Anti-PEG Antibodies Boosted in Humans by SARS-CoV-2 Lipid Nanoparticle mRNA Vaccine.

Humans commonly have low level antibodies to poly(ethylene) glycol (PEG) due to environmental exposure. Lipid nanoparticle (LNP) mRNA vaccines for SARS-CoV-2 contain small amounts of PEG, but it is not known whether PEG antibodies are enhanced by vaccination and what their impact is on particle-immune cell interactions in human blood. We studied plasma from 130 adults receiving either the BNT162b2 (Pfizer-BioNTech) or mRNA-1273 (Moderna) mRNA vaccines or no SARS-CoV-2 vaccine for PEG-specific antibodies. Anti-PEG IgG was commonly detected prior to vaccination and was significantly boosted a mean of 13.1-fold (range 1.0-70.9) following mRNA-1273 vaccination and a mean of 1.78-fold (range 0.68-16.6) following BNT162b2 vaccination. Anti-PEG IgM increased 68.5-fold (range 0.9-377.1) and 2.64-fold (0.76-12.84) following mRNA-1273 and BNT162b2 vaccination, respectively. The rise in PEG-specific antibodies following mRNA-1273 vaccination was associated with a significant increase in the association of clinically relevant PEGylated LNPs with blood phagocytes ex vivo. PEG antibodies did not impact the SARS-CoV-2 specific neutralizing antibody response to vaccination. However, the elevated levels of vaccine-induced anti-PEG antibodies correlated with increased systemic reactogenicity following two doses of vaccination. We conclude that PEG-specific antibodies can be boosted by LNP mRNA vaccination and that the rise in PEG-specific antibodies is associated with systemic reactogenicity and an increase of PEG particle-leukocyte association in human blood. The longer-term clinical impact of the increase in PEG-specific antibodies induced by lipid nanoparticle mRNA vaccines should be monitored. It may be useful to identify suitable alternatives to PEG for developing next-generation LNP vaccines to overcome PEG immunogenicity in the future.

Humans are exposed to poly(ethylene) glycol (PEG) through consumer products and medicines. A proportion of adults have circulating anti-PEG antibodies.[1−5] Incorporation of PEG into the formulation of some systemically administered drugs to improve pharmacokinetics can result in anti-PEG antibodies that ultimately limit the bioavailability of the drug and can cause side effects.[6−8] Lipid nanoparticle (LNP) mRNA vaccines for SARS-CoV-2 contain small amounts of a PEG–lipid conjugate to stabilize the LNPs.[9,10] An estimated 50 and 117 μg (calculated based on total lipid dose and molar lipid ratios[11]) of the PEG–lipid conjugates ALC-0159 and PEG2000-DMG are present per dose for the BNT162b2 (Comirnaty, Pfizer-BioNTech) and mRNA-1273 (Spikevax, Moderna) vaccines, respectively.[12,13] Over a billion doses of SARS-CoV-2 LNP mRNA vaccines containing PEG–lipids have now been given worldwide. However, it is not known whether anti-PEG antibodies are induced or boosted by intramuscular LNP mRNA vaccination for COVID-19 and what effect, if any, this may have on vaccine responses.

Results and Discussion

SARS-CoV-2 Lipid Nanoparticle Vaccine-Induced Anti-PEG Antibody Response

We studied serial blood samples from 130 adults across three separate cohorts, including 55 adults receiving two doses of BNT162b2 vaccines (subject details in Table S1), 20 adults receiving two doses of mRNA-1273 vaccines (subject details in Table S2), and 55 unvaccinated subjects (including 40 SARS-CoV-2 infected convalescent patients and 15 healthy donors, subject details in Table S3). PEG-specific IgG and IgM antibodies were quantified by ELISA first to a 40 kDa PEG molecule as described in the Methods. For BNT162b2 and mRNA-1273 cohorts, PEG-specific IgG was detectable (end point titer >1:10) prior to vaccination in 53 of the 75 subjects (71%), ranging in titer from 1:12 to 1:3000. Of the 75 subjects, 31 subjects (41%) had a titer >1:100. This is consistent with the wide range of prevalence of anti-PEG antibodies previously reported, ranging from <4 to 72%,[3−5] potentially attributed to differences in geographic location, age of the subjects as well as different assay systems used in the individual studies. A small but consistent increase in PEG-specific IgG and IgM was observed after two doses of BNT162b2 vaccinations with a mean fold change of 1.78 (range 0.68–16.6) and 2.64 (range 0.76–12.84), respectively (Figure A). There were nine (16%) or 23 (42%) subjects who had a boost of anti-PEG IgG or IgM with more than 2-fold change after BNT162b2 vaccination, respectively. Only one or two subjects had a boost of anti-PEG IgG or IgM with more than 10-fold change after BNT162b2 vaccination, respectively. A time course analysis showed that subjects with increased PEG antibody responses had a stepwise increase after the first and second doses of BNT162b2 vaccination (Figure S4). PEG antibodies did not increase further out to 3 months in a subset of subjects (n = 10) where late postvaccination samples were available (Figure S4B). A much larger increase in anti-PEG IgG and IgM was observed after two doses of mRNA-1273 vaccinations with a mean fold change of 13.1 (range 1.0–70.9) and 68.5 (range 0.9–377.1), respectively (Figure B). There were 15 (75%) or 18 (90%) subjects who had a boost of anti-PEG IgG or IgM with more than 2-fold change after mRNA-1273 vaccination, respectively. Six (30%) or 10 (50%) subjects had a boost of anti-PEG IgG or IgM with more than 10-fold change after mRNA-1273 vaccination, respectively. In a control group of 55 unvaccinated subjects, anti-PEG IgG and IgM titers did not increase and overall slightly decreased over a 6 month period with a mean fold change of 0.92 and 0.98, respectively (Figure C). Comparing the fold change of anti-PEG antibody levels across the three cohorts (Figure D), we concluded that both anti-PEG IgG and IgM were significantly boosted by mRNA-1273 vaccination and modestly boosted by BNT162b2 vaccination. The difference in PEG immunogenicity between two vaccines has also been observed by another research group,[14] which might be due to a higher dose of PEGylated lipid in the mRNA-1273 vaccine (117 μg vs 50 μg PEG–lipid per dose) compared to the BNT162b2 vaccine. PEG immunogenicity is affected by the terminal group of PEG in that methoxy-PEG is more immunogenic than hydroxy-PEG,[15] although the PEG–lipids in the mRNA-1273 (PEG2000-DMG) and BNT162b2 formulation (ALC-0159) both contain a methoxy-PEG2000. Furthermore, the boost of anti-PEG antibody in mice depends on the rate of PEG–lipid shedding off the LNP.[16] Future animal studies are required to determine the impact of the PEG–lipid structure and LNP formulation on PEG immunogenicity of mRNA LNP vaccines.

