Standardizing pre-existing anti-PEG antibody detection: A key to clarifying their binding features and clinical relevance

Abstract

Polyethylene glycol (PEG) is a widely used pharmaceutical excipient to improve drug solubility and prolong in vivo circulation duration. Although it was initially considered non-immunogenic, recent studies have confirmed the ubiquitous presence of anti-PEG antibodies pre-existing in human, influencing the efficacy and safety of multiple PEGylated medicines. Due to the contradictions of current anti-PEG antibody detection methods and unclear induction mechanism of human pre-existing anti-PEG antibodies, the actual prevalence, binding features, and biological effects of pre-existing anti-PEG antibodies remain elusive, hampering the clinical implementation of personalized PEGylated therapeutics, as well as the rational design of evasion strategies and the development of innovative PEG materials and PEGylated nanocarriers. Herein, we established a robust enzyme-linked immunosorbent assay (ELISA)-based methodology for the detection of human pre-existing anti-PEG antibodies, and defined the positive threshold based on complement activation. Mechanistic analysis revealed that human pre-existing anti-PEG IgG predominantly recognized the PEG backbone, while anti-PEG IgM specifically bound to the methoxy end group. Notably, the weak binding affinity of human pre-existing anti-PEG IgG limited its ability to induce complement activation or promote capture of PEGylated liposomes by mononuclear phagocyte system, in contrast to IgM. Collectively, this study not only provided a reliable detection strategy for human pre-existing anti-PEG antibodies, but also revealed the distinct binding patterns, affinities, and biological effects of pre-existing anti-PEG IgM and IgG, with important implications for the safe and rational applications of PEGylated therapeutics.

Graphical abstract

Highlights

• •Determining the appropriate methods and rational positive thresholds for human pre-existing anti-PEG antibody detection.
• •Discovering of the different binding selectivity and affinity of human pre-existing anti-PEG antibodies.
• •Deciphering the adverse impacts of human pre-existing anti-PEG antibodies on PEGylated liposomes.

1. Introduction

Polyethylene glycol (PEG) is a well-recognized pharmaceutical excipient to shield drugs from enzymatic degradation and fast clearance through kidney [1], and increase the in vivo circulation time of peptides, proteins, and nanoparticles [2,3]. Despite the original intention of utilizing its low immunogenicity [4], PEG can act as a hapten [5] and a growing number of researches have pointed out that anti-PEG antibodies could be induced by PEGylated drugs in rodents, dogs, monkeys, and human [6,7]. Besides drug induced antibodies, pre-existing anti-PEG antibodies were found to be prevalent and rapidly increasing in general population [8] perhaps due to the wide application of PEG containing cosmetics [9] and PEG like pharmaceutical excipients [10]. The pre-existing anti-PEG antibodies were reported to cause various biological effects, including complement activation, accelerated blood clearance (ABC) phenomenon, hypersensitivity reactions, and so on [11], which significantly threatened the efficacy and safety of PEGylated drugs and hampered their application in general population. For example, Khalil et al. [12] reported a high prevalence of pre-existing anti-PEG IgG (13.9%) and IgM (29.1%) in 701 children, which was associated with the efficacy of PEGylated asparaginase in an antibody-concentration dependent manner. In addition, studies by Ju et al. [13] and Guerrini et al. [14] revealed a strong correlation between pre-existing anti-PEG antibodies and allergic reactions following administration of lipid nanoparticle (LNP)-messenger RNA (mRNA) vaccines.

Despite the widespread observations of adverse effects related to pre-existing anti-PEG antibodies, their clinical significance remains sceptical [15], as conflicting findings have been reported regarding their impacts, and the incidence of clinical adverse events appears inconsistent with the reported prevalence of these antibodies. For example, while Okada et al. [16] reported that high levels of pre-existing anti-PEG antibodies might lead to the adverse reactions after first administration of PEGylated granulocyte colony stimulating factor (G-CSF), Tillmann et al. [17] found no correlation between pre-existing anti-PEG antibodies levels and efficacy of PEGylated interferon (IFN). In addition, despite the widespread prevalence of pre-existing anti-PEG antibodies in the general population [18], only about 5.7% of Americans who received severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) mRNA vaccines reported experience of adverse events, even based on a limited sample size (99 out of 1725). Notably, in Israel, only 4 cases of anaphylaxis were reported among 4,755,585 recipients of the first dose of the Pfizer-BioNTech vaccine [19].

Numerous factors contribute to the mystery of pre-existing anti-PEG antibodies, including the unclear induction mechanisms [20,21], the vagueness of their structures, and insufficient exploration of the differences between anti-PEG IgM and IgG. However, the lack of standardized detection methods and inconsistent results of human pre-existing anti-PEG antibodies may play one of the most critical roles, as accurate antibody quantification is essential for evaluating clinical impacts and relevance. Current studies varied greatly in detection methodology and reported antibody positive rates [22]. For instance, from 2016 to 2021, the positive rates of human pre-existing anti-PEG antibodies varied from 40% to 70% by enzyme-linked immunosorbent assay (ELISA) [23,24] and flow cytometry [25], and up to 97.5% in some ELISA-based studies [26]. Also, some studies reported minimum level of pre-existing anti-PEG IgG [27], while others found the concentration of anti-PEG IgG was greater than IgM [28]. These inconsistencies and potentially inflated positive rates obscure the true clinical significance of pre-existing anti-PEG antibodies and may divert attention from the small subset of individuals with sufficiently high antibody levels to trigger adverse biological effects, ultimately hindering the development of coping strategies and personalized treatment approaches.

Up to now, numerous studies have strived to construct novel and precise methods for detecting anti-PEG antibodies, such as surface plasmon resonance (SPR) [29], acoustic membrane microparticle (AMMP) assays [30], and oligo (ethylene glycol) (OEG) ligand-tethering assemblies [31]. However, flow cytometry and ELISA remain the most widely adopted approaches due to their simplicity and accessibility. ELISA is particularly considered as the gold standard owing to its suitability for large-scale screening and relatively low cost compared to PEG-coated magnetic beads-based flow cytometry [32]. Nevertheless, several critical issues must be addressed to resolve the inconsistencies across studies and to establish a reliable foundation for evaluating the biological effects of pre-existing anti-PEG antibodies. First, many studies utilized human-mouse chimeric anti-PEG antibodies as reference standards to quantify human pre-existing anti-PEG antibody [23,33]. However, the binding affinity of chimeric anti-PEG IgM is about 10 times higher than that of IgG [34]. Therefore, without knowing the actual in vivo binding affinity of human pre-existing anti-PEG antibodies, using chimeric antibodies as standards may relatively amplify the content of IgG in human serum in comparison to IgM and further influence the judgement on their clinical impacts. Additionally, the use of detergents and blocking regents has been shown to affect detection sensitivity, especially Tween-based detergents [35]. Some studies argued that the repeating ethylene glycol units in Tween-20 might compete with coating antigens for antibody binding sites [8], while others claimed that Tween-20 in phosphate-buffered saline (PBST) would not influence the detection of anti-PEG antibodies due to the conformational differences in repeating ethylene glycol units between Tween-20 and PEG materials [36]. Therefore, it is essential to distinguish whether, and to what extent, Tween interferes with the detection of pre-existing anti-PEG antibodies. Last but not the least, establishing clinically meaningful positive thresholds is crucial. Most current thresholds are derived from background comparisons or competitive inhibition assays [37], which would not represent the small subset of individuals who experience clinically significant adverse reactions. For example, Yang et al. [24] found that higher threshold values (100 ng/mL vs. 2−15 ng/mL) might be more closely aligned with adverse effects reported in the literature. Therefore, it is imperative to define rational positive boundary based on both the biological effects and clinical adverse effect relevance of pre-existing anti-PEG antibodies.

