Real-time and label-free monitoring of nanoparticle cellular uptake using capacitance-based assays

Nanoparticles have shown great potential as vehicles for the delivery of drugs, nucleic acids, and therapeutic proteins; an efficient, high-throughput screening method to analyze nanoparticle interaction with the cytomembrane would substantially improve the efficiency and accuracy of the delivery. Here, we developed a capacitance sensor array that monitored the capacitance values of nanoparticle-treated cells in a real-time manner, without the need for labeling. Upon cellular uptake of the nanoparticles, a capacitance peak was observed at a low frequency (e.g., 100 Hz) as a function of time based on zeta potential changes. In the high frequency region (e.g., 15–20 kHz), the rate of decreasing capacitance slowed as a function of time compared to the cell growth control group, due to increased cytoplasm resistance and decreased membrane capacitance and resistance. The information provided by our capacitance sensor array will be a powerful tool for scientists designing nanoparticles for specific purposes.


Frequency-dependent capacitance values in cell growth control groups and amine-modified
PNPs-treated HUVEC cells with fitting simulations.
When the AC electric field was applied at a low frequency, the cell membrane effectively insulated the cytoplasm, and the electric ion current flowed only around the cell.
This was due to the ion current mostly passing through the extracellular spaces. The electrical properties, then, were largely affected by the extracellular volume fraction ( Supplementary   Fig. 4a). Therefore, at low frequencies, the ion current flow was unaffected by whether the cells uptake the NPs or not ( Supplementary Fig. 4b). At high frequencies, however, the cell membrane was more conductive with negligible impedance, so the ion current penetrated the cell membrane and flowed within the intracellular spaces ( Supplementary Fig. 4c) 1 . Here, the changes in the composition/volume of the cell cytoplasm and nucleus affected impedance 2 .
Thus, the differences in capacitance values at the high-frequency region between the cellular growth control group and cells that uptake NPs could be distinguished because the content of each cell was different. Compared with the cellular growth control group, the cells with NPs uptake had a higher resistive particle and vesicle content, which influenced the ion current flow within the intracellular space ( Supplementary Fig. 4d).
To prove this, we measured capacitance as a function of frequency (f) while the HUVECs were growing, or after treatment with polystyrene nanoparticles (PNPs). To induce cellular uptake, we seeded 50,000 HUVECs per well; then 10 9 particles of amine-modified PNPs (positively charged, 200 nm, Supplementary Fig. 1; NH 3 + -PNPs) were added to each well. (The NH 3 + -PNPs are NPs whose uptake pathways are known to be via endocytic 3 .) The relationship C  f -a was observed for all cellular states, with two different exponents at low frequencies (a=α; 0.11 kHz) and high frequencies (a=β; 1520 kHz) (C: capacitance and f: frequency). In the log-log plot, a (α or β) value indicates the slope in the relationship logC  alogf. As the cells grew (before the NPs treatment), the capacitance was fitted to the relationship C  f - with ||0.167 at t=24 hours and ||0.175 at t=48 hours, with a difference of +0.008 (Supplementary Fig. 3b). As the frequency increased, the capacitance was fitted to the relationship C  f -β with |β|0.566 at t=24 hours and |β|0.640 at t=48 hours, with a difference of +0.08 ( Supplementary Fig. 3c).
In contrast, with cellular uptake of NPs, the capacitance conformed to the relationship C  f -a with ||0.173 at t=24 hours (immediately before NP treatment) and ||0.183 at t=48 hours, the difference being +0.013 at low frequencies ( Supplementary Fig. 3b); and at high frequencies with |β|0.567 at t=24 hours and |β|0.468 at t=48 hours, the difference being 0.099 (Supplementary Fig. 3c). A comparison made at 24 and 48 hours for the cellular growth control group showed both the difference in the || and |β| values to be positive; however, before and after cellular uptake of the NPs, the difference in the || values was positive but negative for the |β| values. In other words, at a high frequency, the capacitance decreased at a slower rate after cellular uptake of the NPs, as opposed to the cellular growth control group. This is because the cell expresses extremely high dielectric permittivity at low frequencies but gradually decreases at higher frequencies. To better understand the decreased |β| values in the high frequency region, we fitted our data using a theoretical electric circuit 4 ( Supplementary Fig. 3a); the fitted values are represented by the symbols in Supplementary   Figs. 3b and c.
