UV detector at 214 nm.

graphpad prism software (version 10) was used for all statistical analyses. Statistical significance was analyzed using an unpaired Student’s t-test. Data are expressed as mean ± SD, with α set at 0.05.

Fc receptors (FcRs) that recognize IgG immune complexes induce proinflammatory cytokine production through crosstalk with pattern-recognition receptors (PRRs), such as Toll-like receptors (TLRs).¹² Compared to a soluble IgG1 antibody, an immobilized IgG1 antibody on tissue culture plates, in combination with R848 (a TLR7/8 ligand), increased avidity to FcRs and induced a synergistic effect of FcR-TLR crosstalk [Fig. 1(a)], thereby enhancing the production of TNF-α, IL-1β, and IL-6 [Figs. 1(b)–1(d)]. This synergistic effect was reduced in the immobilized IgG1 antibody with a double mutation (Leu234Ala and Leu235Ala, also known as LALA) that diminishes binding to FcRs¹³ [Figs. 1(b)–1(d)].

To study the effect of antibody immobilization and downstream signaling, we utilized a panel of antihuman DCIR mAbs. DCIR is a C-type lectin receptor (CLR) with an intracellular ITIM motif that recognizes carbohydrate structures on pathogens and self-antigens in a Ca²⁺-dependent manner, providing an immune inhibitory function.^9,14 Given CLRs’ role in governing a wide range of immune responses through multivalent interactions with complex carbohydrates and amplifying this affinity through avidity,¹⁵ literature studies have shown that anti-DCIR mAbs can modulate cytokine production in human dendritic cells via immobilized immunoglobulin immunoassays, inducing DCIR clustering.^7,8

In this study, we further explored the effects of a series of human DCIR antibodies with different epitope motifs but the same IgG1 fragment crystallizable region on monocyte maturation induced by the TLR7/8 ligand R848 [Fig. 2(a)]. To trigger DCIR through cross-linking, CD14+ human monocytes were treated with plate-coated DCIR antibodies on tissue culture plates. Supernatants were collected after 24 h, and TNF-α, IL-6, and IL-10 production were measured [Figs. 2(b)–2(d)]. The isotype control, in the presence of R848, demonstrated the synergistic effect of FcR and TLR crosstalk on TNF-α, IL-6, and IL-10 production compared to the R848-alone group. This synergistic effect was diminished when the isotype control was replaced by the isotype with LALA mutations to eliminate FcR binding.

Anti-DCIR antibodies 5D7 and 5E11 increased the production of proinflammatory cytokines TNF-α and IL-6 compared to the isotype control. However, other anti-DCIR antibodies, such as 1D8, 9D9, 1G3, and 3A4, significantly reduced TNF-α and IL-6 production [Figs. 2(b) and 2(c)], although not below the levels seen in the R848-alone group. These results suggest that DCIR triggering may affect the FcR-TLR crosstalk rather than the TLR pathway itself. In terms of anti-inflammatory cytokine production, 5D7 and 5E11 did not affect the IL-10 release compared to the isotype control. However, antibodies such, as 1D8, 9D9, 1G3, and 3A4, increased IL-10 production [Fig. 2(d)]. Collectively, these data indicate that DCIR triggering by anti-DCIR antibodies 1D8, 9D9, 1G3, and 3A4 inhibits TLR7/8-mediated TNF-α and IL-6 production while selectively increasing IL-10 secretion in human monocytes.

Further investigation into the phosphorylation levels of SHP2 and Syk proteins in human monocytes revealed that cross-linking with the isotype control, human IgG1, increased the phosphorylation of SHP2 and Syk, indicating FcR engagement. This aligns with previous studies showing that SHP2 needs to be recruited to phosphorylated ITAMs in FcR to act as a scaffold protein for Syk.¹⁶ Control, 1D8, and 1G3 DCIR antibodies displayed the lowest levels of SHP2 and Syk phosphorylation. In contrast, 5E11 and 5D7 treatments showed the highest phosphorylation levels, while 9D9 and 3A4 exhibited intermediate phosphorylation signals. There was no difference in the DCIR expression in immobilized anti-DCIR antibody-treated monocytes [Figs. 2(e)–2(g)]. In the presence of R848 treatment, the results demonstrated a similar trend [Figs. 2(h)–2(j)].