In the present study, we established an optimized ELISA method for detecting human pre-existing anti-PEG IgM and IgG by systematically evaluating the effects of different blocking reagents and detergents. In addition, the reasonability of using absorbance (optical density at 450 nm (OD₄₅₀)) of 1:4 diluted serum to represent antibody levels was verified, and the positive threshold of anti-PEG IgM was established based on its capacity to activate complement system. We further investigated the differential tolerance of pre-existing anti-PEG IgM and IgG to PBST. Beyond antibody detection, we also assessed the biological effects induced by pre-existing anti-PEG antibodies, which might have meaningful implications for the rational use of PEGylated medicines.

2. Material and methods

2.1. Cell lines, animals, and human serum

Human monocyte THP-1 cells were purchased from Meilun Biotechnology Co., Ltd. (Dalian, China). Female Sprague-Dawley rats (6−8 weeks old) were purchased from SLAC Laboratory Animal Co., Ltd. (Shanghai, China) and were kept under specific pathogen-free (SPF) conditions. All animal experiments were conducted in compliance with the guidelines approved by the Ethics Committee of Fudan University (Shanghai, China). Human serum samples were obtained, transported, and kept in full compliance with the standard operating procedures (SOPs) of Huashan Hospital, Fudan University. In details, the peripheral blood was collected into yellow top tubes, which contains no contaminations, additives, or preservatives, and kept at room temperature for 20 min to separate the serum. After centrifuged at 1000 g for 10 min, the serum was separated into three protein LoBind® tubes (Eppendorf, Hamburg, Germany), gently transported by dry ice boxes, and stored at −80 °C until use without any freeze-thaw cycles. The use of human samples in this study was approved by the Ethics Committee of Huashan Hospital, Fudan University (Approval No.: KY2024-632).

2.2. Reagents and antibodies

Hydrogenated soy phosphatidylcholine (HSPC), cholesterol, 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(PEG)-2000] (MeO-PEG₂₀₀₀-DSPE), OH-PEG₂₀₀₀-DSPE, NH₂-PEG₂₀₀₀-DSPE, and COOH-PEG₂₀₀₀-DSPE were purchased from A.V.T. (Shanghai) Pharmaceutical Co., Ltd. (Shanghai, China). Goat anti-rat IgM mu chain (horseradish peroxidase (HRP) (ab97180), goat anti-human IgM mu chain (HRP) (ab97205), goat anti-human IgG Fc (HRP) (ab97225), recombinant anti-C3 antibody (EPR19394) (ab200999), rabbit monoclonal (PEG-B-47) to PEG (ab51257), human complement C5a ELISA Kit (ab193695), and human complement C3a ELISA Kit (ab279352) were purchased from Abcam (Cambridge, UK). Recombinant human anti-PEG IgG (cHu-3.3-IgG) and human anti-PEG IgM (cHu-AGP4-IgM) were purchased from Institute of Biomedical Sciences at Academia Sinica (Taipei, China). Goat anti-human IgM (heavy chain) secondary antibody, fluorescein isothiocyanate (FITC) (A18836), and goat anti-human IgG (gamma chain specific) secondary antibody FITC (62-8411) were purchased from Thermo Fisher Scientific Inc. (Rockford, IL, USA). Goat anti-rabbit IgG H&L (HRP), 3,3′,5,5′-tetramethylbenzidine (TMB) chromogen solution, gradient precast polyacrylamide gels (4%–20%), goat anti-rabbit IgG (H+L) secondary antibody, FITC (A0562), 3-[(3-cholamidopropyl)-dimethylammonio]-1-propane sulfonate (CHAPS) (purity ≥99%, ST1146), and sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) sample loading buffer (5×) were acquired from Beyotime Biotechnology (Nantong, China). Tentagel M OH (M301000.1G) and M OCH₃ (a customized product whose certificate of analysis is listed in the Supplementary data) flow cytometry beads was purchased from Rapp Polymere GmbH (Tuebingen, Germany).

2.3. Preparation of PEGylated nanocarriers

Empty loaded PEGylated liposomes were prepared using the thin film hydration and extrusion method, as reported in our previous researches. Briefly, a mixture of HSPC/cholesterol/PEG-DSPE with different terminal groups (molar ratio of 56.5:38.1:5.4) was dissolved in chloroform and vacuum-rotary evaporated to form a thin film, then dried overnight under vacuum to remove residual organic reagent. The lipid film was hydrated with PBS at 60 °C. The lipid suspension was extruded through polycarbonate membranes with pore diameters of 200, 100, and 50 nm to generate liposomes with an average particle size of 90 nm. 5-Carboxyfluorescein (FAM) labeled liposomes were prepared by adding 0.5% molar ratio of FAM-PEG₂₀₀₀-DSPE before the formation of lipid thin film and after extrusion, and the external phase was replaced with 0.9% NaCl by Sephadex G50 column (Sigma-Aldrich, St. Louis, MO, USA). Polymer micelles were prepared by thin film hydration method. MeO-PEG₂₀₀₀-DSPE or OH-PEG₂₀₀₀-DSPE were dissolved in ethanol at a concentration of 10 mg/mL and vacuum-rotary evaporated to form a thin film, then dried overnight under vacuum. The dried lipid film was hydrated with PBS at 45 °C to form micelles self-assembly.

2.4. ELISA measurement of anti-PEG antibodies

ELISA was utilized to evaluate the levels of anti-PEG antibody in serum samples. Medium-binding 96-well ELISA plates were coated with 2.0 μg of PEG₂₀₀₀-DSPE per well. After rinses with detergents (0.05% (v/v) PBST, 0.1% (w/v) CHAPS, or PBS) twice, wells were blocked with 5% (w/v) milk/PBS or bovine serum albumin (BSA)/PBS (w/v) at 37 °C for 1 h. Serum was diluted with 1% (w/v) milk/PBS or BSA/PBS (w/v) and incubated in the microtiter wells at 37 °C for 1.5 h. Following rinses with selected detergent thrice, HRP-labeled goat anti-IgM or IgG antibody was added to incubate at 37 °C for 50 min. TMB substrate was added for 8/12 min (IgM/IgG) to visualize HRP after six more wash, and the reaction was terminated using 0.18 M sulfuric acid. Ultraviolet (UV) absorbance at 450 nm was then measured. For inhibitory anti-PEG ELISA, inhibitors were added into microtiter wells together with serum at a final PEG concentration of 1 mg/mL and other protocols were the same as aforementioned.