Figure 1

Impact of SARS-COV-2 LNP mRNA vaccination on the levels of PEG specific antibodies. (A) Comparing plasma anti-PEG IgG and IgM titers before vaccination (Pre-Vax) and 2–7 weeks (mean 27 days, range 12–49) after second dose (Post-Boost) of the BNT162b2 vaccination of healthy cohort (n = 55; see raw data in Figure S1). (B) Comparing plasma anti-PEG IgG and IgM titers before vaccination (Pre-Vax) and 3 weeks (mean 19 days, range 18–27) after second dose (Post-Boost) of the mRNA-1273 vaccination of healthy cohort (n = 20; see raw data in Figure S2). (C) Comparing plasma anti-PEG IgG and IgM titers over a 6 month period (mean 186 days, range 120–253) of unvaccinated control cohort (n = 55, including 40 SARS-CoV-2 infected convalescent patients and 15 healthy donors; see raw data in Figure S3). (D) Cross-comparison of the fold change (log10) of anti-PEG IgG and IgM titers among three cohort: BNT162b2 (Post-Boost/Pre-Vax, n = 55), mRNA-1273 (Post-Boost/Pre-Vax, n = 20), and unvaccinated control cohort (6 month/baseline, n = 55). p values in (A–C) were derived by Wilcoxon’s matched-pairs signed rank test and in (D) were derived by nonparametric Kruskal–Wallis test with Dunn’s multiple comparisons test.

We observed a sex- and age-dependent PEG-specific antibody levels in 130 subjects across three cohorts. Females had higher titers of pre-existing anti-PEG IgG (mean titer 1:73 vs 1:35) and IgM (mean titer 1:28 vs 1:21) than males (Figure A). This is consistent with previous findings and has been postulated to be related to greater exposure of females to PEG-containing cosmetic products.[3] We found the age of subjects was negatively correlated with the pre-existing levels of anti-PEG IgG and IgM (Figure B). Younger subjects have been previously reported to have higher PEG-specific antibodies in a large serosurvey.[3] However, we did not observe an influence of gender or age on the change in anti-PEG IgG or IgM after BNT162b2 or mRNA-1273 vaccination of 75 subjects (Figure C,D).

Figure 2

Impact of sex and age on the levels of PEG-specific antibodies. (A) Comparing the pre-existing (Pre-Vax) anti-PEG IgG or IgM titers between male and female donors (including all the vaccinated and unvaccinated control cohorts, n = 130, 53 male vs 77 female). (B) Spearman correlation between age of donors and the pre-existing (Pre-Vax) anti-PEG IgG or IgM titers (including all the vaccinated and unvaccinated control cohorts, n = 130). (C) Comparing the fold change (log10) of anti-PEG IgG or IgM titers (Post-Boost/Pre-Vax) between male and female donors after two doses of vaccination (including all the BNT162b2 and mRNA-1273 cohorts, n = 75, 29 male vs 46 female). (D) Spearman correlation between age of donors and the fold change (log10) of anti-PEG IgG or IgM titers (Post-Boost/Pre-Vax) after two vaccinations, including all the BNT162b2 and mRNA-1273 cohorts (n = 75). p values in (A,C) were derived by Mann–Whitney U test and in (B,D) were derived by Spearman correlation analysis.