In summary, the phosphorylation levels of SHP2 and Syk align with the cytokine production results in anti-DCIR antibody-treated monocytes. Anti-DCIR antibodies 5D7 and 5E11, which increased TNF-α and IL-6 while not affecting IL-10 production compared to the isotype control, also displayed the highest SHP2 and Syk phosphorylation. Other anti-DCIR antibodies (1D8, 9D9, 1G3, and 3A4), which significantly reduced TNF-α and IL-6 production and increased IL-10, displayed lower phosphorylation levels of SHP2 and Syk compared to 5D7 and 5E11, with 1D8 and 1G3 showing the least phosphorylation relative to the isotype treatment.

Previously, we demonstrated the DCIR expression in monocytes using a mouse antihuman DCIR antibody from the R&D System [Fig. 2(d)]. In this study, we used the human antihuman DCIR antibody 1G3 as the primary antibody and goat antihuman IgG-horseradish peroxidase (HRP) as the secondary antibody to detect the DCIR expression in monocytes by Western blotting.

As shown in Fig. 3, in the presence of R848, goat antihuman IgG-HRP was used to detect the heavy and light chains of anti-DCIR antibodies on the plate after the cell harvesting procedure. Interestingly, antibodies 5E11 and 5D7 exhibited the highest amounts of heavy and light chains compared to other anti-DCIR antibodies. Additionally, 1D8 and 1G3 showed lower signals compared to 9D9 and the isotype control [Figs. 3(a)–3(c)].

This result indicates unequal binding of different anti-DCIR antibodies to the tissue culture plate. It also suggests a correlation that a higher amount of antibody remaining on the plate induces higher inflammatory cytokine production and increased phosphorylation levels of SHP2 and Syk. Therefore, the unequal binding of anti-DCIR antibodies may trigger varying levels of the FcR signaling pathway through Fc region engagement, thereby affecting the measurement of DCIR function.

To directly analyze the binding ability of DCIR antibodies, we performed an ELISA-based assay on tissue culture plates. Anti-DCIR antibodies were coated overnight and then washed with PBS. The remaining amount of antibodies was detected using goat antihuman IgG-HRP.

At a coating concentration of 5 μg/ml, antibodies 1D8 and 1G3 showed the lowest binding ability. Antibodies 9D9 and 3A4 also demonstrated lower binding compared to the isotype control, 5E11, and 5D7, indicating different binding abilities among the antibodies (Fig. 4). As the coating concentration was diluted, 5E11 and 5D7 maintained the highest binding ability. The isotype control, 9D9, and 3A4 showed intermediate binding, while 1D8 and 1G3 continued to exhibit the least binding on the plate. A bispecific T-cell engager (BiTE), an artificial bispecific monoclonal antibody consisting of two single-chain variable fragments, was used as a negative control to demonstrate the specificity of the goat antihuman IgG-HRP.

This result is consistent with our previous data, indicating that a higher binding ability of the antibody correlates with increased inflammatory cytokine production and higher levels of SHP2 and Syk phosphorylation.

Tissue culture plates are often treated with modified polystyrene designed for cell attachment, which may lead to reduced antibody immobilization. The reason for the varying levels of immobilization among antibodies with different epitope motifs but the same IgG1 Fc region remains unclear. Given that the modified polystyrene surface is primarily hydrophilic and negatively charged (due to carboxyl groups), our hypothesis is that the pI of the antibody could be a factor influencing the immobilization level on tissue culture plates.

The experimental pI of various anti-DCIR antibodies was determined using imaged capillary isoelectric focusing (icIEF). Additionally, the surface positive charge of anti-DCIR antibodies was assessed using heparin-affinity chromatography. Theoretical pI values (VH domain, VL domain, VH + VL domain, and whole antibody), icIEF-determined pI values, and heparin elution times are listed in Table I.

