2.1.

Materials and Reagents

Hela cells were obtained from cell bank of Chinese Academy of Sciences (Shanghai, China). AuNPs were purchased from BBI solution (United Kingdom). Fluorescein isothiocyanate-dextran (FITC-D) with MW of 5 kDa, 10 kDa, 40 kDa, 70 kDa and 100 kDa was purchased from Xi’an ruixi Biological Technology Co., Ltd (Xi’an, China). PI and penicillin/streptomycin were purchased from Solarbio Life Sciences, Ltd. (Beijing, China). Dulbecco’s modified eagle medium (DMED, Hyclone, United States), phosphate buffer saline (PBS, Hyclone, United States) and trypsin (Hyclone, United States) were purchased from the agent Xi’an Haimeng Experimental Technology Co., Ltd. (Xi’an, China). Fetal bovine serum (FBS, Gibico) and voltage sensitive dye Di-4-ANEPPS (Invitrogen) were obtained from ThermoFisher Scientific. Dimethyl sulfoxide was obtained from Sigma-Aldrich (United States).

2.2.

Cell Culture

Hela cells were cultured in DMEM containing 10% FBS supplemented with 1% penicillin/streptomycin at 37°C and 5% ${\mathrm{CO}}_2$ atmosphere. Before experiments, cells were plated into a 96-well plate with a density of $0.8\times {10}^4\text{cells}$ per well and incubated for 12 h.

2.3.

Monitoring Membrane Potential Change

A custom-made photoporation experimental system is shown in Fig. 1. The 532 nm pulsed laser beam (Q-smart 450, Quantel, France) was split into two parts by a beam splitter with a split ratio of 10: 90. One part was focused by a 10 × objective (GCO-2131, Daheng Optics, China) to irradiate cells, and the other part was recorded by energy meter (PE50-DIF-C, Ophir Optronics, Israel). The LED with the wavelength of 528 nm and a power of 3 W (GCI-060403, Daheng Optics, China) was used to excite voltage sensitive dye which was added in culture medium with a concentration of $20\mu \mathrm{g}/\mathrm{ml}$ and washed with PBS after 1 hour incubation. Before the experiment, AuNPs solution with an approximate optical density (OD) of 0.025 was added to cell and incubated for 1 h, and then it was washed with PBS to remove the free AuNPs. A cooled electron-multiplying charge coupled devices (EMCCD) was used to collect the fluorescence images with an exposure time of 1 s. Before irradiating the cell, 100 frame fluorescence images were collected as background images with an interval of 1 s. Then, a 6 ns laser pulse was generated to irradiate cells and 380 frame fluorescence images were obtained after irradiating the cell. A digital delay generator DG645 (Stanford Research Systems Inc., United States) was used to synchronize laser and EMCCD (Xion3, Andor, United Kingdom). A Matlab (Matlab R2020a, MathWorks, United States) program was written for handling the fluorescence and automated graphing the changing curve of relative fluorescence intensity in response to changes in membrane potential. The resealing time was defined as the time point that the relative intensity of membrane potential back up to 90% of resting fluorescence intensity after laser treatment. Then, the resealing time of the pore can be calculated and a function relationship between resealing time and laser fluence was established.

Fig. 1

Schematic of photoporation and membrane potential detection system.

2.4.

Photoporation on Hela Cells

The same system shown in Fig. 1 was used for photoporation. A 532 nm 6 ns pulse laser (Q-smart 450, Quantel, France) was employed to irradiate the cell with a repetition frequency of 10 Hz, the laser beam had a Gaussian profile with a radius of $356\mu \mathrm{m}$ at the interface with the cell. Meanwhile, the two-dimension motion stage (MC600, Zolix Instruments Co., Ltd, China) carrying the 96-well moved line by line at a velocity of $1\mathrm{mm}/\mathrm{s}$. The distance between two lines was 0.1 mm. Before laser treatment, 60 nm AuNPs solution with OD of 0.025 was added into cell medium and incubated for 1 h and it attached on cell membrane by cell adhesion effect. After that, PBS was used to wash the free AuNPs out. Then, fresh culture medium mixed with FITC-D ($0.4\mathrm{mg}/\mathrm{ml}$) was supplemented into the well and the cell was treated with the system. Afterwards, cells were cultured in humidified incubator for 30 min, and then washed with PBS. Afterward, fresh culture medium mixed with PI ($1\mu \mathrm{g}/\mathrm{ml}$) was supplemented into the well and washed with PBS after 10 min. Then, the fluorescence image was obtained by a fluorescence microscope (ECLIPSE TI, Nikon, Japan), and the perforation efficiency was quantitatively evaluated with a FACScan flow cytometer (BD Biosciences, United States). As shown in Fig. S1 in the Supplementary Material, the cells located in Q1 were defined as perforated cells, and the percentage of Q1 was defined as loading efficiency. The cells located in Q2 and Q3 were defined as dead cells, and the percentage of Q2 and Q3 was defined as death rate. The permeability and death rate of photoporated cells was estimated as descripted in our previous publication.³¹

