|Home | About | Journals | Submit | Contact Us | Français|
The precise perturbation of gene circuits and the direct observation of signaling pathways in living cells are essential for both fundamental biology and translational medicine. Current optogenetic technology offers a new paradigm of optical control for cells; however, this technology relies on permanent genomic modifications with light-responsive genes, thus limiting dynamic reconfiguration of gene circuits. Here, we report precise control of perturbation and reconfiguration of gene circuits in living cells by optically addressable siRNA-Au nanoantennas. The siRNA-Au nanoantennas fulfill dual functions as selectively addressable optical receivers and biomolecular emitters of small interfering RNA (siRNA). Using siRNA-Au nanoantennas as optical inputs to existing circuit connections, photonic gene circuits are constructed in living cells. We show that photonic gene circuits are modular, enabling sub-circuits to be combined on-demand. Photonic gene circuits open new avenues for engineering functional gene circuits useful for fundamental bioscience, bioengineering, and medical applications.
Gene circuits, consisting of interconnected genes and proteins, give rise to cell functions essential in signaling, growth, differentiation, and apoptosis. Despite some initial success, many important gene circuits remain incompletely understood. Increased understanding of information processing and operation of gene circuits should advance therapeutic strategies for reconfiguring gene circuits involved in disease progression and cancer. A major challenge is probing native gene circuits with high signal fidelity. Top-down probing approaches utilize an external input signal and essentially treat a living cell as an input-output “black box.” Ideally, the output signal should be directly correlated to the input signal. However, signal distortions – time delays, noise, and signal magnitude reductions – can confound this input-output relationship,1 hindering temporally precise modulation of gene circuits and limiting dynamic reconfiguration of gene circuits in living cells.
Intracellular approaches that enable temporally precise modulation of native gene circuits and dynamic reconfiguration of existing circuit connections will have widespread applications in fundamental biology and translational medicine. Optical input signals are a promising interface to gene circuit connections. To date, current optogenetic methods to interface living cells have so far relied on genomic modifications (i.e., mutations, overexpression) to permanently encode living cells with light-responsive genes,2 limiting on-demand reconfiguration of gene circuits in living cells. To enable on-demand circuit reconfiguration, remotely controlled inputs to gene circuits are required. Previously, remote control of biomolecules has been demonstrated by inductive coupling of radio frequencies to nanocrystal antennas.3 Recently, remote optical control of biomolecules has been demonstrated by coupling optical frequencies to resonant optical nanoantennas.4–15 By receiving and focusing freely propagating optical radiation into localized energy, resonant optical nanoantennas efficiently convert absorbed optical energy into surface-localized heat, otherwise known as the photothermal effect.16–21 When biomolecules are functionalized to the surface of the nanoantenna, this optically generated, surface-localized heat can be used to liberate (i.e., emit) biomolecules from the nanoantenna. Inside of living cells, optical silencing of genes (Fig. S9) has been previously demonstrated by liberating interfering oligonucleotides from nanoantennas using optically generated, surface-localized heat.4, 6, 7, 11 However, the optical gene silencing mechanism has not yet been utilized, in the context of gene circuits, to dynamically reconfigure existing gene circuits and optically engineer new gene circuit configurations.
Here, on-demand optical circuit reconfiguration is enabled by optically addressable siRNA-Au nanoantennas functioning as optical receivers and biomolecular emitters of siRNA via the optical gene silencing mechanism. In this manner, siRNA-Au nanoantennas serve as optical inputs to existing circuit connections (Fig. 1a), forming photonic gene circuits (Fig 1c, 1d, 1e). Photonic gene circuits are modular, allowing sub-circuits to be combined on-demand. This work represents a new way to engineer functional gene circuits, as it allows, for the first time, temporally precise modulation and dynamic reconfiguration of native gene circuits.
Among the various nanoantenna structures of current widespread interest, a rod-shaped Au nanoantenna (i.e., Au nanorod) was selected here (Fig. 1b i) because of its large optical absorption cross-section, narrow spectral bandwidth of the longitudinal plasmon resonance band, and tunable longitudinal plasmon resonance wavelength in the near infrared (NIR) regime, where cells are essentially transparent.22 The optical properties of an Au nanorod are dominated by its plasmon resonance. At resonance, the optical near-field shows two regions of high field intensity (Fig. 1b i) due to charge accumulation at each end of the nanoantenna.23 These regions of high field intensity indicate that the nanoantenna functions as an optical receiver to focus freely propagating optical fields down to nanometer-scale dimensions.24 This antenna effect is prominent at the resonance state of the nanoantenna, where the excitation wavelength is matched with the plasmon resonance wavelength. In Figure 1b i, this antenna effect is seen to be “on” at the resonance state and “off” at the non-resonance state of a nanoantenna, where the resonance and non-resonance states depend on aspect ratio. We exploited this antenna effect to design nanoantennas, based on aspect ratios, which can be optically addressed with minimal antenna crosstalk for use in constructing photonic gene circuits.
