Relationship between bio-effects and energy transduction during nanoparticle-mediated photoporation

Many targets for therapeutic and diagnostic medical purposes, as well as basic research investigation, are located inside cells. Intracellular delivery systems have been widely researched to reach these targets. Many intracellular delivery systems have relied on viral vectors to transfect cells with nucleic acids.¹ Although there are a few U.S. Food and Drug Administration-approved products on the US market,^2,3 immunogenic responses and other safety concerns have limited widespread use of viral vectors.^1,4 Researchers have explored alternative methods using polymers,⁵ peptides,⁶ cationic lipids,⁷ and dendrimers⁸ for intracellular delivery; and this has resulted in an expansion of delivery molecules from nucleic acids to a variety of proteins, synthetic nucleases, and molecular probes.⁹ However, these delivery methods rely on the endocytic pathway to cross the cell membrane, which often lead to molecular degradation.¹⁰ 

Several other routes of intracellular delivery have been investigated, including the application of external stress causing direct poration of the cell membrane, which avoids endocytosis. Electroporation¹¹ has been the most common method, which destabilizes the cell membrane via build-up of charge on the cell surface generated by a pulsed electric field to create transient membrane pores. Additionally, researchers have used sonoporation,¹² magnetofection,¹³ and microfluidics¹⁴ among other methods to explore the use of external stress application for intracellular delivery. However, physical delivery methods often encounter trade-offs between cell viability and molecular uptake.^15,16

Recently, nanoparticle-mediated photoporation has been presented as a novel method to create transient membrane pores that enable the delivery of molecules to up to ∼90% of cells with insignificant viability loss.¹⁷ Using this approach, a suspension of cells is mixed with carbon black (CB) nanoparticles and desired delivery molecules and then, irradiated with a nanosecond (or faster) pulsed near infrared (NIR) laser beam of 1064 nm wavelength. The CB nanoparticles absorb the laser energy and then, dissipate it in the form of thermal and acoustic outputs as well as fluid flow. This energy concentration and subsequent transduction generate short-lived vapor bubbles around each nanoparticle as long as the temperature is sufficiently high to vaporize surrounding water.^17,18 The resulting energy interactions can create external stress on the cell membrane, leading to membrane disruption that allows desired molecules to diffuse through the transient pores. In optimal conditions, the cell survives this process but sub-optimal conditions can lead to loss of cell viability.

The entire process involves several energy transfer steps from laser to nanoparticle, from nanoparticle to surrounding media, and then from media to the cell membrane. We can break this process down into its constituent steps (Fig. 1). First, energy is emitted from the laser, which is controlled by parameters such as laser fluence, wavelength, pulse length, and number of pulses (i.e., laser operating parameters).

Next, energy is absorbed by CB nanoparticles (which raises their temperature); absorption is controlled by nanoparticle related parameters such as their optical, thermal, and physical properties (e.g., material, size, shape), as well as nanoparticle concentration (i.e., nanoparticle parameters). At the 1064 nm wavelength, very little laser energy is absorbed by water, cells, or anything other than the CB nanoparticles in the cell suspension. The amount of energy absorbed by the nanoparticles can be characterized in multiple ways, such as the total energy absorbed, energy absorbed per nanoparticle, and energy absorbed per nanoparticle mass (i.e., energy absorption parameters).

Finally, energy absorbed by the CB nanoparticles is emitted to the surrounding water, which can be vaporized, and transferred to nearby cells to cause bio-effects. The magnitude and extent of bio-effects are controlled by vapor bubble size, temperature, and number of bubbles among other parameters (i.e., energy transduction parameters).

In this study, we consider how these various parameters influence cellular bio-effects such as intracellular delivery, loss of cell viability, and cell fragmentation, which are believed to be the result of progressively stronger forms of energy transduction to the cells. Prior work examining relationships between bio-effects and laser operating parameters has demonstrated correlations.¹⁹ In this study, we seek to examine relationships between bio-effects and parameters that are closer in the energy transfer process to the cellular impact: namely, the nanoparticle absorption parameters, energy parameters, and energy transduction parameters. We hypothesized that the energy transduction parameters would correlate best with bio-effects: they are likely more closely related to the mechanism of action since they characterize the energy transduced to the cells rather than further upstream processes.

DU145 human prostate carcinoma cells (American Type Culture Collection, Manassas, VA) were cultured in adherent tissue culture flasks. Roswell Park Memorial Institute (RPMI) 1640 Medium (Cellgro, Herndon, VA) with 10% v/v fetal bovine serum (Corning, Palo Alto, CA) and 1% penicillin–streptomycin (Cellgro) were used as media for cell growth during the incubation period. Culture flasks containing cells were incubated at 37 °C, 5% CO₂, and 98% relative humidity for growth. After 85% of surface confluency was reached, cells were harvested using 0.25% Trypsin/EDTA (Cellgro). Subsequently, these cells were suspended in RPMI at a concentration of 10⁶ cells/ml for experiments.