2.5.

Model of Loading Efficiency Based on Brownian Motion

To estimate the probability that the foreign fluorescence molecule of FITC-D entered into the cell, a model based on Brownian motion was established by Monte Carlo simulation.

In the simulation, the cell was set as a ball in three-dimensional space coordinate, marks was appeared randomly on the surface that indicated pores on the cell membrane. Each of foreign materials was treated as an independent point without any associated area. The point of foreign materials was randomly generated and dispersed at $t=0$. Each of the foreign material randomly moved obeying Brownian motion, and the moving step was determined according to MW of foreign materials. The probability that each point diffused into the coordinate of pore was defined as delivery efficiency.

According to Brownian motion materials size determined the diffusion velocity of materials, it was quantified by the step size of movement in our program. The relationship between moving velocity and material size was deduced as follow: the foreign materials randomly diffused in culture medium with the manner of Brownian motion, and the diffusion coefficient can be determined as the following equation:

After pulse laser irradiating on gold nanoparticle, the temperature of gold nanoparticle and surrounding medium reached to steady state in several nanoseconds.³² Therefore, we deem that the temperature was constant in the process that foreign materials entered into the cell. In this work, the concentration of foreign materials was constant and there was little difference between the MW of foreign materials used in experiments. So, we assumed the viscosity was constant. Thus, we can conclude that diffusion coefficient was inversely proportional to the radius of foreign materials from Eq. (1), and the radius of foreign materials was proportional to the cube root of MW from Eq. (3). In a certain interval of time, the velocity of foreign molecule was proportional to diffusion coefficient.³³ Therefore, we can conclude the following equation:

3.1.

Measurement of Cell Membrane Resealing Time

Cell membrane resealing time can be obtained by monitoring the relative fluorescence intensity ($\mathrm{\Delta }F/F$), which changes in respond to the membrane voltage changes. $\mathrm{\Delta }F/F$ denotes the change in fluorescence intensity relative to resting fluorescence intensity. The fluorescence image of cells [Fig. 2(a)] was captured by EMCCD in a custom-made experimental system (Fig. 1). We captured 100 frames of background firstly and 380 frames of fluorescence images after one pulse of laser irradiation at imaging speed of 1 frame/second. The background images were used as calibration. Considering the laser irradiation for photoporation was only last 6 ns (laser pulse width), the fluorescence photobleaching was mainly due to the LED. A linear correction method was used to reduce the effect of photobleaching (Fig. S2 in the Supplementary Material). Figure 2(a) is an example of the fluorescence images captured after the laser treatment. Figures 2(b) and 2(c) were the relative fluorescence intensity of a perforated and an unperforated cell [marked as a and b in Fig. 2(a)] over time, respectively, corresponding to a change in membrane voltage with time. It can be seen clearly that when the laser pulse turned on at $\sim 100\mathrm{s}$, there is a dip appeared on fluorescence intensity from the cell membrane of the perforated cell [Fig. 2(b) and Fig. S2 in the Supplementary Material], whereas no significant change can be observed on that come from the unperforated cell [Fig. 2(c)]. This dip in fluorescence intensity can be attributed to the formation of a pore on the cell membrane after laser irradiation. Subsequently, the pore underwent recovery once the irradiation ceased. The presence of the pore disrupted the static potential of the cell membrane and influenced the curve representing the relative fluorescence intensity of the membrane. As showed in Fig. 2(d), the resealing time was defined as the time from laser on (100’th seconds) to relative fluorescence intensity recovering to 90% of resting intensity,³⁴ and a computer program was coded to determine the resealing time. It increased while the irradiation fluence increased and the increasing rate can be defined by the function of $y=106.{3\mathrm{e}}^{x/2.7}-80.6$ [Fig. 2(e)].

