Journal: bioRxiv

Article Title: Resolving heterogeneity of targeted lipid nanoparticles through solution-based biophysical analyses

doi: 10.64898/2026.03.31.715590

Figure Lengend Snippet: ( a ) SAXS profiles of NT LNPs and tLNPs, showing the characteristic Bragg peak feature associated with internal lipid–RNA organization. ( b ) Fitting of the Bragg peak feature using a multiple-Lorentz model where red is the first-order Bragg peak fit, green represents higher-order particle disorder and blue is the cumulative fit of the two features. (c) In-line AF4-UV-Vis contour maps (200–300 nm) depicting wavelength-resolved absorbance of LNPs during separation. The 260 nm absorbance signal is characteristic of encapsulated RNA, enabling identification of RNA-containing LNP populations across the AF4 elution profile. Overlapping spectral features indicate the presence of multiple co-eluting populations with distinct compositional profiles, motivating subsequent chemometric deconvolution. ( d ) Absorbance at 260 nm (RNA-associated signal) and 280 nm (protein-associated signal) from ( c ) plotted against the corresponding 260:280 ratio for each LNP formulation. Deviations in the 260:280 ratio across elution time indicate heterogeneity in RNA and protein content, suggesting the presence of compositionally distinct subpopulations that cannot be resolved by bulk measurements alone. These data were further subjected to chemometric analysis (see Supplemental Figure 6). The determined ( e ) R h profiles derived from in-line DLS and ( f ) molar mass profiles derived from MALS analysis for NT LNPs (beige) and tLNPs (colors) overlaid with UV fractograms from in-line AF4 separation. ( g ) Peak 260:280 ratios from ( d ), shown for comparison across LNP groups. ( h ) In-line DLS R h and MALS-derived ( i ) mass, ( j ) radius of gyration, and ( k ) polydispersity plotted for comparison across LNP groups. Measurements are reported mean ± standard error for ( h–k).

Article Snippet: 100 μL of LNP at a particle concentration of ∼10 particles were injected and eluted isocratically from an Eclipse TM NEON field flow fractionation (AF4) instrument with dilution control module using a variable-height short channel with a 275 μm spacer and a 10 kDa regenerated cellulose membrane (Wyatt Technology), connected to a 1260 Infinity II HPLC system with a G711A quaternary pump and a G7167A multisampler (Agilent Technologies).

Techniques: Formulation, Derivative Assay, Comparison

Journal: bioRxiv

Article Title: Resolving heterogeneity of targeted lipid nanoparticles through solution-based biophysical analyses

doi: 10.64898/2026.03.31.715590

Figure Lengend Snippet: ( a ) SAXS elution profiles of NT LNPs and tLNPs. Dotted boxes indicate the main LNP elution region, expanded in ( b ). Evolving factor analysis (EFA) ( b ) applied to AF4-SAXS data, highlighting frames corresponding to discrete evolving species (C1–C3) within each formulation. The resolved components correspond to distinct LNP subpopulations that differ in size and internal organization and a third component (when present) reflecting a minor population of highly heterogeneous or higher-mass particles. The temporal evolution of EFA-selected frames across the elution peak indicates that these components arise from partially overlapping populations that co-elute but differ in hydrodynamic size and composition. ( c ) Singular value decomposition (SVD) of SAXS datasets, indicating the number of independent signals present in each formulation. The presence of multiple significant singular values indicates that the SAXS signal cannot be described by a single particle population, necessitating decomposition into independent scattering components. (d) Volatility of ratio (V r ) analysis used to assess the statistical uniqueness of deconvoluted component subpopulations relative to the ensemble-averaged population. ( e-h ) Regularized alternating least squares (REGALS) deconvolutions of SAXS data yields component-resolved scattering profiles. Shown are the ( e ) ensemble-averaged SAXS profile over the elution peak, and the REGALS-derived scattering profiles for the ( f ) C1, ( g ) C2, and ( h ) C3 subpopulations. These structural components are consistent with compositionally distinct populations inferred from UV-based chemometric analysis, linking RNA-associated heterogeneity to differences in particle size and morphology. Together, these analyses demonstrate that the ensemble SAXS signal arises from multiple, structurally distinct tLNP subpopulations that can be resolved and interpreted through chemometric decomposition.

