Layer-by-Layer Polymer Functionalization Improves Nanoparticle Penetration and Glioblastoma Targeting in the Brain

Abstract

Glioblastoma is characterized by diffuse infiltration into surrounding healthy brain tissues, which makes it challenging to treat. Complete surgical resection is often impossible, and systemically delivered drugs cannot achieve adequate tumor exposure to prevent local recurrence. Convection-enhanced delivery (CED) offers a method for administering therapeutics directly into brain tumor tissue, but its impact has been limited by rapid clearance and off-target cellular uptake. Nanoparticle (NP) encapsulation presents a promising strategy for extending the retention time of locally delivered therapies while specifically targeting glioblastoma cells. However, the brain's extracellular structure poses challenges for NP distribution due to its narrow, tortuous pores and a harsh ionic environment. In this study, we investigated the impact of NP surface chemistry using layer-by-layer (LbL) assembly to design drug carriers for broad spatial distribution in brain tissue and specific glioblastoma cell targeting. We found that poly-l-glutamate and hyaluronate were effective surface chemistries for targeting glioblastoma cells in vitro. Coadsorbing either polymer with a small fraction of PEGylated polyelectrolytes improved the colloidal stability without sacrificing cancer cell selectivity. Following CED in vivo, gadolinium-functionalized LbL NPs enabled MRI visualization and exhibited a distribution volume up to three times larger than liposomes and doubled the retention half-time up to 13.5 days. Flow cytometric analysis of CED-treated murine orthotopic brain tumors indicated greater cancer cell uptake and reduced healthy cell uptake for LbL NPs compared to nonfunctionalized liposomes. The distinct cellular outcomes for different colayered LbL NPs provide opportunities to tailor this modular delivery system for various therapeutic applications.

Keywords: MRI imaging; cellular targeting; convection-enhanced delivery; glioblastoma; layer-by-layer; nanoparticles; tumor penetration.

Figure 1:. Co-layered LbL NPs balance colloidal stability with cellular associations.

A) Size, polydispersity, and zeta potential of LbL NPs with outer layers of poly-L-glutamate (PLE), sodium hyaluronate (HA), and poly-L-glutamate-b-PEG (PLE-PEG) measured using dynamic light scattering. All NPs display similar size and zeta potential when measured in DI water, but only PLE-PEG NP maintain their size in artificial cerebrospinal fluid (aCSF) after overnight incubation at 37 °C. B) Schematic showing the approach to balancing colloidal stability with cellular uptake by combining outer layer polyanions with PLE-PEG. C, D) NP characterization using dynamic light scattering of PLE (C) and HA (D) NPs with different % (w/w) of PLE-PEG present during NP layering. The presence of 40% (w/w) PLE-PEG in the layering buffer is sufficient to form NPs that maintain colloidal stability in aCSF. E) Flow cytometry histograms showing the decreasing association of fluorescently-tagged NPs with GBM22 glioblastoma cells with increasing weight fractions of PLE-PEG in the outer layer. Histograms are pooled results from 3 technical replicates with * p<0.05 by one-way ANOVA. Error bars in panels A, C, and D represent the standard deviation from the arithmetic mean of three technical replicates.

Figure 2:. Co-layered LbL NPs maintain selective interactions with glioblastoma cells, leading to internalization.

A) Schematic of NPs used in NP-cell association studies. B) Column-normalized heat map of median fluorescence intensity values from flow cytometry after 4 and 24 hours of incubation. Hierarchical clustering reveals LbL NPs have specificity for glioblastoma cell lines over healthy cells in vitro. The lower box plot shows the overall NP-associated fluorescence for each formulation pooled over all cancerous and all healthy cell lines. * p<0.05 by Mann-Whitney U Test for non-parametric data with a false discovery rate of 0.05. C, D) Relative flow cytometry association of NPs with and without exogenous PLE (C) or HA (D) to block interactions with cell membrane receptors. * p<0.05, ** p<0.01 by one-way ANOVA with Tukey post-hoc multiple comparisons. E) GBM22 cells were incubated with NPs for 24 hours and then fixed and analyzed by confocal microscopy. Relative to the liposome core, all LbL NPs showed substantially higher NP signal. PLE NPs were present mainly on the outer cell membrane, while HA and both co-layered formulations were internalized into the cytoplasm.

Figure 3:. Co-layered LbL NPs penetrate 3D glioblastoma spheroids.

