Real-time visualization of clustering and intracellular transport of gold nanoparticles by correlative imaging

Mechanistic understanding of the endocytosis and intracellular trafficking of nanoparticles is essential for designing smart theranostic carriers. Physico-chemical properties, including size, clustering and surface chemistry of nanoparticles regulate their cellular uptake and transport. Significantly, even single nanoparticles could cluster intracellularly, yet their clustering state and subsequent trafficking are not well understood. Here, we used DNA-decorated gold (fPlas-gold) nanoparticles as a dually emissive fluorescent and plasmonic probe to examine their clustering states and intracellular transport. Evidence from correlative fluorescence and plasmonic imaging shows that endocytosis of fPlas-gold follows multiple pathways. In the early stages of endocytosis, fPlas-gold nanoparticles appear mostly as single particles and they cluster during the vesicular transport and maturation. The speed of encapsulated fPlas-gold transport was critically dependent on the size of clusters but not on the types of organelle such as endosomes and lysosomes. Our results provide key strategies for engineering theranostic nanocarriers for efficient health management.


wide-field DFM image containing approximately 100 spots of different colors. (b)
The ex-situ SEM image of the fPlas-gold particles recorded in a. The red rectangular area in a and b were zoomed in as shown in (c) and (e), respectively, in which the bright yellow spot in the red circle contained 8 fPlas-gold single particles as indicated in (d). Scale bar represents 2 μm. scattering spectra of intracellular fPlas-gold. Green (i), yellow (ii) and bright yellow (iii) dots in the DFM image represented single particles (n=1), small clusters (n=2-5) and large clusters (n>5), respectively. The scattering spectra of the three intracellular fPlas-gold nanoparticles were distinguishable, which shifted from 560 nm to 600 nm with the increase of cluster size. Scale bar represents 2 μm. fPlas-gold were localized in various organelles. (a) One single particle was located in cytosol; (b) two single particles were wrapped by a coated pit connected with cell surface by a neck, which was supposed to be a clathrin-coated vesicle; (c) a single particle was located in a coated vesicle with outer diameter of~100 nm, which was also supposed to be a clathrin-coated vesicle; (d) a single particle was wrapped by a sphere-shape structure with outer diameter of~70 nm without an electron-dense coat, which was supposed to be a caveolae; (e) a single particle located in a ruffle with diameter >1 μm, which was a macropinsome. Scale bar represents 200 nm.

Endocytic mechanisms
TEM studies provided direct evidence for the endocytic mechanisms. The majority of fPlas-gold was trapped in vesicles, with only few found in the cytosol (Supplementary Figure 13a). Further analysis of these vesicle structures revealed the formation of clathrin-coated pits (CCPs) that connected to the cellular surface via a neck (Supplementary Figure 13b) or an electron-dense clathrin-coated vesicle with an outer diameter around 100 nm (Supplementary Figure 13c); caveosomes with non-electron-dense spherical vesicles with an outer diameter less than 100 nm (Supplementary Figure 13d); and macropinosome ruffles with a diameter larger than 1 μm ( Supplementary Figure 13e), suggesting the occurrence of multiple types of internalization pathways including caveole-and clathrin-mediated endocytosis as well as macropinocytosis.

Categorization of fPlas-gold dots with different colors
The green, yellow and bright yellow colors of fPlas-gold in DFM images are the colors that we visualized under the microscope, which are further corroborated by collecting their scattering spectra (see supplementary Figure 7b). We then compared our naked-eye classification of randomly-selected 20 fPlas-gold clusters with the results based on scattering spectra, and found a good agreement (see supplementary Figure 7d).
We also performed finite-difference time-domain (FDTD) simulation and found that the simulated scattering spectra were generally consistent with the measured ones (Supplementary Figure 7c).
Correlative imaging with SEM and DFM established the dependence of the color change on the aggregation states of fPlas-gold. A single particle of fPlas-gold exhibited green color in DFM images, whereas the color of clustered fPlas-gold gradually turned to yellow along with the increased number (n) of particles (from n=1 to n=10) (Supplementary Figure 7a). Yellow spots in DFM images are small clusters containing 2-5 single particles under SEM imaging and bright yellow spots are large clusters containing more than 5 particles.
Next, we performed a quantitative analysis on the correlation between the colors of fPlas-gold in DFM images with their aggregation states. A wide-field DFM image containing approximately 100 particles was recorded (Supplementary Figure 6a). This image was subsequently used as a pattern recognition template during the SEM analysis to locate and correlate particles of different aggregation states (Supplementary Figure 6b). This study confirmed the robustness of color classification.

Preparation of fPlas-gold
Citrate-stabilized 50nm AuNPs were synthesized according to previous literature [1] and characterized using UV-Vis and TEM.
The synthesized aqueous AuNPs were mixed with thiol modified oligonucleotide S1 with a final AuNPs:S1 molar ratio of 1:3000 in milliQ water and incubated at room temperature overnight to form ssDNA-AuNPs. Then PB concentration was adjusted to 10 mM and NaCl concentration was increased to 50 mM. The resultant solution was sonicated for~10 s followed by a 30 min incubation at room temperature, and this salting process was repeated till NaCl concentration reached 300 mM. The resultant solution was incubated overnight at room temperature and then centrifuged. The supernatant was removed and the precipitates were resuspended in 10 mM PB (pH 7.4). This washing process was repeated for three times and then the ssDNA-AuNPs were resuspended in 1×PBS (pH 7.4) for further uses.
To a 1nM ssDNA-AuNPs solution in 100 mM PBS (pH 7.4), the complementary oligonucleotide sequence S2-CY3 in 100 mM PBS was added with a final AuNPs:S2-CY3 molar ratio of 1:3000. This mixture was incubated at 37 o C for more than 30 min to yield Cy3 tagged dsDNA-AuNPs (i.e. fPlas-gold) then centrifuged and washed following procedures described above. The final product was resuspended in 1×PBS (pH 7.4) for further uses.

