Decoupling Physisorption from Chemisorption in Clickable Lipid Nanoparticles

Abstract

Antibody conjugation is essential for targeted lipid nanoparticle (LNP) delivery, but here we show that click-chemistry produces artifacts that confound accurate measurement of covalent antibody-LNP bonding. We demonstrate that hydrophobic interactions between the most common click alkyne linker, dibenzocyclooctyne (DBCO), and the inherently hydrophobic LNP surface drive extensive nonspecific antibody physisorption, even in the absence of LNP azide groups. This physisorption yields artificially high apparent conjugation efficiencies measured by chromatographic methods. In contrast, less hydrophobic liposomes exhibit azide-dependent conjugation, highlighting a consequence of nanoparticle surface chemistry. Plasma incubation rapidly displaces physisorbed antibodies from LNPs, confirming their weak, noncovalent association, whereas covalently bound antibodies remain attached and enable effective in vivo targeting. Substituting DBCO with the less hydrophobic bicyclononyne (BCN) also reduces nonspecific associations. Our findings reveal hydrophobicity as a hidden variable in antibody–LNP conjugation and establish new standards for quantitative and reproducible measurement of targeted LNPs.

Targeted lipid nanoparticles (LNPs) are central to the next generation of nucleic acid therapeutics, enabling cell-type specific delivery beyond the liver. Recent breakthroughs, such as Capstan Therapeutics’ in vivo CAR-T platform, which delivers CAR-encoding mRNA directly to circulating T cells, demonstrate the transformative promise of extrahepatic LNP targeting and have attracted major pharmaceutical investment, including AbbVie’s acquisition of Capstan. Despite these advances, the broader translation of targeted LNPs remains constrained by the bottleneck of achieving reproducible, high-efficiency conjugation of targeting ligands to the LNP surface.

One of the most effective strategies for achieving receptor-specific targeting of LNPs is covalent conjugation of antibodies. Antibody-LNP conjugates enable selective cellular uptake and have been used in various therapeutic applications. − Numerous conjugation chemistries have been explored, including thiol–maleimide coupling, EDC/NHS amide formation, and click reactions. , Among these, copper-free click chemistry, particularly the strain-promoted azide–alkyne cycloaddition (SPAAC) between azide and the alkyne dibenzocyclooctyne (DBCO), is popular due to high reaction efficiency, biocompatibility, and ease of manipulation under mild conditions.

For targeted nanocarriers, the valency of antibody conjugation, that is, the number of antibodies per particle, is crucial in determining binding avidity and targeting performance. , Therefore, reliable purification and quantitative analysis of antibody-LNP conjugation are essential for accurate evaluation of targeting systems. For this purpose, chromatographic methods are effective at separating unbound antibodies from conjugated nanoparticles. Our laboratory has previously developed and reported extensively on nanoparticle–antibody conjugation and characterization using azide–DBCO chemistry, which offered consistent and reproducible conjugation results in liposomes.

However, when this strategy was translated from liposomes to LNPs, we observed an unexpected deviation: Conjugation efficiency appeared high even in the absence of azide functional groups on the LNPs. We hypothesized that this discrepancy arises from nonspecific hydrophobic interactions between DBCO and the LNP surface, leading to physisorption rather than chemisorption. Physisorption refers to the reversible, noncovalent adherence of molecules through hydrophobic or van der Waals forces, whereas chemisorption, or chemical adsorption, involves the formation of a covalent bond, as expected from click conjugation. The presence of physisorbed antibodies can lead to both inaccurate quantification of conjugation efficiency and the instability of the antibody-LNP conjugate upon coating in plasma, leading to only transient association in vivo and reduced tissue targeting.

In this study, we systematically quantify and dissect the parameters governing antibody-LNP conjugation using click chemistry. We reveal that interactions between hydrophobic DBCO linkers and the LNP surface lead to substantial nonspecific physisorption of antibodies on LNPs, confounding typical conjugation measurements. This hypothesis is validated by several methods, including incubating LNPs in plasma and replacing DBCO (CLogP 4.849) with the less hydrophobic alkyne group bicyclononyne (BCN, CLogP 3.065), both of which reduce nonspecific associations between the antibody and LNP. These insights provide a framework for improving the accuracy and reproducibility of antibody-LNP conjugation for targeted delivery applications.