For use as biomolecular emitters, resonant siRNA-Au nanoantennas were designed such that the photothermal heat transfer from the nanoantenna surface to the surrounding cellular environment is highly localized, decaying exponentially25 and therefore is thought to have minimal adverse effects on cells. To further minimize adverse effects, the plasmon resonance wavelength is also tuned to the NIR since cells are essentially transparent in the NIR regime.22 In this work, the nanoantenna surface is modified with a cationic phospholipid bilayer,26 and negatively charged siRNA is adsorbed to the surface. Optically generated, surface-localized heat is used to emit siRNA from the nanoantenna. The calculated temperature profiles (Fig. S1) show this heat generated is highly localized to the surface and decays exponentially within 100 nanometers. The calculated concentration profiles in Figure 1b ii show siRNA is emitted from the nanoantenna when the antenna effect is “on” at the resonance state. We utilized this resulting outcome of the antenna effect to design siRNA-Au nanoantennas which can be optically addressed to emit siRNA with minimal antenna crosstalk for use in constructing photonic gene circuits.
The siRNA-Au nanoantennas were synthesized and experimentally characterized as functional optical receivers and biomolecular emitters of siRNA (further information in Methods, Fig. S2 – S8). Inside living cells, interfering siRNA silences intracellular genes in a sequence-specific manner, but alone, lacks the temporal control necessary for precise modulation. The siRNA-Au nanoantennas combine the benefits of sequence-specificity with spatiotemporal control.
Darkfield microscopy was used to visualize internalized siRNA-Au nanoantennas (Fig. 2a). Human cervical carcinoma HeLa cells were illuminated with unpolarized white light from an oblique angle using a darkfield condenser lens and scattered light was collected using a transmission-mode darkfield microscope. To locate cells’ boundaries and nuclei, DIC images were overlaid with DAPI-stained images and placed adjacent to darkfield scattering images. Using darkfield microscopy, scattered light from individual nanoantennas can be visualized. In Figure 2c, individual nanoantennas appear uniforrmly distributed in the cytosol of HeLa cells. It is clearly evident from Figure 2 that scattered light from cells containing nanoantennas (Fig. 2c) is easily differentiated from cells lacking nanoantennas (Fig. 2b). Long-term viability/cytotoxicity and proliferation studies were previously conducted to ensure that internalized nanoantennas26 and optical excitation4 caused no adverse effects on cell behavior.
The circuit diagram for an OFF-switch photonic gene circuit is shown in Figure 2d i. Gene X (gX) is transcribed into messenger RNA X (mX) which is then translated into protein X (pX). The siRNA-Au nanoantenna is introduced into the circuit diagram as an optical input. Circuit connections are notated in the legend in Figure 1a. When the siRNA-Au nanoantenna is optically addressed, siRNA targeting X is emitted from the siRNA-Au nanoantenna, thereby turning off mX and subsequently pX due to optical gene silencing of X (Fig. S9). To experimentally demonstrate the OFF-switch circuit in HeLa cells, the activated isoform nuclear factor κB-p65 (p65) was chosen to represent X, and flow cytometric analysis was used to quantify the OFF-switch of p65 (i.e., decrease in p65 protein levels) in single HeLa cells. In Figure 2d, p65 protein levels decreased by Δ = 40% as a result of the OFF-switch photonic gene circuit. No decrease in p65 protein levels was observed when optically addressed, non-resonant siRNA-Au nanoantennas failed to emit p65 siRNA (Fig. 2d ii). No decrease in p65 protein levels were observed when optically addressed siRNA-Au nanoantennas emitted scrambled siRNA (Fig. 2d iii). It was observed in the positive control (Fig. S10a) that siRNA alone targeting p65 decreased protein expression levels by 60%. We surmise that the difference between optical gene silencing and the positive control could be due to the efficiency of siRNA to escape out of endosomes and into the cytosol after photothermal destabilization of the endosomal membrane. Future in-depth imaging studies of photothermal destabilization of endosomal membranes could help to better understand and improve efficiencies of siRNA escape from photothermally destabilized endosomes. A logic table summarizing the OFF-switch behavior is shown in Figure 2e, where “0” (off-state) and “1” (on-state) represent low and high concentrations, respectively. In Figure 2e, HeLa cells immunostained for p65 showed an overall high fluorescence when the input was “0” and an overall low fluorescence when the input was “1” (optically addressed OFF-switch). OFF-switch photonic gene circuits presented in Figure 2 are not limited to p65 and can be constructed for virtually any protein-of-interest. OFF-switch photonic gene circuits are also modular, and can be combined with other naturally occurring sub-circuits and/or other photonic sub-circuits.
