Light triggered nanoscale biolistics for efficient intracellular delivery of functional macromolecules in mammalian cells

Design, characterization, and light-triggered activation of NBs

The NB design consists of a thermal core particle that is surrounded by smaller nanoparticles that act as nanoprojectiles upon light-triggered activation (Fig. 2a). For the NB’s core we used commercially available 0.5 µm magnetic beads, consisting of a polymeric core that is surrounded by a layer of iron-oxide and coated with a covalently attached streptavidin monolayer. We will further refer to these particles as iron oxide nanoparticles (IONP), as the iron oxide shell is the photothermal component needed for VB generation²⁰. Biotinylated fluorescent polystyrene nanospheres (100, 200, and 500 nm) were initially selected as nanoprojectiles, as they can be conveniently detected with fluorescence techniques (e.g., confocal microscopy). According to the projectile size, we will refer to those NBs as 100-NBs, 200-NBs, and 500-NBs, respectively. Self-assembly of the NBs was achieved by simply mixing the streptavidin functionalized IONPs with the biotinylated nanospheres (ratio IONP/biotin-nanoprojectiles = 1/1750) overnight under constant stirring, after which they were purified by magnetic isolation. This IONP/nanoprojectile ratio was chosen to have a sufficient excess of 100 nm nanoprojectiles to cover the entire surface of IONP core particles. The same ratio was used for larger nanoprojectile sizes as well for the same reason. Figure 2b shows the DLS characterization of the IONP core particles, 200 nm polystyrene beads and the assembled 200-NBs. A considerable increase in hydrodynamic size was indeed observed upon NB formation. The zeta potential of the 200-NBs became slightly less negative compared to the IONP core particles, which is another indication of the attachment of biotinylated nanoprojectiles. Similar measurements for 100-NBs and 500-NBs are shown in Supplementary Fig. 1a. Figure 2c shows representative SEM images of 200-NBs, further confirming successful NB formation with polystyrene nanospheres being clearly visible around the IONP surface. Image analysis showed an average NB size of 1206 ± 86 nm, quite close to the DLS result. After synthesis the NBs remained stable in PBS buffer for at least three weeks (Supplementary Fig. 1b).

Following successful NB synthesis, we evaluated the formation of laser-induced VB, which is expected to propel the attached nanoprojectiles away from the core particles. VB can be triggered from photothermal nanoparticles upon irradiation with a short laser pulse (typically <10 ns) of sufficient fluence²¹. The VB threshold is commonly defined as the fluence level (J/cm²) at which 90% of the irradiated particles generate a VB. This threshold can be determined experimentally by detecting and quantifying the formation of VBs using dark-field microscopy as a function of the applied laser fluence (Supplementary Fig. 1c)^21,22. During their lifetime, VB strongly scatter the incident light from the dark-field microscope’s light source, resulting in localized bright flashes of light soon after arrival of the laser pulse. The number of detected VB will gradually increase with increasing laser pulse fluence, until all particles in the irradiated area form VB. This results in the typical sigmoidal curve as shown in Fig. 2d for 200-NBs (7 ns laser pulse, λ = 561 nm). After fitting of a Boltzmann curve, the VB threshold was found to be 1.22 J/cm² for the 200-NBs. Similar curves can be found for 100-NBs and 500-NBs in Supplementary Fig. 1di, ii, from which the VB threshold was determined as 1.14 J/cm² and 1.26 J/cm², respectively. The fact that these VB thresholds are similar for all NBs is to be expected since they are all based on the same 0.5 µm photothermal core particle. In addition to ns laser pulses, we also confirmed VB formation upon irradiation with ps laser pulses at the same wavelength (2 ps laser pulse, λ = 561 nm). In this case, the VB threshold for 200-NBs was 0.49 J/cm² (Supplementary Fig. 1diii). Not surprisingly, this is lower than for ns laser pulses as ps laser pulses provide more efficient energy transfer into photothermal nanoparticles²³.