Antibody-LNP conjugation via SPAAC is driven by an azide group (here on the surface of the LNP) and a cyclic alkyne group (here attached to an antibody) (Figure A). Azide-functionalized LNPs (azide-LNPs) are formulated as described in Figure C, with a mixture of the PEG lipids DMG-PEG(2000) and DSPE-PEG(2000)-Azide combining to make up 1.5% of the molar composition of the LNP. To prepare full length immunoglobulin G (IgG) for conjugation, antibodies are modified with a PEG-containing cyclic alkyne group (e.g., DBCO or BCN), then mixed overnight with azide-LNPs to allow the click chemistry reaction to take place (Figure A). To quantify association between antibodies and nanoparticles, the mixture is passed through a size exclusion column, collecting 24 one milliliter fractions (Figure B). Conjugation efficiency is calculated by measuring the signal from a fluorescent or radiolabeled antibody in the nanoparticle fractions (typically 7–8 mL) compared to the combined signal in the nanoparticle and free antibody peaks, with the majority of unbound antibodies eluting between fractions 14–16 mL (Figure E, Table S1). This conjugation and purification strategy is effective, and has been shown to work for both monoclonal antibodies (mAbs) and antibody fragments (Fabs), at different antibody valencies, and varying LNP concentrations, although conjugation efficiency decreases at lower LNP concentrations (Figure S1 A-C)

1.

Alkyne-antibody association with azide-LNPs depends on antibody modification chemistry, not azide density. (A) Graphical schematic depicting strain-promoted azide–alkyne cycloaddition (SPAAC) between azide-LNPs and dibenzocyclooctyne (DBCO, top) or bicyclononyne (BCN, bottom) modified antibodies. These graphics are idealized images strictly for the purpose of visualization, as they depict monomolecular, spatially oriented uniform conjugations that likely do not reflect the orientation and distribution of antibodies in vivo. (B) Schematic depicting purification and quantification of conjugation by running the antibody-nanoparticle mixtures through a size exclusion column (SEC) with collection of ≥24 fractions of 1 mL volume. Nanoparticles elute in fractions 7–8, while free, unbound antibodies elute in fractions 14–16. (C) Lipid formulation, by molar percentage, of mRNA-LNPs. (D) Quantification of the association, or apparent conjugation efficiency, measured by SEC between IgG (functionalized with a 5x molar excess of DBCO) and LNPs containing different molar percentages of DSPE-PEG(2000)-Azide at a molar ratio of 50 IgG/LNP. (E) Sample SEC elution curve measuring the signal (arbitrary units, AU) of radiolabeled or fluorescent IgG in elution fractions. IgG associated with nanoparticles is quantified in fractions 7–8 (first peak), while unbound IgG elutes in fractions 14–16 (second peak). (F-G) Quantification of the association, or apparent conjugation efficiency, measured by SEC between either 0.75% azide-LNPs (blue) or 0% azide-LNPs (black) and (F) ¹²⁵I IgG functionalized with different amounts of DBCO (D0-D5 = 0–5x molar excess of DBCO) per antibody or (G) ¹²⁵I IgG functionalized with different amounts of BCN (B0–B20 = 0–20x molar excess of BCN) per antibody at a molar ratio of 50 IgG/LNP. For panel (D), one-way ANOVA with Tukey’s multiple comparisons test was performed. For panels (F), and (G) two-way ANOVA with Tukey’s multiple comparisons test was performed. All data include n = 3 and represent mean ± SEM; * = p < 0.05, ** = p < 0.01, *** = p < 0.001, ns = not significant (not shown). Figures (F) and (G) have additional significant multiple comparisons not shown in the graph for clarity, but are reported in Table S3. Figure contains images created in BioRender. Brenner, J. (2026) https://BioRender.com/d3dsre0.