The fitted capacitance values (symbols) fit relatively well to the experimental data (curved line) for the cellular growth control group (Set 1; cell growth at 24 and 48 hours) and the NPs-treated group (Set 2; cell growth at 24 and 24 hours after internalization of 10 9 particles). The fitted values related to the cell are intracellular bulk resistance (R cyto ), cell membrane capacitance (C m ) and reactive charge transfer resistance on the cell membrane (R m ), which are colored red in the electric circuit in Supplementary Fig. 3a. The parameters related to the electrode are reactive charge transfer resistance on the electrode surface (R dl ); resistance developed on the electrode due to polarization (R e ); diffuse double layer capacitance across the electrodes (C dl ); and constant phase element due to polarization on the electrode surface (CPE e : CPE e -T and CPE e -P). We ignored the resistance and capacitance of the solution because we assumed that the cell and electrode were in very close proximity. In addition, we also ignored membrane polarization because it was very small compared to the resistance and capacitance of the total cell membrane.
In practice, CPE e -P should be in the range of 0 to 1. If it is equal to 1, then CPE e would be identical to the capacitor 5, 6 . In our study, our fitted values, CPE e -P, obtained from the Set 1 and Set 2 experiments were 0.8978 and 0.8936, respectively. Thus, the CPE e values obtained in this work are close to the capacitor values, although they are less than 1 due to polarization on the electrode surface (C e , capacitance of electrode surface). Since the experiment electrodes for Set 1 and 2 were different, we obtained different fitted values from the data at 24 hours for each set. However, for each set, the parameter values related to the electrode (R dl , C dl , R e , CPE e -T and CPE e -P) were fixed to determine the changes to intracellular bulk resistance (R cyto ), cell membrane capacitance (C m ) and reactive charge transfer resistance on the cell membrane (R m ) at 48 hours.
For the cellular growth control group without NPs treatment, R cyto decreased by 9.6%. In contrast, once the NPs were internalized, R cyto increased by 25.5%, as the internalized NPs reduced the volume of the ion solution inside the cytoplasm and thereby could be considered electric resistant elements. In other words, the NPs' internalization into the cells increased cytoplasm resistance. R m and C m value changes for cellular growth without NPs treatment were -18.6% and -9.2% but after the NPs uptake these changes were -54.6% and -24.8%, respectively. Thus, the NPs' internalization also changed the cellular charge transfer resistance and membrane capacitance. Based on these results, we concluded that changes in Only R cyto , R m and C m were adjusted according to the percentages (9.6%, -18.6% and -9.2%, respectively) of those obtained from Set 1 at 48 hours ( Supplementary Fig. 5b)  In theory, when the cell-cell or cell-matrix interaction decreases, C m should also decrease. In addition, since a weaker interaction causes increased exposure of the electrode to the ionic solution, C dl and CPE e should decrease. We simulated changes in the cell-cell or cell-matrix interactions when we fixed the parameter values associated with resistance (R cyto , R m, R dl and R e ) to the values obtained from Set 1 at 24 hours. Only C m , C dl and CPE e were adjusted ( Supplementary Fig. 6). The |α| values at the low frequency region decreased from 0.167 to 0.146 (a change of 12.6%) when C m , C dl and CPE e were adjusted to -31%, -92.5% and -12.6 %, respectively, but the |β| values increased from 0.566 to 0.604 (a change of +6.71%).
The theoretically simulated parameter values are summarized in Supplementary Table 2a: Cell-cell or cell-matrix interactions changes/Simulation. This observation is similar to our experimental results; endothelial cell permeability increased due to the TNF-α treatment and the cell surface adhesion weakened due to the siRNA-mediated depletion of CD44 (Figs. 4b and 5b).
In conclusion, cell permeability or adhesion changes can be observed by measuring the decrease of the |α| values at the low frequency region, and cellular uptake of NPs can be distinguished by measuring the decrease of the |β| values at the high frequency region.