As shown in Fig. 5(a), the theoretical pI of the VH + VL variable domain displays a strong correlation with the experimental pI determined by icIEF. Interestingly, the theoretical pI of the VH domain shows some correlation with icIEF pI, while no correlation is observed between the theoretical pI of the VL domain and icIEF pI [Fig. 5(b)].

Heparin elution time was used to evaluate the surface positive charge of anti-DCIR antibodies, as it may influence immobilization on a hydrophilic and negatively charged surface. It was revealed that heparin elution time correlates with the theoretical pI of the VH + VL variable domain [Fig. 5(c)] and with icIEF pI [Fig. 5(d)] to some extent. These results suggest that the immobilization level of anti-DCIR antibodies on hydrophilic and negatively charged surfaces is affected by the surface positive charge and pI of the VH + VL variable domain, particularly the VH domain. Increasing the hydrophilicity of plates has resulted in an inconsistent surface density of immobilized DCIR antibodies.

The structure of polystyrene consists of a long carbon chain with pendant benzene rings on every other carbon, making it highly hydrophobic. When polystyrene is molded into plates, this hydrophobicity is retained. A nontreated plate (nontreated polystyrene) surface is inherently hydrophobic. Antibodies attach to the plate through passive adsorption, driven by hydrophobic interactions between the large hydrophobic regions of the antibodies and the surface.¹⁷ Additionally, polystyrene can be treated (radiated) to become a high-binding plate, wherein carboxylic acid groups are incorporated onto the accessible carbons of the benzene rings. This modification allows for ionic interactions between the carboxyl groups and the positively charged groups of the antibodies.¹⁸ Therefore, immobilization occurs through both hydrophobic and ionic interactions.

We analyzed the binding ability of anti-DCIR antibodies using an ELISA-based assay on both nontreated and high-binding plates. Antibodies were selected based on their previous binding ability data: high binding (5E11, 5D7), medium binding (isotype control, 9D9), and low binding (1D8). Antibodies were coated onto the plates overnight and then washed with PBS. The remaining amount of antibody was detected using goat antihuman IgG-HRP. The initial concentration of coated antibodies was 10 μg/ml, and further dilutions were made down to 0.3125 μg/ml.

Even at the lowest concentration (0.3125 μg/ml), there was no significant difference in the retained amount of the antibody between different anti-DCIR antibodies and the isotype control on both nontreated [Fig. 6(a)] and high-binding [Fig. 6(b)] plates. The BiTE antibody was used as a negative control to demonstrate the specificity of antihuman IgG-HRP.

This result demonstrates that through the mechanism of passive adsorption, including hydrophobic and ionic interactions, antibodies can be equally immobilized. This binding ability is independent of the pI profile of the antibodies.

Next, we conducted the same experiment as in Fig. 2 using nontreated plates to analyze the remaining levels of antibodies from the cell lysates by Western blotting, detecting heavy and light chains. Human monocytes were treated with a coated isotype or anti-DCIR antibodies for 30 min and then harvested for Western blotting analysis.

There was no significant difference in the expression of heavy and light chains between the isotype, 1D8, 9D9, and 5E11 antibodies [Figs. 7(a)–7(c)]. This result shows that antibodies coated on the nontreated plate overnight provide an equal binding level across different treatment groups. This suggests that monocytes receive an equal level of FcR pathway stimulation through the antibody Fc region. Consequently, the further cellular effects and the role of the DCIR pathway can be evaluated through the engagement of the anti-DCIR antibody Fv region.

DCIR has been shown to inhibit TLR7/8-mediated cytokine production in myeloid dendritic cells.⁸ Here, we tested the function of various anti-DCIR antibodies in TLR7/8-induced human monocytes. The anti-DCIR antibodies 1D8, 9D9, 1G3, 3A4, 7G5, and 3F7 reduced TNF-α and IL-1β production compared to the isotype treatment [Figs. 8(a) and 8(b)]. Interestingly, antibodies 1D4, 4D3, and 5D7 exhibited higher TNF-α and IL-1β production compared to the isotype treatment. Only 3F7 showed a significant reduction in IL-6 compared to the isotype treatment [Fig. 8(c)]. In the anti-inflammatory cytokine analysis, none of the anti-DCIR antibodies significantly increased IL-10 production [Fig. 8(d)]. Some antibodies, such as 2H5, 3F7, 3B4, and 1D4, even significantly reduced IL-10 production.