Antibody Specificity to Lipid Components of BNT162b2 and mRNA-1273 Vaccines

To determine whether the PEG antibodies detected specifically recognized the PEG–lipid or other synthetic lipid components of the BNT162b2 and mRNA-1273 vaccine, we assessed a subset of 18 vaccines where larger plasma samples were available for studying antibody response to the ALC-0159 (BNT162b2 PEG–lipid), PEG2000-DMG (mRNA-1273 PEG–lipid), ALC-0315 (BNT162b2 ionizable lipid), SM-102 (mRNA-1273 ionizable lipid), and DSPC (BNT162b2 and mRNA-1273 helper lipid). We found that there are various levels of anti-human IgG that was specific to the ALC-0159 or PEG2000-DMG, consistent with the anti-PEG IgG detected by the 40 kDa PEG (Figure A). In contrast, the other lipid components, ALC-0315, SM-102, and DSPC, elicited no antibody responses (Figure A). After vaccination, the increase of PEG–lipid-specific IgG correlated with the increase of PEG-specific IgG from the same subjects detected using the 40 kDa PEG (Figure B,C). Furthermore, a humanized monoclonal anti-PEG IgG was used to assess specificity to the synthetic lipid components of BNT162b2 and mRNA-1273 vaccine. We found the monoclonal anti-PEG IgG specifically recognized the ALC-0159, PEG2000-DMG, and 40 kDa PEG in a concentration-dependent manner, while it did not recognize ALC-0315, SM-102, or DSPC (Figure D). Taken together, these results suggest anti-PEG antibodies can specifically recognize PEG in the PEG–lipid component of the BNT162b2 and mRNA-1273 vaccines.

Figure 3

Specificity of PEG-specific antibodies to PEG–lipid and other synthetic lipid components of BNT162b2 and mRNA-1273 vaccine. (A) Specificity of anti-human IgG to 40 kDa PEG, ALC-0159 (BNT162b2 PEG–lipid), PEG2000-DMG (mRNA-1273 PEG–lipid), ALC-0315 (BNT162b2 ionizable lipid), SM-102 (mRNA-1273 ionizable lipid), and DSPC (BNT162b2 and mRNA-1273 helper lipid) in comparison to background (no precoating) of plasma from 18 healthy subjects (donors 1–18) at day 14 post-boost BNT162b2 vaccination. ELISA was performed by coating individual lipid components (ALC-0159, PEG2000-DMG, ALC-0315, SM-102, or DSPC) or 40 kDa PEG on hydrophobic plates, and the plasma were diluted at 1:10 in 5% skim milk. Data are shown as the mean of two independent measurements. (B) Spearman correlation between the fold change (log10) of anti-PEG–lipid IgG titers detected by ALC-0159 and the fold change of anti-PEG IgG titers detected by 40 kDa PEG at day 14 post-boost BNT162b2 vaccination compared to pre-vax (n = 18, donors 1–18 from BNT162b2-vaccinated cohort). (C) Spearman correlation between the fold change (log10) of anti-PEG–lipid IgG titers detected by PEG2000-DMG and the fold change of anti-PEG IgG titers detected by 40 kDa PEG at day 14 post-boost BNT162b2 vaccination compared to pre-vax (n = 18, donors 1–18 from BNT162b2-vaccinated cohort). (D) Specificity of monoclonal anti-PEG IgG to 40 kDa PEG, ALC-0159, PEG2000-DMG, ALC-0315, SM-102, and DSPC, which were precoated as antigen on hydrophobic ELISA plates. Data are shown as mean of two independent measurements.

Impact of Anti-PEG Antibodies on Nanoparticle–Immune Cell Interactions in Human Blood

PEG is a hydrophilic compound commonly formulated with therapeutics to reduce nonspecific immune clearance. Anti-PEG antibodies can reduce the bioavailability and efficacy of PEG-containing therapeutics.[6] To determine the potential of anti-PEG antibodies induced by SARS-CoV-2 LNP mRNA vaccination to promote immune-mediated clearance of nanomaterials containing PEG, we used a previously developed ex vivo human blood assay to assess the impact of plasma on particle–immune cell association (experimental design shown in Scheme ).[17] The assay incubates nanoparticles for 1 h with fresh primary human blood cells containing monocytes, granulocytes, and B cells in the presence of plasma before and after vaccination and measures nanoparticle and immune cell association by flow cytometry (details in Methods, gating strategy shown in Figure S5). We studied three PEGylated nanoparticles, including 100 nm PEGylated mesoporous silica (PEG-MS) nanoparticles, PEGylated doxorubicin-encapsulated liposomes that were formulated with the same lipid composition and drug/lipid ratio as U.S. Food and Drug Administration (FDA)-approved liposomal doxorubicin agent (Doxil) for anticancer chemotherapy,[18] and PEGylated LNPs that contain the same molar composition of lipids as clinically used Onpattro (Patisiran) LNP formulation for treating transthyretin-mediated amyloidosis (characterization in Figure and Table S4).[19] Both Doxil and Onpattro LNPs are intravenously administered and therefore are relevant nanomedicine models for studying bio–nano interactions in human blood. All three nanoparticles displayed a range of plasma donor-dependent association with immune cells (Figures S6–S8), consistent with our previous findings.[20] We observed that the association of PEGylated nanoparticles with human blood immune cells strongly correlated with the baseline levels of anti-PEG IgG and IgM. Specifically, the association of Onpattro LNPs with granulocytes and monocytes correlated with pre-existing anti-PEG IgG and IgM (Figure A,B). Similarly, the association of Doxil and PEG-MS nanoparticles with monocytes and granulocytes both correlated with the pre-existing levels of anti-PEG IgG, while the association of Doxil and PEG-MS nanoparticles with B cells correlated with pre-existing anti-PEG IgM and IgG, respectively (Figures S9 and S11).

