Nanoparticle-Binding Immunoglobulins Predict Variable Complement Responses in Healthy and Diseased Cohorts

Abstract

Systemic administration of nanomedicines results in the activation of the complement cascade, promoting phagocytic uptake and triggering proinflammatory responses. Identifying the biomarkers that can predict the "risk" of abnormally high complement responders can improve the safety and efficacy of nanomedicines. Polyethylene glycol (PEG) and dextran are two types of clinically approved polymer coatings that trigger complement activation. We performed a multifaceted analysis of the factors affecting the complement activation by PEGylated liposomal doxorubicin (PLD) and dextran-coated superparamagnetic iron oxide nanoworms (SPIO NWs) in plasma from patients with different inflammatory disease conditions and healthy donors. The complement activation (measured as deposition of the complement protein C3) varied greatly, with 29-fold and 26-fold differences for PLD and SPIO NWs, respectively. Chronic inflammation, acute infection, use of steroids, and sex had minor effects on the variable complement activation, whereas age inversely correlated with the complement activation. C-reactive protein level was not predictive of high (top 20th percentile) complement responses. Plasma concentrations of the main complement factors, as well as total IgG and IgM, showed no correlation with the activation by either nanoparticle. On the other hand, plasma concentrations of anti-PEG IgG and IgM showed a strong positive correlation with the activation by PLD. Particularly, titers of anti-PEG IgM showed the best predictive value for the "risk" of high complement activation by PLD. Titers of antidextran IgG and IgM showed a lower correlation with the activation by SPIO NWs and poor predictive value of the top 20% complement responses. Nanoparticle-bound immunoglobulins showed the best correlation with complement activation and a strong predictive value, supporting the critical role of immunoglobulins in inciting complement. The opsonization of PLD with C3 in plasma with high anti-PEG antibodies was predominantly via the alternative pathway. Characterizing the nature of nanoparticle-binding antibodies has important implications in mitigating and stratifying nanomedicine safety.

Keywords: CRP; anti-PEG antibody; antidextran antibody; complement; infection; inflammation; nanoparticles.

Figure 1.

Between-subject variability of complement activation. (A) Overall study design. (B) Measurements of C3 deposition on SPIO NWs (left) and PLD (right) in lepirudin plasma (raw data in Supplemental Figure 1). Blue bars show C3 deposition in lepirudin plasma; red bars are with the addition of 10 mM EDTA (total complement inhibitor). The red dotted line is the mean value. Black dotted lines show the 80th and the 95th percentile of the responses. Values are the integrated densities of the C3 signal of the dot-blot immunoassay. (C) Western blotting confirming the deposition of C3 on nanoparticles in high vs low activation plasma. Surface-deposited C3 shows the cleaved α′2 chains of iC3b and the α′2-chain (just below 75 kDa), as well as some high-molecular-weight bands. (D) C3 on SPIO NWs and PLD showing a low correlation (Pearson's r) in the same donor.

Figure 2.

Associations of disease status, steroid use, complement factors, and age with complement activation. In all subfigures: left panel, SPIO NWs; right panel, PLD. (A) Percentage differences and p-values (when significant) between donor groups. *p < 0.05, **p < 0.01. Expanded data in Supplemental Table 3. (B) “Risk” prediction (area under the receiver operator curve (ROC)) of complement activation based on cutoff CRP values. The top 20% for C3 deposition was chosen as a “risk” threshold. Whiskers mark a 95% CI. (C) Correlation between plasma complement factors and C3 deposition. Plasma with high and low levels of C3 deposition on each particle type was selected (n = 13; Supplemental Figure 2A,B for raw data). None of the correlations showed statistical significance (Pearson's r and 95% CI). (D) Inverse correlation between C3 deposition and age. (E) “Risk” prediction (area under the ROC) based on age. Whiskers mark 95% CI.

Figure 3.

Correlation between total plasma IgG/IgM and C3 deposition. (A) Plasma IgG/IgM vs C3 on SPIO NWs; and (B) plasma IgG/IgM vs C3 on PLD. There is a small correlation between plasma IgM and C3 on SPIO NWs. IgG and IgM values are the integrated densities of the dot-blot immunoassay.

Figure 4.

Levels of anti-PEG and antidextran IgG and IgM antibodies were measured with specific ELISA. Plasma was screened at 1:100 dilution. Data were plotted after subtracting residual values after inhibition with PEGylated liposomes in anti-PEG ELISA (A) and free dextran in antidextran ELISA (B). See Supplemental Figure 3 for raw data. The red dotted line is the mean value. Black dotted lines show the 80th and 95th percentiles. Values are the O.D. of the ELISA. (C) No correlation between antidextran and anti-PEG antibodies in the same donor.

Figure 5.

Correlation between anti-PEG/antidextran antibodies and C3 deposition. (A) antidextran IgG and IgM; (B) anti-PEG IgM and IgG; and (C) “risk” prediction of complement activation based on antidextran and anti-PEG antibody levels. The top 20% for C3 deposition was chosen as a “risk” threshold. The area under the ROC is the largest for anti-PEG IgM. Whiskers mark a 95% CI.

Figure 6.

(A),(B) Correlation between nanoparticle-bound immunoglobulin and C3. Immunoglobulin was measured by a dot-blot assay using an antibody that reacts with IgG, IgM, and other immunoglobulin classes. There was a strong correlation (Pearson’s r) between C3 deposition and bound immunoglobulin for both particle types. Immunoglobulin values are integrated density of the dot-blot immunoassay. (C) “Risk” prediction of complement activation based on bound immunoglobulin. Whiskers mark a 95% CI.

Figure 7.

Pathway of complement activation by PLD in plasma with high anti-PEG antibodies. Effect of the inhibitors of the classical (sutimlimab) and alternative (CR2-CR1) pathways on C3 (A) and C4d deposition (B). There was some activation of the classical pathway, but most of the C3 deposition proceeds via an alternative pathway. Means of three replicates and s.d. *p < 0.05, ***p < 0.001.