2.5. Flow cytometry measurement of anti-PEG antibodies

Flow cytometry method was utilized to confirm the binding of anti-PEG antibodies with PEG and their termini-selectivity found by ELISA, based on previous study with some modifications. Briefly, Tentagel M OH or M OCH₃ particles were suspended in 1× PBS at a concentration of 0.01 g/mL, using ultrasonic to disrupt any particulate aggregates. Then 25 μL of particle suspension was incubated with 150 μL of diluted human serum or chimeric anti-PEG antibodies for 1 h at room temperature. Incubation with 150 μL of 1% BSA/PBS was used as the negative control. After rinsing with different detergents (0.05% PBST or PBS) twice for 200 g, 2 min, the particles were suspended in 500 μL of FITC labeled secondary antibody dilutent according to the ratio recommended by the manufacturer for another 1 h incubation in darkness at room temperature. Finally, after rinsing with different detergents above thrice for 200 g, 2 min, the particles were suspended in 100 μL of PBS and the mean fluorescence intensity of 1 × 10⁶ particles at least was tested by flow cytometry. For inhibitory anti-PEG flow cytometry, inhibitors were added together with serum at a final PEG concentration of 1.0 mg/mL or washed using 6% PEG₈₀₀₀/PBS and other protocols were the same as aforementioned.

2.6. Detergent inhibition test of anti-PEG antibodies

Detergent inhibition test was performed on ELISA and flow cytometry. For PBST inhibition ELISA, PBST was added into microtiter wells for 15 min incubation after sample incubation, concentration ranging from 1% to 0% (w/v) at four times dilution. For CHAPS/PBS inhibition ELISA, CHAPS was added into microtiter wells for 15 min incubation after blocking (-BLOCK-) or human serum incubation (-primary Ab-), concentration ranging from 1% to 0% at four times dilution. Other procedures were the same as above. For PBST inhibition flow cytometry, PBST was mixed with Tentagel M OCH₃ particles for 15 min incubation after human serum incubation, concentration ranging from 0.05% to 0 at five times dilution. Other procedures were the same as above.

2.7. Isothermal titration calorimetry (ITC)

To minimize heat of dilution release from solvent dilution due to the high concentration of salt ions and proteins in the serum, we added blank rat serum to PEG materials-DSPE to a final concentration of 20%. Approximately 300 μL of 20% serum sample was carefully injected into the sample cell of MicroCal PEAQ-ITC (Malvern Instruments, Malvern, UK), and 60 μL of blank serum mixture was loaded to the syringe for titration into the serum sample. Experimental parameters were set as follows: running temperature 25 °C, paddle speed 750 rpm, total drop 19, drop interval 60 s, and initial delay 60 s. The volume of each drop was 2 μL, except for the first drop (0.4 μL). A 20% blank rat serum was used as the reference, with a reference power set to 10 μcal/s. Since there is no standard for pre-existing anti-PEG antibodies at present, the concentration of anti-PEG antibodies in the human serum was unknown; therefore, the sample concentration is set based on the total antibody IgM and IgG concentration, approximately 20 μM and the concentration of PEG-DSPE was 200 μM.

2.8. The complement activation of liposomes in serum

Empty loaded liposomes were mixed with anti-PEG antibodies positive/negative human serum or antibody negative serum containing 10 μg/mL chimeric anti-PEG antibodies at a final PEG-DSPE concentration of 1.6 mg/mL and incubated at 37 °C for 1 h. The mixture was then diluted 100 times with 10 mM ethylene ∖diamine tetraacetic acid (EDTA) to terminate complement activation. C3a level in the mixture was measured via human C3a ELISA kit, according to the manufacturer's instructions. For C3 cleavage, the mixture was mixed with 5× loading buffer. After heating for 15 min at 95 °C, the sample was loaded onto a 4%–20% gradient polyacrylamide gel for electrophoresis and transferred to a polyvinylidene fluoride (PVDF) membrane. Nonspecific binding sites on the PVDF membrane were blocked by incubating in 0.1% PBST containing 5% non-fat milk powder at room temperature for 2 h, followed by overnight incubation at 4 °C with recombinant anti-C3 antibody (1:2000). After three washes with 0.1% PBST (10 min each) to remove residual primary antibody, HRP-labeled goat anti-rabbit IgG H&L (1:5000) was added. The signals of C3 and cleavage products were imaged using a ChemiScope 6000 (Clinx Science Instruments Co., Ltd., Shanghai, China), and the data were analyzed with ImageJ software.

2.9. Cell culture and THP-1 derived macrophage differentiation

Human monocytic cell line THP-1 were cultured in calcium-free Roswell Park Memorial Institute (RPMI) 1640 containing 10% fetal bovine serum (FBS), 100 units/mL penicillin, 100 μg/mL streptomycin, and 0.05 mM β-mercaptoethanol with 5% CO₂ at 37 °C. THP-1 cells were seeded in 24-well plates at a density of 200,000 cells per well. Phorbol-12-myristate-13-acetate (PMA) (100 ng/mL) was added in culture medium to induce the differentiation of THP-1 into macrophages. After two days, the medium was changed to complete culture medium without PMA to rest the cells for three days before the experiment.

2.10. Cellular uptake of liposomes by human macrophages

Human macrophages were derived from THP-1 cells through treatment with PMA, following the method mentioned above. Liposomes at a final MeO-PEG₂₀₀₀-DSPE concentration of 1.6 mg/mL labeled with FAM were pre-incubated with an equal volume of anti-PEG antibodies positive/negative human serum or antibody negative serum containing 10 μg/mL chimeric anti-PEG antibodies at 37 °C for 1 h, and then incubated with cells in pure RPMI 1640 medium at a PEG-DSPE concentration of 15.95 μg/mL for 4 h at 37 °C. After digesting and washing thrice with 1× PBS, cells were harvested and the mean fluorescence intensity of 1 × 10⁴ cells at least was tested by flow cytometry.

2.11. Statistical analysis

Data are presented as means ± standard deviations (SDs) and analyzed by Student's t-test, one-way analysis of variance (ANOVA) or two-way ANOVA followed their respective applicable principles with GraphPad Prism 9.0. P < 0.05 was considered to be statistically significant (i.e., not significant (n.s.), ^∗P < 0.05, ^∗∗P < 0.01, ^∗∗∗P < 0.001, and ^∗∗∗∗P < 0.0001).

3. Results

3.1. Tween-20 affects ELISA detection of human pre-existing anti-PEG IgG but not IgM

Among various anti-PEG antibody detection methods, ELISA offers advantages in cost, throughput, and simplicity. Currently, numerous studies employed ELISA methods to detect pre-existing anti-PEG antibodies, and the details were summarized in Table S1 [13,23,24,26,27]. Among those, it mainly varied in blocking reagents, detergents, and results of anti-PEG IgG levels. To optimize ELISA detection of human pre-existing anti-PEG antibodies, we compared blocking with BSA vs. 5% non-fat milk powder and detergents PBST (0.05% (v/v) Tween-20 in PBS) vs. CHAPS (0.1% (w/v) in PBS). First, with PBST, blocking reagents showed no difference in detecting chimeric or human anti-PEG antibodies (Figs. S1A−C). However, CHAPS reduced chimeric antibody regression and increased nonspecific binding of human anti-PEG antibodies with BSA (Figs. S1D−F). Further tests using different concentrations of CHAPS (fourgxeh6id6Ye23 times dilution from 1% to 0% (w/v)) to soak the wells for 15 min after blocking step (-BLOCK-) or serum incubation step (-primary Ab-) on diethyl pyrocarbonate (DEPC) coated plates confirmed that 5% non-fat milk powder could minimize nonspecific signals (Figs. S1G and H). Therefore, 5% non-fat milk powder was chosen as the appropriate blocking regent for anti-PEG ELISA detection.