Scheme 1

Schematic Illustration of Human Blood Assay to Assess the Impact of Plasma on PEGylated Nanoparticle Association with Human Immune Cells

Fresh human blood from a healthy donor was washed by centrifugation with serum-free media multiple times to completely remove plasma. PEGylated nanoparticles were preincubated with plasma from multiple donors before vaccination (Pre-Vax) and post-boost of BNT162b2 or mRNA-1273 vaccination and then incubated with washed blood in serum-free media for 1 h at 37 °C, followed by phenotyping cells with antibody cocktails and analysis by flow cytometry. Created with BioRender.com.

Figure 4

Characterization of PEGylated nanoparticles. (A) Dynamic light scattering of PEG-MS nanoparticles, Doxil, and Onpattro LNPs. (B) Transmission electron microscopy images of PEG-MS nanoparticles. (C,D) Cryogenic transmission electron microscopy images of Doxil and LNPs. Scale bars are 100 nm in (B–D).

Figure 5

Impact of pre-existing anti-PEG antibody levels on the association of Onpattro LNPs with immune cells in human blood. (A,B) Spearman correlation between pre-existing (Pre-Vax) anti-PEG IgG or IgM titers (n = 20, donors 56–75 from mRNA-1273-vaccinated cohort) and LNP association with monocytes, granulocytes, or B cells in human blood. Cell association (%) refers to the proportion of each cell type with positive fluorescence, above background, stemming from fluorescence-labeled particles (see gating strategy in Figure S5.). Cell association (%) data are shown as the mean of three independent experiments (using the same batch of plasma from each donor), with at least 150,000 leukocytes analyzed for each experimental condition studied (see raw data in Figure S6).

Importantly, we observed a significant increase in the association of Onpattro LNPs with blood phagocytes (i.e., granulocytes and monocytes) in plasma obtained after mRNA-1273 vaccination (Figure A). The increase in anti-PEG IgG and IgM antibodies after mRNA-1273 vaccination correlated with the increase in association of Onpattro LNPs to granulocytes and monocytes (Figure B). Similar to Onpattro LNPs, we observed an increase in Doxil–monocyte association and PEG-MS nanoparticle–B cell association mediated by plasma after BNT162b2 vaccination (Figures S10A and S12A), which correlated with the level of changes in anti-PEG IgM levels in plasma (Figures S10B and S12B). The nanoparticle–B cell association is likely mediated by complement receptors on B cells since heat inactivation of plasma to destroy complement activity has previously been shown to inhibit this association.[21] In contrast to mRNA-1273 vaccination, the increase in blood immune cell association with PEGylated nanoparticles was lower for most plasma after BNT162b2 vaccination (Figures S10 and S12), likely due to the smaller boost of anti-PEG antibodies following BNT162b2 vaccination. These ex vivo results across three independent particle systems and two mRNA LNP vaccinated cohorts demonstrate anti-PEG antibodies boosted by mRNA LNP vaccination can increase immune cell interactions with other PEG-based nanomaterials. Further, the increased level of anti-PEG antibodies induced by mRNA-1273 vaccination significantly enhanced human blood immune cell association of the PEG-containing therapeutic Onpattro. These findings could be biologically relevant given that Onpattro and Doxil are administered intravenously and will be opsonized by plasma PEG antibodies if present and then associate with blood immune cells. Plasma proteins are known to opsonize foreign nanomaterials in the blood, forming so-called “biomolecular coronas”,[22−24] which leads to recognition and inactivation of nanomaterials by the immune system.[25−28] The formation of biomolecular coronas on Doxil in vivo in human systemic circulation has been previously described.[23] Recently, the composition of biomolecular coronas was also found to regulate fate and/or utility of LNPs.[29,30] In our previous studies, we found that the enrichment of immunoglobulins and complement proteins in biomolecular coronas is correlated with donor-specific nanoparticle association with human blood immune cells.[20] The studies herein demonstrate that anti-PEG antibodies are probably one of the key immunoglobulins that drive the donor-dependent immune cell association of PEG-containing nanoparticles in human blood.

Figure 6

Impact of change in anti-PEG antibody after mRNA-1273 vaccination on the association of Onpattro LNPs with immune cells in human blood. (A) Comparing the impact of plasma collected before vaccination (Pre-Vax) and post-boost of the mRNA-1273 vaccination (n = 20, donors 56–75) on Onpattro LNP association with monocytes, granulocytes, or B cells in human blood. p values were derived by Wilcoxon’s matched-pairs signed rank test. (B) Spearman correlation between the fold change (log10) of anti-PEG IgG or IgM titers (Post-Boost/Pre-Vax) after two doses of mRNA-1273 vaccination (n = 20) and change in Onpattro LNP association with monocytes, granulocytes, or B cells in human blood. Cell association (%) refers to the proportion of each cell type with positive fluorescence, above background, stemming from fluorescence-labeled particles. Cell association (%) data are shown as the mean of three independent experiments (using the same batch of plasma from each donor), with at least 150,000 leukocytes analyzed for each experimental condition studied (see raw data in Figure S6). Change in cell association (%) refers to post-boost cell association (%) minus pre-vax cell association (%).