Fig. 2

Establishing the function relationship between laser fluence and resealing time. (a) The fluorescence image of Hela cells stained by Di-4-ANEPPS after laser treatment; green boxes labeled “a” and “b” show a representative perforated and unperforated cell, respectively. Relative fluorescence intensity ($\mathrm{\Delta }F/F$) of (b) a perforated cell [marked as “a” in panel (a)] and (c) an unperforated cell [marked as “b” in panel (a)], in response to a change in membrane voltage with time, respectively. (d) Relative fluorescence ($\mathrm{\Delta }F/F$) intensity in response to a change in membrane voltage with time after cell was treated with different laser fluence. The laser fluence was in the range of 0.064 to $1.6\mathrm{J}/{\mathrm{cm}}^2$. (e) The relationship between laser fluence and resealing time. The error bars show the ranges of laser fluence and membrane resealing time from at least ten independent measurements; the red line is the fitted line, which can be expressed as $y=106.{3\mathrm{e}}^{x/2.7}-80.6$. Ten cells from different independent experiments ($n\ge 3$) were selected to reduce error for each laser fluence.

3.2.

Evaluation of Cell Photoporation

FITC-D with a MW of 10 kDa (FD10) was employed as foreign material to study the VNBs induced photoporation. 60 nm gold nano-sphere particles with an approximate OD of 0.025 was chosen to assist photoporation because 60 nm is the optimizing diameter for vapor bubble generation according our previous research.^35,36 Cells were irradiated by different fluence at 0.064, 0.32, 0.64, 0.96, 1.28, and $1.6\mathrm{J}/{\mathrm{cm}}^2$, respectively.

Fluorescence images of laser treated cells are shown in Fig. 3(a), and the quantitative statistic results are shown in Fig. 3(b) where 5000 cells are analyzed (Fig. S1 in the Supplementary Material). The results demonstrated a significant increase in loading efficiency when using laser fluence levels above $0.96\mathrm{J}/{\mathrm{cm}}^2$. It can reach to 39.4% at $1.6\mathrm{J}/{\mathrm{cm}}^2$; however, at this point, the death rate was significantly increased to 37.83%. Based on the loading efficiency of experimental data and the relationship between resealing time and laser fluence, the loading efficiency as a function of resealing time was achieved [Fig. 3(c)].

Fig. 3

Delivering the FD10 into Hela cells by VNBs induced photoporation. (a) Fluorescence images of intracellular delivery of FD10 after treated with different laser fluence; FITC, fluorescein isothiocyanate and PI, propidium iodide. (b) The percentage of perforated cells and dead cells evaluated by flow cytometry at different laser fluence. Perforated cells were the FITC positive and PI negative cells. Dead cells were the PI positive cells. Statistically significant differences were indicated by *** ($p<0.001$) and * ($p<0.05$). Error bars represent standard deviation from the mean of independent experiments ($n=3$). (c) The loading efficiency as the function of resealing time by using experimental data. The error bars show the range of loading efficiency from the mean of independent experiments ($n=3$); the red line is the fitted line, which can be expressed as $y=46.{89\mathrm{e}}^{x/251.87}-48.68$.

To study the impact of foreign material size on perforation efficiency, we irradiated the cells with different size of molecules, under certain laser fluence. Laser fluence of $0.96\mathrm{J}/{\mathrm{cm}}^2$ was selected for its higher delivery efficiency and lower death ratio with minimal cell disturbance (i.e., cell deformed under bright field) compared to that under other laser fluence. FITC-Ds with different MWs of 5 kDa, 10 kDa, 40 kDa, 70 kDa, and 100 kDa were used in the experiments. The fluorescence images and the quantitative statistical results were shown in Figs. 4(a), 4(b), and 4(c), respectively. It clearly showed that FITC-Ds with small MW were more easily to be delivered into cells than that of large MW. The perforation efficiency of 5 kDa can reach to 42.03% at $0.96\mathrm{J}/{\mathrm{cm}}^2$, while it was 19.76% in the control group without laser treatment indicating some of the FITC-D (5 kDa) were entered into cell by endocytosis. Indeed, the percentage of perforated cells was significant decreased compared to that under 5 kDa when MW of FITC-Ds was greater than or equal to 10 kDa in control groups. Similarly, the loading efficiency of VNBs induced photoporation was decreased while increasing MW. Based on the loading efficiency of experimental data, the loading efficiency as a function of MW was achieved [Fig. 4(d)].