To discern the effect azide has on conjugation efficiency, we formulated LNPs with different molar percentages of DSPE-PEG(2000)-Azide and mixed each formulation with ¹²⁵I-labeled IgG modified with 5 DBCO groups per antibody (D5 IgG) at a ratio of 50 IgG per LNP. We chose D5 because higher DBCO content on the antibody leads to unwanted antibody aggregation and complement activation in vivo. , Surprisingly, the molar percentage of azide had no effect on the association between the D5 IgG and LNP, with every formulation having an apparent “conjugation efficiency” of ∼80% (Figure D). For 0% azide LNPs, an apparent 80% conjugation efficiency cannot be driven by covalent reactions, as the LNP is missing the azide necessary for SPAAC. We hypothesized that D5 IgG can associate to the surface of 0% azide-LNPs via reversible, noncovalent physisorption driven by hydrophobic interactions. In contrast, azide-LNPs can form covalent bonds with D5 IgG driven by a true click chemistry reaction (chemisorption).

To probe the role of hydrophobicity in nonspecific physisorption, we compared antibodies modified with DBCO, a highly hydrophobic alkyne, versus BCN, a less hydrophobic analogue. Antibodies were functionalized with 0–5 DBCO (D0-D5) or 0–20 BCN (B0–B20) groups per IgG and incubated with LNPs formulated with either 0.75% or 0% DSPE-PEG(2000)-Azide at a ratio of 50 IgG per LNP. A higher BCN per IgG was chosen since azide-DBCO reactions are reported to be more efficient than azide-BCN. Even a single DBCO group per antibody (D1 IgG) produced ∼55% apparent conjugation with associated increases in LNP size, indistinguishable from higher DBCO modifications (D5 IgG) (Figure F, Table S2 A-B). This high association occurred regardless of azide presence (Table S3A). In contrast, BCN-modified antibodies exhibited azide-dependent binding (Figure G). Conjugation efficiency improved as the BCN:IgG ratio was increased in reactions with azide-LNPs, while BCN-IgG interactions with nonazide LNPs remained nonsignificant up until the B20 density (Table S3B). These findings support the hypothesis that hydrophobic DBCO linkers strongly interact with the LNP surface, driving nonspecific physisorption that cannot be discerned from covalent click conjugation using SEC alone, but can be mitigated by using the less hydrophobic BCN linkers.

To determine whether hydrophobicity of the nanoparticle surface contributes to nonspecific antibody binding, we performed a parallel study using conventional liposomes. Liposomes were formulated with increasing molar percentages of azide-PEG lipid (0–6 mol %) and mixed with D5 IgG at a ratio of 50 IgG per nanoparticle. To ensure that surface PEG density was comparable between formulations, we estimated the fraction of PEG lipids exposed on the outer leaflet. Because liposomes have a bilayer structure with ∼50% of PEG groups facing inward, a 1.5 mol % azide-PEG liposome corresponds to ∼ 0.75% PEG exposed at the surface. For LNPs, small-angle neutron scattering (SANS) studies report that ∼70% of PEG lipids are surface-exposed during formation; thus, our 0.75 mol % azide-LNPs yield ∼ 0.5% surface PEG, comparable to liposomes. Despite similar surface PEG densities, LNPs showed azide-independent antibody association, whereas liposomes did not. Antibody association with liposomes was strongly azide-dependent with concurrent increases in size (Figure A, Table S2C).

2.

LNP hydrophobicity drives nonspecific DBCO-antibody binding. (A) Quantification of the association, or apparent conjugation efficiency, measured by SEC between D5 IgG and liposomes containing different molar percentages of DSPE-PEG(2000)-Azide. (B) Surface hydrophobicity (quantified by the ratio of the rose bengal to nile blue adsorption) of SM-102 LNPs vs liposomes. (C) Elution curves measuring association between 0.75% azide-LNPs (blue) or 0% azide-LNPs (black), formulated using the 4A3-SC8 ionizable lipid in place of SM-102, and ¹²⁵I D5 IgG mixed overnight at a ratio of 50 IgG per LNP. (D) Elution curves measuring the association between 0% azide-LNPs and ¹²⁵I D5 IgG (black) or ¹²⁵I D5 mouse serum albumin (MSA, blue) mixed overnight at a ratio of 50 protein per LNP. , In panel (A), a one-way ANOVA with Tukey’s multiple comparisons test was performed. In panel (B), an unpaired t test with Welch’s correction was performed. All data include n = 3 and represent mean ± SEM; **** = p < 0.0001, ns = not significant (not shown). Figure contains images created in BioRender. Brenner, J. (2026) https://BioRender.com/d3dsre0. Structures of albumin (PDB ID: 1BM0) and IgG (PDB ID: 1IGY) were modified from the RCSB Protein Data Bank.