Since Tween-20 contains repeating ethylene glycol units, we assessed its interference by comparing CHAPS vs. PBST in detecting human pre-existing anti-PEG IgM and IgG. Results in Figs. 1A and B showed that CHAPS would significantly increase the positive rate (1:4 diluted serum whose OD₄₅₀ exceeded 1.0 was defined as positive sample) of human anti-PEG IgG, while the positive rate of anti-PEG IgM basically remained the same with a minor reduction. Specificity of pre-existing anti-PEG antibodies was confirmed by inhibition with polymeric micelles composed of MeO-PEG₂₀₀₀-DSPE (MeO-PM) (Figs. 1C and D). To validate the ELISA results, flow cytometry was introduced to detect anti-PEG IgG screened by CHAPS and IgM screened by PBST using customized Tentagel M OCH₃ beads. Chimeric anti-PEG antibodies (Table S2) were utilized to test the feasibility of the flow cytometry assay and Tentagel M OCH₃ beads could bind with both chimeric antibodies, which would be inhibited by MeO-PM incubation or PEG₈₀₀₀ washing (Figs. 1E−G). Similarly, pre-existing anti-PEG IgG detected by CHAPS could bind with Tentagel M OCH₃ beads and was inhibited by MeO-PM incubation or PEG₈₀₀₀ washing. Comparison of anti-PEG IgG screened by both PBST and CHAPS and only by CHAPS further shows that PBST could cause false negative signals of anti-PEG IgG detection (Fig. S2). However, pre-existing anti-PEG IgM turned out not to be influenced by PEG₈₀₀₀ washing (Figs. 1H−J). In a nutshell, pre-existing anti-PEG IgM was barely influenced by detergents containing repeating ethylene glycol units while IgG was significantly reduced by PBST. Therefore, the appropriate method for anti-PEG IgG detection should be 5% non-fat milk powder with CHAPS while for IgM was 5% non-fat milk powder with PBST, since PBST was more efficient in washing unconjugated secondary antibodies (Fig. S3).

Fig. 1.

Determination on appropriate detergents for the enzyme-linked immunosorbent assay (ELISA) detection of human pre-existing anti-polyethylene glycol (PEG) antibodies. (A, B) Detection of human anti-PEG IgG (n = 93) (A) and anti-PEG IgM (n = 104) (B) using 3-[(3-cholamidopropyl)-dimethylammonio]-1-propane sulfonate (CHAPS) and 0.05% (v/v) Tween-20 in phosphate-buffered saline (PBS) (PBST). (C, D) Inhibition of human anti-PEG IgG (n = 13) (C) and anti-PEG IgM (n = 22) (D) by polymeric micelles composed of methoxy-PEG₂₀₀₀-1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N (DSPE) (MeO-PM). (E) Flow cytometry detection of chimeric anti-PEG IgG (c3.3) and cAGP4, and inhibition by MeO-PM and PEG₈₀₀₀. (F, G) Mean fluorescence intensity (MFI) values of bovine serum albumin (BSA), 3.8 μg/mL c3.3, 3.8 μg/mL c3.3 + 1.0 mg/mL MeO-PM, and 3.8 μg/mL c3.3 using 6% PEG₈₀₀₀ as detergent (F) and BSA, 1.4 μg/mL cAGP4, 1.4 μg/mL cAGP4 + 1.0 mg/mL MeO-PM, and 1.4 μg/mL cAGP4 using 6% PEG₈₀₀₀ as detergent (G) (n = 3). (H) Flow cytometry detection of human pre-existing anti-PEG IgG detected by CHAPS and IgM detected by PBST, and their specificity by inhibition with MeO-PM and PEG₈₀₀₀ washing. (I, J) MFI values of BSA, anti-PEG IgG negative samples, anti-PEG IgG positive samples, and inhibition by 1.0 mg/mL MeO-PM and using 6% PEG₈₀₀₀ as detergent (I) and BSA, anti-PEG IgM negative samples, anti-PEG IgM positive samples, and inhibition by 1.0 mg/mL MeO-PM and using 6% PEG₈₀₀₀ as detergent (J) (n = 3). Data are shown as dots, lines and mean ± standard deviation (SD). Statistical significance is evaluated using Student's t-test or one-way analysis of variance (ANOVA) with Prism 9.0. ^∗∗∗∗P < 0.0001. n.s.: not significant. FITC: fluorescein isothiocyanate.

3.2. Human pre-existing anti-PEG IgM exhibits PEG-termini selectivity while anti-PEG IgG shows repeating-units selectivity

It is noteworthy that in Fig. 1J, the binding of human pre-existing anti-PEG IgM with Tentagel M OCH₃ beads was not reduced by washing with PEG₈₀₀₀. Considering that PEG₈₀₀₀ mainly differed from MeO-PM in its terminal group and chimeric anti-PEG IgG (c3.3) and IgM (cAGP4) targeted at the repeating ethylene glycol units of PEG (Table S2), binding selectivity of pre-existing anti-PEG IgM might be different from anti-PEG IgG and chimeric antibodies. Therefore, PEG₂₀₀₀-DSPE of different terminal groups including methoxyl, hydroxyl, amino, and carboxyl were coated on medium binding plates and were incubated with human serum samples. The heat map clearly demonstrated that human pre-existing anti-PEG IgM was mostly PEG-termini selective that barely bound with OH-PEG (Fig. 2A), while anti-PEG IgG was primarily repeating-units selective that bound equally with PEG of different terminal groups (Fig. 2B). To easily distinguish their binding selectivity, anti-PEG antibody recognizing the repeating units was named as anti-EGs antibody while anti-PEG antibody specific to the methoxy group was termed as anti-MeO antibody. Anti-EGs antibodies screened by the aforementioned ELISA could be fully inhibited by MeO-PM and polymeric micelles composed of OH-PEG₂₀₀₀-DSPE (OH-PM). In contrast, anti-MeO antibodies could be inhibited only by MeO-PM but not OH-PM (Figs. 2C and D). In addition, the conclusion was validated on flow cytometry assay using Tentagel M OCH₃ and Tentagel M OH, which only differed from each other in the PEG termini. Through comparing binding difference between two beads, the flow cytometry assay was capable of distinguishing cAGP4 (specific to the repeating units of PEG) and PEG-B-47 (specific to the methoxy group of PEG) (Table S2 and Fig. 2E). As expected, human pre-existing anti-EGs antibodies bound both flow beads while anti-MeO antibodies only bound Tentagel M OCH₃ bead (Figs. 2F and G). In conclusion, different binding patterns between human pre-existing anti-PEG IgM and IgG were discovered by ELISA coated with different PEG materials and confirmed by inhibitory ELISA and flow cytometry assay.

Fig. 2.