Anti-PEG Antibody-Mediated Complement Protein Binding

Studies have shown the enrichment of immunoglobulins in biomolecular coronas can trigger complement activation and opsonization of nanoparticles.[31] To understand the role of anti-PEG antibody in complement deposition, we developed an antibody-dependent complement deposition assay to assess anti-PEG antibody-mediated complement component 1q (C1q) binding in plasma collected before and after mRNA-1273 vaccination (see Methods section). We found that anti-PEG antibody-mediated C1q binding was negligible in plasma collected before vaccination but increased to various levels in plasma collected after two doses of mRNA-1273 vaccination from 11 out of 20 subjects (Figure A). The level of C1q binding significantly correlated with levels of anti-PEG IgG and IgM (Figure B), illustrating anti-PEG antibody-dependent C1q binding.

Figure 7

Anti-PEG antibody-mediated C1q binding. (A) ELISA absorbance showing the PEG-specific antibody-mediated C1q binding of plasma collected before vaccination (Pre-Vax) and post-boost of the mRNA-1273 vaccination (n = 20, donors 56–75). ELISA was performed by coating 40 kDa PEG on ELISA plates and the plasma were diluted at 1:5 in 1% bovine serum albumin (BSA), followed by washing and incubating with biotinylated C1q and horseradish peroxidase (HRP)-conjugated streptavidin. Non-PEG-precoating wells were treated the same as PEG-precoated wells and used as background for absorbance subtraction (see raw absorbance data in Figure S13). Data are shown as mean of two independent measurements. (B) Spearman correlation between anti-PEG IgG or IgM titers and C1q binding absorbance of plasma collected before vaccination (Pre-Vax) and post-boost of the mRNA-1273 vaccination (n = 40 from 20 donors, donors 56–75, two time points).

Impact of Anti-PEG Antibody on SARS-CoV-2 Specific Neutralizing Antibody Response

Many PEG-based therapeutics are delivered intravenously to promote biodistribution throughout the body, and anti-PEG antibodies can interfere with this process. Intramuscular vaccination, however, exerts its effect primarily via the vaccine traversing lymphatics and stimulating immunity in the draining lymph node. This process may be less impacted by systemic anti-PEG antibodies. The impact of PEG-specific antibodies on human LNP vaccination was analyzed by quantifying SARS-CoV-2-specific neutralizing antibody levels 2–7 weeks after the second vaccination of BNT162b2 or mRNA-1273 using a live virus neutralization assay, as neutralization responses have emerged as a key correlate of protective immunity.[32] All 75 vaccinated subjects (55 BNT162b2 and 20 mRNA-1273 vaccines) generated neutralizing antibody responses to SARS-CoV-2 after the second vaccine (median 346, range 82–1848 ID50). There was no correlation between the absolute levels or change of PEG-specific IgG or IgM antibodies before or after vaccination and the neutralizing antibody titers (Figure S14). This suggests PEG antibodies induced by two doses of BNT162b2 or mRNA-1273 vaccination unlikely interfere with vaccine immunogenicity.

Impact of Anti-PEG Antibody on Vaccine Reactogenicity

There has been speculation that anti-PEG antibody responses may be involved in reactogenicity and anaphylaxis associated with COVID-19 lipid mRNA nanoparticle vaccination.[33−37] Anaphylaxis and other manifestations of hypersensitivity are rare adverse events after SARS-CoV-2 LNP mRNA vaccination,[11] while systemic and local reactogenicity after the mRNA-1273 and BNT162b2 vaccination is common.[38] We assessed self-reported local and systemic reactions after two doses of the BNT162b2 and mRNA-1273 vaccination from a subset of 63 of the 75 vaccines where reactogenicity data were available. Local (at the injection site), systemic (constitutional symptoms), and total reactogenicity was scored as the sum of self-reported local, systemic, and total adverse events, respectively (Tables S5 and S6). Vaccine reactions were generally mild, and no subjects experienced anaphylaxis or myocarditis in our cohort. We found the increase in levels of anti-PEG IgG and IgM correlated with a higher rate of systemic (and total) reactogenicity post-boost of vaccination (Figure A,B). In contrast, no association between local reactogenicity and change in anti-PEG antibody was observed (Figure A,B). Our study suggests that boosting of anti-PEG antibodies by mRNA LNP vaccination is associated with higher systemic reactogenicity of SARS-CoV-2 LNP mRNA vaccination. Larger cohort studies and animal studies are required to confirm the causal relationship between reactogenicity and change in anti-PEG antibodies.

Figure 8

Impact of change in PEG specific antibody on reactogenicity after two doses of SARS-CoV-2 LNP mRNA (BNT162b2 or mRNA-1273) vaccination. (A,B) Spearman correlation between the fold change (log10) of anti-PEG IgG or IgM titers (Post-Boost/Pre-Vax) and post-boost reactogenicity (n = 63, including 44 BNT162b2 and 19 mRNA-1273 vaccines that have reactogenicity reports are available in Tables S5 and S6).