We hypothesized that differences in surface hydrophobicity underlie the LNP vs liposome conjugation differences. To test this hypothesis, we quantified hydrophobicity using two complementary approaches. First, we calculated the composite logP (cLogP) of all lipid components, which positively correlates with hydrophobicity (Table S4). LNPs displayed a higher cLogP than liposomes, consistent with a more hydrophobic formulation. Second, we experimentally validated surface hydrophobicity using a dye-partitioning assay with hydrophilic rose bengal and hydrophobic nile blue dyes. In this method, nanoparticles are titrated into dilute dye solutions and bound versus free dye fractions are quantified spectroscopically to determine a partitioning coefficient (Figure S3). The ratio of rose bengal to nile blue adsorption (RB/NB) serves as a relative measure of surface hydrophobicity. LNPs exhibited a significantly higher RB/NB ratio than liposomes, confirming that their surfaces are more hydrophobic (Figure B). Furthermore, zeta potential measurements were comparable between the two nanoparticles, indicating that differences in electrostatic interactions do not account for the observed behavior (Table S5). These data support the hypothesis that enhanced physisorption of DBCO-antibodies arises from hydrophobic interactions between the linker and the LNP surface.

Nonspecific associations between LNPs and DBCO are independent of the ionizable lipid of the nanoparticle, as 0% azide-LNPs formulated with ionizable lipid 4A3-SC8, ALC-0315, and DLin-MC3-DMA all experienced physisorption at similar rates to the SM-102 LNPs explored in all other experiments from this paper (Figure C, Figure S2). Likewise, physisorption is independent of the DBCO modified protein, as D5 mouse serum albumin (MSA), an ∼65 kDa plasma protein, experienced nonspecific association to the surface of 0% azide-LNPs similar to D5 IgG (Figure D). In totality, these results support the hypothesis that nonspecific physisorption is due to hydrophobic interactions between the LNP and DBCO, not the conjugating protein or lipid formulation.

We next investigated whether physisorption interactions between D5 IgG and LNPs would persist in plasma. We hypothesized that, when exposed to plasma, physisorbed antibodies would dissociate from the LNP surface (Figure B). To test this, 1.5% or 0% azide-LNPs were incubated with radiolabeled D5 IgG and exposed to human plasma at 37 °C for 30 min prior to elution. The elution curves demonstrate that exposure to plasma causes substantial dissociation of D5 IgG from nonazide LNPs, as most antibodies eluted in post-LNP fractions (10–16 mL) (Figure S4). With azide-LNPs, the majority of D5 IgG eluted in LNP fractions, indicating chemisorption that is stable during plasma incubation. The percent of D5 IgG associated with LNPs pre- and postplasma incubation is quantified in Figure A. Nonazide LNPs experienced a significant decrease in D5 IgG association, dropping from ∼85% pre- to ∼40% postplasma, indicating that D5 IgG dissociates in physiological environments. Azide-LNPs also experienced a significant decrease in D5 IgG association following plasma exposure, although it was less severe, dropping from ∼85% pre- to ∼70% postplasma.

3.