Binding patterns of human pre-existing anti-polyethylene glycol (PEG) antibodies. (A, B) The binding of anti-PEG IgM to PEG₂₀₀₀-1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N (DSPE) (A) and anti-PEG IgG to PEG₂₀₀₀-DSPE (B) with different terminal groups (the ordinates represent each human serum sample tested and the abscissas indicate PEG₂₀₀₀-DSPE of different terminal groups) (n = 40). (C, D) Inhibition of anti-PEG IgM (C) and anti-PEG IgG (D) using polymeric micelles composed of methoxy-PEG₂₀₀₀-DSPE (MeO-PM) and OH-PEG₂₀₀₀-DSPE (OH-PM) (n = 4). (E–G) The binding of cAGP4 and PEG-B-47 (E), anti-PEG IgG targeting at the repeating ethylene glycol units (anti-EGs antibodies) and methoxy group (anti-MeO antibodies) (F), and anti-EGs IgM and anti-MeO IgM (G) to Tentagel M OH and OCH₃ beads (n = 3). Data are shown as dots, lines and mean ± standard deviation (SD). Statistical significance is evaluated using two-way analysis of variance (ANOVA) or Student's t-test with Prism 9.0. ^∗P < 0.05 and ^∗∗∗∗P < 0.0001. n.s.: not significant. MFI: mean fluorescence intensity; BSA: bovine serum albumin.

3.3. Human pre-existing anti-PEG IgM shows greater binding affinity than anti-PEG IgG

In addition to human pre-existing anti-PEG antibodies, anti-PEG antibodies could also be produced after repeated injections of PEGylated nanocarriers in animals [38]. Nowadays, most preclinical experiments utilized anti-PEG antibodies stimulated by empty PEGylated nanocarriers to simulate pre-existing antibodies [39] due to the limited understanding on the originality and production of human pre-existing antibodies. To explore the binding selectivity of anti-PEG antibodies induced by PEGylated nanocarriers in animals, SD rats were intravenously injected with empty PEGylated liposomes (see Section 2.3 for the preparation, 2.5 mg HSPC/kg) to generate anti-PEG antibodies and blood was sampled after six days. Binding of anti-PEG IgM with PEG of different terminal groups was detected using 5% non-fat milk powder with PBST. Unlike human pre-existing anti-PEG antibodies, anti-PEG IgM in rats exhibited no PEG-termini selectivity (Fig. 3A) and it was not affected by PBST (Fig. 3B) contrary to human anti-PEG IgG showing similar binding selectivity, which drove us to find why human anti-PEG IgG was much easier influenced by Tween-20. On one hand as stated above, Tween-20 would indeed compete with PEG materials for the binding sites of anti-PEG IgG targeting at the repeating units. The hypothesis was confirmed by Tween-20 inhibitory ELISA (see Section 2.6) which soaked the wells into different concentrations of Tween-20 after incubation of serum. As shown in Fig. 3C, the levels of anti-EGs antibodies decreased with the increase of Tween-20 concentration. In contrast, the levels of anti-MeO antibodies were not influenced as expected. Meanwhile, it was verified by measuring the binding of different anti-PEG antibodies with Tentagel M OCH₃ beads using PBS or PBST as the detergent. PEG-B-47 and human pre-existing anti-PEG IgM, which were specific to the methoxy group of PEG, were not influenced by PBST washing. Nevertheless, c3.3 and human pre-existing anti-PEG IgG were significantly inhibited by the use of Tween-20 (Fig. 3D).

Fig. 3.

Human pre-existing anti-polyethylene glycol (PEG) IgG was neglected by Tween-20 for its repeating units selectivity and low binding affinity. (A) Sprague-Dawley (SD) rats were intravenously injected with empty PEGylated liposomes (hydrogenated soy phosphatidylcholine (HSPC) (2.5 mg/kg)) to generate anti-PEG antibodies. After six days, blood was sampled and the serum was separated. The binding patterns of rat anti-PEG IgM towards PEG with different terminal groups were measured by enzyme-linked immunosorbent assay (ELISA) (n = 3). (B) Binding of rat anti-PEG IgM with methoxy (MeO)-PEG₂₀₀₀-1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N (DSPE) was detected using 0.05% (v/v) Tween-20 in phosphate-buffered saline (PBST) and 0.1% (w/v) 3-[(3-cholamidopropyl)-dimethylammonio]-1-propane sulfonate (CHAPS)/phosphate-buffered saline (PBS) as detergent (n = 3). (C) Tween-20 inhibition (four times serial dilution from 1% to 0% (v/v)) of chimeric antibodies (cAGP4 and c3.3 for repeating units-selective and PEG-B-47 for methoxy group selective), rat anti-PEG IgM detected by 0.05% (v/v) PBST and human anti-PEG antibodies (n = 3). (D) Binding of chimeric anti-PEG antibodies and human anti-PEG antibodies to Tentagel M OCH₃ beads using PBS and PBST as the detergent (n = 3). (E) Binding affinity comparison of c3.3 (four times serial dilution from 3.8 μg/mL to 0.9 ng/mL) and cAGP4 (four times serial dilution from 2.8 μg/mL to 0.7 ng/mL). (F, G) Binding affinity comparison of human anti-PEG antibodies (total antibody concentration as the standard with four times serial dilution from 0.2 mg/mL to 48.8 ng/mL) by ELISA detection (n = 9) (F) and flow cytometry detection (n = 6) (G). Data are shown as dots, lines and mean ± standard deviation (SD). Statistical significance is evaluated using Student's t-test with Prism 9.0. ^∗∗∗∗P < 0.0001. n.s.: not significant. EGs: ethylene glycol units; MFI: mean fluorescence intensity; BSA: bovine serum albumin.

On the other hand, binding affinity difference between anti-PEG IgM and IgG was responsible for the high tolerance of anti-PEG IgM to PBST. The distinctive pentamer structure of anti-PEG IgM which possessed more antigen binding sites than IgG would cross-link with PEG to enhance its binding force, possibly making it less influenced by PBST washing [40]. The chimeric cAGP4 showed greater binding affinity to PEG than c3.3 (Fig. 3E), which was consistent with previous reports [34]. At the same time, as shown in Fig. 3C, anti-EGs IgM (rat and cAGP4) were influenced by PBST only at more than 0.05% while anti-EGs IgG (human and c3.3) decreased at lower concentrations. Also, in Fig. 3D, chimeric anti-EGs IgM cAGP4 was not affected by PBST washing in contrast with chimeric anti-EGs IgG c3.3, which illuminated that human pre-existing anti-PEG antibodies might exhibit same binding affinity difference as chimeric antibodies. Due to the fact that no chimeric antibodies could be used as standards to quantify the levels of human pre-existing anti-PEG antibodies, total antibody concentration was set to be the same level at 0.2 mg/mL to draw the binding curves using samples whose 1:4 diluted OD₄₅₀ was at about 2.0. The results from ELISA (Fig. 3F) and flow cytometry (Fig. 3G) showed that binding affinity of human pre-existing anti-PEG IgM was significantly higher than that of IgG, which was similar to chimeric antibodies.

