Despite decades of research and drug development for cancer treatment, cancer still remains one of the major causes of human deaths in the world.1 Human cancers consist of a mixed population of malignant cells that carry multiple genetic mutations. It has been estimated that there are around 40 DNA mutations that result in amino acid changes in an individual tumor of glioblastoma2 or pancreatic cancer,3 and twice as many mutations in breast and colorectal cancers.4 Many of them are driver mutations that determine tumor initiation, progression, and metastasis. The genomic landscapes indicate that multiple signal transduction pathways determine the fate of a cancer cell, and it is almost impossible to treat cancer with a single therapeutic agent. It is thus not surprising that many of the recently developed targeted therapy drugs, such as the EGFR inhibitors for non-small-cell lung-cancer treatment and Herceptin for breast-cancer therapy, are effective in only a certain set of cancer patients;5, 6 and in many cases, drug resistance arises from additional genetic and epigenetic alterations. It has been reported that mutations within the EGFR gene,7, 8 the KRAS gene,9, 10 and up-regulation of other signaling pathways11, 12 could all cause resistance to EGFR-targeted therapy. Clearly, there is an urgent and persistent motivation to develop novel therapeutics independent of the genetic background of the cancer in the fight against this deadly disease.
Nanotechnology has played a crucial role in the development of cancer therapeutics.13 Doxil and Abraxane14–17 are just two examples of nanoformulated drugs. Gold nanoparticles (AuNPs) are currently being explored to induce hyperthermic cytotoxicity.18–22 When exposed to light at the correct wavelength, the conduction-band electrons of the nanoparticle generate heat that is transmitted to the cells and surrounding tissues. Thermal therapy has the advantage of killing cancer cells without causing resistance regardless of the genetic background, and thus can be applied to all cancer patients. However, successful application of this approach requires adequate accumulation of AuNPs in the tumor and sufficient tumor penetration of the excitation energy.23 Due to lack of effective delivery, tumor delivery of AuNPs in most studies has been relied on the enhanced permeability and retention (EPR) effect, a result of tumor blood vessel leakiness due to a state of ongoing angiogenesis, and thus has not been very efficient.24 It has been reported that less than 5% of the total injected dose of PEGylated AuNPs could ultimately reach the tumor tissue.22 The amount of AuNPs needed for each treatment makes it impractical for clinical therapies. Besides, AuNPs tend to accumulate unevenly in the tumor tissue dependent on particle size, surface charge, and other factors,25–27 which makes it difficult to eradicate the whole tumor tissue with this approach.
We have previously developed a multistage vector delivery system based on porous silicon (pSi). This system is composed of two delivery carriers: the first stage is a degradable pSi vector and the second stage nanoparticles are loaded into the pores of pSi. The nanoparticles can be liposomes or micelles incorporated with therapeutics (third stage) or diagnostic agents.28 A major advantage of this system is that the size, shape, and surface properties of pSi can be tailored so that maximal tissue-specific localization and release of the therapeutics at the target tumor can be achieved.28–31 We have recently demonstrated successful application of this system to deliver siRNA therapeutics for cancer treatment with two mouse models of ovarian cancers.32, 33
Here, we designed a pSi nanoassembly with hollow gold nanoshells (HAuNSs) to explore photothermal effects for therapeutics, taking advantage of the nanoscale effect. HAuNSs are favorable over solid AuNPs, since hollow gold responds to near-infrared (NIR) light, which has a deep penetration inside the body and causes less damage comparing to a shorter wavelength light to tissues due to less absorbance by the tissue chromophores. We used human and mouse breast cancer lines to test cell killing in vitro and the mouse model of 4T1 mammary tumor for in vivo studies. The potential mechanism of action with the new therapeutics will be discussed.
pSi particles were fabricated in three major steps: the formation of porous silicon films, photolithographic patterning of particles, and reactive ion etching. The porous structure of the particles was tailored by electrochemical etching to a mean pore size of 60 nm and a porosity of about 80%, while the particle sizes were precisely defined by photolithography to 1000 nm in diameter and 400 nm in thickness. Scanning electron microscopy (SEM) images revealed that the pores were evenly distributed across the whole area (Figure 1A and B). Since the surface of the silicon particles was conjugated with 3-aminopropyltriethoxysilane (APTES), these particles were positively charged, which facilitated loading of the negatively charged HAuNSs by favorable electrostatic interactions (Figure 1C). Multiple HAuNSs particles could be found in each pore across the whole silicon microparticle.