We take advantage of this modularity to construct an ON-switch photonic gene circuit. In the circuit diagram in Figure 3b, genes (gX, gY) are transcribed into messenger RNA (mX, mY) and are then translated into proteins (pX, pY), where pY inhibits the active form of pX, represented by pXnuc, by sequestering pXnuc in the cytoplasm (off-state). To construct an ON-switch of pXnuc, a modular OFF-switch sub-circuit is introduced into this circuit diagram to generate a double-negative signal (i.e., inhibit the inhibitor pY). Therefore, when the siRNA-Au nanoantenna is optically addressed, pY is turned off, thereby enabling pXnuc to turn on and freely translocate to the nucleus (on-state). To experimentally demonstrate this ON-switch circuit configuration in HeLa cells, p65 was chosen to represent X and the inhibitor κB (IκB) was selected to represent Y. Firstly, flow cytometric analysis was used to quantify the IκB OFF-switch sub-circuit (i.e., decrease in IκB protein levels) in single HeLa cells. In Figure 3a, IκB protein levels decreased by Δ = 40% as a result of the OFF-switch sub-circuit. Having confirmed the IκB OFF-switch sub-circuit, the ON-switch of p65nuc was studied using immunofluorescence imaging. A logic table summarizing the ON-switch behavior is shown in Figure 3b. In the off-state, IκB and p65 form an inactive complex such that p65 is sequestered in the cytoplasm. Therefore, HeLa cells immunostained for p65 showed an overall low fluorescence in the nucleus when the input was “0”. Conversely, HeLa cells showed a strong nuclear presence of p65 (p65nuc) when the input was “1” (optically addressed ON-switch). To locate cells’ boundaries and nuclei, DIC images and DAPI images were placed adjacent to p65 immunostained images. Similar to OFF-switch circuits, ON-switch photonic gene circuits are also modular and can potentially be combined with multiple sub-circuits to construct more sophisticated photonic gene circuits.
To construct more sophisticated photonic gene circuits from multiple sub-circuits, modular sub-circuits should function independently of each other. This can be accomplished with siRNA-Au nanoantennas that operate at distinct optical wavelengths (Fig 4b), based on aspect ratio (A.R.), with minimal antenna crosstalk. Selective optical release of two distinct DNA strands has been previously demonstrated,14 but not yet in the context of engineering gene circuits. Fluorescence analysis was conducted to measure antenna crosstalk between 4.0 A.R. and 2.5 A.R. siRNA-Au nanoantennas. Using λ1= 785 nm, a statistically significant increase in fluorescence intensity was seen when optically addressed resonant siRNA-Au nanoantennas (4.0 A.R.) emitted fluorescently labeled siRNA, while no increase in fluorescence intensity was observed when non-resonant nanoantennas (2.5 A.R.) were optically addressed (Fig. 4b, Fig. S6a), indicating no antenna crosstalk using λ1= 785 nm. It was observed that when using λ2= 660 nm (Fig. S6b), a statistically significant increase in the fluorescent intensity was seen with resonant nanoantennas (2.5 A.R.), but a small but measurable increase in fluorescent intensity was also observed with non-resonant nanoantennas (4.0 A.R.). In this case, signal interference occurred with non-resonant nanoantennas, but this can be circumvented in the future by designing nanoantennas to achieve a wider spectral separation between their longitudinal plasmon resonance bands.
Having confirmed that siRNA-Au nanoantennas can selectively receive optical signals and preferentially emit siRNA with minimal antenna crosstalk, we constructed a PULSE-switch photonic gene circuit from multiple, independently operating sub-circuits. In the circuit diagram in Figure 4a, ON-switch and OFF-switch sub-circuits are combined to form a PULSE-switch circuit. The ON-switch sub-circuit functions to initiate the pulse of target protein pZ at time t1. In this ON-switch sub-circuit, siRNA-Au nanoantennas (4.0 A.R.) are optically addressed at time t1 using λ1 = 785 nm to emit siRNA targeting Y. This causes pY to turn off, pXnuc to turn on, and subsequently target pZ to turn on. The OFF-switch sub-circuit then operates to terminate the pulse of target pZ at time t2. In this OFF-switch sub-circuit, siRNA-Au nanoantennas (2.5 A.R.) are optically addressed at time t2 using λ2 = 660 nm to emit siRNA targeting X. This causes pX to turn off and subsequently target pZ to turn off. ON-switch and OFF-switch sub-circuits should operate separately, with minimal signal interference from the other circuit. However, since some signal interference was observed with λ2 = 660 nm (Fig. S6b) while no signal interference was observed with λ1 = 785 nm, here, the order of operation became critical. The ON-switch was first designed to operate at λ1 = 785 nm since no signal interference occurred using λ1 = 785 nm (Fig. 4b and S6a). The OFF-switch was then designed to operate at λ2 = 660 nm. In this order of operation, both circuits were able to operate independently of each other. In future versions, signal interference will be minimized by designing nanoantennas with a wider spectral separation between their longitudinal plasmon resonance bands or by designing new geometries of nanoantennas with narrower spectral bandwidths.