Conclusions

We investigated serial plasma samples from 130 adults and observed that PEG-specific IgG and IgM antibodies were boosted by mRNA-1273 vaccination and to a lesser extent by BNT162b2 vaccination. Although pre-existing levels of anti-PEG antibody were sex- and age-dependent, the boost level of anti-PEG antibody following vaccination was not affected by sex or age. The impact of vaccine-induced anti-PEG antibody on nanoparticle–immune cell interactions was studied using an ex vivo human blood assay. We found significant increases in blood cell association with clinically relevant PEGylated LNPs mediated by the anti-PEG antibodies in human plasma after mRNA-1273 vaccination. The PEG-specific antibodies induced by mRNA-1273 vaccination were also able to bind the complement protein C1q in a proportion of subjects, which could potentially result in complement opsonization of nanoparticles. Our study suggests it will be important to monitor the impact of mRNA LNP vaccination on the biology of other PEGylated medicines. We further studied the clinical impact of anti-PEG antibodies induced by the vaccination. PEG-specific antibodies present prior to or following BNT162b2 or mRNA-1273 vaccination did not negatively impact the neutralization immune response to two doses of the vaccines. However, we found the increase in anti-PEG antibodies correlated with higher systemic reactogenicity after 2 doses of SARS-CoV-2 LNP mRNA (BNT162b2 and mRNA-1273) vaccination. Larger and longer studies are needed to analyze the longer-term impact of boosting anti-PEG antibodies by LNP mRNA vaccination. Booster doses of SARS-CoV-2 LNP mRNA vaccines, now commonly recommended to enhance immunity[39] and for emerging variants of SARS-CoV-2,[40] as well as new mRNA LNP vaccines and therapies for other infectious diseases or cancers[41] may further increase PEG-specific antibodies. Generating next-generation LNPs with alternatives to PEG[42] may be useful for overcoming PEG immunogenicity in the future.

Methods

Human Subjects

Seventy-five healthy adults were recruited to donate blood samples before and after receiving the BNT162b2 (n = 55) or mRNA-1273 (n = 20) LNP mRNA vaccine intramuscularly. Healthy unvaccinated subjects (n = 15) followed over time were recruited prior to the COVID-19 pandemic with two plasma samples stored at approximately a 6 month interval. In addition, COVID-19 infected subjects (n = 40) followed over time with two plasma samples stored at approximately a 6 month interval were also studied as unvaccinated controls. Subject details and timing of sampling are shown in Tables S1–S3. All subjects provided written informed consent, and the studies were approved by the University of Melbourne human research and ethics committee (approvals 2056689, 21198153983, and 2021-21626-15756-1) and the Royal Melbourne Hospital human research and ethics committee (approvals 6309/MH-2020, 2013.290, and 68355/MH-2020). For all participants, whole blood was collected with sodium heparin anticoagulant. Plasma was collected, aliquoted, and stored at −80 °C. Adverse events were collected from subjects after vaccination using a self-reported questionnaire (Tables S5 and S6). Reactogenicity scores were calculated by the summing the number of the positive adverse events.

ELISA to Detect Anti-PEG Antibodies in Human Plasma

The ELISA was conducted based on a modified previously developed method.[3] The 8-arm PEG-NH2 (40 kDa, 200 μg mL–1, JenKem Technology, USA) in PBS was coated onto MaxiSorp 96-well plates (Nunc, Denmark) for 18 h at 4 °C, followed by washing with PBS four times. Plates were blocked with 5% (w/v) skim milk powder in PBS for 5 h when detecting anti-PEG IgG or 22 h when detecting anti-PEG IgM, followed by adding serially diluted human plasma in 5% skim milk in duplicate for 1 h at 22 °C. Plates were washed with 0.1% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS, Sigma-Aldrich, USA)/PBS buffer twice and PBS four times prior to addition of a horseradish peroxidase (HRP)-conjugated anti-human IgG (Dako Agilent, USA) at 1:20,000 dilution or HRP-conjugated anti-human IgM (Jackson ImmunoResearch Laboratories, USA) at 1:10,000 for 1 h at 22 °C. Plates were washed as above and then developed using 3,3′,5,5′-tetramethylbenzidine (TMB) liquid substrate (Sigma-Aldrich, USA). Reaction was stopped with 0.16 M H2SO4 and read at 450 nm. End point titers were calculated as the reciprocal plasma dilution giving signal 2× background using a fitted curve (4-parameter log regression) and reported as a mean of duplicates. Background was detected by adding the diluted plasma samples (at 1:10 dilution in 5% skim milk) to the non-PEG-coated wells, followed by the same ELISA procedure.

ELISA to Detect Plasma Antibody Response to the Synthetic Lipid Components of BNT162b2 and mRNA-1273 Vaccine

Lipid components 2-[(polyethylene glycol)-2000]-N,N-ditetradecylacetamide (ALC-0159, Sinopeg, China), 1,2-dimyristoyl-rac-glycero-3-methoxypolyethylene glycol-2000 (PEG2000-DMG, Avanti Polar Lipids, USA), (4-hydroxybutyl)azanediyl)bis(hexane-6,1-diyl)bis(2-hexyldecanoate) (ALC-0315, Sinopeg, China), heptadecan-9-yl 8-((2-hydroxyethyl)(6-oxo-6-(undecyloxy)hexyl)amino)octanoate (SM-102, Sinopeg, China), and 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC, Avanti Polar Lipids, USA) in 100% ethanol were coated onto PolySorp 96-well plates (Nunc, Denmark) at 100 μg mL–1 for 20 h at 37 °C, followed by washing with PBS four times. Plates were blocked with 5% (w/v) skim milk powder in PBS for 5 h, followed by adding serially diluted human plasma in 5% skim milk in duplicate for 1 h at 22 °C. The humanized monoclonal anti-PEG IgG (Hu-6.3-IgG, Institute of Biomedical Sciences at Academia Sinica, Taiwan) was serially diluted in 5% skim milk in duplicate and added to PolySorp 96-well plates for 1 h at 22 °C. Plates were then washed and treated in the same way as the section above.