We carried out a spectrum scan to determine absorption profiles. The HAuNS particles of 28 nm in diameter showed a plasma resonance peak at 750 nm (Figure 1D), which has previously been observed for most HAuNS particles of similar size.34, 35 This peak disappeared when the HAuNS particles were loaded into pSi. There was a small peak around 950 nm indicating a red shift of the absorbance from pSi/HAuNSs (Figure 1D), while empty pSi particles did not have any significant absorption in the 400–1100 nm range. We have also measured the absorption spectra of solid AuNPs with a plasma resonance peak at 528 nm (Supporting Information, Figure S1). Loading of AuNPs into pSi also resulted in the disappearance of the peak and a red shift of the small peak in the 600–750 nm range. Simply mixing AuNPs with silicon did not result in a disappearance of the plasma resonance peak (data not shown). The most plausible explanation for vanishing absorption from the pSi/HAuNSs nanoassembly is a scattering effect from the pSi microparticle.
A water suspension of the particles was used to measure the heat generation triggered by NIR laser irradiation. pSi particles alone did not show any heat generation, as expected, and constantly remained at room temperature during continous exposure to NIR laser irradiation (Figure 2). The temperature in the HAuNS colloidal suspension increased by 10.9 °C and reached a steady level of 34.7 °C within 10 min. A bigger increase in temperature was observed in the pSi/HAuNS suspension. The temperature reached 45.0 °C within 7 min, an overall increase of 20.6 °C from room temperature, which corresponds to almost twice the temperature increase found for the same amount of colloidal HAuNSs. The time constants for heat generation kinetics were calculated at 3.1 s for HAuNSs and 1.9 s for pSi/HAuNSs.
To test whether the enhanced thermal generation could be translated into efficient cell killing, we treated cancer cells with free HAuNSs or pSi/HAuNSs and monitored cell growth using a MTT assay. pSi and AuNP were used as controls. The AuNP particles were not expected to have any effect on thermal cytotoxicity, as the NIR laser used in the study with a wavelength of 808 nm did not have any impact on the solid gold. We loaded different amount of HAuNSs into a fixed number of silicon particles (2 × 109 HAuNSs or 2 × 1010 HAuNSs in 1 × 108 pSi), so that any changes in cell growth would arise from the impact of HAuNSs, but not from silicon particles. As expected, neither free HAuNSs nor pSi/HAuNSs had any significant impact on cell growth if there was not enough HAuNSs for heat generation (Figure 3A). However, when the number of gold particles increased, the pSi/HAuNSs was very efficient in killing cancer cells, while the effect from free HAuNSs was mild (Figure 3A). Further increase of the HAuNS particle numbers resulted in cell killing from both free gold and the gold-silicon nanoassembly (data not shown). A similar trend was observed with MDA-MB-231 and SK-BR-3 human breast cancer cells (Figure 3A, upper and middle panels) and 4T1 murine mammary tumor cells (Figure 3A, bottom panel). These results indicate that the thermal ablation effect is general, since these cell lines carry significantly different genetic background and diverse mutation spectra. For example, the SK-BR-3 cells overexpress the HER2 gene, while MDA-MB-231 is a triple-negative cell line lacking the expression of estrogen receptor, progesterone receptor, and HER2.