To experimentally demonstrate this PULSE-switch photonic gene circuit in HeLa cells, IκB and p65 were chosen to represent Y and X, respectively. To test the kinetic behavior of the PULSE-switch photonic gene circuit, temporally regulated genes were used as a functional assay of kinetics. Two NF-κB regulated genes (IP-10 and RANTES) that display different transcriptional activation profiles were selected to represent Z (as Z1 and Z2, respectively). IP-10, an early response gene, is known to be immediately activated within 30 minutes of stimulation,27 whereas RANTES, a late response gene, requires more than 3 hours of stimulation to turn on.28 We used these two temporally regulated genes to functionally validate the 2 hour PULSE-switch shown in Figure 4a. We reasoned that if a 2 hour PULSE-switch was constructed, IP-10 should turn on while RANTES should remain off. To implement the 2 hour PULSE-switch, we constructed a logic table for IP-10 and RANTES summarizing four possible conditions (Fig. 4c). These conditions listed in the logic table were shown to function correctly using flow cytometric analysis of IP-10 and RANTES (Fig. 4d). Notably, Figure 4d iv confirmed that IP-10 is turned on while RANTES is turned off as a result of the 2 hour PULSE-switch. In addition to functional assays of kinetics, visual assays using fluorescently labeled proteins can also be utilized in the future to characterize photonic gene circuits.
In this work, we have demonstrated the construction of a PULSE-switch circuit using two independent sub-circuits. This was accomplished with two independently operating nanoantennas. In the future, more sophisticated circuit configurations can be conceivably constructed in the future by combining multiple, independent sub-circuits. Due to the narrow spectral bandwidths of Au nanorods, four sub-circuits could potentially operate independently of each other in the visible and near-infrared wavelength regime. More sub-circuits could potentially be combined by synthesizing new geometries of nanoantennas with more narrow spectral bandwidths in the visible and near-infrared wavelength regime. As more sub-circuits are combined, signal strength should be addressed to improve circuit performance. Signal distortions, such as amplitude reductions, occur due to inherent dilution, degradation, and diffusion effects in the intracellular environment. These signal distortions inherently occur in native gene circuits. Inevitably, photonic gene circuits are subject to similar signal distortions as native gene circuits. Improving RNA interference efficiency could help to overcome the effects of signal amplitude reduction. For example, designing siRNA to bind to optimal target sites could help to improve RNA interference efficiency. Incorporating other geometries of RNA, such hairpin-shaped or dumb bell-shaped RNA, with nanoantennas which may be less prone to degradation could also help to improve RNA interference efficiency in the intracellular environment.
In closing, we have demonstrated the construction of photonic gene circuits using optically addressable siRNA-Au nanoantennas in living cells. Photonic gene circuits are a promising approach to systematically study native gene circuits in complex, naturally occurring living systems. Since native gene circuits remain genomically unaltered, photonic gene circuits are a promising alternative to synthetic circuits for studying temporal dynamics29 in native gene circuits. Towards in vivo applications, we envision constructing photonic gene circuits using optically addressable siRNA-Au nanoantennas in optically transparent whole model organisms, such as C. elegans and zebrafish embryos, to study temporal dynamics and to probe, identify, and reconfigure malfunctioning gene circuits as a therapeutic strategy. We envision that light also could be delivered transdermally or by fiber optics in other model organisms that are not optically transparent. We expect that photonic gene circuits will play a pivotal role in engineering functional gene circuits as they enable, for the first time, dynamic reconfiguration and temporally precise modulation of native gene circuits.
Finite element method (COMSOL Multiphysics software) was used to model a 3-D Au nanorod (4.0 and 2.5 A.R.) suspended in water and to achieve a solution to the Helmholtz wave equation: . The Au nanorod was constructed using a cylinder with hemispheres on each end of the cylinder. The relative permeability of Au was assumed to be μr = 1 and the complex permittivity of Au εr was assumed to be a function of wavelength λ. A spherical perfectly matched layer and an integration layer, modeled by concentric spheres, were used to reach perfect absorption at the outer boundary and minimize spurious reflections. A plane wave was used for excitation (λ = 780 nm or λ = 650 nm). The adaptive mesh was refined until the maximum electric field converged.
We solved the bioheat transfer equation: . Time-average resistive heat (W/m3) was used for the spatial heat source Qext under the assumption that electromagnetic energy was converted to heat by resistive heating. The thermal conductivity of gold, density of gold, and specific heat of gold were assumed to be kg = 320 W/m-K, ρg = 19300 kg/m3, and Cg = 129 J/kg-K, respectively.30, 31 The thermal conductivity of water, density of water, and specific heat of water were assumed to be kw = 0.61 W/m-K, ρw = 1000 kg/m3, and Cw = 42000 J/kg-K, respectively. The metabolic heat source Qmet and the perfusion rate ωb were assumed to be insignificant.
To investigate biomolecular emission of siRNA, we solved the thermal diffusion equation: . Electromagnetic energy was assumed to be converted to surface localized heat and was used here for temperature T. It was assumed that siRNA dissociated from the surface of the nanoantenna at elevated temperatures, and mass flux occurred due to the temperature gradient in addition to the siRNA concentration gradient. The diffusion coefficient and the thermal diffusion coefficient of siRNA were estimated to be to be D = 36.00 × 108 cm2/s and DT = 0.45 × 108 cm2/s-K, respectively, based on double-stranded DNA of similar length.32 Using the surface density (9.0 × 1012 molecules/cm2) and the size dimensions, the initial concentrations of siRNA were calculated to be c0 = 1.81 × 10−14 mol/m3 for 4.0 A.R. nanoantennas and c0 = 2.11 × 10−14 mol/m3 for 2.5 A.R. nanoantennas, respectively.