ELISA to Detect Anti-PEG Antibody-Mediated C1q Binding

Complement C1q protein was purchased from MP Biomedicals (USA) and was conjugated with biotin using Sulfo-NHS-LC-LC-Biotin (Thermo Scientific, USA) according to the manufacturer’s instruction. The 8-arm PEG-NH2 (40 kDa, 200 μg mL–1, JenKem Technology, USA) in PBS was coated onto MaxiSorp 96-well plates (Nunc, Denmark) for 18 h at 4 °C, followed by washing with PBS four times. Plates were blocked with 5% (w/v) bovine serum albumin (BSA) in PBS for 18 h, followed by adding diluted human plasma at 1:5 in 1% BSA in duplicate for 1 h at 22 °C. Biotinylated C1q and HRP-conjugated streptavidin (Thermo Scientific, USA) were mixed at a molar ratio of 4:1 for 30 min at 22 °C. After plasma incubation, plates were washed with 0.1% CHAPS/PBS buffer twice and PBS four times prior to addition of the premixed biotinylated C1q and HRP-conjugated streptavidin at 2 ug mL–1 of C1q for 2 h at 22 °C. Plates were washed as above and then developed using TMB liquid substrate (Sigma-Aldrich, USA). Reaction was stopped with 0.16 M H2SO4 and read at 450 nm. Non-PEG-precoated wells were treated the same as PEG-precoated wells and used as background for absorbance subtraction.

Particle Preparation

Alexa Fluor 647 (AF647)-labeled PEGylated mesoporous silica (PEG-MS) nanoparticles were prepared using a previously published method.[20] Dioctadecyl-3,3,3,3-tetramethylindodicarbocyanine (DiD)-labeled Doxoves, PEGylated doxorubicin-encapsulated liposomes that were formulated with the same lipid composition and drug/lipid ratio as FDA-approved liposomal doxorubicin agent (Doxil), in addition to 0.2 mol % DiD, were purchased from FormuMax Scientific Inc., USA. DiD-labeled PEGylated LNPs were formulated with (6Z,9Z,28Z,31Z)-heptatriaconta-6,9,28,31-tetraen-19-yl 4-(dimethylamino)butanoate (DLin-MC3-DMA, DC Chemical, China), DSPC (Avanti Polar Lipids, USA), cholesterol (Sigma-Aldrich, USA), and PEG2000-DMG (Avanti Polar Lipids, USA) with the same molar composition of lipids used in the US FDA-approved Onpattro (Patisiran) formulation, in addition to 0.1 mol % DiD (Thermo Fisher Scientific, USA), using the NanoAssemblr platform (Precision NanoSystems, Canada). Particles were loaded with a nonimmunogenic nucleic acid cargo (firefly luciferase plasmid DNA, PlasmidFactory GmbH & Co. KG, Germany) in order to regulate lipid packing and particle size.

Particle Characterization

Transmission electron microscopy (TEM) images were acquired using an FEI Tecnai TF20 instrument at an operation voltage of 120 kV under liquid nitrogen cooling. PEG-MS particle suspensions were dropped and air-dried on Formvar carbon-coated copper grids (plasma-treated). Cryogenic TEM (Cryo-TEM) images were acquired using a FEI Tecnai Spirit instrument at an operation voltage of 120 kV under liquid nitrogen cooling under low dose conditions. Doxil and LNP suspensions were applied to glow-discharged lacey carbon grids and vitrified using a Vitrobot (FEI) system. Vitrified sample grids were stored in liquid nitrogen prior to imaging. Dynamic light scattering (DLS) analysis of the particles was performed on a Zetasizer Nano-ZS (Malvern Instruments, UK) instrument. Zeta-potential measurements of the particles were performed at pH 7.4 in phosphate buffer (5 mM) using a Zetasizer Nano-ZS (Malvern Instruments, UK). The mass of the PEG-MS particles was determined by weighing the freeze-dried particles. The concentration of Doxil (based on lipid) was provided by the manufacturer. The concentration of encapsulated pDNA in Onpattro LNPs (97 μg/mL) was determined using the Quant-iT PicoGreen dsDNA Assay kit; encapsulation efficiency of LNPs utilized in this study was >96%.