A) Schematic of flow cytometry assay used to measure the extent of NP uptake and penetration in spheroids. Nanoparticles are represented by green shading. B) After incubation with NPs, spheroids were dissociated into single cells and analyzed for percent NP-positive cells and their overall NP-associated fluorescence. Co-layered LbL NPs increase the fraction of NP+ cells compared to PLE and HA LbL NPs while increasing the NP-associated fluorescence compared to PLE-PEG NPs. C) BT145 spheroids were incubated with NPs for 24 hours, fixed, and cryosectioned into 10 μm sections for confocal microscopy. Representative cross-section shows that co-layered PLE NPs have more uniform penetration throughout spheroids than PLE NPs. D) Quantification of the radial average intensity as a function of distance from the spheroid center. E, F) Representative cross sections of spheroids incubated with either HA or HA/PLE-PEG NPs reveal no significant change in penetration. Shaded areas in panels D and E represent the standard deviation from the arithmetic mean of three technical replicates.

Figure 4:. Co-layered LbL functionalization of Gd-functionalized liposomes increases distribution and retention following CED in healthy mice.

A) Top: Schematic showing the incorporation of 5 mol% 18:0 PE-DTPA(Gd) into the liposomal bilayer. Bottom: In-vitro MRI images of agarose phantoms with different concentrations of NPs, showing concentration-dependent T1 contrast. B) Top: T1-weighted coronal MRI images of healthy mice 24 hours after infusion with NPs. Bottom: Fluorescent micrographs of cryosectioned brains at approximately the same coronal plane as MRI images. C) Quantification of the segmented 3-D volume of NP T1 MRI contrast, expressed as a ratio of the volume of distribution (V_D) by the volume of infusion (V_I). Statistical analysis of n = 5 biological replicates was performed using one-way ANOVA with Welch’s correction. D) T1-weighted coronal MRI scans of healthy mice showing the gradual loss of Gd-labelled albumin or NPs over 21 days. E) Quantification of the volume of distribution expressed as a ratio of V_D/V_I over 21 days. An exponential fit of V_D/V_I as a function of time for each group indicates that both LbL formulations have a longer lifetime than liposomes. A two-way mixed-effects model indicated a significant effect of time (p<0.001) and formulation on the volume of NP distribution (p<0.001). F) Haematoxylin and eosin (H&E) staining of brains 21 days after administration of i) 5% dextrose, ii) liposomes, iii) co-PLE NPs, and iv) co-HA NPs. All NP groups showed trauma and focal necrosis at the needle tip, with infiltration of macrophages around the site of injection. Error bars in panels B, D, and E represent the standard deviation from the arithmetic mean of n = 6 biological replicates.

Figure 5:. Co-layered LbL functionalization increases the volume of distribution and glioblastoma targeting following CED in tumor-bearing mice.

A) Top row: T2 weighted coronal MRI images of mice 24 hours before CED. Regions of T2 hyperintensity outlined in red circles represent the tumor bulk. Middle row: T1-weighted coronal MRI images of mice 24 hours after CED of NPs. Bottom row: Fluorescent micrographs of cryosectioned brains at approximately the same coronal plane as MRI images. B) Quantification of the segmented 3-D volume of NP T1 MRI contrast, expressed as a ratio of V_D/V_I. Error bars show the mean and standard deviation of n = 9 biological replicates. Statistical analysis was performed using one-way ANOVA with Welch’s correction. C) Schematic showing the preparation of mouse brain tissue for flow cytometric analysis of NP content. D) Representative flow cytometry histograms showing the NP-associated fluorescence in tumor cells. Percentages represent the portion of all tumor cells that were detected as NP-positive from pooled biological replicates. N = 6 for liposome and co-PLE NPs, n = 5 for co-HA NPs. E) Cellular composition of dissociated mouse brains from the different NP-treatment groups. F) Cellular composition of the fraction of cells that were first gated as NP+. Both LbL formulations show significant enrichment of the tumor cell population compared to the liposome group and overall population of live cells. Error bars show the mean and SEM for 6 biological replicates for liposomes and co-PLE and 5 replicates for co-HA. For each cell type, bars with a unique symbol are statistically significant from other NP groups (*p <0.05, two-way ANOVA with Tukey post hoc test). G, H) Immunofluorescence of cryosectioned brain slices, showing myeloid cells and microglia in white (iba1), human tumor cells in green (hVimentin), nuclei in blue (DAPI), and NPs in red for co-HA (G) and co-PLE NPs (H).