Article Snippet: 100 μL of LNP at a particle concentration of ∼10 particles were injected and eluted isocratically from an Eclipse TM NEON field flow fractionation (AF4) instrument with dilution control module using a variable-height short channel with a 275 μm spacer and a 10 kDa regenerated cellulose membrane (Wyatt Technology), connected to a 1260 Infinity II HPLC system with a G711A quaternary pump and a G7167A multisampler (Agilent Technologies).

Techniques: Formulation, Single Particle, Derivative Assay

Journal: bioRxiv

Article Title: Resolving heterogeneity of targeted lipid nanoparticles through solution-based biophysical analyses

doi: 10.64898/2026.03.31.715590

Figure Lengend Snippet: ( a ) Guinier analyses with corresponding residuals for SVD-resolved C1 and C2 components of NT LNPs and tLNPs, with the exception of F(ab’) 2 tLNPs, where only C3 is shown. White regions indicate components for which Guinier analysis failed due to large size (q min R g > 1.3). ( b ) P(r) analyses normalized by I(0) for the average profile and individual components (C1–C3) for NT LNPs and tLNPs. ( c ) Radius of gyration (R g ) and ( d ) maximum dimension (D max ) of the average profile and individual components derived from GNOM analysis for NT LNPs and tLNPs. Blank regions denote populations where GNOM analysis was invalid (q min D max > 4). ( e ) LNP shape factor calculated as D max / R g , where values of ∼2.58 and ∼3.0 correspond to spherical and prolate ellipsoid geometries, respectively. DENSS ab initio electron density reconstructions from the AF4-UV-DLS-MALS-SAXS profiles for ( f ) NT LNPs, ( g ) nanobody tLNPs, ( h ) DAPRin tLNPs, ( i ), F(ab’) 2 tLNPs, and ( j ) antibody tLNPs.

Article Snippet: 100 μL of LNP at a particle concentration of ∼10 particles were injected and eluted isocratically from an Eclipse TM NEON field flow fractionation (AF4) instrument with dilution control module using a variable-height short channel with a 275 μm spacer and a 10 kDa regenerated cellulose membrane (Wyatt Technology), connected to a 1260 Infinity II HPLC system with a G711A quaternary pump and a G7167A multisampler (Agilent Technologies).

Techniques: Derivative Assay

Journal: bioRxiv

Article Title: Resolving heterogeneity of targeted lipid nanoparticles through solution-based biophysical analyses

doi: 10.64898/2026.03.31.715590

Figure Lengend Snippet: Targeted mRNA delivery to the placenta is driven by tLNP structural subspecies. ( a-d ) DiR-labeled NT LNPs and tLNPs containing mCherry mRNA were incubated with placental BeWo b30 trophoblasts at a dose of 150 ng of mRNA per 150,000 cells. After ( a ) 1 h, ( b ) 4 h, and ( c ) 24 h, cellular accumulation was quantified. After ( d ) 24 h, mCherry expression was also quantified. Normalized DiR and mCherry MFI was calculated by normalizing to cells treated with NT LNPs. ( e–m ) NT LNPs and tLNPs containing FLuc mRNA were administered intravenously via retroorbital injection into pregnant and nonpregnant mice at a dose of 12 µg mRNA per mouse. After 6 h, mice were euthanized, and major organs were dissected. For pregnant mice, luminescence imaging of ( e ) livers and spleens and ( f ) placentas and fetuses were performed via an in vivo imaging system (IVIS). Luminescence from ( e-f ) was quantified via region of interest (ROI) analysis to obtain luminescence flux in the ( g ) liver, ( h ) spleen, ( i ) placentas, and ( j ) fetuses of pregnant mice. For nonpregnant mice, luminescence imaging of ( k ) livers and spleens was performed. Luminescence from ( k ) was quantified via region of interest (ROI) analysis to obtain luminescence flux in the ( m ) liver and ( m ) spleen of nonpregnant mice. Signal is reported mean ± SD from n = 3 biological replicates for ( a–d ) and n = 4 biological replicates for ( e–m). One-way ANOVA with post hoc Student’s t-tests using the Holm–Sídak correction for multiple comparisons was used to compare fluorescence in for ( a–d ) and luminescence in ( g–h, l–m ) across treatment groups. Nested one-way ANOVA with post hoc Student’s t-tests using the Holm–Sídak correction for multiple comparisons was used to compare luminescence in ( i–j ) across treatment groups. ( n–q ) Spearman correlations for ( n ) placental, ( o ) pregnant hepatic, and ( p ) nonpregnant hepatic luminescence values using the physicochemical parameters from traditional characterization methods, static SAXS analyses, and AF4-UV-DLS-MALS-SAXS analyses. ( q ) Heatmap representing the entire dataset. For Spearman correlation graphs, dotted lines represent r = –0.6 and 0.6.