Internalization of fPlas-gold
HeLa cells were cultured in 60 mm Petri dishes overnight, thenthe culturing supernatant was removed and the cells was washed with 1×PBS buffer (pH 7.4) twice. The cells were subsequently incubated for different time in fresh DMEM medium with 0.1 nM fPlas-gold. The uptake process was stopped by washing cells with PBS buffer twice, and the cells were cultured in fresh DMEM again for DFM imaging. To investigate effect of temperature and pharmacological inhibitors, cells were incubated in four different conditions for 30 min (a, 4 o C; b, 10 μg ml -1 chlorpromazine; c, 2.5 mM MβCD; d, fresh DMEM medium), respectively, then exposed to 0.1 nM fPlas-gold for 1 hour before DFM imaging.
For ICP-AES measurements, cells with internalized fPlas-gold were washed with 1×PBS three times and then trypsizined and centrifuged at 5000 rpm for 3 min. Cell pellets were digested with aqua regia (HCl:HNO3=3:1) at room temperature overnight and the content of Au-197 of resultant solution was measured with an Optima 8000 ICP-OES spectrometer (PerkinElmer).
For TEM imaging, cells with internalized fPlas-gold were washed, trypsizined and centrifuged. Then cells were resuspended and fixed with 2.5% glutaraldehyde in 1 ×PBS buffer (pH 7.4) and stained with 1% OsO4 at 4 ºC. After gradual dehydration with ethanol and acetone, cell pellets were embedded in Epon 812 resins (Electron Microscopy Science) and sliced to pieces with a thickness of 70nm then stained with uranyl acetate. Images of cell slices were taken with a FEI Tecnei G2-205 Twin transmission electron microscope using a beam voltage of 80 kV.

Co-localization of fPlas-gold with endosomes and lysosomes
For endosome staining, HeLa cells were cultured in 35 mm Petri dishes overnight. Lipo solution was prepared by diluting lipofectamin®3000 in opti-MEM Medium; master solution was prepared by diluting DNA (Rab5-GFP for Early Endosomes and Rab7 for Late Endosomes) in opti-MEM Medium and add P3000® Reagent to master solution. Then, we mixed lipo solution and master solution well, incubated them for 5 minute, and then added complex solution in fresh MEM Medium.The culturing supernatant of Hela cells was removed and the cells was washed with 1×PBS buffer (pH 7.4) twice. Then the cells were incubated in mixed MEM Medium for above 24 hours. HeLa cells were cultured in 35 mm Petri dishes overnight. Then the culturing supernatant was removed and the cells was washed with 1×PBS buffer (pH 7.4) twice. The cells were subsequently incubated for different time in fresh MEM medium with 0.1 nM fPlas-gold. Finally, to observe co-localization of fPlas-gold with endosomes by using confocal and DFM, cells were washed with 1×PBS buffer and fixed in 4% (wt/vol) paraformaldehyde and 4% (wt/vol) sucrose in 1×PBS buffer at room temperature for 20 min. To observe dynamic co-localization of fPlas-gold with endosomes, cells were washed with 1 ×PBS buffer (pH 7.4) and cultured in fresh MEM for confocal imaging.
For lysosome staining, cells were first incubated with 0.1 nM fPlas-gold for different time, then washed and incubated in fresh medium containing 50 μM probes (LysoTracker® Green DND-26, Invitrogen) for 5 min. Finally cells were washed with 1 ×PBS buffer (pH 7.4) and cultured in freshMEM for confocal imaging.

Single-particle tracking of fPlas-gold using fluorescence imaging
Cells were incubated in fresh MEM medium containing 0.1 nM fPlas-gold for 30 min, then washed and visualized using TIRF microscope. To investigate effect of pharmacological inhibitors on particle movement, HeLa cells were incubate with 0.1 nM fPlas-gold for 6 h, and then incubated with media containing 60 µM nocodazole and 20 µM cytochalasin B for 30 min for confocal imaging, respectively. The drugs were maintained in the cell culture throughout the experiments.
To observe the movement of fPlas-gold along microtubules, HeLa cells were first incubated with staining solution of tubulin (CellLight® Tubulin-GFP, BacMam 2.0, Life Technologies, 15 µL in 200 µL medium) overnight, then incubated with 0.1 nM fPlas-gold for 2 h and washed before confocal imaging.

Single-particle tracking of fPlas-gold using DFM imaging
Cells were incubated in fresh MEM medium containing 0.1 nM fPlas-gold for 1 h, washed and then visualized using DFM.

Image analysis
Fluorescence images were first analyzed using ImageJ software (US National Institutes of Health). To quantify the co-localization efficiency of two fluorescent signals, tMr values(the thresholded Mander's coefficients) indicating the percentage of Cy3 signals co-localized with green signals in merged images were calculated. Values represent mean ± SE based on analysis of randomly selected 10 cells from three independent experiments. For single particle tracking, the trajectories of Cy3 signals were built by pairing spots in each frame using single-particles tracking plug-in of ImageJ.
Speed calculation and mean square displacement (MSD) analysis were performed using user-written program with MATLAB (The MathWorks) software. MSD data of each particle was calculated following formula shown below [2]: , in which Δt is the time interval between two successive recorded images, N is the total number of frames, n is a positive integer that determines the time increase, and r is displacement. The upward and downward relationships of the MSD over time plots indicate the movement is the manner of directed motion and anomalous diffusion, respectively.