Based on the TNF-α and IL-1β production profile, we selected antibodies 1D8, 9D9, 1G3, 3F7, and 5E11 for further analysis of the Syk and NF-κB (p65) pathways in anti-DCIR antibody-treated monocytes. Phosphorylation of SHP2, Syk, and p65 increased in the isotype treatment compared to the control, indicating FcR pathway engagement [Figs. 8(e)–8(h)]. However, treatments with antibodies 1D8, 9D9, 1G3, 3F7, and 5E11 reduced the phosphorylation of SHP2 and Syk. Among these antibodies, 1D8, 9D9, and 5E11 showed a reduction in phosphorylated p65 compared to the isotype treatment.

In summary, we demonstrated that DCIR reduces TLR7/8-mediated cytokine production through a pathway involving crosstalk between the SHP2, Syk, and p65 signaling pathways.

C-type lectin receptors (CLRs) recognize carbohydrate moieties on glycan structures of both pathogens and self-antigens. The “C” in CLR signifies calcium (Ca²⁺), as CLRs bind carbohydrate ligands in a Ca²⁺-dependent manner.¹⁹ Previous literature reports that in the innate immunity, CLRs serve as pattern-recognition receptors (PRRs) by recognizing conserved pathogen-associated molecular patterns (PAMPs)²⁰ and play a pivotal role in regulating immune homeostasis and inflammation.²¹ These CLRs share one or more carbohydrate-recognition domains (CRDs). Due to the characteristically low affinity of CLR-carbohydrate interactions, this low affinity is overcome by a multivalent ligand presentation on cell or pathogen surfaces.^22, In vitro, researchers often perform cross-linking with a plate-coated antibody/ligand and CLR to enhance cellular functions. Targets of CLR in these assays include DCIR,⁷ Dectin-1,²³ Dectin-2,²⁴ and Mincle.²⁵ Some researchers have conducted immobilized assays using tissue culture plates,^26,27 demonstrating that a high retained level of an immobilized antibody is not always required or desirable in inhibition assays. However, our study reveals that a series of anti-DCIR antibodies with different epitope motifs, but the same IgG1 Fc region exhibit different binding behaviors on tissue culture plates.

The surface of tissue culture plates is modified polystyrene, prepared using a high-energy plasma treatment process under oxidative conditions. These plates are hydrophilic and negatively charged, with a high ratio of carboxyl groups to hydroxyl groups for cell attachment purposes. We performed heparin-affinity chromatography to evaluate the binding ability of these antibodies on a hydrophilic and negatively charged surface. Heparin elution time correlates with both the theoretical pI of the variable domain (VH + VL) and icIEF pI. This result suggests that the pI of the antibody, particularly the pI of the variable domain, is crucial for binding to a hydrophilic and negatively charged surface. However, the pI may not be the only factor to affect the binding. Different antibody formats, such as F(ab′)2, also affect the binding ability. We have observed a similar pI between DCIR Abs, with IgG1 and F(ab′)2 formats, but noted varying residue levels on the plate.