Article Snippet: 100 μL of LNP at a particle concentration of ∼10 particles were injected and eluted isocratically from an Eclipse TM NEON field flow fractionation (AF4) instrument with dilution control module using a variable-height short channel with a 275 μm spacer and a 10 kDa regenerated cellulose membrane (Wyatt Technology), connected to a 1260 Infinity II HPLC system with a G711A quaternary pump and a G7167A multisampler (Agilent Technologies).

Techniques: Labeling, Incubation, Expressing, Injection, Imaging, In Vivo Imaging, Fluorescence

Journal: bioRxiv

Article Title: Resolving heterogeneity of targeted lipid nanoparticles through solution-based biophysical analyses

doi: 10.64898/2026.03.31.715590

Figure Lengend Snippet: ( a-b ) NT LNPs and tLNPs containing FLuc mRNA were administered intravenously via retroorbital injection into pregnant and nonpregnant mice at a dose of 12 µg mRNA per mouse. After 6 h, mice were euthanized, and serum was collected. Serum levels of C3a, TNF, IFN-γ, IL-6, ALT, and AST were quantified in ( a ) pregnant and ( b ) nonpregnant mice via ELISA. Measurements are reported mean ± SD from n = 3–4 biological replicates. One-way ANOVA with post hoc Student’s t-tests using the Holm–Sídak correction for multiple comparisons was used to compare cytokine levels across treatment groups. ( c–e ) Spearman correlations for ( c ) TNF, ( d ) IFN-γ, and ( e ) IL-6 serum levels in pregnant (top) and nonpregnant (bottom) mice using the physicochemical parameters from traditional characterization methods, static SAXS analyses, and AF4-UV-DLS-MALS-SAXS analyses. ( f ) Heatmap representing the entire dataset. For Spearman correlation graphs, dotted lines represent r = –0.6 and 0.6.

Article Snippet: 100 μL of LNP at a particle concentration of ∼10 particles were injected and eluted isocratically from an Eclipse TM NEON field flow fractionation (AF4) instrument with dilution control module using a variable-height short channel with a 275 μm spacer and a 10 kDa regenerated cellulose membrane (Wyatt Technology), connected to a 1260 Infinity II HPLC system with a G711A quaternary pump and a G7167A multisampler (Agilent Technologies).

Techniques: Injection, Enzyme-linked Immunosorbent Assay

Journal: bioRxiv

Article Title: Resolving heterogeneity of targeted lipid nanoparticles through solution-based biophysical analyses

doi: 10.64898/2026.03.31.715590

Figure Lengend Snippet: ( a ) SAXS profiles of NT LNPs and tLNPs, showing the characteristic Bragg peak feature associated with internal lipid–RNA organization. ( b ) Fitting of the Bragg peak feature using a multiple-Lorentz model where red is the first-order Bragg peak fit, green represents higher-order particle disorder and blue is the cumulative fit of the two features. (c) In-line AF4-UV-Vis contour maps (200–300 nm) depicting wavelength-resolved absorbance of LNPs during separation. The 260 nm absorbance signal is characteristic of encapsulated RNA, enabling identification of RNA-containing LNP populations across the AF4 elution profile. Overlapping spectral features indicate the presence of multiple co-eluting populations with distinct compositional profiles, motivating subsequent chemometric deconvolution. ( d ) Absorbance at 260 nm (RNA-associated signal) and 280 nm (protein-associated signal) from ( c ) plotted against the corresponding 260:280 ratio for each LNP formulation. Deviations in the 260:280 ratio across elution time indicate heterogeneity in RNA and protein content, suggesting the presence of compositionally distinct subpopulations that cannot be resolved by bulk measurements alone. These data were further subjected to chemometric analysis (see Supplemental Figure 6). The determined ( e ) R h profiles derived from in-line DLS and ( f ) molar mass profiles derived from MALS analysis for NT LNPs (beige) and tLNPs (colors) overlaid with UV fractograms from in-line AF4 separation. ( g ) Peak 260:280 ratios from ( d ), shown for comparison across LNP groups. ( h ) In-line DLS R h and MALS-derived ( i ) mass, ( j ) radius of gyration, and ( k ) polydispersity plotted for comparison across LNP groups. Measurements are reported mean ± standard error for ( h–k).