Various nanoparticle dispersions were prepared using powdered nanoparticles and de-ionized (DI) water. The following nanoparticles were all procured in powdered form: CB nanoparticles with primary particle size of 25 nm (Black Pearls 470, Cabot, Boston, MA), CB nanoparticles with primary particle size of 90 nm (Asbury Carbon, Asbury, NJ), CB nanoparticles with average primary particle size of 367 nm (Continental Carbon Company, Houston, TX), copper nanoparticles (Cu, primary particle size of 25 nm, Sigma-Aldrich, St. Louis, MO), silica nanoparticles (primary particle size of 10–20 nm, Sigma-Aldrich), silicon carbide nanoparticles (SiC, primary particle size of 40 nm, SkySpring Nanomaterials Inc., Houston, TX), zinc oxide nanoparticles (ZnO, primary particle size of 10–30 nm, SkySpring Nanomaterials), graphite nanoparticles (primary particle size of 3–4 nm, SkySpring Nanomaterials), diamond nanoparticles (primary particle size of 3–4 nm, SkySpring Nanomaterials), multi-walled carbon nanotubes (MWCNT1020, an outer diameter of 10–20 nm, SkySpring Nanomaterials), multi-walled carbon nanotubes (MWCNT2030, an outer diameter of 20–30 nm, SkySpring Nanomaterials), and single-walled carbon nanotubes (SWCNT1020, a diameter of 1–2 nm, SkySpring Nanomaterials).

Nanoparticles dispersions were prepared at a concentration of 400 mg/l (unless otherwise mentioned) in DI water containing 0.013% v/v Tween-80 (Sigma-Aldrich), which were used to prevent particle aggregation (and are not expected to have adverse effects on the cells²⁰). The dispersions were then homogenized through 35 min sonication in an ultrasonic water bath (FS3OH, Fisher Scientific, Pittsburg, PA) and then, 1 min sonication with an ultrasonic needle (Sonics Ultracell, Sonics & Materials, Newton, CT), which we have found is sufficient time to break apart large clusters of aggregates (data not shown). Dynamic light scattering (DLS) measurements (ZetaSizer Nano, Malvern Instruments, Malvern, UK) are provided by aggregate size in Table I. See Fig. S1 in the supplementary material for an example of DLS measurements.

Zeta potential values (measured using Zetasizer) in Table I suggested that not every nanoparticle dispersion was stable. Therefore, fresh dispersions were prepared for each nanoparticle before conducting laser exposure experiments. In addition, prior to actual laser irradiation, samples (including sham and cells-only controls) were vortexed to ensure homogeneous distribution of cells and nanoparticles.

Exposure samples were prepared with 520 μl of DU145 cells (10⁶/ml) suspended in RPMI, 37 μl of given nanoparticle dispersion, and 5.5 μl of calcein (Molecular Probes, Eugene, OR) solution (at a final concentration of 10 μM) in a 1.5 ml Eppendorf tube. To minimize cellular functionality during storage, samples were kept in an ice bath before and after laser exposure.

Cells were exposed to a laser beam in a 2 mm wide and 21 mm diameter cylindrical Pyrex glass cuvette (37-PX-2, Starna Cells, Atascadero, CA) with an Nd:YAG solid-state laser (Continuum Powerlite II Plus, Continuum, San Jose, CA) that generated 5–9 ns pulses of 1064 nm wavelength at a frequency of 10 Hz. Laser beams were shaped and focused to irradiate the entire cuvette surface area (21 mm diameter) with a homogeneous energy profile. Briefly, a 12 mm beam was first passed through an 8 mm aperture, thereby blocking the tapered edges of the beam. Cutting the edges helped obtain a flat top beam profile with uniform energy distribution across the beam diameter. This beam was then passed through a diverging lens that allowed the beam to irradiate the whole cuvette surface. Immediately after exposure, samples were transferred back to the Eppendorf tubes and stored on ice. “Sham” samples with cells suspended in RPMI, nanoparticles, and calcein, experienced similar procedures except laser irradiation, and were used as negative controls.

Propidium iodide (PI) (Invitrogen, Grand Island, NY), added at a concentration of 13.4 μM, was used to label non-viable cells after laser irradiation. Excess calcein and CB nanoparticles were removed from cell suspension through centrifugation at 500 g and subsequent washing (repeated three times) with phosphate buffer saline before analysis.