Interestingly, successful propelling of projectiles after NB activation could be visualized by darkfield imaging as well. Figure 2e shows such dark field images of 200-NBs in water before (t₀ = 0 s) and after (t₁ = 10 µs) application of a single laser pulse that was directed to the NB indicated by the orange arrow. Apart from intensely scattered light by the VB, nanoprojectiles (green arrows) can be seen as well as they are propelled away from the nanobomb over a distance of a few 10 s of µm.

SEM images of 100-NBs, 200-NBs, and 500-NBs before and after laser irradiation at the VB threshold are shown in Supplementary Fig. 2a. After laser irradiation, the IONP core particles are found to be highly deformed, likely as a consequence of the high temperatures that are reached within the particles upon laser irradiation in combination with the mechanical stress induced by VB expansion and collapse. The nanoprojectiles on the other hand seem to remain mostly intact.

To quantify the degree of nanoprojectile release, 200-NBs were irradiated (7 ns, λ = 561 nm, 1.22 J/cm²), and the released beads were separated from the IONP core particles by a magnetic wash. The percentage of released nanoprojectiles was quantified by fluorimetry based on a previously determined calibration curve (Supplementary Fig. 1bi). As a positive control, we included samples incubated with 10% trypsin, which is an enzyme able to degrade streptavidin and is expected to release all nanoprojectiles. As can be seen in Fig. 2f, almost 80% of the nanoprojectiles were successfully released upon laser activation of the 200-NBs. Similar results were obtained for 100-NBs and 500-NBs (Supplementary Fig. 1bii–iv).

To evaluate if the propelled nanoprojectiles can penetrate a physical barrier, their penetration was investigated in a tissue phantom gel made of agarose (0.5 % w/v) with mechanical properties comparable to biological tissue¹⁷. As shown in the schematic representation of Fig. 2g, the gel was fluorescently labeled by incorporation of FITC-Dextran 10 kDa (FD10) into the gel matrix during solidification, allowing visualization by confocal microscopy. As a control, 200-NBs (1.3 × 10⁸ NPs/mL) were added to the solidified gel and incubated for 1 h, after which they remained mostly on top of the gel (Fig. 2h). Next, when activating NBs after just 5 min incubation by irradiation at the VB threshold (Fig. 2i), nanoprojectiles penetrated up to 15 µm into the gel matrix with an average penetration distance of 8 ± 3 µm. Since there is virtually no heat transfer into the surrounding upon VB formation^1,2, this observation cannot be attributed to mere heating and melting of the agarose gel. To verify this more explicitly, we proceeded to irradiate the NBs after 5 min incubation at a laser pulse fluence of 10% of the VB threshold (0.37 J/cm²). In that case almost no VB can be formed and all heat from the NBs will diffuse into the gel. As can be seen in Fig. 2j, the fluorescent nanobeads attached to the surface of NBs did not penetrate into the gel, providing further confirmation that bead penetration as observed in Fig. 2i is a consequence of the mechanical forces induced by VBs and not just because of thermal heating and melting of the agarose gel. In addition, to verify that penetration of beads was not merely due to passive diffusion of released nanoprojectiles into the gel, 200 nm biotinylated beads (1% v/v) were incubated for 1 h on the gel. Since they remained at the gel surface (Fig. 2k) this confirms that penetration into the gel must have happened by an active force exerted by the light-triggered NBs. The experiment was repeated for 500-NBs as well, with virtually identical results (Supplementary Fig. 2d). Also beads released from 100-NBs were found to penetrate into the gel upon VB generation (Supplementary Fig. 1ci), although in this case results are confounded by the fact that those small nanoparticles could diffuse passively into the gel as well (Supplementary Fig. 2cii). Nevertheless, together these results confirm the ability of light-triggered NBs to propel their nanoprojectiles across a physical barrier.

Cell membrane penetration by light-triggered NBs

Next, we evaluated to which extent nanoprojectiles ejected from laser-activated NBs can penetrate cell membranes. 200-NBs were added to HeLa cells (1.3 × 10⁸ NBs/mL) and immediately irradiated with laser pulses at the VB threshold. 3D confocal microscopy was used to determine the location of the released nanoprojectiles. For this, we stained the cytoplasm with CellTracker Deep Red and used green fluorescent beads as projectiles. Figure 3a shows an exemplary horizontal confocal section of HeLa cells after activation of the 200-NBs. Different vertical sections are shown as well along the y (i and ii) and x (iii and iv) directions. Beads were primarily found in the cytoplasmic stained regions or at the top of the cells, as indicated by the arrows in the z-projections (Fig. 3ai–iv). To study this more quantitatively on a large number of cells, entry of fluorescent nanoprojectiles in cells was evaluated by flow cytometry at different laser pulse fluences (0.5×, 1× and 1.5× VB threshold) for a NB concentration of 1.3 × 10⁸ NBs/mL. Figure 3b shows the percentage of cells that were positive for the presence of fluorescent nanoprojectiles. A clear increase in the percentage of cells positive for nanoprojectiles was observed for the two highest laser fluences (1× and 1.5× VB threshold), while for the lowest laser fluence, which is below the VB threshold, this was not the case. At a laser fluence of half the VB threshold fewer VB are formed which are also smaller in size and, therefore, less powerful. It is likely due to this combined effect that hardly any cells are positive for beads below the VB threshold. Upon 5 min incubation with NBs (1.3 × 10⁸ NBs/mL) without laser irradiation, no noticeable fluorescence could be detected in the cells, showing that there is no spontaneous association to or uptake of NBs by cells. As an additional control, cells were incubated for 5 min with pristine IONP core particles (1.3 × 10⁸ NPs/mL) together with an excess of free nanoprojectiles (ratio beads:IONP 300) and irradiated with laser light. In this case a slight increase in the number of cells positive for nanoprojectiles was observed. As it was virtually independent of the applied laser fluence, likely this stems from spontaneous binding and uptake of free nanoprojectiles by cells. Together this shows that (1) VB formation is essential to propel the nanoprojectiles, and (2) fully assembled NBs are needed to induce cell membrane permeabilization and not just a mixture of its components.

Additionally, we investigated to which extent (fragments of) IONP enter cells. For this purpose, inductively coupled plasma-mass spectrometry (ICP-MS) was used to quantify the iron content of cells before and after NB activation. HeLa cells were exposed to 200-NBs and irradiated at the VB threshold fluence. Next, as schematically shown in Fig. 3ci, cells were washed with PBS to remove any remaining extracellular IONP. Following trypsinization of the cells and digestion with aqua regia (3:1 mixture of hydrochloric acid and nitric acid), the iron content was determined using ICP-MS and normalized to the total protein content, as determined by BSA quantification assay (Supplementary Fig. 3), which is a measure for the total number of cells. As a negative control, we included untreated cells, and as a positive control, we included cells with their cell medium after incubation with NBs (i.e., without PBS washing step so that all added NBs are still present). While the positive control showed a much higher Fe content compared to untreated cells, there was no significant increase for the NB-treated cells (Fig. 3cii), showing that IONP fragments do not (substantially) enter cells upon laser activation.

Finally, to confirm that nanoprojectiles from activated NBs effectively cause membrane disruption, we evaluated if a cell-impermeable molecule can enter cells after light-triggered NB treatment. For this, we used propidium iodide (PI) which becomes more fluorescent when entering the cytosol by interacting with intracellular nucleic acids. As a control, cells were irradiated after incubating with IONP core particles (1.3 × 10⁸ NPs/mL) mixed with free nanoprojectiles (1% v/v). As can be seen in a representative confocal image in Fig. 3d, this resulted in only a few cells that became permeable to PI (yellow). This is according to expectation since we already observed before that a simple mixture of uncoupled core particles and projectiles did not cause much penetration of projectiles in cells (cfr. Fig. 3b). In contrast, when cells were treated with intact NBs, the number of PI positive cells increased tremendously. The high magnification image in Fig. 3d shows that cells positive for PI (yellow) also show the presence of fluorescent beads (green; highlighted with white arrows). Vice versa, cells negative for PI had no or very few nanoprojectiles associated with them, clearly establishing the link between PI influx and the presence of nanoprojectiles at the cell membrane.

The experimental data so far suggest that laser activation of NBs above the VB threshold can propel the nanoprojectiles over a distance in space and cause membrane permeabilization in nearby cells. Yet, as explained in the Theory and Simulations section of the Supplementary Information, directed long-range movement of nanoparticles through a viscous medium is far from evident given the large drag force that they experience. Indeed, even if a 200 nm NP is ejected with a substantial initial velocity of 10 m/s, it would come to a halt within a few 10 s of nanometers (Supplementary Movies 1 and 2), even for high density particles (Supplementary Movies 5 and 6). As our numerical simulations in the Supplementary Information show, long distance displacement of NPs requires the presence of a persistent force that carries the particles along. In particular, our simulations show that the force should be present at least until the particle reaches the cell membrane (Supplementary Movies 3 and 7), and even slightly beyond (Supplementary Movies 4 and 8), in order for the particle to effectively penetrate through the cell membrane. As it has been described before that fluid flows can emerge around oscillating and collapsing bubbles in fluids²⁴, we therefore hypothesize that long-range displacement of the ejected nanoprojectiles from our nanobombs is enabled by such fluid flows that emerge around the nanobombs upon optical activation and VB formation and which carry along the nanoprojectiles over long distances. It is important to note that the fluid flows by themselves are apparently insufficient to penetrate the cell membrane, since our control experiments with mixtures of core particles and bullets did not lead to PI influx upon laser irradiation (and hence VB formation from the core particles). Therefore, we conclude that the observed membrane penetration is effectively caused by the ballistic impact of ejected nanoprojectiles, which are dragged along in the generated fluid streams around the nanobombs. To understand this better it will be of interest in future work to investigate theoretically and experimentally the nanobomb mechanism in more detail.

Intracellular delivery in adherent and suspension cells

Next, we proceeded to evaluate the extent to which NBs are capable of mediating intracellular delivery of large macromolecules. As a model molecule we selected FITC dextran 500 kDa (FD500) for first experiments. Intracellular delivery efficiency was quantified by flow cytometry, while cell viability was determined in parallel using the CellTiter-Glo metabolic assay. Initial experiments were performed on HeLa cells, which is an often-used model cell line for delivery studies. Figure 4a shows a schematic overview of these experiments: (i) cells are co-incubated with cargo molecules (FD500) and NBs for a certain period of time in optiMEM, after which the NBs are activated by scanning of the laser beam over the cell sample; (ii) FD500 will enter into the cytosol after penetration of the nanoprojectiles through the cell membrane; (iii) 5 min later, which should be sufficient for cells to repair the inflicted membrane damage⁴, cells are washed with PBS and supplemented with fresh cell medium for further analysis. Laser activation was performed with the ps irradiation set-up, which is equipped with galvo-mirrors for fast scanning of the laser beam, allowing fast evaluation of several experimental conditions. Starting with 200-NBs, the effect of NB concentration was evaluated. Figure 4bi shows the results in case laser irradiation was immediately applied after adding FD500 and NBs to the cell medium (0 min incubation). Both the percentage of positive cells and the amount of FD500 delivered per cell (rMFI) gradually increased for an increasing NB concentration, while cell viability gradually decreased. At the highest concentration of 1.3 × 10⁸ NBs/mL, 47% FD500 positive cells were obtained for a cell viability of 75%. As a control we used uncoupled IONP core particles at the highest concentration tested, mixed with polystyrene beads (1% v/v). In this case only 13% positive cells were obtained, showing once more that the NB effect is not simply due to the sum of its parts, but requires fully assembled NBs in order to work optimally. Next, for a fixed concentration of 1.3 × 10⁸ NBs/mL, we evaluated the effect of the nanoprojectile size by comparing NBs synthetized with 100 nm (100-NBs), 200 nm (200-NBs) or 500 nm (500-NBs) polystyrene beads (Fig. 4bii). While again virtually no delivery was seen for the controls (IONPs mixed with free polystyrene beads), an increasing trend in delivery efficiency was seen for intact NBs as a function of the nanoprojectile size, which went hand by hand with a decrease in cell viability. However, as cell viability decreased to 47% for 500-NBs, while for 100-NBs and 200-NBs it remained above 75%, we opted to continue with 200-NBs for further experiments in which we tested the influence of NB incubation time (0–20 min) before applying laser irradiation. As shown in Fig. 4biii, both the percentage of positive cells and their rMFI increased by extending the incubation time to 10 min, with no further improvement for 20 min incubation. However, cell viability dropped to below 50% starting from 10 min incubation, so that 5 min incubation was selected as the best condition to continue with. The reason why cell viability decreases with increasing laser fluence is due to more NBs being in close contact with cells or even being endocytosed by cells, leading to gradually more cell damage. Note that for the optimization of these different experimental parameters (NB design, incubation, time, concentration, etc.), delivery efficiency and cell viability was determined 2 h after irradiation. Based on these results, for further experiments on HeLa cells it was decided to continue with 200-NBs at a concentration of 1.3 × 10⁸ NBs/mL, and 5 min incubation time. Next, we confirmed that the delivery efficiency and cell viability does not depend on the laser pulse duration as long as irradiation at the VB threshold is applied (7 ns pulse: VB threshold = 1.22 J/cm², 2 ps pulse: VB threshold = 0.47 J/cm²). Since both lasers operate at λ = 561 nm, we additionally performed experiments on a laser irradiation set-up equipped with a 532 nm 5 ns pulsed laser. As this laser set-up was not equipped with dark field imaging (needed to determine the VB threshold), we empirically determined that a laser fluence of 0.84 J/cm² yields similar results as for the other two lasers. The results in Fig. 4biv show equal intracellular delivery efficiency and similar effects on cell viability in all cases, demonstrating the versatility of the NBs with respect to the type of pulsed laser used to activate them.

Next, we evaluated the delivery efficiency as a function of the molecular weight of the cargo molecules. To do this we performed delivery experiments in HeLa cells using the above optimized conditions with FITC-dextran of 10, 150 and 500 kDa (FD10, FD150, FD500) all at the same effective concentration of 2 mg/mL. As shown in Fig. 4c, while the percentage of positive cells only increases slightly for increasing molecular weight, the rMFI decreases more steeply. This can be expected since larger molecules diffuse more slowly, so that fewer of them can enter into the cells while the pores are open.

To measure the time frame that cells need to repair the pores created by NBs, we performed a FD500 delivery experiment by adding the cargo at different timepoints (0, 1, 2, 3, 5, and 7 min) after irradiation of HeLa cells in the presence of 200-NBs (using previously optimized conditions). As shown in Supplementary Fig. 4a, FD500 delivery efficiency decreases as it is added longer after cell membrane permeabilization as a result of membrane resealing. After 3 min, there was no significant difference in FD500 influx any more compared to the non-treated control cells, showing that pores reseal within this timeframe. For traditional photoporation, we reported before that pores formed on HeLa cells reseal in 1 min when using FD500 as a marker²⁵. The fact that it is longer for pores generated by NBs could be due to the pores being larger in size, requiring more time to repair.

Next, we proceeded to evaluate the effect of membrane fluidity on the delivery efficiency after NB activation. It is well known that cholesterol plays a crucial role in cell membrane tension regulation and influences mechanical parameters like bending rigidity and elastic modulus²⁶. In previous work by Biswas et al., it was demonstrated that biochemical agents like methyl-beta-cyclodextrin (CD) can be used for depleting cholesterol in HeLa cells, leading to a direct increase in the membrane tension, and thus, a decrease in the membrane fluidity²⁷. In order to get insight regarding the possible impact of membrane fluidity on pore formation by NBs, we pre-incubated HeLa cells for 1 h with CD at a concentration of 4 mg/mL (determined based on toxicity studies), followed by FD500 delivery with NBs according to the optimized conditions discussed above. The results in Supplementary Fig. 4b show that decreasing membrane fluidity did not influence the effective delivery efficiency, while it did negatively affect cell viability.

Finally, we proceeded to evaluate if nanoprojectile density has any impact on cell membrane permeabilization. For this, we prepared 200-NBs with higher density nanoprojectiles prepared from poly(D,L-lactide-co-glycolide (PLGA, ρ_PLGA = 1.3 g/cm³) and TiO₂, ρ_TiO2 = 4.23 g/cm³), which have ~1.3× and 4× higher mass density as compared to polystyrene, respectively. The hydrodynamic size and zeta potential of those NBs and their building blocks are shown in Supplementary Fig. 5a and b. Being a metal-oxide, one may wonder if TiO₂ NPs themselves may form VBs, similar to the IONP core particles. While this is indeed the case (Supplementary Fig. 5c), their VB threshold (2.58 J/cm²) is about 2.5× higher than for the NBs (1.22 J/cm²). As indicated by the dashed vertical line, at a fluence of 1.22 J/cm² TiO₂ NPs do not form VB, so that any observed effects are due to NB activation, and not side-effects due to VB formation from TiO₂ NPs. When we used those three types of NBs for delivery of FD500 in HeLa cells, no significant difference was found in delivery efficiency (Fig. 4d). Therefore, the here proposed NBs can work independent of the mass density of the nanoprojectiles used, offering great flexibility in their composition. Similar results were obtained by the numerical simulations presented and discussed in the supporting information.

Finally, after evaluation of we verified if light-triggered nanobombs can be used for intracellular delivery in suspension cells as well. For this, we used Jurkat suspension cells, which is an immortalized cell line of human T lymphocytes and a widely used model for hard-to-transfect primary human T cells²². Similar to HeLa cells, we delivered FD500 using 200-NBs with polystyrene beads as nanoprojectiles. Keeping the concentration of NBs the same as for HeLa cells (1.3 × 10⁸ NBs/mL), we determined what is the optimal incubation time. As shown in Supplementary Fig. 6a, the delivery efficiency and rMFI increased gradually for longer incubation times, reaching a FD500 delivery efficiency of 49% at 78% viability after 20 min, which again is much better than what is achieved when unassembled IONP and 200 nm beads are used as a control (Supplementary Fig. 6b). We also checked again for this condition to which extent IONP (fragments) may be present in those cells. Supplementary Fig. 6c shows that again a small but non-significant increase was observed for the NB-treated cells, showing that the IONP core particles do not (substantially) enter the cells upon laser activation, similar to HeLa cells (Fig. 3c). Together, these results confirm that the NB system proposed here can be equally applied to suspension or adherent cells, testifying to the flexibility and broad usability of this system.

It is interesting to see that the optimal NB incubation time for HeLa cells is 5 min, while it is 20 min for Jurkat cells. In case of adherent HeLa cells, when NBs are added to the cell medium they will gradually sediment on top of the cells, ensuring good contact with cells. Instead, in case of Jurkat suspension cells the NBs are mixed with the cell suspension which is a more dynamic system which may reduce the probability of NBs associating tightly with the cells. To get a better view on this we prepared NBs with fluorescently labeled nanoprojectiles and measured the level of cell-associated fluorescence as a function of the incubation time. As can be seen in Supplementary Fig. 7, NBs associate indeed more quickly with HeLa’s as compared to Jurkat cells. Upon laser irradiation at the VB threshold, it can be seen for HeLa’s that the nanoprojectiles are associated with the cells already after 5 min, while for Jurkat cells 20 min incubation was needed, in line with the optimal incubation times based on the delivery experiments.

mRNA and pDNA transfection in adherent and suspension cells

The results so far have shown that light-triggered NBs can be used for intracellular delivery of macromolecular compounds in adherent as well as suspension cells. Encouraged by these results we proceeded to evaluate functional delivery of large nucleic acids, including mRNA and pDNA. In particular we used mRNA (996 nucleotides) and pDNA (5757 base pairs) encoding for enhanced green fluorescent protein (eGFP), always at a concentration of 0.1 μg/μL unless specified otherwise. Rather than using fluorescent polystyrene beads, we made use of 200-NBs with PLGA NPs as nanoprojectiles since PLGA has excellent biocompatibility and biodegradability and, therefore, was considered to be a more relevant system to continue with²⁸.

A first question is if NBs may perhaps damage such delicate biological molecules upon activation by laser irradiation? To test this, a mixture was prepared of pDNA (0.1 μg/μL) and NBs (1.3 × 10⁸ NBs/mL) and irradiated with increasing laser fluences. pDNA integrity was subsequently analyzed by gel electrophoresis. As can be seen in Supplementary Fig. 8, no degradation products of pDNA were observed, confirming that even at the highest laser fluences there is no noticeable damage to pDNA. For the highest laser fluences (>0.86 J/cm²) it was noted that the bands corresponding to nicked and linear pDNA became fused, which may indicate that VB generation can induce a conformational change of pDNA. However, this had no apparent influence on transfection efficiency.

Next we proceeded with cell transfections. Conditions were used as optimized for FD500 delivery (1.3 × 10⁸ NBs/mL, 5 min incubation for HeLa cells and 20 min incubation for Jurkat cells). Here we used the laser irradiation set-up with 5 ns laser (fluence of 0.84 J/cm²) which is also equipped with a galvo scanning mirror for fast scanning of the laser beam, thus offering higher throughput (10⁴–10⁵ cells/s). The read-out of eGFP expression by flow cytometry or microscopy was performed 24 h after transfection at which point cell viability was measured with the CellTiter-Glo assay as well. Figure 5ai shows the results for the transfection of HeLa cells with eGFP-mRNA. Flow cytometry results indicated 48% eGFP positive cells with a viability of 70%. Moreover, the percentage of transfected cells could be further increased to 61% when using 0.3 μg/μL mRNA, in which case also the rMFI values increased by more than twofold. Figure 5aii shows a large field-of-view confocal image of mRNA transfected cells in a well from a 96 titer plate, as well as a zoomed-in section showing both the transmission and fluorescence channel. Figure 5aiii, finally, shows the results after transfection with pDNA, in which case 22% of the cells were positive for eGFP with a cell viability of 70%. While increasing the pDNA concentration in this case did not translate in an increase in the number of eGFP+ cells, it did result in a more than twofold increase in rMFI.

Performing similar experiments on Jurkat cells revealed that mRNA transfections resulted in 32% eGFP+ cells with a cell viability of 78% (Fig. 5bi). As this means there is still a fairly large fraction of living untransfected cells, we also tried exposing the cells two times to the transfection procedure. This increased the percentage of cells expressing GFP up to 53%, which can be visually appreciated from the confocal images in Fig. 5bii, with only a small decrease of the percentage of viable cells (68%). For comparison we transfected Jurkat cells with mRNA using electroporation as the most explored non-viral transfection method for engineering of hard-to-transfect T-cells⁴. Electroporation of Jurkat cells was performed with a 4D Nucleofector (Lonza) using the protocol recommended by the manufacturer (Pulse code: CL-120; SE cell line solution). As frequently reported^6,29,30, electroporation resulted in a drastic loss of cell viability with only about 10% of the cells surviving the treatment (measured after 24 h by CellTiter-Glo). Most of the surviving cells were nevertheless transfected (~80%), with a level of expression per cell (rMFI) similar to cells treated by NBs. We summarized these results in Fig. 5biii in which we show the percentage of non-viable cells (gray), the percentage of viable untransfected cells (blue) and the percentage of viable transfected cells (green). The latter is often referred to as the cell transfection yield, and is calculated from multiplying the cell viability (which is calculated relative to the starting amount of cells) with the percentage of transfected cells as determined by flow cytometry (which is gated on living cells). This clearly shows that the mRNA transfection yield for NBs (25%) is markedly higher as compared to electroporation (4%). For the samples that were transfected twice with NBs, the transfection yield even reached 36%. This represents a six-fold increase in transfected cell yield for NBs as compared to electroporation, and a nine-fold increase when the treatment is repeated twice. Note that in this study we asses cell viability making use of an ATP detection assay (i.e., CellTIter-Glo). As it measures metabolic activity it is a more sensitive measure for the health status of a cell instead of a simple live/dead staining which only indicates if a cell is dead or alive. Highly stressed cells may be found to be alive, but can have reduced metabolic activity and, therefore, an altered functionality. This becomes clear in Supplementary Fig. 9, in which we compare the viability determined by live-dead staining using DAPI with the one determined by CellTiter-Glo on Jurkat cells transfected with eGFP-mRNA by electroporation. As can be seen in the figure, after treatment with electroporation, 60% of the cells become positive for GFP expression with a viability of 25% according to live-dead staining and quantification by flow cytometry, similarly as previous reports³¹. Nevertheless, when measuring cell viability based on an ATP metabolic assay (CellTiter-Glo), the observed ‘viability’ was clearly lower. It shows that many of the viable cells according to a simple live/dead stain are actually highly stressed and not very healthy, a finding which has been reported before for electroporated cells^6,32.

Finally, we also performed pDNA transfections in Jurkat cells with light-triggered NBs (Fig. 5biv), leading to 20% eGFP+ cells with a cell viability of 71%. Electroporation, on the other hand, resulted in 60% eGFP+ cells but with a cell viability of only 6%. The corresponding transfection yield values show that 19% living pDNA transfected cells are obtained with NBs, which is 7.6x more than electroporation (Fig. 5bv).

From a procedural point of view, transfecting cells with laser-activated nanobombs is similar to nanoparticle-sensitized photoporation^{9,23,33,34,35,36}. In photoporation cells are incubated with photothermal nanoparticles, mostly AuNPs of ~70 nm, which associate with the cell membrane. Upon applying laser irradiation of sufficient intensity, VB are formed from the AuNPs whose mechanical force directly creates pores in the cell membrane. However, so far it has proven challenging for this ‘traditional’ type of photoporation to efficiently transfect cells with large nucleic acids like mRNA and pDNA. To see if our nanobombs form an improvement compared to AuNP sensitized photoporation, we performed comparative photoporation experiments on HeLa and Jurkat cells according to conditions that we optimized and published recently³⁰. On HeLa’s, photoporation resulted in only 23% transfected cells for mRNA (Supplementary Fig. 10a), similar to previously published results³⁰, which is about half the efficiency of what we obtained with nanobombs. For pDNA photoporation with AuNPs performed even worse, resulting in a negligible amount of only ∼2% eGFP+ cells, while this was about 10x better with nanobombs. On Jurkats mRNA transfections with photoporation resulted in 15% eGFP+ cells (Supplementary Fig. 10b), again only about half the efficiency of nanobombs. For pDNA photoporation only gave 3% eGFP+ cells, while it was 20% with nanobombs. Together this clearly shows the superiority of nanobombs over photoporation with 70 nm AuNPs as the most commonly used photothermal sensitizers.

Evaluation of spatially resolved mRNA transfections

Since activation of NBs happens by scanning the cell sample with the pulsed laser, it should be possible to transfect cells in a spatially controlled manner by scanning the laser beam according to a pre-defined pattern (Fig. 6a). The confocal image in Fig. 6b shows that cells could indeed be transfected with eGFP-mRNA according to a pattern resembling the letter J (visualized 24 h after treatment). Following this successful proof-of-concept experiment, we proceeded to apply this possibility to a more complex system: cell-selective reprogramming by delivering mRNA encoding for the site-specific recombinase “CRE” in a reporter cell line³⁷. For this, HeLa cells were previously transduced with a lentiviral construct containing the eFS-LoxP-DsRed Express II-rev(eGFP)-PxoL. This system presents an easy readout, as untransfected cells will only express the red fluorescent protein DsRed Express II, whereas in cells transfected with CRE mRNA the DsRed Express II-stop cassette will be inverted between the LoxP sites, hence switching on the expression of eGFP (Fig. 6ci). A suitable time point for reading out the conversion of DsRed Express II to eGFP expression was determined to be 72 h after treatment (Supplementary Fig. 11). Spatial-selective cell reprogramming was evaluated by confocal microscopy (Fig. 6cii). Three channels were recorded, being Hoechst (nuclear stain, blue), DsRed Express II (red), and eGFP (green). The first row corresponds to a control where cells are incubated with CRE mRNA (0.1 µg/µL) and NBs but without laser irradiation. As expected, only red staining of the cell’s cytoplasm can be seen corresponding to DsRed Express II expression. Instead, when cells are irradiated in the presence of NBs and CRE-mRNA, expression of CRE recombinase induces eGFP expression according to the “J” pattern. Together these experiments confirm that light-triggered NBs can be used for delivering compounds into cells in a spatially resolved manner by patterned scanning of the laser beam.