Article Snippet: The AF4 system was controlled by VISION 3.2.0 software (Wyatt Technology).

Techniques: Formulation, Derivative Assay, Comparison

Journal: bioRxiv

Article Title: Resolving heterogeneity of targeted lipid nanoparticles through solution-based biophysical analyses

doi: 10.64898/2026.03.31.715590

Figure Lengend Snippet: ( a ) SAXS elution profiles of NT LNPs and tLNPs. Dotted boxes indicate the main LNP elution region, expanded in ( b ). Evolving factor analysis (EFA) ( b ) applied to AF4-SAXS data, highlighting frames corresponding to discrete evolving species (C1–C3) within each formulation. The resolved components correspond to distinct LNP subpopulations that differ in size and internal organization and a third component (when present) reflecting a minor population of highly heterogeneous or higher-mass particles. The temporal evolution of EFA-selected frames across the elution peak indicates that these components arise from partially overlapping populations that co-elute but differ in hydrodynamic size and composition. ( c ) Singular value decomposition (SVD) of SAXS datasets, indicating the number of independent signals present in each formulation. The presence of multiple significant singular values indicates that the SAXS signal cannot be described by a single particle population, necessitating decomposition into independent scattering components. (d) Volatility of ratio (V r ) analysis used to assess the statistical uniqueness of deconvoluted component subpopulations relative to the ensemble-averaged population. ( e-h ) Regularized alternating least squares (REGALS) deconvolutions of SAXS data yields component-resolved scattering profiles. Shown are the ( e ) ensemble-averaged SAXS profile over the elution peak, and the REGALS-derived scattering profiles for the ( f ) C1, ( g ) C2, and ( h ) C3 subpopulations. These structural components are consistent with compositionally distinct populations inferred from UV-based chemometric analysis, linking RNA-associated heterogeneity to differences in particle size and morphology. Together, these analyses demonstrate that the ensemble SAXS signal arises from multiple, structurally distinct tLNP subpopulations that can be resolved and interpreted through chemometric decomposition.

Article Snippet: The AF4 system was controlled by VISION 3.2.0 software (Wyatt Technology).

Techniques: Formulation, Single Particle, Derivative Assay

Journal: bioRxiv

Article Title: Resolving heterogeneity of targeted lipid nanoparticles through solution-based biophysical analyses

doi: 10.64898/2026.03.31.715590

Figure Lengend Snippet: ( a ) Guinier analyses with corresponding residuals for SVD-resolved C1 and C2 components of NT LNPs and tLNPs, with the exception of F(ab’) 2 tLNPs, where only C3 is shown. White regions indicate components for which Guinier analysis failed due to large size (q min R g > 1.3). ( b ) P(r) analyses normalized by I(0) for the average profile and individual components (C1–C3) for NT LNPs and tLNPs. ( c ) Radius of gyration (R g ) and ( d ) maximum dimension (D max ) of the average profile and individual components derived from GNOM analysis for NT LNPs and tLNPs. Blank regions denote populations where GNOM analysis was invalid (q min D max > 4). ( e ) LNP shape factor calculated as D max / R g , where values of ∼2.58 and ∼3.0 correspond to spherical and prolate ellipsoid geometries, respectively. DENSS ab initio electron density reconstructions from the AF4-UV-DLS-MALS-SAXS profiles for ( f ) NT LNPs, ( g ) nanobody tLNPs, ( h ) DAPRin tLNPs, ( i ), F(ab’) 2 tLNPs, and ( j ) antibody tLNPs.

Article Snippet: The AF4 system was controlled by VISION 3.2.0 software (Wyatt Technology).