Cellular bio-effects were quantified using a bench-top flow cytometer (BD Accuri, BD Biosciences, San Jose, CA) and were categorized as either viable cells with intracellular uptake or non-viable cells (i.e., intact cells with PI staining). Fluorescence of cells with calcein uptake was measured using a 530/28 nm bandpass filter with excitation at 488 nm, and fluorescence of cells with PI stain was measured using a 670 nm longpass filter with an excitation at 535 nm. Cell samples were run at a constant flow rate of 35 μl/min for 1 min. A “cells-only” negative control in RPMI was used to gate intact cells in the forward-scattered and side-scattered analyses. Fragmented cells were calculated as the difference between the number of viable and non-viable cells counted in a laser-exposed sample and the number of viable cells counted in sham samples.

1064 nm wavelength light absorbance of nanoparticles was measured using a spectrophotometer (Evolution 220, Thermo Scientific, MA). After nanoparticle dispersions were prepared, they were mixed with RPMI at a given concentration and transferred to a cuvette provided as a sample holder. A baseline measurement was done with only RPMI; all the absorbance values reported here had that baseline value subtracted to provide the true absorbance of each nanoparticle dispersion. It is important to note that the values given by a spectrophotometer include both absorbance and scattering. Therefore, the absorbance data presented in this study should be interpreted as the maximum possible absorbance, while the actual absorbance values were smaller. There are methods to measure actual absorbance²¹ but this level of analysis was beyond the scope of this study.

For bulk temperature-rise measurements, a suspension of 560 μl nanoparticle dispersion in RPMI at a concentration of 26.3 mg/l was irradiated with 1064 nm wavelength laser pulses for 1 min at 88 mJ/cm² laser fluence. Temperature measurements were taken using a standard J-type thermocouple immediately before and after the laser exposure. To account for light absorbance by the suspension media, a baseline temperature-rise measurement was taken for RPMI only (exposed to 88 mJ/cm² laser fluence for 1 min). All the temperature values reported here had that baseline value subtracted to provide the true rise because of the presence of each nanoparticle.

Calculation methods to determine laser energy absorption by nanoparticles; number, radius, and temperature of bubbles generated by laser irradiation of nanoparticles; and other parameters are shown in Sec. S1 in the supplementary material. A minimum of three replicates was used for each experimental condition. Mean and standard error of the mean were calculated using the three replicates. One-way analysis of variance (ANOVA, α = 0.05) was performed to compare three or more experimental conditions using Graphpad Prism 8 (GraphPad Software, La Jolla, CA) followed by post hoc Tukey's pairwise comparison. A value of p < 0.05 was considered statistically significant.

Correlations between bio-effects and experimental parameters were generated by combining data from the experiments presented in the Results section (performed over a range of laser fluence and nanoparticle parameters that effectively varied peak nanoparticle temperature, total number of bubbles, and peak bubble radius) using a power-law relation. Furthermore, the exponents of power-law were optimized for a better fit by maximizing the R² value of Langmuir isotherm type correlation between the given parameter and total bio-effects.

Our initial studies sought to understand the relationship between nanoparticle absorption parameters and bio-effects by examining the effects of changing nanoparticle composition and geometry. Cell suspensions mixed with different nanoparticle materials were irradiated with pulsed NIR laser. Subsequently, through flow cytometer analysis, cells were found to fall into one of four categories: (i) viable cells with little or no intracellular uptake of calcein that appeared to be unaffected by the treatment (i.e., no fluorescent staining); (ii) viable cells containing calcein [i.e., the green fluorescence of the uptake marker but no red fluorescence of the marker of non-viable cells (PI)], hereafter termed as “uptake cells;” (iii) non-viable cells (i.e., red fluorescence of PI); and (iv) fragmented cells (i.e., cells identified as “lost” based on reduced cell concentration). We categorized the cells in the latter three groups as having bio-effects from the laser exposure, and these are the categories plotted in the stacked bars shown in subsequent figures. We interpret these bio-effects to follow a continuum, where cells with uptake had milder bio-effects, non-viable cells experienced stronger bio-effects, and fragmented cells felt the strongest bio-effects.

When cells were irradiated in the presence of silica, ZnO, SiC, Cu, and diamond nanoparticles (all at the same concentration on a mass basis), less than 20% of cells had bio-effects and less than 11% of cells showed calcein uptake [Fig. 2(a)].