Additionally, ITC was utilized to directly quantify the interaction between different PEG materials and anti-PEG antibodies in serum [41]. The measurement was taken under constant temperature conditions by titrating PEG-DSPE into serum (see Section 2.6). MeO-PEG was titrated to anti-PEG antibody positive (stimulated by intravenously injecting 2.5 mg HSPC/kg PEGylated liposome in rats six days ago) or negative (wild-type) rat serum to determine whether ITC was capable of detecting binding of anti-PEG antibodies to PEG materials. As shown in Fig. S4, the heat release from titration of anti-PEG antibody positive rat serum with MeO-PEG was significantly higher than that from the titration of negative serum. As the antibody in sample cell became saturated with PEG materials, the heat signal gradually decreased, whereas the heat release in antibody-negative rat serum was minimal and did not vary with the increase of PEG materials concentration. These results indicated that ITC could distinguish between antibody-positive and antibody-negative serum. Therefore, the binding affinity difference was further confirmed by ITC analysis. As shown in Figs. S5A and B, human serum samples identified as antibody negative by ELISA (OD₄₅₀ < 0.5) weakly interacted with both MeO-PEG and HO-PEG, exhibiting low heat release and no trend of decreasing heat release as the proportion of PEG in the sample cell increased. In contrast, titration of MeO-PEG in pre-existing anti-PEG IgM positive sera released significantly more heat than OH-PEG, and the heat release gradually decreased as the reaction approached saturation (Figs. S5C and D), which indirectly proved the PEG-termini selectivity of anti-PEG IgM. The detected dissociation constant (K_D) ranged from 3 to 15 μM. While no significant heat release or specific binding was observed during titration of OH-PEG in IgM positive serum, thus the K_D values could not be fitted (Table S3). Meanwhile, titration of PEG materials in anti-PEG IgG positive sera showed similar trends as antibody negative samples, which indicated that binding affinity of anti-PEG IgG was greatly lower than that of anti-PEG IgM (Figs. S5E and F). Notably, the Tween-20 inhibitory results by ELISA and flow cytometry (Figs. S6A and B) showed that pre-existing anti-PEG IgM bound to PEG materials more stable than IgG, also resulting in it less influential to PBST washing (Fig. S6C).

3.4. Human pre-existing anti-PEG IgM but not IgG induces prominent biological effects

Complement activation was a main biological effect induced by anti-PEG antibodies and played an important part in subsequent pharmacological processes [42]. Therefore, complement activation induced by chimeric anti-PEG IgM and IgG was firstly examined. Considering the low proportion of anti-PEG antibodies in human serum, 10 μg/mL chimeric antibodies were incubated with empty PEGylated liposomes (1.6 mg/mL PEG-DSPE) and same volume of PBS incubation was measured as baseline. As shown in Figs. 4A and B, c3.3 induced no complement activation while cAGP4 greatly activated the C3 component (Western blot (WB) images of complement activation induced by different samples were shown in Fig. S7). Furthermore, human pre-existing anti-PEG IgM detected by PBST and IgG detected by PBST and CHAPS were chosen to weigh their influences on complement activation. Complement activation of anti-PEG IgG screened by different detergents showed no significant difference with anti-PEG antibody negative samples (1:4 diluted OD₄₅₀ less than 0.2) (Figs. 4C and D). However, when it came to anti-PEG IgM positive samples (C3 index in Figs. 4E and F; C5 index in Fig. S8), complement activation level was significantly higher than those of anti-PEG IgG positive and antibody negative samples. In summary, in contrast to antibody negative samples, only incubation of anti-PEG IgM positive samples with liposomes could significantly increase complement activation level than PBS control. (WB images of complement activation induced by different samples were shown in Figs. S9 and S10).

Fig. 4.

Comparison of biological effects mediated by anti-polyethylene glycol (PEG) IgM and IgG. (A) Complement activation of negative human serum, 10 μg/mL c3.3, and 10 μg/mL cAGP4 with empty loaded liposome (1.6 mg/mL PEG-1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N (DSPE)) or equal volume of PBS by Western blot (WB) analysis using iC3b/C3 (PEGylated liposome (sLip)/phosphate-buffered saline (PBS)) as the indicator (n = 3). (B) Determination of the complement activation product C3a by enzyme-linked immunosorbent assay (ELISA) (n = 3). (C) Complement activation of negative human serum, anti-PEG IgG positive serum detected by 0.05% (v/v) Tween-20 in PBS (PBST), and by 3-[(3-cholamidopropyl)-dimethylammonio]-1-propane sulfonate (CHAPS) with empty liposome (1.6 mg/mL PEG-DSPE) or equal volume of PBS by WB analysis using iC3b/C3 (sLip/PBS) as the indicator (n = 4). (D) Determination of the complement activation product C3a by ELISA (n = 4). (E) Complement activation of negative human serum, anti-PEG IgG positive serum, and anti-PEG IgM positive serum with empty liposome (1.6 mg/mL PEG-DSPE) or equal volume of PBS by WB analysis using iC3b/C3 (sLip/PBS) as the indicator (n = 4). (F) Determination of the complement activation product C3a by ELISA (n = 4). (G) Macrophage uptake of 5-carboxyfluorescein (FAM) labeled liposomes mediated by 10 μg/mL c3.3 and cAGP4 (n = 3). (H) Macrophage uptake of FAM labeled liposomes mediated by anti-PEG IgG positive samples detected by PBST and CHAPS and anti-PEG IgM positive samples detected by PBST (n = 3) (each dot was the mean value of two replicates). Data are shown as dots and mean ± standard deviation (SD). Statistical significance is evaluated using one-way analysis of variance (ANOVA) with Prism 9.0. ^∗P < 0.05, ^∗∗P < 0.01, and ^∗∗∗P < 0.001. n.s.: not significant. Ab: antibody; MFI: mean fluorescence intensity; BSA: bovine serum albumin.

Besides the complement activation effects, macrophage uptake of liposomes mediated by anti-PEG antibodies could cause the accelerated clearance of liposomal drugs from circulation, which would hamper the efficacy and biosafety of PEGylated liposomes [43]. Therefore, macrophage uptake of fluorochrome labeled liposomes induced by different kinds of anti-PEG antibodies were compared. Briefly, 2 × 10⁵ THP-1 cells per well were seeded into 24 well plates and PMA was added to differentiate them into adhesive macrophages. Chimeric antibodies and human serum samples were incubated with FAM labeled PEGylated liposomes (1.6 mg/mL PEG-DSPE) at 37 °C for 1 h and then transferred to 24 well plates in fresh RMPI 1640 cell culture medium for additional 4-h incubation at 37 °C. Cells were digested, rinsed by PBS, and harvested to measure by flow cytometry. The results showed that cAGP4 significantly increased the macrophage uptake of liposomes while c3.3 did not (Fig. 4G). Similarly, only human anti-PEG IgM could dramatically increase the cellular uptake of liposomes while both kinds of anti-PEG IgG did not (Fig. 4H). At the same time, after using EDTA to prevent classical pathway and alternative pathway of complement activation in serum containing chimeric or human anti-PEG IgM, macrophage uptake level was greatly reduced (Fig. S11), proving that complement activation effect was closely associated with the downstream biological effects.