Techniques: Derivative Assay

Journal: bioRxiv

Article Title: Resolving heterogeneity of targeted lipid nanoparticles through solution-based biophysical analyses

doi: 10.64898/2026.03.31.715590

Figure Lengend Snippet: Targeted mRNA delivery to the placenta is driven by tLNP structural subspecies. ( a-d ) DiR-labeled NT LNPs and tLNPs containing mCherry mRNA were incubated with placental BeWo b30 trophoblasts at a dose of 150 ng of mRNA per 150,000 cells. After ( a ) 1 h, ( b ) 4 h, and ( c ) 24 h, cellular accumulation was quantified. After ( d ) 24 h, mCherry expression was also quantified. Normalized DiR and mCherry MFI was calculated by normalizing to cells treated with NT LNPs. ( e–m ) NT LNPs and tLNPs containing FLuc mRNA were administered intravenously via retroorbital injection into pregnant and nonpregnant mice at a dose of 12 µg mRNA per mouse. After 6 h, mice were euthanized, and major organs were dissected. For pregnant mice, luminescence imaging of ( e ) livers and spleens and ( f ) placentas and fetuses were performed via an in vivo imaging system (IVIS). Luminescence from ( e-f ) was quantified via region of interest (ROI) analysis to obtain luminescence flux in the ( g ) liver, ( h ) spleen, ( i ) placentas, and ( j ) fetuses of pregnant mice. For nonpregnant mice, luminescence imaging of ( k ) livers and spleens was performed. Luminescence from ( k ) was quantified via region of interest (ROI) analysis to obtain luminescence flux in the ( m ) liver and ( m ) spleen of nonpregnant mice. Signal is reported mean ± SD from n = 3 biological replicates for ( a–d ) and n = 4 biological replicates for ( e–m). One-way ANOVA with post hoc Student’s t-tests using the Holm–Sídak correction for multiple comparisons was used to compare fluorescence in for ( a–d ) and luminescence in ( g–h, l–m ) across treatment groups. Nested one-way ANOVA with post hoc Student’s t-tests using the Holm–Sídak correction for multiple comparisons was used to compare luminescence in ( i–j ) across treatment groups. ( n–q ) Spearman correlations for ( n ) placental, ( o ) pregnant hepatic, and ( p ) nonpregnant hepatic luminescence values using the physicochemical parameters from traditional characterization methods, static SAXS analyses, and AF4-UV-DLS-MALS-SAXS analyses. ( q ) Heatmap representing the entire dataset. For Spearman correlation graphs, dotted lines represent r = –0.6 and 0.6.

Article Snippet: The AF4 system was controlled by VISION 3.2.0 software (Wyatt Technology).

Techniques: Labeling, Incubation, Expressing, Injection, Imaging, In Vivo Imaging, Fluorescence

Journal: bioRxiv

Article Title: Resolving heterogeneity of targeted lipid nanoparticles through solution-based biophysical analyses

doi: 10.64898/2026.03.31.715590

Figure Lengend Snippet: ( a-b ) NT LNPs and tLNPs containing FLuc mRNA were administered intravenously via retroorbital injection into pregnant and nonpregnant mice at a dose of 12 µg mRNA per mouse. After 6 h, mice were euthanized, and serum was collected. Serum levels of C3a, TNF, IFN-γ, IL-6, ALT, and AST were quantified in ( a ) pregnant and ( b ) nonpregnant mice via ELISA. Measurements are reported mean ± SD from n = 3–4 biological replicates. One-way ANOVA with post hoc Student’s t-tests using the Holm–Sídak correction for multiple comparisons was used to compare cytokine levels across treatment groups. ( c–e ) Spearman correlations for ( c ) TNF, ( d ) IFN-γ, and ( e ) IL-6 serum levels in pregnant (top) and nonpregnant (bottom) mice using the physicochemical parameters from traditional characterization methods, static SAXS analyses, and AF4-UV-DLS-MALS-SAXS analyses. ( f ) Heatmap representing the entire dataset. For Spearman correlation graphs, dotted lines represent r = –0.6 and 0.6.

Article Snippet: The AF4 system was controlled by VISION 3.2.0 software (Wyatt Technology).

Techniques: Injection, Enzyme-linked Immunosorbent Assay