The OFF-switch (Eq. 1–3), ON-switch (Eq. 4–8), and PULSE-switch (Eq. 9–17) circuit configurations were modeled (Matlab software). Genes (gX, gY, gZ) were transcribed into messenger RNA (mX, mY, mZ) which were then translated into proteins (pX, pY, pZ), where gX, gY, gZ, mX, mY, mZ, pX, pY, and pZ denote concentrations. Degradation of messenger mRNA by optical gene silencing using gold nanoantennas functionalized with siRNA (NAsirna1, NAsirna2) were modeled based on Michaelis-Menton/Hill kinetics.
Kinetic rates and initial values are listed in supporting information Table 1 and were approximated similarly to those previously reported in literature for a eukaryotic model.33 It was assumed no siRNA was emitted by non-resonant siRNA-Au nanoantennas. It was also assumed that when siRNA-Au nanoantennas were optically addressed with the correct optical wavelength, all siRNA was emitted. Since the model’s timescale was on the order of hours, it was assumed that nanoantennas did not degrade in this timeframe and therefore the concentration of nanoantennas remained the same. Due to the tight packing of siRNA on the nanoantennas, steric hinderances inhibited nuclease degradation34 of siRNA while siRNA was bound to the nanoantennas. Once siRNA was emitted from nanoantennas, the siRNA degraded with degradation rates as reported previous literature.35
Gold nanorods of aspect ratios 2.5 and 4.0 (Fig. S2) were synthesized by adapting a seed-mediated growth method36, 37 to an RNase-free environment.26 All solutions were prepared using 0.2 µm filtered DEPC-treated water. All glassware and metalware were baked at 240°C for 24 hours to remove exogenous RNases. All pipetting devices and counter space was treated with 70% ethanol. All disposable plastic pipette tips and centrifuge tubes were certified to be free of RNase. The resultant CTAB-coated gold nanoantennas were ensured to be free of RNases by detecting RNase activity over time (Fig. S3). RNase activity in the supernatant solution was detected using an RNase activity kit (AM1964, Ambion) and was quantitatively measured over time using a fluorometer (Fluoromax-3, Horiba Jobin Yvon). The concentration of CTAB-coated nanorods was confirmed by adjusting to an absorbance of 1 at the longitudinal plasmon resonance wavelength using UV-VIS spectroscopy (8453, Hewlett Packard). Aspect ratios were determined by scanning electron microscopy and transmission electron microscopy.
Gold nanoantennas functionalized with siRNA were synthesized as previously described.26 In summary, the CTAB surfactant on the nanoantennas’ surface was exchanged with a cationic phospholipid bilayer to form biologically functional cationic phospholipid-gold nanoantennas. To remove excess CTAB surfactant, 500 µL unmodified CTAB-coated nanoantennas (UV-VIS absorbance of 1) were centrifuged at 5000 rpm for 10 minutes. A 10 µL pellet was transferred to a new microcentrifuge tube, redispersed in 500 µL of nuclease-free water, briefly vortexed, and sonicated for 1 minute. To replace CTAB surfactant with a phospholipid bilayer membrane at the nanoantennas surface, nanoantennas were then centrifuged again at 5000 rpm for 10 minutes. A 10 µL pellet was transferred to a new microcentrifuge tube, resuspended in 50 µL of Oligofectamine, briefly vortexed, and sonicated for 1 minute.
After the CTAB coating was replaced with a cationic phospholipid coating, siRNA was then conjugated to the nanoantennas. IκB siRNA (Qiagen) was conjugated to nanoantennas (plasmon resonance wavelength 780 nm). p65 siRNA (Qiagen) was conjugated to nanoantennas (plasmon resonance wavelength 650 nm). To 500 µL of nanoantennas solution, 2 µL of 100 µM siRNA was added. The solution was vortexed and allowed to incubate for 30 minutes. To remove excess siRNA from solution, nanoantennas were washed with nuclease-free water by centrifugation at 5000 rpm for 10 minutes and finally resuspended in 500 µL of nuclease-free water. After preparation of siRNA-Au nanoantennas, an absorbance of 0.2 was measured by UV-VIS (8453, Hewlett Packard). By comparing with the original nanoantennas’ UV-VIS absorbance of 1, the concentration of siRNA-Au nanoantennas was estimated to be approximately 1/5 the original concentration (approximately 6 µg/mL or 1.4E11 nanoantennas/mL).
For internalization, the siRNA-Au nanoantennas were then coated with cationic phospholipids to improve entry into living cells. Firstly, 1000 µL of siRNA-Au nanoantennas were concentrated into a 10 µL pellet by centrifugation at 5000 rpm for 20 minutes. The pellet was transferred to a new microcentrifuge tube and then resuspended in 25 µL of Oligofectamine. The solution was vortexed and allowed to incubate for 20 minutes. For higher concentrations of conjugates, the multiple tubes of conjugates were prepared the same as described above. After 20 minutes, 175 µL of nuclease-free water was added to each tube to dilute the concentration of free cationic phospholipids in solution. The conjugates were concentrated into a 0.5 µL pellet by centrifugation for internalization.