Interactions between antibodies and LNPs dissociate upon exposure to plasma. (A) Quantification of the association, or apparent conjugation efficiency, measured by SEC between mixtures of either 1.5% azide-LNPs (blue) or 0% azide-LNPs (black) and ¹²⁵I D5 IgG before (solid) and after (striped) a 30 min incubation in human plasma at 37 °C. (B) Graphical schematic depicting the hypothesis that, in vitro (top) DBCO modified IgG experience a mixture of physisorption and chemisorption. Upon exposure to plasma in vivo (bottom), physisorption associations are disrupted, while the covalent bonds remain intact. (C) Experimental diagram portraying mixtures of 1.5% azide-LNPs (top) or 0% azide-LNPs (bottom) with AF488 D5 IgG incubated in plasma and analyzed via fluorescent NTA, with IgG dissociating from the surface of 0% azide-LNPs. (D) Absolute change in the average normalized fluorescent intensity per nanoparticle (arbitrary units) of either 1.5% azide-LNPs (blue) or 0% azide-LNPs (black) with AF488 D5 IgG before and after plasma incubation. (E,F) Plots of the size (nanometers, nm) vs normalized fluorescent intensity (arbitrary units, AU) of individual particles tracked via fluorescent NTA from mixtures of either (E) 0% azide-LNPs or (F) 1.5% azide-LNPs with AF488 D5 IgG before (gray) and after (green) plasma incubation. (G) Fluorescent NTA measured fluorophore-positive concentrations (particles/mL) of either 1.5% azide-LNPs (blue) or 0% azide-LNPs (black) with AF488 D5 IgG through a 30 min plasma incubation. (H) Average measured concentration (particles/mL) of particles before (solid) and after (striped) plasma incubation. For panels (A) and (H), a two-way ANOVA with Tukey’s multiple comparisons test was performed. In panel (D), an unpaired t test with Welch’s correction was performed. All data include n = 3 and represent mean ± SEM; * = p < 0.05, ** = p < 0.01, **** = p < 0.0001, ns = not significant (not shown). Figure contains images created in BioRender. Brenner, J. (2026) https://BioRender.com/d3dsre0.

To further evaluate the effect plasma has on physisorption, we incubated azide or nonazide LNPs with Alexa Fluor 488-labeled D5 IgG in human plasma at 37 °C and monitored the mixtures by fluorescent nanoparticle tracking analysis (NTA). Fluorescent NTA measured size, concentration, and fluorescence intensity of the antibody-bound nanoparticles by tracking Brownian motion of nanoparticles detectable by fluorescent signal. Because the system only detects particles > 20 nm, free antibodies are not tracked. Consequently, the technique exclusively assesses nanoparticles associated with fluorescently labeled antibodies, while unbound nanoparticles or free antibodies remain undetected. We hypothesized that, once incubated in plasma and exposed to the flow conditions in the NTA, D5 IgG on the surface of 0% azide-LNPs would dissociate from the particle, while D5 IgG chemisorbed to azide-LNPs remain bound (Figure C). Therefore, the apparent size, fluorescence intensity, and concentration of nonazide LNPs measured by the fluorescent NTA should decrease over time.

At the 30 min end point, NTA revealed marked differences between the two formulations. 0% azide LNPs exhibited a pronounced reduction in normalized fluorescence intensity, reflecting loss of surface-bound antibodies, whereas azide-LNPs maintained nearly constant fluorescence (Figures D-F). 0% azide LNPs also showed a greater decrease in measured particle size, though not statistically significant, compared with azide-LNPs (Figure S5B). Consistent with these end point results, the concentration of fluorescently detectable azide-LNPs remained stable throughout the 30 min analysis, while that of nonazide-LNPs dropped significantly (Figures G–H, S5A). Collectively, these data demonstrate that weakly adsorbed antibodies are displaced under physiological conditions, underscoring the need for plasma-incubation assays as a critical quality-control step to distinguish chemisorption from nonspecific adsorption in antibody–nanoparticle conjugates.

To determine whether physisorbed antibodies influence active targeting in vivo, we performed biodistribution studies using 0% azide and 0.75% azide-LNPs conjugated to mAbs against platelet endothelial cell adhesion molecule-1 (PECAM). PECAM is a transmembrane glycoprotein abundantly expressed on the surface of endothelial cells, particularly within the lung microvasculature. Previous studies from our group demonstrated that conjugation of αPECAM antibodies to LNPs can effectively redirect LNP delivery from the liver to the lungs. , In this study, mice were intravenously injected with equivalent doses of either azide or 0% azide LNPs that had been mixed overnight with αPECAM mAbs (∼50 mAbs per particle). To track nanoparticle biodistribution independently of antibody targeting, each formulation also contained a small amount of radiolabeled untargeted IgG (∼5 per particle, labeled with Na¹²⁵I). This tracer was chosen because, unlike αPECAM, untargeted IgG does not bind to the endothelium and remains largely in circulation. Thus, loss of antibody from the nanoparticle surface would result in elevated blood radioactivity rather than apparent lung accumulation.