3.5. Serum titer appropriately represents levels of human pre-existing anti-PEG antibodies based on biological effects

The aforementioned exploration on biological effects mediated by different anti-PEG antibodies illuminated that pre-existing anti-PEG IgM rather than IgG would influence the behavior of PEGylated nanocarriers. Therefore, fast screen of human pre-existing anti-PEG IgM levels and definition on a rational positive threshold of anti-PEG IgM by its biological effects are of great significance. Considering that serum titers were often used to represent antibody levels, complement activation of anti-PEG IgM samples with different levels of serum titers (calculated with 2.1 times of background values in ELISA by analyzing duplicate serial 4-fold dilutions from 1:4 to 1:65536) were detected by WB (Fig. 5A) and ELISA (C3 index in Fig. 5B; C5 index in Fig. S12), with samples whose titer less than 1:10 as negative control and more than 1:1500 as positive control. The results showed that only samples whose anti-PEG titer exceeding 1:100 showed significantly stronger complement activation than negative control. Additionally, a larger number of samples whose titers distinguished at 1:100 were examined for their complement activation levels by WB (Fig. 5C) and ELISA (C3 index in Fig. 5D; C5 index in Fig. S13), and the results further verified the above finding. As aforementioned above, we roughly set the positive threshold of anti-PEG antibody on 1:4 diluted serum whose OD₄₅₀ at 1.0 to perform preliminary screening. To examine the reasonability of our preliminary screening results, serial dilutions of anti-PEG IgM of 43 samples were prepared (Fig. 5E) and the correlation between their OD₄₅₀ (1:4 dilution) values and IgM titers was analyzed (Fig. 5F). Surprisingly, titers of samples whose 1:4 diluted OD₄₅₀ at 1.0 were distributed around 1:100. Also, as shown in Fig. 5G, IgM titer of the positive group (1:4 diluted serum OD₄₅₀ more than 1.0) was remarkably higher than that of the negative group (1:4 diluted serum OD₄₅₀ less than 1.0) (over 1:100 vs. below 1:50), exhibiting prominent variation among groups. Also, the OD₄₅₀ values of 1:4 diluted samples were of good linear correlation with their IgM titers, validating the former as a suitable indicator for large population screening to save time and increase efficiency. The same verification was also performed on anti-PEG IgG (Fig. S14) and showed a relatively worse linear correlation than IgM because the weaker binding affinity of IgG resulted in extremely low readings at high dilutions. In conclusion, these results illuminated that serum titer above 1:100 might serve as a preliminary positive threshold of pre-existing anti-PEG IgM ELISA detection, considering its biological effects and 1:4 diluted serum whose OD₄₅₀ at 1.0 was suitable for large-scale screening of positive anti-PEG antibody samples in the population. Further correlation analysis on antibody-complement activation-downstream biological effects is urgently needed to assess the rationality of the positive threshold we established.

Fig. 5.

Reasonability on using serum titers to represent levels of human pre-existing anti-polyethylene glycol (PEG) IgM. (A, B) Comparison of Western blot (WB) (n = 3) (A) and enzyme-linked immunosorbent assay (ELISA) (n = 4) (B) results about complement activation of anti-PEG IgM samples with different levels of serum titers. (C, D) Comparison of WB results (n = 28) (C) and ELISA results (n = 32) (D) about complement activation of anti-PEG IgM samples distinguished by serum titer at 1:100 (1:4 diluted serum ultraviolet absorbance (optical density at 450 nm (OD₄₅₀) at 1.0). (E) Serial dilutions of anti-PEG IgM in 43 human samples. (F) The relationship between 1:4 diluted serum OD₄₅₀ values and titrations of anti-PEG IgM. Titration was calculated as –log10(dilution) (whose OD₄₅₀ was 2.1 times of the control buffer) using the embedded equation of specific binding in GraphPad Prism 9.0. Linear regression line (y = 0.7562x + 1.247, R² = 0.9472) was shown in red line (n = 43). Grey area showed 95% prediction intervals. (G) Comparison on the titers of anti-PEG IgM in positive samples (1:4 diluted serum OD₄₅₀ exceeded 1.0) and negative samples (1:4 diluted serum OD₄₅₀ below 1.0) (n = 43). Data are shown as dots, lines, and mean ± standard deviation (SD). Statistical significance is evaluated using one-way analysis of variance (ANOVA) or Student's t-test with Prism 9.0. ^∗P < 0.05, ^∗∗∗P < 0.001, and ^∗∗∗∗P < 0.0001. n.s.: not significant. sLip: PEGylated liposome; PBS: phosphate-buffered saline.

4. Discussion

In this study, we aimed to optimize the ELISA method for pre-existing anti-PEG antibody detection, given the controversial variability in methodology and results across current studies. We specifically investigated the effects of different detergents, PBST and CHAPS, on the detection of pre-existing anti-PEG antibodies, which represents a significant paradox in anti-PEG ELISA detection. Our findings demonstrated that CHAPS markedly increased the detection signal of pre-existing anti-PEG IgG, whereas IgM levels remained unaffected, and PBST underestimates the presence of pre-existing anti-PEG IgG, while having minimal influence on IgM detection. Meanwhile, comparative analysis using both chimeric and human pre-existing anti-PEG antibodies revealed that BSA as a blocking reagent led to increased nonspecific signals combined with CHAPS, as further confirmed by DEPC-coated ELISA plates. Although the underlying mechanism remains unclear, this effect may be attributed to the large molecular size, partially unfolded ellipsoid structure, and dimeric state of BSA in neutral salt solutions, which may promote nonspecific interactions with macromolecules [44]. Based on these findings, 5% non-fat milk powder was selected as the optimal blocking reagent for subsequent assays.

Previous researches have suggested that Tween-20 may interfere with anti-PEG antibody detection by competing with PEG for antibody binding due to the presence of shared ethylene glycol units [8]. To find out whether it was the reason for the neglection of pre-existing anti-PEG IgG by PBST, we changed PEG coating materials from MeO-PEG-DSPE to OH-PEG-DSPE, NH₂-PEG-DSPE, and COOH-PEG-DSPE. Unexpectedly, we found distinct binding preferences between antibody isotypes: pre-existing anti-PEG IgM preferentially recognized the methoxy terminus of PEG, whereas IgG predominantly targeted the repeating ethylene glycol units. Based on these observations, we classified human pre-existing anti-PEG antibodies into two categories: anti-EGs antibodies which bound the PEG backbone, and anti-MeO antibodies which specifically recognized the methoxy group. Using competitive ELISA with MeO- and OH-terminated PEG micelles (MeO-PM and OH-PM), composed solely of PEG polymers to eliminate liposomal structural interference, we found that anti-MeO antibodies were only inhibited by MeO-PM, whereas anti-EGs antibodies were equally inhibited by both micelles. This distinction was further validated using commercial Tentagel M OCH₃ and OH beads, where anti-MeO antibodies selectively bound to OCH₃-modified beads, while anti-EGs antibodies bound to both bead types. To our best knowledge, this is the first study reporting binding selectivity difference of human pre-existing anti-PEG antibodies, which might guide the rational design of evading strategies and new-style PEGylated therapeutics. Although binding selectivity of pre-existing anti-PEG antibodies was deciphered, anti-PEG antibodies induced by PEGylated drugs in animals were unknown for its binding epitopes. For instance, Shimizu et al. [45] found that OH-PEG modified liposomes (OH-sLip) induced fewer anti-PEG antibodies in mice than MeO-sLip, which might due to the fact that OH-sLip could only induce anti-EGs antibodies while MeO-sLip could elicit both anti-EGs antibodies and anti-MeO antibodies. However, further studies are needed to determine whether this selectivity applies to drug-induced anti-PEG antibodies in humans.