Prior to optically addressing siRNA-Au nanoantennas, the controlled thermal liberation of siRNA from the surface was first demonstrated. If siRNA dissociates from the cationic phospholipid bilayer at elevated temperatures, this unbinding event should change the dielectric constant of the medium locally surrounding the nanoantennas and therefore result in an observable shift in nanoantennas’ longitudinal plasmon resonance wavelength.38, 39
A UV-VIS spectrometer containing a thermal-jacketed cell was used to evaluate the temperature-dependent absorbance of the nanoantenna-containing solutions. To each separate cell of an 8-cell micro sample holder (208–92086, Shimadzhu), 50 µL of siRNA-Au nanoantennas or unconjugated nanoantennas were added. Samples were simultaneously heated from 20°C to 70°C (10°C increment, 5 minutes) using a temperature-controlled UV-VIS spectrometer (2501, Shimadzhu). Absorbance spectra were collected for each sample at each temperature increment.
Control nanoantennas lacking siRNA cargo showed no shift in the longitudinal plasmon resonance wavelength at elevated temperatures (Fig. S4b), indicating that the phospholipid bilayer remained stable at elevated temperatures. In contrast, siRNA-Au nanoantenna conjugates showed a marked blue-shift in the longitudinal plasmon resonance wavelength at elevated temperatures due to the dissociation of siRNA from the surface (Fig. S4a). Additionally, supplemental Figure S4c shows that the longitudinal plasmon resonance wavelength of the siRNA-Au nanoantennas blue-shifts with incrementing temperature until it eventually matches the longitudinal plasmon resonance wavelength of the control nanoantennas, strongly suggesting the complete dissociation of siRNA from the intact phospholipid bilayer at 70°C.
An 80 mW 785 nm CW diode laser (Newport Corp.) and a 60 mW 660 nm CW diode laser (Newport Corp.) were optically co-aligned such that their beams simultaneously overlapped (Fig. S5). Firstly, both lasers were positioned parallel to each other on a manually-adjustable xyz stage. A dichroic mirror (Omega Optical) was positioned below the 785 nm laser. A mirror (Omega Optical) was placed below the 660 nm laser. To make the two beams orthogonal to each other, the 660 nm laser beam was reflected 90° towards the direction of the other laser by the mirror. The dichroic mirror transmitted the 785 nm laser beam and reflected the 660 nm laser beam another 90° such that both beams simultaneously overlapped and co-aligned with each other. To circularly polarize the light, an achromatic quarter-wave plate (CVI Laser Corp.) was placed below the co-aligned beams.
Having established that siRNA can be thermally dissociated from the cationic phospholipid bilayer, siRNA-Au nanoantennas of different aspect ratios were then selectively addressed. Because their narrow longitudinal plasmon resonance bands are spectrally separated (Fig. S2a), rod-shaped Au nanoantennas can be optically addressed to selectively emit siRNA.
6-carboxyfluorescein (FAM; excitation 495 nm, emission 520 nm) labeled siRNA were bound to nanoantennas and unbound FAM-siRNA were then removed from the background solution by centrifugation. To a 2 mm path length quartz cuvette (3-2.45-Q-2, Starna Cells Inc.), 50 µL of FAM-siRNA-Au nanoantennas (2.5 A.R.) or FAM-siRNA-Au nanoantennas (4.0 A.R.) were added. Using the optical setup shown in Figure S5, samples were illuminated from a top with 50 mW of either 785 nm or 660 nm light with a spot size of 2 mm. During illumination, fluorescence emission spectra were collected at 2 minute intervals for each sample using a fluorometer (Fluoromax-3, Horiba Jobin Yvon). A shortpass filter was positioned in front of the fluorometer’s detector to block the detector from the laser sources.
To calibrate the fluorescent intensity to the concentration of siRNA liberated from nanoantennas carriers into solution, the fluorescent intensities of known concentrations of FAM-siRNA were also measured (Fig. S7). As seen in supporting Figure S7, nanoantennas (concentration 1E11 nanoantennas/mL based on UV-VIS measurements) released approximately 0.05 µM siRNA after illumination. As a positive control, Triton X-100 detergent was used to disrupt the cationic phospholipid bilayer around nanoantennas, thereby ensuring the complete liberation of all bound siRNA from nanoantennas into solution. The concentration of siRNA released using Triton X-100 matched closely to the concentration of optically liberated siRNA, strongly suggesting complete liberation of siRNA from optically addressed nanoantennas.
The human cervical carcinoma cell line HeLa was purchased from the American Type Culture Collection (ATCC). Dulbecco’s modified eagle’s media (DMEM) formulated with high glucose and GlutaMAX was purchased from Invitrogen and was supplemented with 10% heat-inactivated fetal bovine serum. Cells were seeded at an initial concentration of 20,000 cells/well in a 96-well plate, cultured in the supplemented media, and maintained in a 37 °C incubator with 5% CO2 humidified air.
siRNA-Au nanoantennas were internalized into HeLa human cervical cancer cells. For visualization purposes, HeLa cells were seeded onto 12 mm gridded glass coverslips (Electron Microscopy Sciences) at 30,000 cells/well in a 24-well plate for 24 hours before use. HeLa cells were then washed once with Optimem media. The 0.5 µL concentrated pellet of siRNA-Au nanoantennas was resuspended in 100 µL of Optimem media, gently mixed, and added to each well of the 24-well plate. The cells were allowed to incubate for 4 hours at 37°C. Cells were then fixed by incubating cells with 2% paraformaldehyde (Electron Microscopy Sciences) in 1X PBS per well for 10 minutes. Cell nuclei were stained with DAPI (Invitrogen) by incubating cells in 300 nM DAPI in 1X PBS per well for 5 minutes. 1X PBS was used to twice-wash the cells. The coverslip containing fix, adhered cells was then placed facedown and adhered to a microscope slide. Cells were located using the grids imprinted on the coverslips.
Darkfield microscopy was used to visualize internalized siRNA-Au nanoantennas (Fig. 2a). Darkfield scattering was visualized using an inverted microscope (Axiovert, Zeiss) at 40X magnification. Broadband white light was shined onto the adhered cells from an oblique angle using a darkfield condenser lens. The scattered light alone was collected using a microscope objective lens with a numerical aperture (NA) of 0.65 that was smaller than the NA (1.2–1.4) of the illumination condenser lens. To locate cells’ boundaries and nuclei, DIC images were overlaid with DAPI-stained images and placed adjacent to darkfield scattering images. DIC and DAPI were visualized using an upright fluorescence microscope (Axio Imager, Zeiss) at 40X magnification.
To estimate the amount of siRNA liberated from nanoantennas into the intracellular space, fluorescently-labeled FAM-siRNA were bound to nanoantennas and unbound FAM-siRNA were then removed from the background solution by centrifugation. Known concentrations of FAM-siRNA-Au nanoantennas (7E11, 4E11, and 3E11 nanoantennas/mL based on UV-VIS measurements) were internalized in HeLa cells. Fluorescent intensities of individual cells were then measured by flow cytometry and a standard concentration curve of internalized FAM-siRNA-Au nanoantennas was constructed (Fig. S8b). Fluorescence quenching by nanoantennas was not observed.26 To correlate the fluorescent intensities to FAM-siRNA concentration, control cells were incubated with known concentrations of FAM-siRNA (50 nM, 100 nM, and 200 nM) for 5 hours. A standard concentration curve of internalized FAM-siRNA was then constructed based on flow cytometry analysis (Fig. S8a). These standard curves were utilized to estimate the concentration of siRNA-Au nanoantennas necessary for optical gene silencing.
HeLa cells were then washed once with Optimem media. 0.5 µL concentrated pellet of siRNA-Au nanoantennas (2.5 A.R.) functionalized with p65 siRNA was resuspended in 100 µL of Optimem media, gently mixed, and added to each well of the 96-well plate. The cells were allowed to incubate for 4 hours at 37°C. After internalization of siRNA-Au nanoantennas for 4 hours, the media was replaced with fresh supplemented DMEM culture media. The 96-well plate was placed in a CO2-filled, sealed container, containing a high transmission glass window (Edmund Optics). Wells were illuminated from a top with 50 mW of 660 nm CW diode laser (Newport Corp.) with a spot size of 2 mm (one quadrant of a well in a 96-well plate) for 15 minutes. After illumination, cells were allowed to incubate for an additional 72 hours at 37 °C. Cells were then immunostained for p65, and analyzed by flow cytometry and by immunofluorescence imaging.
HeLa cells were then washed once with Optimem media. 0.5 µL concentrated pellet of siRNA-Au nanoantennas (4.0 A.R.) functionalized with IκB siRNA was resuspended in 100 µL of Optimem media, gently mixed, and added to each well of the 96-well plate. The cells were allowed to incubate for 4 hours at 37°C. After internalization of siRNA-Au nanoantennas for 4 hours, the media was replaced with fresh supplemented DMEM culture media. The 96-well plate was placed in a CO2-filled, sealed container, containing a high transmission glass window (Edmund Optics). Wells were illuminated from a top with 50 mW of 785 nm CW diode laser (Newport Corp.) with a spot size of 2 mm (one quadrant of a well in a 96-well plate) for 15 minutes. After illumination, cells were allowed to incubate for an additional 72 hours at 37 °C. Cells were either immunostained for IκB and analyzed by flow cytometry, or immunostained for p65 and analyzed by immunofluorescence imaging.
HeLa cells were then washed once with Optimem media. 0.5 µL concentrated pellet of siRNA-Au nanoantennas (4.0 A.R.) functionalized with IκB siRNA was added to 0.5 µL concentrated pellet of siRNA-Au nanoantennas (2.5 A.R.) functionalized with p65 siRNA and was resuspended in 200 µL of Optimem media, gently mixed, and added to each well of the 96-well plate. The cells were allowed to incubate for 4 hours at 37°C. After internalization of siRNA-Au nanoantennas for 4 hours, the media was replaced with fresh supplemented DMEM culture media. The 96-well plate was placed in a CO2-filled, sealed container, containing a high transmission glass window (Edmund Optics). Wells were illuminated from a top with 50 mW of 785 nm CW diode laser (Newport Corp.) with a spot size of 2 mm (one quadrant of a well in a 96-well plate) for 15 minutes. Two hours after initial illumination with 785 nm light, wells were illuminated from a top with 50 mW of 660 nm CW diode laser (Newport Corp.) with a spot size of 2 mm (one quadrant of a well in a 96-well plate) for 15 minutes. After illumination, cells were allowed to incubate for an additional 72 hours at 37 °C. The media was replaced with fresh supplemented DMEM culture media containing 3 µm monensin (Sigma) 48 hours after illumination. The cells were allowed to incubate for an additional 24 hours at 37 °C. Cells were then immunostained for IP-10 or RANTES and analyzed by flow cytometry.
Fluorescently-labeled antibodies recognizing p65 and IκB were purchased from Santa Cruz Biotechnologies. Fluorescently-labeled normal mouse isotype antibodies were also purchased from Santa Cruz Biotechnologies and were used as negative controls. Fluorescently-labeled antibodies recognizing IP-10 and RANTES, and the isotype control were purchased from R&D Systems. Permeabilization buffer, for use with p65 and IκB antibodies, was prepared by adding 0.1% (w/v) saponin, 0.3% (w/v) Triton-X, and 0.1% (w/v) NaN3 to Hank’s Balanced Salt Solution (Invitrogen). Permeabilization buffer, for use with IP-10 and RANTES antibodies, was prepared by adding 0.1% (w/v) saponin and 0.06% (w/v) NaN3 to Hank’s Balanced Salt Solution (Invitrogen).
Cells were harvested, resuspended in 250 µL of 1X PBS, and fixed with 250 µL of 4% paraformaldehyde for 10 minutes. After 10 minutes, excess paraformaldehyde was removed by centrifuging and resuspending cells in 400 µL of permeabilization buffer (repeated twice). Cells were then counted to ensure all samples contained the same number of cells prior to immunostaining. For 50,000 cells, 10 µL of antibody or isotype antibody were added. Cells were gently mixed and incubated at room temperature for 45 minutes. After 45 minutes, excess antibodies were removed by centrifuging and resuspending cells in 400 µL of permeabilization buffer. To remove permeabilization buffer, cells were finally centrifuged and resuspended in 500 µL of 1X PBS. LSRII flow cytometer (BD Biosciences) and FlowJo software (Tree Star, Ashland, Oregon) were used to analyze samples.
72 hours after optical gene silencing of IκB or p65, cells (adhered to glass coverslips) were fixed in cold 50% methanol for 3 minutes on ice followed by cold 100% methanol for 15 minutes on ice. For three times, cells were washed with and incubated in 1X PBS for 5 minutes on a rocker at speed ~100 rpm. Cells were blocked with 5% normal mouse serum (01-6501, Invitrogen) in 1X PBS for 30 minutes on a rocker at speed ~100 rpm. For two times, cells were washed with and incubated in 1X PBS for 5 minutes on a rocker at speed ~100 rpm. Cells were then incubated in 300 nM DAPI (Invitrogen) in 1X PBS for 5 minutes in the dark on a rocker at speed ~100 rpm. Cells were washed with and incubated in 1X PBS for 5 minutes on a rocker at speed ~100 rpm. To 200 µL of 1X PBS, 60 µL of AF488-labeled anti-p65 (Santa Cruz Biotechnologies) was added. Cells were allowed to incubate for 2 hours in the dark on a rocker at speed ~100 rpm. For three times, cells were washed with and incubated in 1X PBS for 5 minutes on a rocker at speed ~100 rpm. Coverslip containing fixed, stained adhered cells was finally placed facedown on a microscope slide, sealed, and imaged using fluorescence microscopy.
The authors acknowledge the National Institutes of Health (NIH) Nanomedicine Development Center for the Optical Control of Biological Function (PN2 EY018241) and the Center for Nanostructured Materials and Technology (CNMT) of the Korea government for financial support of the project. S.E. Lee was supported by the Siebel Scholarship (Siebel Foundation) for the initial phase and the NIH Ruth L. Kirschstein National Research Service Award (F32 EB013972) for the final phase of the project. D.Y. Sasaki was supported by the Division of Materials Science and Engineering in the Department of Energy Office of Basic Energy Sciences at Sandia National Laboratories, a multiprogram laboratory operated by Sandia Corp., a Lockheed Martin Co., for the U.S. Department of Energy (contract no. DE-AC04-94AL85000). M.J. Bissell was supported by the U.S. Department of Energy, Office of Biological and Environmental Research, and Low Dose Radiation Program (contract no. DE-AC02-05CH1123) and the Bay Area Physical Sciences–Oncology Center, University of California, Berkeley, California 94720, USA (NCI U54CA143836). The authors thank Prof. Han Lim (UC Berkeley) for insightful discussion on gene circuits. The authors also thank Ann Fischer and Michelle Yasukawa of the UC Berkeley Tissue Culture Facility for long-term maintenance of the HeLa cell line.