In comparison, cells suspended with graphite and CB nanoparticles (at the same mass concentration) had significantly higher bio-effects with more than 90% of cells showing some effect (one-way ANOVA, p < 0.0001 for uptake, non-viable, and fragmented cells). We interpreted bio-effects in the presence of CB nanoparticles as the strongest since a greater percentage of cells had fragmented (and thus experienced stronger bio-effects). Overall, CB nanoparticles can be comparatively seen as the most efficient transducer of energy leading to bio-effects.

For the next study, carbon nanoparticles of different shapes were suspended with cells and irradiated with laser pulses. Figure 2(b) shows that spherical CB nanoparticles were most efficient in causing photoporation followed by SWCNT (one-way ANOVA, p < 0.0001 for uptake, non-viable, and fragmented cells). Approximately, 85 ± 2% of cells lost viability due to cell death or fragmentation when exposed to laser pulses in the presence of CB nanoparticles.

In comparison, approximately 64 ± 4% of cells lost viability when suspended with SWCNT, and <50% of cells lost viability when suspended with MWCNTs. Both the MWCNTs performed similarly with no statistically significant differences in bio-effects (Tukey's multiple comparison tests, p > 0.05).

The first energy transfer in nanoparticle-mediated photoporation takes place from the laser beam to nanoparticles. Therefore, we expected that nanoparticles with high absorbance of 1064 nm wavelength laser light would show higher bio-effects. Figure 3(a) shows the 1064 nm wavelength laser light absorbance (+ scattering) measurements for different nanoparticles at the same mass concentration. Silica, ZnO, SiC, Cu, and diamond nanoparticles had insignificant differences in their measured absorbance values (Tukey's multiple comparisons test, p > 0.05), which were close to 0. Conversely, SWCNT and CB nanoparticles had similar absorbance values (p > 0.05) that were higher than the rest of the nanoparticles.

Viewing Figs. 2 and 3(a) together indicates that the nanoparticles with greater laser energy absorption also caused greater bio-effects. However, absorbance (+ scattering) measurements only assess the nanoparticles’ ability to absorb the laser energy and do not provide information about energy transfer from the nanoparticle to surrounding water and cells. Additionally, what we report as nanoparticle absorbance in fact includes both absorbed and scattered light. It is important to note that scattered light would not be expected to play a role in the energy transfer steps of nanoparticle-mediated photoporation.

To address this issue, we took bulk temperature-rise measurements of the various nanoparticle suspensions after laser irradiation. Theoretically, these measurements should correlate with the ability of the nanoparticles to absorb and dissipate energy, heating up the surrounding fluid, and should not include energy associated with scattered light. No significant differences (Tukey's multiple comparisons test, p > 0.05) were observed in bulk temperature-rise values of nanoparticle suspensions made of silica, ZnO, SiC, Cu, diamond, or graphite [Fig. 3(b)], which were measured at <2 °C. In comparison, the bulk temperature rise was measured as significantly higher for MWCNTs, SWCNT, and CB nanoparticles (Tukey's multiple comparisons test, p < 0.05), achieving 6 °C or greater temperature rise. However, no significant differences were seen in measured temperature-rise values among these four carbon-based nanoparticles (Tukey's multiple comparisons test, p > 0.05).

When data from Figs. 3(a) and 3(b) were replotted as temperature rise vs absorbance in Fig. 4(a), low absorbance values corresponded to lower bulk temperature rise and high absorbance values corresponded to higher bulk temperature rise. However, there was no direct dependence of temperature rise on absorbance: instead, we see a binary behavior, indicating that the CB and CNTs all absorbed the laser light similarly [Fig. 3(b)] but scattered it differently [Fig. 3(a)]. It can be assumed that higher absorbance can lead to greater bio-effects when the energy transferred in all the subsequent energy transfer steps is equal.

Based on these considerations, we plotted bio-effects as a function of nanoparticle energy absorbance (+ scattering) and as a function of bulk temperature rise, which is a more accurate measure of energy absorbance (i.e., that does not include energy from scattered light), in Figs. 4(b) and 4(c), respectively. While the bio-effects did depend on these two factors (one-way ANOVA, p < 0.0001), there was considerable scatter in the data, which did not show a clear trend or dependence. This may be because different nanoparticles absorbed and released the same amount of energy but the manner in which they released the energy (e.g., kinetics) might be different due to properties such as nanoparticle size, shape, heat capacity, thermal conductivity, etc.

So far, we have considered changes to nanoparticle parameters by changing nanoparticle material, size, and shape to influence the process of converting laser energy into bio-effects. To better understand this process, we accounted for the combination of laser operating parameters and nanoparticle parameters that lead to energy absorption parameters. Our goal was to vary laser and nanoparticle parameters to test three hypotheses to identify the dominant nanoparticle energy absorption parameter associated with bio-effects.