Blood Assay to Determine Nanoparticle Association with Human Immune Cells

The human blood assay was conducted based on a modified previously developed method.[17] The fresh human blood was collected from a healthy volunteer (donor 16) into sodium heparin vacuettes (Greiner Bio-One). The blood was centrifuged (900g, 15 min, no brake), followed by removing plasma in the top layer. The remaining blood cells were further washed with serum-free RPMI 1640 medium five times to remove any residual plasma proteins. The absence of plasma proteins was confirmed by the lack of absorbance of the supernatant at 280 nm (Nanodrop 2000, Thermo Fisher Scientific, USA). Washed blood cells were finally dispersed in serum-free RPMI 1640 medium and counted with a CELL-DYN Emerald analyzer. Nanoparticles (3.2 μg of PEG-MS, 3.2 μg of Doxil, or 1 μg of LNP) were preincubated in plasma (100 μL) from different subjects (donors 1–18 from BNT162b2-vaccinated cohort or donors 56–75 from mRNA-1273-vaccinated cohort, collected before vaccination vs post-boost vaccination) at 37 °C for 1 h. Individual plasma from different subjects at each time point (pre-vax vs post-boost) were used to coat the nanoparticles, forming a personalized protein corona. The nanoparticles (0.2 μg) with a personalized plasma protein corona were subsequently added to washed blood cells (5 × 105) in serum-free RMPI 1640 medium (100 μL) at a particle concentration of 2 μg mL–1 for 1 h at 37 °C. After incubation, the red blood cells were lysed using Pharm Lyse buffer (4 mL) and removed by centrifugation (500g, 5 min). After being washed twice with flow cytometry staining (FACS) wash buffer (PBS containing 2 mM EDTA and 0.5% w/v BSA), the cells were phenotyped at 4 °C for 1 h using an antibody cocktail consisting of CD3 AF700 (SP34-2, BD), CD14 APC-H7 (MΦP9, BD), CD66b BV421 (G10F5, BD), CD45 V500 (HI30, BD), CD56 PE (B159, BD), lineage-1 (Lin-1) cocktail FITC (BD), HLA-DR PerCP-Cy5.5 (G46-6, BD), and CD19 BV650 (HIB19, BioLegend) in titrated concentrations. The cells were subsequently washed twice with FACS wash buffer (4 mL, 500g, 7 min) to remove unbound antibodies and fixed with 1% w/v formaldehyde in PBS and directly analyzed by flow cytometry (LSRFortessa, BD Bioscience). The data were processed using FlowJo V10. Cell association (%) refers to the proportion of each cell type with positive fluorescence, above background, stemming from fluorescence-labeled nanoparticles (gating strategy shown Figure S5). Data were reported as a mean of three independent experiments (using the same batch of plasma from each donor), with at least 150,000 leukocytes analyzed for each experimental condition studied. Cell association data of T cells, NK cells, and dendritic cells were uniformly low (data not shown) and thus excluded from further data analysis.

SARS-CoV-2 Live Virus Neutralization Assay

A SARS-CoV-2 isolate (CoV/Australia/VIC/01/2020) was passaged in Vero cells and stored at −80 °C. The 96-well flat bottom plates were seeded with Vero cells (20,000 cells per well in 100 μL). The next day, Vero cells were washed once with 200 μL serum-free DMEM prior to the addition of 150 μL of infection media (serum-free DMEM with 1.33 μg/mL TPCK trypsin). Eight 2.5-fold serial dilutions of heat-inactivated sera (from 1:20 to 1:12,207) were incubated with SARS-CoV-2 virus at 2000 TCID50/mL at 37 °C for 1 h. Next, sera-virus mixtures (50 μL) were added to Vero cells in duplicate and incubated at 37 °C for 48 h. “Cells only” and “virus + cells” controls were included to represent 0 and 100% infectivity, respectively. After 48 h, all cell culture media were carefully removed from wells, and 200 μL of 4% formaldehyde was added to fix the cells for 30 min at room temperature. The plates were then dunked in a 1% formaldehyde bath for 30 min to inactivate any residual virus prior to removal from the BSL3 facility. Cells were washed once in PBS and then permeabilized with 150 μL of 0.1% Triton-X for 15 min. Following one wash in PBS, wells were blocked with 200 μL of blocking solution (4% BSA with 0.1% Tween-20) for 1 h. After three washes in PBST (PBS with 0.05% Tween-20), wells were added with 100 μL of rabbit polyclonal anti-SARS-CoV N antibody (Rockland, #200-401-A50) at a 1:8000 dilution in dilution buffer (PBS with 0.2% Tween-20, 0.1% BSA, and 0.5% NP-40) for 1 h. Plates were then washed six times in PBST and added with 100 μL of goat anti-rabbit IgG (Abcam, #ab6721) at a 1:8000 dilution for 1 h. After six washes in PBST, plates were developed with TMB and stopped with 0.15 M H2SO4. OD values read at 450 nm were then averaged for duplicate samples and used to calculate % neutralization with the following formula: (“virus + cells” – “sample”) ÷ (“virus + cells” – “cells only”) × 100. Inhibitory dilution 50 (ID50) values were determined using four-parameter nonlinear regression in GraphPad Prism with curve fits constrained to have a minimum of 0% and maximum of 100% neutralization.

Statistical Analyses

Pair comparison in Figures A–C, 6A, S4B (between 1 month post-boost and 3 month post-boost), S10A, and S12A was derived by Wilcoxon’s matched-pairs signed rank test using Prism 9.0 (GraphPad). Cross-comparison between the three cohorts in Figure D were derived by nonparametric Kruskal–Wallis test with Dunn’s multiple comparisons test using Prism 9.0 (GraphPad). Statistical significance in Figure A,C was assessed by Mann–Whitney U test using Prism 9.0 (GraphPad). Longitudinal comparison in Figure S4 was derived by nonparametric Friedman’s test with Dunn’s multiple comparisons test in Prism 9.0 (GraphPad). Correlation analysis in Figures B,D, 3B,C, 5, 6B, 7B, 8, S9, S10B, S11, S12B, and S14 were assessed using nonparametric Spearman correlation tests in Prism 9.0 (GraphPad).