Using this design, we found that αPECAM azide-LNPs exhibited significantly higher lung accumulation than αPECAM nonazide LNPs (Figure A). In contrast, αPECAM nonazide-LNPs primarily localized to the liver and spleen and displayed substantially higher blood retention, consistent with the antibodies dissociating from the nanoparticle surface and subsequently remaining in circulation. As a result, the lung localization ratio of αPECAM azide-LNPs, which accounts for blood retention (lung%ID/g normalized by whole blood%ID/g) was ∼35x higher than αPECAM nonazide LNPs (Figure S6) These findings confirm that, in vivo, noncovalent physisorption between DBCO-modified antibodies and the 0% azide-LNP surface is insufficient to mediate stable targeting, whereas covalent αPECAM chemisorption to azide-LNPs effectively redirects LNP biodistribution to the lungs.

4.

Chemisorption enables effective antibody-LNP targeting in vivo. (A) 30 min biodistribution from mice treated with either 0.75% azide-LNPs (blue) or 0% azide-LNPs (black) with D5 αPECAM mAbs (∼50 mAbs/LNP). LNPs were tracked with ¹²⁵I D5 IgG (∼5 IgG/LNP). %ID/g represents the percent of total injected dose detected in each organ, normalized to organ mass. (B) Comparison of the total percentage of eGFP positive cells in the lungs, liver, and spleen of Ai6 mice 24 h post treatment with either αPECAM 0.75% azide-LNPs (green) or 0% azide-LNPs (black) containing Cre mRNA. (C) Histogram showing eGFP expression across all lung (gray) and liver (black) cells from mice treated with αPECAM 0% azide-LNPs, demonstrating increased eGFP expression in the liver. (D) Histogram showing eGFP expression across all lung (light green) and liver (dark green) cells from mice treated with αPECAM 0.75% azide-LNPs, demonstrating increased eGFP expression in the lungs. (E) Comparison of the total percentage of eGFP positive cells in different cell types in the lungs of both αPECAM 0% azide or 0.75% azide-LNPs treated mice. For panels (A), (B), and (E), two-way ANOVA with Tukey’s multiple comparisons test was performed. All data include n = 3 and represent mean ± SEM; * = p < 0.05, ** = p < 0.01, **** = p < 0.0001, ns = not significant (not shown).

We used flow cytometry to further probe the mRNA expression of azide and nonazide LNPs. For these flow studies, Ai6 mice, which express an enhanced green fluorescent protein (eGFP) following Cre-mediated recombination of a STOP cassette, were treated with either αPECAM 0% azide or 0.75% azide-LNPs loaded with Cre recombinase mRNA. At 24 h after treatment, the lungs, liver, and spleen of the mice were processed into single cell suspensions, then stained with antibodies against general immune cells, monocytes and macrophages, neutrophils, endothelial cells, and epithelial cells (Figure S7). Analyzing the percentage of eGFP expressing cells by organ (denoting the percentage of cells transfected with Cre mRNA), the lungs of αPECAM azide-LNP treated mice had a significantly higher percentage of eGFP positive cells than the 0% azide-LNP treated group (Figures B-D). Rather, the majority of eGFP positive cells in αPECAM 0% azide-LNP treated mice were in the liver and spleen, corresponding to nonspecific clearance of the nanoparticle rather than targeted delivery. Further breaking down expression by cell type, the lungs of αPECAM azide-LNP treated mice had a significantly larger percentage of eGFP positive endothelial cells than the nonazide group, once again confirming only azide-dependent chemisorption of αPECAM mAbs drives endothelial-specific transfection (Figure E). Together, these results establish that covalent conjugation of antibodies is required for functional, cell-specific mRNA delivery and expression in vivo, whereas noncovalent physisorption fails to achieve meaningful targeting.

