Surface Presentation of Hyaluronic Acid Modulates Nanoparticle-Cell Association

Abstract

Nanoparticle (NP) drug carriers have revolutionized medicine and increased patient quality of life. Clinically approved formulations typically succeed because of reduced off-target toxicity of the cargo. However, increasing carrier accumulation at disease sites through precise targeting remains one of the biggest challenges in the field. Novel multivalent ligand presentations and self-assembled constructs can enhance cell association, but an inability to draw direct comparisons across formulations has hindered progress. Furthermore, how nanoparticle structure influences function often is unclear. In this report, we leverage the well-characterized hyaluronic acid (HA)-CD44 binding pair to investigate how the surface architecture of modified NPs impacts their association with ovarian cancer cells that overexpress CD44. We functionalized anionic liposomes with 5 kDa HA by either covalent conjugation via surface coupling or electrostatic self-assembly using the layer-by-layer (LbL) adsorption method. Comparing these two methods, we observed a consistent enhancement of NP-cell association with the self-assembly LbL technique, particularly with higher molecular weight (≥10 kDa) HA. To further optimize association, we increased the surface-available HA. We synthesized a bottlebrush glycopolymer composed of a polynorbornene backbone and pendant 5 kDa HA and layered this macromolecule onto NPs. Flow cytometry revealed that the LbL HA bottlebrush NP outperformed the LbL linear display of HA. Cellular visualization by deconvolution optical microscopy corroborated results from all three constructs. Using exogenous HA to block NP-CD44 interactions, we found the LbL HA bottlebrush NP had a 4-fold higher binding avidity than the best-performing LbL linear HA NP. We further observed that decreasing the density of HA bottlebrush side chains to 75% had minimal impact on LbL NP stability or cell association, though we did see a reduction in binding avidity with this side-chain-modified NP. Our studies indicate that LbL surfaces are highly effective for multivalent displays, and the mode in which they present a targeting ligand can be optimized for NP cell targeting.

Figure 1.

Schematic illustration of the three distinct hyaluronic acid (HA) nanoparticle constructs. A) HA is coupled to the surface of liposomes via copper-catalyzed azide-alkyne click chemistry or B) self-assembled via the layer-by-layer technique with either linear HA or a synthetic bottlebrush HA. C) Chemical structures of the reducing end of each HA derivative.

Figure 2.

Cellular association of two distinct HA NP surface presentations. A) Size and zeta potential measurements of the covalently coupled and noncovalent, linear layered 5 kDa HA NPs. Flow cytometry was used to evaluate the association of the layered, self-assembled HA NPs with B) COV362 or OVCAR8 ovarian cancer cells relative to the median fluorescence intensity (MFI) of the initial, bare liposomes. Error bars represent standard deviation for N = 3 technical repeat measurements in A, and N = 5 technical replicates for B. Fold increase of liposome MFI is defined as the MFI of each formulation normalized to the liposome MFI at a given timepoint. Kruskal-Wallis non-parametric tests with a false-discovery rate of 0.05 were used, in which * q < 0.05.

Figure 3.

Comparison of how terminal layer HA molecular weight impacts LbL NP assembly and NP-ovarian cancer cell association. A) LbL NP zeta potential and z-average diameters as a function of wt. eq. for each given molecular weight. Cell association is normalized by fold-increase over bare liposomes in both B) COV362 and OVCAR8 cell lines. Error bars represent standard deviations for N = 3 technical measurements for A, and N = 5 technical measurements for B. N.D. stands for no data. Kruskal-Wallis non-parametric tests with a false-discovery rate of 0.05 were used, in which * q < 0.05, ** q < 0.01, and *** q < 0.001.

Figure 4.

Micrographs obtained from live cell imaging of COV362 cells treated with the three different HA NP architectures. A) First 30 min following NP addition is shown. NPs are pseudo-colored magenta, and Cell Tracker Blue signal, used to visualize the nucleus, is pseudo-colored gray. B) Representative images at 60 min post-NP addition with LysoTracker, used to visualize endo-lysosomes pseudo-colored cyan. Co-localization between the NP and LysoTracker signal is shown in white. Scale bars = 10 μm.

Figure 5.

Synthesis, particle assembly, and cellular association of bottlebrush HA-layered NPs. A) Synthetic scheme for bottlebrush HA-substituted polymers with PEG side-chain co-addition, with a schematic illustration of the three different bottlebrush compositions as outer layered NPs. B) Bottlebrush LbL NP zeta potential and z-average diameters as a function of wt. eq. at each given bottlebrush formulation. Arrows indicate the ratio at which the LbL NPs aggregated, as defined by > 100 nm increase in Z-average diameter as compared to the parent nanoparticle. Error bars represent standard deviations for N = 3 technical measurements. C) Cell association in median fluorescent intensity normalized to untreated cells for both COV362 and OVCAR8 cell lines. Error bars represent standard deviations for N = 5 technical replicates. Kruskal-Wallis non-parametric tests with a false-discovery rate of 0.05 were used, in which * q < 0.05 and *** q < 0.001. D) Reduction in OVCAR8 association as a function of exogenous 100 kDa HA added. Error bars represent standard deviations for N = 4 technical measurements.