To verify that Tween-20 would compete with PEG for antibody binding, serial Tween-20 inhibition tests were performed using classified anti-EGs and anti-MeO antibodies. The results supported our hypothesis, showing that Tween-20 selectively inhibited anti-EGs antibodies, consistent with findings from flow cytometry using Tween-20 as the detergent. In addition to the binding selectivity of human anti-PEG antibodies, interspecies difference was examined and anti-PEG IgM produced in rats by PEGylated liposomes was confirmed to be repeating-units selective. However, despite this similarity, Tween-20 did not affect the detection of rat-derived anti-PEG IgM, indicating that factors beyond repeating-units selectivity contribute to the differential detergent sensitivity observed in humans. Considering the finding of affinity difference between chimeric antibodies [34], binding affinity and stability of human anti-PEG antibodies were also measured and compared. Due to the fact that no anti-PEG antibody standard was manufactured nowadays, it was impossible to quantify the anti-PEG antibody in circulation. Moreover, when chimeric antibodies were used as standards, the same OD₄₅₀ readings corresponded to significantly higher concentrations of IgM (by at least 100-fold) compared to IgG. Therefore, binding curves were generated starting from equivalent total antibody concentrations. Regardless of the detection method, binding affinity of anti-PEG IgM was superior than IgG, which added reasons into why IgM was less influenced by Tween-20 based detergent. We further assessed antibody stability via Tween-20 inhibition assays and flow cytometry. Even when using samples specifically targeting the methoxy group, anti-MeO IgG showed marked signal reduction after incubation with high concentrations of Tween-20, in contrast to the stable binding observed with IgM. These findings confirm that IgM exhibits superior binding stability to PEG compared with IgG.

The mechanistic basis for the binding features of human anti-PEG IgM and IgG remains unclear. We speculate that they may originate from divergent induction pathways and structural properties. In details, human pre-existing anti-PEG IgM might be mainly derived from T cell-independent (TI) pathway that preferentially recognized the methoxy group, while IgG might primarily be produced through T cell-dependent (TD) pathway that recognized the repeating units of PEG chains [46]. Moreover, IgM's pentameric structure provides it with a higher valency and the ability to crosslink PEG molecules more effectively than IgG [40]. Reports also suggest that antigen-binding sites within IgM can interact cooperatively to enhance overall apparent affinity, particularly when binding multivalent haptens [47]. Given the fact that human anti-PEG IgM and IgG were not illustrated properly by proteomics, explicit interpretation and analysis on their structures might also explain why there were such differences between their binding features and guide the establishment of more rational detection methods.

Based on above findings, a few insights could be concluded as: i) PBST is recommended for detecting anti-PEG IgM, as it does not interfere with IgM binding while offering effective washing performance. ii) CHAPS is preferred for detecting anti-PEG IgG, as PBST significantly underestimates IgG levels. iii) Human pre-existing anti-PEG IgM primarily recognizes PEG terminal groups, whereas anti-PEG IgG predominantly targets the repeating ethylene glycol units. iv) The underestimation of pre-existing anti-PEG IgG by PBST is attributed not only to its epitope selectivity but also to its relatively weak binding affinity and reduced stability compared to IgM.

The final purpose of anti-PEG antibodies detection was to figure out how it would influence the effectiveness and safety of PEGylated therapeutics. Therefore, we explored the biological effects mediated by different anti-PEG antibodies in the last part. Complement activation, a key immunological event triggered by antibody-PEG interactions, has been implicated in adverse events such as ABC phenomenon and hypersensitivity reactions. Considering the current ELISA methods mainly varied from each other in anti-PEG IgG detection, we firstly compared complement activation induced by anti-PEG IgG detected by PBST and CHAPS, respectively. Through comparing the WB and ELISA detection of complement lysates, we discovered that no anti-PEG IgG could significantly increase the level of complement activation in contrast to antibody negative samples. Furthermore, we compared activation abilities of pre-existing anti-PEG IgM and IgG. In contrast, anti-PEG IgM induced robust complement activation, consistent with its superior C1q-fixation ability and higher binding affinity [48]. In addition, we expanded the study into downstream biological effects such as the macrophage uptake, a hallmark of the ABC phenomenon. Compared with negative serum, only anti-PEG IgM positive samples would promote the uptake of liposomes by human macrophages and the promotion significantly relied on the complement activation effects, which also proved the central role of complement activation in adverse effects mediated by anti-PEG antibodies. Noteworthily, the macrophage uptake of liposomes induced by anti-PEG IgM was not completely inhibited by EDTA possibly due to the fragment crystallizable (Fc) receptor mediated phagocyte by anti-PEG IgM [49], which indirectly showed that anti-PEG IgM itself prevailed IgG in inducing biological effects.

To support practical application of anti-PEG antibody detection and simplify the detection method for large sample screening, we established a correlation between single point OD₄₅₀ values (at 1:4 serum dilution) and antibody titers. A strong linear relationship was observed for both IgM and IgG, although the detection limit for IgG had to be increased to 2.9-fold above background due to weak signals at high dilution. Moreover, complement activation assays stratified by anti-PEG IgM levels showed that only samples with titers more than 1:100 and OD₄₅₀ > 1.0 induced significant activation comparable to negative and positive controls. Due to the fact that anti-PEG IgG could not efficiently induce complement activation, the verification was not established on different levels of anti-PEG IgG. Above findings warned us that anti-PEG IgM was an indicator of possible adverse events using PEGylated drugs and personal monitoring of anti-PEG IgM combined with the complement activation levels is essential for PEGylated drug application [50].

Despite these findings, several critical questions remain unresolved. First, the purification of human anti-PEG antibodies and further interpretation on their sequence information, structures, and binding modes with antigens were still a major task to reveal the mystery of anti-PEG antibodies and their induction mechanism and to produce a generic standard. Second, a more reasonable positive threshold of pre-existing anti-PEG IgM based on the clinical situations was urgent to be established to warn possible threats caused by anti-PEG IgM. Finally, the biological consequences of pre-existing anti-PEG antibodies, particularly their role in hypersensitivity reactions and their influence on the pharmacokinetics and efficacy of PEGylated protein therapeutics, warrant further investigation. Solving these problems and precisely measuring clinical anti-PEG antibodies would be beneficial for the rational use of PEGylated drugs and promote the clinical translation of PEGylated therapies.

5. Conclusion

In the present research, an optimized ELISA method for detecting human pre-existing anti-PEG IgM and IgG was established by comparing effects of different blocking reagents and detergents. CHAPS, instead of PBST, was verified to be more sensitive in detecting IgG. In addition, the differential tolerance of pre-existing anti-PEG IgM and IgG to PBST was investigated, which turned out to be attributed to their distinctive binding selectivities and affinities. Meanwhile, reasonability of using OD₄₅₀ of 1:4 diluted serum to represent antibody levels was confirmed, and the positive threshold of anti-PEG IgM was established based on its capacity to activate complement system. Beyond antibody detection methodology, the biological effects induced by human pre-existing anti-PEG antibodies were carefully compared, which might have significant implications for the rational use of clinical PEGylated medicines.

CRediT authorship contribution statement

Jiaru Fu: Writing – review & editing, Writing – original draft, Resources, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Tianhao Ding: Writing – review & editing, Writing – original draft, Methodology, Funding acquisition, Formal analysis, Data curation, Conceptualization. Guanghui Li: Investigation, Data curation. Ercan Wu: Visualization, Data curation. Anqi Wei: Data curation. Mengyuan Liu: Data curation. Bin Wang: Supervision, Methodology, Conceptualization. Changyou Zhan: Writing – review & editing, Writing – original draft, Validation, Supervision, Project administration, Methodology, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article: