Nanocluster size and physical properties, especially interparticle spacing, are controlled by varying particle volume fractions during solvent evaporation. The changes in electrostatic, van der Waals (VDW), depletion, and steric interactions upon the concentration of the gold nanoparticles and polymer micelles govern the kinetic assembly of the nanoclusters, as well as their disassembly after polymer degradation. Without the presence of a polymeric stabilizer, irregular micron-sized aggregates of gold nanoparticles have been formed by varying the particle charge upon adjusting pH, for particles capped with lysine,22, 23
ligands . Alternatively, nanoclusters of metals maybe be formed by equilibrium self-assembly with polymer templates.26–28
However, the high polymer concentrations required for the equilibrium assembly leads to metal loadings and interparticle spacings that are not sufficient for strong NIR absorbance. Additionally, the templating agents are highly specialized. In contrast, our kinetically-controlled assembly platform requires only small concentrations of common copolymers as stabilizers and simple biocompatible capping ligands on gold, such as citrate and/or lysine.
Cluster growth is controlled through mediation of the interactions between ligand-capped gold particles with the biodegradable polymer, as shown in . Gold nanoparticles stabilized with citrate ligands were synthesized based on a previously published method.29
A solution of 1% lysine in pH 8.4 phosphate buffer (10 mM) solution was added to 1.2 mL of a 3.0 mg/mL colloidal gold solution to yield a final lysine concentration of 0.4 mg/mL and an average diameter of 4.1±0.8 nm ( and ). The dispersion was stirred for 12 hours.23
PLA(2K)-PEG(10K)-PLA(2K) (Sigma Aldrich Co., St. Louis, MO ) (60 mg) was added to the aqueous gold dispersion, yielding a final polymer concentration of 50 mg/mL. The dispersion was sonicated in a bath sonicator for 5 minutes, during which the dispersion changed from ruby red color to a darker red-purple color. Upon evaporation of ~80% of the solvent, the dispersion turned blue, indicating absorption in the red. Complete solvent evaporation over two hours produced a smooth blue film. Reconstitution of the film with deionized (DI) water to a concentration of ~0.3 mg/mL, yielded a dark blue dispersion. The fact that this dispersion consists of sub-100 nm clusters composed of primary gold nanoparticles is indicated by scanning electron (SEM) and transmission election microscopy (TEM) (). TEM images taken at various angles reveal closely-spaced primary gold nanoparticles throughout the porous cluster. The average hydrodynamic diameter measured by dynamic light scattering (DLS) was 83.0±4.6 nm (, ), in agreement with the TEM results. In the SEM image (), a polymer-rich shell a few nanometers thick is apparent on the exterior of the clusters, which potentially provides steric stabilization of the dispersion.
Electron microscopy characterization of gold nanoclusters
Characterization of gold primary particles and nanoclusters
Degradation of gold nanoclusters in solution
Thermogravimetric analysis indicated that the nanoclusters contained only 20 ± 5% organic material (polymer and ligands), 10–15% of which was polymer. These low polymer loadings allow close-spacing of the gold primary particles. Interparticle distances between constituent gold particles within the cluster were estimated to be 1.80 ± 0.6 nm based on the more discernible particles in the periphery of TEM images (Supporting Information Figure S1
). This spacing is consistent with the length scale of a lysine-lysine dipeptide in solution of 1.49 nm.22
The ability to simultaneously control cluster size and spacing of gold nanoparticles within the nanocluster was achieved through manipulation of particle volume fractions upon solvent evaporation, as well as control of the electrostatic, VDW, depletion, and steric interactions between the gold primary particles and the polymer (see Supporting Information
). This approach provides greater control over the nanocluster morphology, relative to nanoclusters formed during reduction of gold precursors without a polymer stabilizer.30, 31
Control of the depletion interactions, along with the propensity of hydrophilic PEG blocks to migrate to the exterior of the clusters at the interface with water, facilitates low polymer loadings within the clusters.
Optical extinction spectra changed markedly upon cluster formation as shown in . The initial dispersion of gold nanoparticles had a maximum absorbance at 520 nm, which is characteristic of isolated gold spheres. The blue nanocluster dispersion had a broad, relatively constant absorbance in the NIR region from 700 to 900 nm. The extinction coefficient at the maximum absorbance, ε703
, was 0.020 cm2
/µg of gold for a 56 µg/mL gold dispersion. Assuming that the gold nanoparticles are in a closest packed state (based on SEM and TEM images in ), the estimated nanocluster extinction cross section was ~10−14
(see Supporting Information
), comparable to the value for nanoshells,1, 8
The high NIR absorbance observed for nanoclusters may be attributed to a combination of the following factors: close interparticle spacing between constituent gold nanoparticles, non-uniform spatial distribution of the constituent particles within the nanocluster, roughness of the nanocluster boundary surface, and finally, any deviation in the overall aspect ratio of the entire nanocluster from that of a sphere. The close spacing between gold particles is well within the fraction of the particle diameter known to produce significant red-shift in the SPR.8, 32–37
TEM images () also show that short oligomers of primary particles and sub-cluster domains can be clearly observed, indicative of a non-uniform volume-packing distribution of the constituent particles. When combined with strong plasmonic coupling, this non-uniform distribution significantly enhances red-shift of the SPR.8, 34
The NIR absorbance per total particle mass is much higher compared to previous composite particles with smaller amounts of gold nanoparticles templated with liposomes, block copolymer micelles or DNA.26, 28, 38
Nanoclusters with strong NIR absorbance were also produced by clustering more negatively charged citrate-capped gold nanoparticles, which possess a zeta potential of −44.0±4.9 mV, compared to −30.1±2.4 mV for the lysine/citrate-capped nanoparticles (). The remainder of the paper will focus on the lysine/citrate-capped nanoclusters due to their slightly smaller sizes.
The initial stability and degradation of the nanoclusters were examined at pH 7.4 and 5, simulating normal cellular environments and the interior of cellular lysosomes, respectively.39
After storage for 1 week in pH 7.4 buffer, the DLS peak shifted modestly towards smaller sizes (). This limited degradation is consistent with the long half-life of PLA (MW=2K) of about 4 weeks at neutral pH. In contrast, upon incubation at pH 5, upon addition of 0.1N HCl, for one week, nearly complete nanocluster deaggregation was observed by TEM (, ). The mean particle size of the deaggregated particles was 4.3±0.1 nm (over 100 particles analyzed), comparable to the initial size of the ligand-capped gold particles of 4.1±0.8 nm (Supporting Information Figure S2
). Upon degradation of the polymer, a combination of steric repulsion due to the capping ligands and remaining polymer fragments, electrostatic repulsion due to the negative charge of the gold particles, and effective entropic forces are sufficient to completely redisperse the primary gold particles. The associated extinction spectra undergo a substantial shift towards the original spectrum of the colloidal gold spheres and the color of the dispersion changes back towards red (). The remaining discrepancy between these spectra is attributed to the presence of a small number of clusters (7%) still present in the dispersion. Deaggregation to constituent nanoparticles was also observed for the clusters produced using citrate-only capped nanoparticles ().
Nanocluster biodegradation was also assessed in a murine macrophage cell line, (J477A.1, American Type Culture Collection, Manassas, VA). Scattering spectra from hypespectral images of cells (), dark-field reflectance (DR) (, top row), and hyperspectral (HS) images (, bottom row) were acquired at 24, 96, and 168 hours time points after cells were treated with nanoclusters. High nanocluster uptake was evident in the DR images, where nanoclusters strongly scattered illumination light; overall scattering intensity decreased over time as macrophages divided and nanoclusters were distributed between daughter cells (, top row). A significant increase in the red-NIR scattering signal of the labeled cells was seen compared to unlabeled cells (compare , dark blue curve, and c), consistent with the high scattering efficiency of the nanoclusters in solution (, light blue curve). The relative intensity of the red-NIR scattering signal decreased after 96 hours and the scattering from labeled cells showed a marked blue shift to ~550 nm that is consistent with scattering from the constituent lysine/citrate-capped gold nanoparticles (Supporting Information Figure S3
). Hyperspectral images showed a gradual progression from very strong scattering in the 650–700 nm region at t=24 hours to a less intense scattering signal predominantly in the 500–550 nm region at t= 168 hours (, bottom row). The endogenous scattering for the control cells did not significantly change with time (). Note that scattering from the control cells is approximately six times weaker compared to the labeled cells. In addition, most of the pixels in the hyperspectral images of untreated cells do not exhibit any distinct scattering peaks that results in black appearance of the images in , bottom row.
Biodegradation of gold nanoclusters inside live cells
The biodegradation of nanoclusters inside live cells was further confirmed by TEM (). After 24 hours, large ~100 nm nanoclusters can be observed throughout the interior of cells (, 24 hours), whereas after 168 hours, cells contain only particles less than 5 nm in diameter (, 168 hours). These TEM results are in excellent agreement with optical measurements and with deaggregation results in solution, providing unambiguous proof of essentially complete biodegradation of the initial ~100 nm nanoclusters into sub-5 nm primary particles (Supporting Information Figure S2