Fig. 4

Delivering the FITC-D with different MWs (5 kDa, 10 kDa, 40 kDa, 70 kDa, and 100 kDa) into Hela cells by VNBs induced photoporation. (a) Fluorescence images of the control group, cells were incubated with AuNPs, and FITC-D and PI were added for imaging. (b) Fluorescence images of the treated group (laser fluence: $0.96\mathrm{J}/{\mathrm{cm}}^2$). (c) The percentage of perforated cells and dead cells determined by flow cytometry (laser fluence: $0.96\mathrm{J}/{\mathrm{cm}}^2$). Perforated cells were the FITC positive and PI negative cells. Dead cells were the PI positive cells. Statistically significant differences were indicated by *** ($p<0.001$), ** ($p<0.01$), and * ($p<0.05$). Error bars represent standard deviation from the mean of independent experiments ($n=3$). (d) The loading efficiency as the function of MW using experimental data. The error bar shows the range of loading efficiency from the mean of independent experiments ($n=3$); the red line is the fitted line, which can be expressed as $y=23.{5\mathrm{e}}^{-x/20.3}+4.5$.

3.3.

Estimating the Delivery Efficiency of Photoporation

To find out the relationship between loading efficiency and resealing time, a diffusion-based model was established to predict the loading efficiency and the dynamic process that extracellular materials diffusing into cell is shown in Video 1. FITC-Ds with a MW of 10 kDa was used as a model (of foreign materials) to study the loading efficiency changes due to the certain pore resealing time (opening time) that determined by corresponding laser fluence according to the function relationship in Fig. 2(e). Based on the diffusion-based model (see Sec. 2), the loading efficiency as a function of resealing time can be achieved with a constant MW of exogenous molecules [Fig. 5(a)]. The experimental and simulated results were presented an exponential growth while increasing the resealing time within a range of 28.6 to 163.8 s. The $R^2$ were 0.9604 and 0.9944 for experimental and simulated data, respectively. Based on the above results (Fig. 4), we can see that the molecular weight/size is another important parameter for loading efficiency. FITC-Ds with different MW (5 kDa, 10 kDa, 40 kDa, 70 kDa, and 100 kDa) and constant resealing time were also implemented into the model. When using the loading efficiency corresponding to different MW at $0.96\mathrm{J}/{\mathrm{cm}}^2$ (obtained from the previous experiments), the loading efficiency as a function of MW can be achieved [Fig. 5(b)]. The $R^2$ were 0.9891 and 0.9918 for experimental and simulated data, respectively. It can be seen that the loading efficiency has a downward trend when the molecular weight was increased. To evaluate the accuracy of simulation data, the comparison between the loading efficiency of simulated data and the experimental data were presented in Figs. 5(c) and 5(d). It clearly showed that the loading efficiency of simulation data has a consistent variation trend with the loading efficiency of experimental data.

Fig. 5

Comparison of loading efficiencies between experimental and simulated data. (a) The loading efficiencies as the functions of resealing time using experimental and simulated data, respectively. (b) The loading efficiency as a function of MW using experimental and simulated data, respectively. (c) Simulated versus measured loading efficiency at different resealing time. (d) Simulated versus measured loading efficiency for different MW. The red line is the fitted line of experimental data, and the black line is the fitted line of simulated data in panels (a) and (b); the red circles are the experimental data and the black squares are the simulated data in panels (a) and (b); and the red line is the fitted line in panels (c) and (d). Error bars represent standard deviation from the mean of independent experiments ($n=3$) for experimental data and the mean of independent simulation experiments ($n=10$) for simulation data.