Additionally, the binding ability of an anti-DCIR antibody on such a surface may be influenced not only by surface density but also by the orientation of the antibody.²⁸ Antibodies may be immobilized in various orientations, such as “end-on” (Fc facing the substrate), “head-on” (Fv facing the substrate), “side-on,” and “lying-on.”²⁹ On a hydrophilic and negatively charged surface, the orientation of the antibody may not be randomly distributed due to the antibody’s pI.³⁰ Literature also shows the strength of electrostatic repulsion between deprotonated antibodies and negatively charged surface carboxyl groups.³¹ 

We found that the variable domain pI affects the binding ability of anti-DCIR antibodies. It is plausible that different variable domain pIs lead to different orientations of anti-DCIR antibodies, subsequently affecting the accessibility of the Fc region in the cellular function and ELISA-based assays. For example, 5E11, which has the highest variable domain pI, provides the highest positive charge on the variable domain, favoring a “head-on” orientation. This orientation enhances the accessibility of the Fc region, thus amplifying the signal in FcR pathway activation assays and binding with goat antihuman IgG-HRP in ELISA-based assays. Conversely, 1G3, which has the lowest variable domain pI, may prefer an “end-on” orientation due to the negative charge on the variable domain, resulting in reduced FcR pathway activation and lower binding with goat antihuman IgG-HRP.

DCIR is expressed on various immune cells, including monocytes, macrophages, neutrophils, dendritic cells, and B cells. However, the mechanisms by which the DCIR pathway regulates immune responses and the specific ligands of DCIR are still not well understood. Therefore, we investigated the cellular functions of a series of DCIR antibodies with different epitope motifs on the Fv region but the same IgG1 Fc region in the immobilized immunoglobulin immunoassay. We found that the variable domain pI affects the binding ability of antibodies on hydrophilic and negatively charged surfaces. Nontreated or high-binding polystyrene surfaces are hydrophobic and suitable for antibody immobilization through passive interactions due to the large hydrophobic regions and ionic groups of antibodies. This immobilized anti-DCIR antibody immunoassay provides an alternative and accurate method for analyzing the role of DCIR in regulating immune responses. However, the analysis of Ab orientation is further required even though we have seen the equal binding density. Other groups have engineered the mAb with different Ab immobilization chemistries (cross-linkers) to achieve the uniform orientation.³² 

Our study explores how the charge properties of the variable region of mAbs affect their immobilization on hydrophilic, negatively charged surfaces. Through DCIR humanized mAbs with varying epitope motifs, it is found that the pI of an antibody’s variable domain significantly influences its immobilization level, impacting density and orientation on tissue culture plates. This affects effector responses by modulating FcR and DCIR signaling and downstream pathways, such as SHP2, Syk, and p65. However, the underlying mechanisms and their translation to a therapeutic approach require further study.

Our findings suggest that nontreated or high-binding polystyrene surfaces can improve antibody immobilization for diagnostic and therapeutic applications, enhancing our understanding of immune modulation via antibody engineering. These insights lay the groundwork for expanding antibody-based diagnostics and therapies.

We would like to appreciate former AbbVie employees, Subramanya Hegde and Tariq Ghayur, for their invaluable contributions to this research. Their expertise and guidance during the scientific discussion, data reviewing, and hypothesis conceptualization have been instrumental in shaping the significance of this research. This work was generously supported by AbbVie.

The authors have no conflicts to disclose. Hsi-Ju Wei, Jeffrey Barbon, Nancy Crosbie, and Eric Dominguez are employees of AbbVie. Jun Zhang was an employee of AbbVie at the time of the study. The design, study conduct, and financial support for this research were provided by AbbVie. AbbVie participated in the interpretation of data, review, and approval of the publication.

Ethics approval is not required.

H.-J.W. and J.Z. designed the experiments and wrote the manuscript. Most of the in vitro experiments and analyses were performed by H.-J.W. and J.Z. All contributing authors discussed the results in the final manuscript and agreed to submit this manuscript for publication.

Hsi-Ju Wei: Conceptualization (lead); Data curation (lead); Formal analysis (lead); Methodology (lead); Supervision (lead); Validation (lead); Visualization (lead); Writing – original draft (lead); Writing – review & editing (lead). Jun Zhang: Data curation (equal); Methodology (equal); Writing – original draft (supporting); Writing – review & editing (supporting). Jeffrey Barbon: Data curation (supporting); Formal analysis (supporting). Nancy Crosbie: Data curation (supporting). Eric Dominguez: Data curation (supporting).

The data that support the findings of this study are available from the corresponding author upon reasonable request.