The first hypothesis was that the total energy absorbed is the dominant parameter and correlates with bio-effects. This assumes that it does not matter how the energy is distributed among different numbers and sizes of nanoparticles; the primary factor is the total energy that is absorbed and then, released by those nanoparticles. The second hypothesis was that the energy absorbed per nanoparticle is the critical parameter. This takes a more localized view that each individual nanoparticle has the opportunity to interact with a nearby cell, and, thus, it is the amount of energy absorbed and then released by each individual nanoparticle that determines bio-effects. The third hypothesis was that energy absorbed per nanoparticle mass is most closely associated with bio-effects. This, in effect, means that each nanoparticle is heated to the same temperature (assuming all nanoparticles have the same heat capacity) and that might be the critical parameter driving bio-effects.

To identify the dominant parameter, each of the three nanoparticle energy absorption parameters was analyzed individually by keeping that parameter constant and varying other parameters by changing experimental conditions, including nanoparticle size, total number of nanoparticles, total mass of nanoparticles, and laser fluence using CB nanoparticles.

The first nanoparticle energy absorption parameter analyzed was total energy absorbed with the hypothesis that total absorbed energy by the whole system is the dominant energy parameter. For this study, the number/concentration of CB nanoparticles and the laser fluence were varied to maintain constant total energy absorbed, which caused the energy per nanoparticle and energy absorbed per nanoparticle mass to vary.

Figure 5(a) reveals that total bio-effects spanned from 18 ± 1% to 95 ± 3% of cells while keeping total absorbed energy constant (one-way ANOVA, p < 0.0001). Additionally, cellular uptake percentages varied from 16 ± 1% to 71 ± 3%, non-viable cell percentages from 3 ± 1% to 18 ± 1%, and fragmented cell percentages from 0% to 6 ± 1% (one-way ANOVA, p < 0.0001 for uptake, non-viable, and fragmented cells). The variability in these data suggests that the total absorbed energy was not the dominant parameter.

Theoretical calculations of energy absorbed, number of nanoparticles (or bubbles), and peak nanoparticle temperature for the conditions used in Fig. 5(a) showed that while total energy absorbed was kept constant at 3.7 mJ, the number of nanoparticles (or bubbles) varied from 1.7 × 10⁹ to 8.5 × 10⁹. An increase in the number of bubbles suggests more opportunities for interaction with cells, which should increase bio-effects. However, as the number of nanoparticles (or bubbles) increased, bio-effects decreased [Fig. 5(a)]. This might be explained by the decrease in calculated peak nanoparticle temperature (from 1810 °C to 380 °C) and vapor bubble radius (from 670 nm to 370 nm) with increasing nanoparticle concentration because the constant total energy absorbed was distributed among more nanoparticles.

Next, we considered our second hypothesis that the energy absorbed per nanoparticle is the dominant nanoparticle energy absorption parameter. For this study, CB nanoparticle size and number and laser fluence were varied to maintain constant energy absorbed per nanoparticle, which caused the total energy absorbed and energy absorbed per nanoparticle mass to vary.

Figure 5(b) shows that total bio-effects varied from 7 ± 1% to 97 ± 4% while keeping energy absorbed per nanoparticle constant (one-way ANOVA, p < 0.0001). Additionally, cellular uptake percentages varied from 4 ± 1% to 14 ± 2%, non-viable cell percentages from 3 ± 1% to 37 ± 3%, and fragmented cell percentages from 0% to 46 ± 3% (one-way ANOVA, p < 0.0001 for uptake, non-viable, and fragmented cells). The variability of these data indicates that energy absorbed per nanoparticle was not the dominant parameter.

Theoretical calculations showed that when energy absorbed per nanoparticle was kept constant at 2 × 10⁻⁹ mJ, the number of nanoparticles (or bubbles) varied from 2.1 × 10⁸ to 8.5 × 10⁹, peak nanoparticle temperature varied from 65 °C to 1810 °C, and vapor bubble radii ranged from 0 to 673 nm in the conditions used in Fig. 5(b). The conditions with higher numbers of nanoparticle (or bubbles), greater nanoparticle peak temperatures, and larger bubble radius corresponded to the conditions with the greatest bio-effects, which can explain the data.

Finally, we examined our third hypothesis that energy absorbed per nanoparticle mass is the dominant energy parameter. For this study, CB nanoparticle size, number and laser fluence were varied to maintain constant energy absorbed per nanoparticle mass, which caused the total energy absorbed and energy per nanoparticle to vary.