In this study, we systematically examined antibody-LNP conjugation using copper-free click chemistry and uncovered an important, previously overlooked artifact of this widely adopted bioconjugation approach; DBCO-modified antibodies exhibited significant nonspecific association with LNPs even in the absence of azide functional groups. This phenomenon arises from hydrophobic interactions between the DBCO linker and the inherently hydrophobic LNP surface, resulting in physisorption rather than true chemisorption via covalent click reaction.

These findings have broad implications for the field of targeted nanomedicine. First, they reveal that conventional chromatographic purification alone can lead to systematic overestimation of conjugation efficiency, particularly when hydrophobic linkers such as DBCO are used. Because the apparent conjugation efficiency increases with the degree of DBCO modification, due to enhanced hydrophobic interactions rather than increased chemical coupling, studies that rely solely on chromatography-based separation may misinterpret conjugation yields and targeting performance. Second, this work provides a standardized framework for quantitative and reproducible assessment of antibody-LNP conjugation efficiency. Such standardization is critical not only for academic reproducibility but also for industrial and regulatory translation, where consistent antibody surface density and covalent stability are key parameters for product quality control and eventual regulatory approval. Our findings therefore bridge an important gap between research-scale optimization and manufacturing-level reproducibility.

Nevertheless, several limitations should be acknowledged. The mechanistic comparison between DBCO and BCN was limited to selected experimental conditions, and future studies could systematically quantify the relationship between linker hydrophobicity, nanoparticle composition, and nonspecific binding across different antibody isotypes and LNP formulations. Future work would also benefit from incorporating more advanced in situ models to better replicate and evaluate the behavior of azide and nonazide LNPs in circulation. Likewise, antibody orientation on the nanoparticle surface was not directly assessed and may differ depending on the nature of the interaction, which could also contribute to differences in targeting performance.

Additionally, the current work focuses solely on azide–DBCO click chemistry, as it is widely used for a number of targeted drug delivery applications. However, other coupling chemistries, such as tetrazine-TCO or thiol-maleimide, may also exhibit nonspecific interactions, and merit further investigation in future work. Finally, although plasma preincubation effectively reduced nonspecific adsorption in our system, the generality of this strategy for different biological matrices or large-scale manufacturing workflows remains to be fully evaluated.

Overall, this study highlights a critical yet previously underappreciated factor in antibody-LNP conjugation: the impact of hydrophobic interactions on apparent conjugation efficiency. By identifying and addressing this confounding variable, we provide a foundation for more accurate, reproducible, and regulatory-compliant design of targeted LNP therapeutics.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsnanoscienceau.5c00170.

• Experimental methods, calculation of conjugation efficiency and bound antibodies from raw elution data; additional antibody-LNP conjugations under varying parameters; size of different antibody-nanoparticle conjugates; additional Tukey’s multiple comparisons tests; calculated LogP of liposomes vs LNPs; measured zeta potentials of liposomes and LNPs; antibody-LNP conjugations using different ionizable lipids; liposomes vs LNP hydrophobicity titrations; LNP elution curves following plasma incubation; additional fluorescent NTA measurements; lung localization ratios; gating strategies for flow cytometry (PDF)

#.

T.V.B. and E.D.H contributed equally to this work. CRediT: Taylor V. Brysgel conceptualization, data curation, formal analysis, methodology, visualization, writing - original draft, writing - review & editing; Elizabeth D Hood conceptualization, data curation, formal analysis, methodology, visualization, writing - original draft, writing - review & editing; Aleksa Milosavljevic data curation, formal analysis, writing - review & editing; Nicolas Marzolini data curation, formal analysis, writing - review & editing; Marco E Zamora data curation, formal analysis; Jichuan Wu data curation, resources; Vladimir V Shuvaev resources; Tea Shuvaeva data curation, resources; Michael Kegel data curation, resources; Jenna Muscat-Rivera data curation, resources; Drew Weissman funding acquisition, resources; Jilian R Melamed resources; Jacob W. Myerson investigation, methodology, writing - review & editing; Jacob S. Brenner funding acquisition, investigation, methodology, writing - review & editing; Vladimir R. Muzykantov funding acquisition, investigation, methodology, supervision, writing - original draft; Jia Nong conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, supervision, visualization, writing - original draft, writing - review & editing.

The authors declare no competing financial interest.