The data in Fig. 5(c) show that total bio-effects varied between 33 ± 1% and 84 ± 2% while keeping energy absorbed per nanoparticle mass constant (one-way ANOVA, p < 0.0001). Additionally, cellular uptake percentages ranged from 22 ± 1% to 51 ± 2% and non-viable cell percentages spanned from 5 ± 1% to 28 ± 1% (one-way ANOVA, p < 0.0001 for both uptake and non-viable cells) but fragmented cell percentage did not significantly vary (Tukey's multiple comparisons test, p > 0.05). Once again, these variable data suggest that energy absorbed per nanoparticle mass was not the dominant parameter.

Theoretical calculations revealed that when energy absorbed per nanoparticle mass was kept constant at 8 × 10² mJ/mg, the number of nanoparticles (or bubbles) varied from 2.1 × 10⁸ to 8.5 × 10⁹. In Fig. 5(c), bio-effects increased with an increasing number of bubbles, which might explain the data. However, calculated peak nanoparticle temperature remained constant (at 1170 °C) and the vapor bubble radius increased from 580 nm to 1970 nm as bio-effects decreased, which appears inconsistent with the expectation that larger bubbles should have greater bio-effects. It appears that the number of bubbles is more important than bubble size in causing bio-effects.

Overall, the data from Figs. 5(a)–5(c) suggest that all three of our hypotheses are incorrect and that there is no single nanoparticle energy absorption parameter that dominates nanoparticle-mediated photoporation. Theoretical analysis of the data suggested that energy transduction parameters like number of bubbles, bubble size, and bubble temperature maybe provide better correlations.

This analysis led us to design experiments to look for relationships between bio-effects and energy transduction parameters associated with bubbles created during photoporation. The hypothesis for this study was that increasing the total number of bubbles, peak bubble radius, and peak nanoparticle temperature all increase bio-effects. For the first case, the total number of bubbles was varied (from 1.7 × 10⁹ to 8.5 × 10⁹) by changing the CB nanoparticle concentration while keeping peak nanoparticle temperature (1810 °C) and peak bubble radius (670 nm) constant. We assumed that each nanoparticle generated a bubble, and, thus, the number of bubbles was equal to the number of nanoparticles. To aid this assumption, laser fluence was chosen such that peak nanoparticle temperature was always well above 100 °C and hence, heat dissipation would be expected to cause bubble formation from each nanoparticle. Prior theoretical analysis showed that nanoparticle heating can be approximated as adiabatic because nanoparticle cooling takes much longer than nanoparticle heating during the 5–9 ns laser pulse.²² This finding allowed us to calculate peak nanoparticle temperature based on adiabatic heating of the nanoparticles.^22,23

Given that the cell concentration is kept constant, an increasing number of bubbles should decrease the average bubble-to-cell distance. Results in Fig. 6(a) show that bio-effects increased with an increasing total number of bubbles as shown by increases in non-viable and fragmented cell percentages from 24 ± 2% to 90 ± 3% (one-way ANOVA, p < 0.0001). As cells were killed, percentages of uptake cells decreased from 71 ± 8% to 7 ± 1% (one-way ANOVA, p < 0.0001). These data are consistent with the total number of bubbles positively correlating with bio-effects.

Next, the peak bubble radius was varied between 670 nm and 2300 nm while keeping peak nanoparticle temperature (1800 °C) and total number of bubbles (5.5 × 10⁸) constant. Corresponding results in Fig. 6(b) show that total bio-effects increased with increasing peak bubble radius as shown by increases in non-viable and fragmented cell percentages from 24 ± 1% to 90 ± 5% with increasing peak bubble size (one-way ANOVA, p < 0.0001). Percentages of uptake cells correspondingly decreased from 66 ± 4% to 6 ± 1% (one-way ANOVA, p < 0.0001). These data indicate that larger bubbles caused greater bio-effects.

Finally, peak nanoparticle temperature was varied from 140 °C to 1800 °C while keeping total number of bubbles (5.5 × 10⁸) and peak bubble radius (670 nm) constant. Corresponding results in Fig. 6(c) show that total bio-effects increased from 7 ± 1% to 89 ± 4% (one-way ANOVA, p < 0.0001). Additionally, uptake cell percentages increased from 4 ± 1% to 70 ± 4%, non-viable cell percentages increased from 3 ± 1% to 15 ± 1%, and fragmented cell percentages increased from 0.5 ± 0% to 3 ± 0% with increasing peak nanoparticle temperature (one-way ANOVA, p < 0.0001 for uptake, non-viable, and fragmented cells). These data show that nanoparticles heated to higher temperatures caused greater bio-effects.

We showed that three energy transduction parameters associated with bubbles correlated with bio-effects when the other parameters were held constant. Our next step was to determine how these parameters correlated with bio-effects when they were all allowed to vary. For this analysis, we used data from more than 100 different experimental conditions and looked for correlation with total bio-effects, defined as the sum of uptake cells, non-viable cells, and fragmented cells.

This analysis shows that there was a poor correlation between bio-effects and the individual energy transduction parameters: peak nanoparticle temperature [Fig. 7(a)], total number of bubbles [Fig. 7(b)], and peak bubble radius [Fig. 7(c)]. However, when we combined the values of all three parameters, a reasonable correlation emerged [Fig. 7(d)]. The combined parameter, P, accounts for (i) the total number of bubbles, which relates to the total number of bubble–cell interactions; (ii) the nanoparticle temperature, which is related to the bubble temperature that may drive the bio-effect interaction; and (iii) the bubble radius, which influences the lifetime of the bubble–cell interaction since larger bubbles contain more energy with a smaller surface-to-volume ratio and, therefore, dissipate their energy over a longer period of time.

It is notable, however, that P does not weigh the three energy transduction parameters equally. The correlation of bio-effects with the direct product of the three parameters was only fair (R² = 0.75). A much better correlation (R² = 0.94) was found when peak nanoparticle temperature (T) was raised to the power 1.9, number of bubbles (N) to the power 0.55, and peak bubble radius (R) to the power 0.3,

While we should not try to read too much mechanistic information into this equation or claim that its form is fully optimized, it does suggest that peak nanoparticle temperature is the most important parameter that correlates with bio-effects, whereas the number of bubbles and the bubble radius have progressively less significant impact. Further analysis and correlations can be found in Figs. S2–S10 in the supplementary material.

Nanoparticle-mediated photoporation has been explored as a novel molecular delivery platform in the past either to investigate the underlying mechanism^18,22,24,25 or to study the effects of varying operating conditions to attain optimum delivery conditions.^17,25–27 While these studies have varied laser operating parameters, previous work has not focused on nanoparticles’ role as energy transducers by looking for relationships between nanoparticles, energy absorption, and energy transduction parameters and the resulting bio-effects. This study was motivated by the need to understand the role of the energy transduction process from laser to nanoparticle to medium and cell.

Nanoparticle-mediated photoporation has three crucial energy transfer steps: (i) from laser to nanoparticle, (ii) from nanoparticle to the surrounding medium, and (iii) from medium to the cell membrane. Nanoparticles are directly involved in the first two of those energy transfer steps since they absorb the laser energy and transduce it into a thermal form that is then dissipated in thermal and mechanical forms from the nanoparticle to the surrounding medium. By using a variety of nanoparticle materials and shapes, we learned that bio-effects caused by nanoparticle-mediated photoporation are dependent on their absorbance values. In general, a lower laser light absorbance corresponded to smaller bio-effects and a higher laser light absorbance corresponded to greater bio-effects.

Because our initial analysis of light absorbance also included scattered light, we performed a second analysis that accounted for just the energy absorbed by the nanoparticles by measuring system temperature change. In general, higher temperatures were associated with greater bio-effects but a clear trend was not evident.

We next considered three hypotheses to explain the relationship between photoporation parameters and bio-effects. These hypotheses proposed that bio-effects would correlate with (i) total energy absorbed, (ii) energy absorbed per nanoparticle, or (iii) energy absorbed per nanoparticle mass. Total energy absorbed corresponds to the cumulative energy absorbed by all the nanoparticles and is available to be transferred into the surrounding fluid to cause photoporation. Energy absorbed per nanoparticle corresponds to the localized absorption and release of energy and, thus, controls bubble formation. Energy absorbed per nanoparticle mass corresponds to the energy absorbed per unit mass of CB nanoparticles and, thus, controls the temperature increase of the nanoparticles.

This study was designed to investigate if any one of these energy parameters is dominant and correlates strongly, by itself, with photoporation. These energy parameters are directly related to operating conditions that include both nanoparticle composition and laser fluence and are hypothesized to be mechanistically closer to the photoporation phenomenon. However, our study found that none of the three nanoparticle energy absorption parameters alone correlated with bio-effects, which prompted rejection of our three hypotheses regarding the existence of a single, dominant energy parameter.

Our final analysis addressed phenomena after the transduction of energy from laser light to heat to bubble formation by focusing on the energy transduction parameters that may be mechanistically closest to causing bio-effects. We found that total number of bubbles, bubble size, and peak nanoparticle temperature each correlated with bio-effects when the other parameters were held constant.

The generation of more bubbles implies more sources of thermal, acoustic, and fluid mechanical outputs that can effectively transfer energy to cell membranes to cause poration. Thermal output comes from the bubbles being hot (i.e., many hundreds and even thousands of degrees Celsius). Acoustic and fluid mechanical outputs come from the rapidly expanding bubble generated by heat transfer from the nanoparticle to the surrounding fluid. It is not surprising that more sources of energy output can increase the chances of a close encounter between a nanoparticle and a cell, which results in bio-effects. A similar dependence of cellular uptake and viability loss on cavitation bubble nucleation site concentration was observed for ultrasound-mediated intracellular delivery.²⁸ 

Increasing bubble size increased bio-effects. Bigger bubbles may bring bubbles closer to neighboring cells and may also generate greater acoustic and fluid mechanical effects. Bigger bubbles (at the same temperature) also have more energy that can impact cells and are released over a longer time due to reduced surface-to-volume ratio. Bubble size dependence has been previously investigated in cavitation-based intracellular delivery methods that showed greater bio-effects for in vitro applications^29,30 and better penetration depths for in vivo applications.³¹ 

Finally, higher peak nanoparticle temperature, which is related to peak bubble temperature, caused greater bio-effects. Higher peak nanoparticle temperature may impact cells by (i) increasing the rate of heat transfer from nanoparticle to the surrounding fluid, thus affecting acoustic wave production, and (ii) increasing the chances of direct heat transfer to cell membranes that could melt the lipids or otherwise disrupt lipid bilayer structures.

A plausible way to understand the role of transient photoporation parameters on bio-effects is depicted in Fig. 8. Essentially, bio-effects would be least for case I, when no bubbles are formed during laser exposure due to nanoparticle heating below 100 °C. Even in this case, there could be bio-effects due to direct thermal energy transfer from nanoparticle to cell membranes. Next, when the bubbles do form (case II), bio-effects can be increased by increasing the number of bubbles (case III), increasing bubble size (case IV), and increasing peak nanoparticle (bubble) temperature (case V).

Guided by these findings, we combined data from more than 100 different experimental conditions and found a correlation between bio-effects and the product of the three energy transduction parameters, P. It is notable that the best correlation involved a power-law relationship with bio-effects scaling with peak nanoparticle temperature approximately squared, and the total number of bubbles and peak bubble radius being raised to powers less than one. This correlation showed a Langmuir isotherm fit with bio-effects initially increasing with an increasing P value and then saturating around 100%. A higher power for peak nanoparticle temperature suggests that this parameter is a more important determinant of bio-effects compared to number and radius of bubbles. This correlation can be used for determining operating conditions in order to obtain a desired bio-effect (which indicates that different combinations of the three parameters can have similar cellular responses).

Note that this correlation does not account for the internal variations in bio-effects such as the ratios of uptake cells, non-viable cells, and fragmented cells, which place limitations on its ability to provide delivery efficiency (see Figs. S2–S5 in the supplementary material). Additionally, only data on calcein delivery (at a fixed concentration) were used to generate this correlation and the exposure time was kept constant at 1 min (i.e., 600 pulses). Therefore, the correlation may not be applicable to estimate bio-effects under conditions not included in this study.

In this work, we examined the dependence of cellular bio-effects caused by nanoparticle-mediated photoporation on a variety of parameters expected to influence outcomes. We first found that nanoparticle composition strongly affects bio-effects, where CB and carbon nanotubes were much more effective than graphite, diamond, or non-carbon materials. These differences were partially explained by different laser absorption properties. Energy absorption by nanoparticles was characterized as total energy absorbed, energy absorbed per nanoparticle, and energy absorbed per nanoparticle mass but none of these parameters were found to dominate the energy transduction process leading to bio-effects.

We finally looked at energy transduction parameters more closely related to the energy form transferred to cells in terms of the temperature, size, and number of vapor bubbles generated by the nanoparticles heated by the laser. These three parameters each individually correlated with bio-effects when the other two were held constant. An overall correlation with data from more than 100 different experimental conditions was obtained by accounting for the number of bubbles, size of bubbles, and peak nanoparticle temperature with the best correlation coming from a power-law relationship that indicated that peak nanoparticle temperature was the strongest determinant of bio-effects among the three energy transduction parameters.

This study provides a better understanding of the physical parameters at play during nanoparticle-mediated photoporation, how they influence bio-effect outcomes, and which ones are most important. This understanding also facilitates the design of photoporation parameters that achieve a desired bio-effect.

See the supplementary material for energy transduction parameters calculations and additional bio-effects correlation plots.

We acknowledge the National Science Foundation (NSF) (Contract No. 1510028) for funding this work. We also acknowledge Donna Bondy for administrative support and Petit Institute Cellular Analysis Core Facilities for their services and shared resources.

The data that support the findings of this study are available from the corresponding author upon reasonable request.