We used two dyes to detect cell viability after exposure to NIR laser. The nonfluorescent calcein AM was converted to the intensely green fluorescent calcein in live cells. Ethidum homodimer-1 (EthD-1) entered dead/dying cells through damaged membranes and bound to nucleic acid, producing a bright-red fluorescence. Cells were treated with a low dose (1×, 6.25 × 109 HAuNSs per well) and a high dose (4×, 2.5 × 1010 HAuNSs per well) free HAuNSs or pSi/HAuNSs. Around 20% cells were positive for EthD-1 staining without NIR laser treatment (Supporting Information, Figure 2). The free HAuNS-treated cells did not undergo significant cell death at the 1× dose (Figure 3B, upper left). The amount of EthD-1-positive cells doubled with the 4× HAuNS dose (Figure 3B, upper right); however, the percentage of dead cells was not as high as in the sample treated with the 1× dose pSi/HAuNSs (Figure 3B, upper right vs. bottom left). Up to 88% cancer cells treated with 4× pSi/HAuNSs were EthD-1-positive 24 h after NIR treatment, comparing to less than 35% EthD-1-positive cells treated with 4× HAuNSs (Figure 3B, bottom right). Quantitative analysis of cell death from these treatments is shown in Figure 3C. This result was in perfect agreement with the thermal generation result, which showed that the increase of temperature was three times as fast from the same amount of HAuNSs in the nanoassembly as from free HAuNSs. These results indicate that enhanced cancer-cell killing can be achieved through packaging AuNS into a pSi nanoassembly.
In a proof-of-principle study, we generated murine 4T1 tumor mice to test the efficacy of thermal ablation by pSi/HAuNSs on tumor growth. When the tumors reached sizes of 150–200 mm3, free HAuNSs or pSi/HAuNSs were delivered by intratumor injection. The tumors were treated with NIR laser the next day, and tumor growth was monitor over the following 10 days before the mice were sacrificed. A single treatment of thermal ablation from both free HAuNSs and pSi/HAuNSs significantly inhibited tumor growth (Figure 4). However, the pSi/HAuNSs-treated mice did not have much tumor growth, while the tumor weight in the free HAuNSs group more than doubled during the same period of time. In the pSi control group, the average tumor weight almost quadrupled to 1 g per tumor. This result clearly supports the application of the pSi/HAuNSs assembly in cancer therapy.
Previous studied have demonstrated the potential application of pSi as a multistage vector for the delivery of therapeutics and imaging agents.32, 33, 36 Due to the nature of the size and porosity, pSi can be used as a cargo to deliver a large quantity of therapeutic agents and can achieve tumoritropic accumulation independent of the EPR effect. In this study, pSi is used not only for enhanced tumor localization but also as an essential part of the therapeutic complex for the enhancement of thermal ablation, making it a part of therapeutic mechanism. We have analyzed the kinetics of heat generation by HAuNSs with respect to heat exchange. Theoretical models of AuNP suspensions have been developed,37, 38 where the energy balance was governed by laser-induced heat through electron–phonon relaxations on HAuNS surfaces and heat dissipation. The temperature profile generated by laser was derived by introducing the rate of energy adsorption, A [K s−1], and the rate of heat loss, B [s−1]37:
We have fitted Equation 1 to the temperature profiles in Figure 2, where A = 4.17 K s−1 and B = 0.36 s−1 for HAuNSs and A = 12.26 K s−1 and B = 0.56 s−1 for pSi/HAuNSs. The rate of heat generations was three times as high for pSi/HAuNSs as for free HAuNSs. We had anticipated that both HAuNSs and pSi/HAuNSs would have similar heat loss rates, since the only difference was the inclusion of pSi in the nanoassembly. Interestingly, the heat dissipation rate was also higher for the pSi/HAuNSs nanoassembly. Thus, the model in Equation 1 that was validated for colloidal gold could not explain the pSi/HAuNSs temperature increase curve using the same heat-loss rates. This suggests that the thermal equilibrium of pSi/HAuNSs is defined by different thermal properties of the system.
The clear difference between HAuNSs and pSi/HAuNSs was the distribution of the AuNPs, which is summarized in Figure 5. Based on the amount of HAuNSs and pSi used in this study, we concluded that the average gold interparticle distance in colloidal HAuNSs was 1.7 μm, and free pSi particles were separated by 11 μm. In the pSi/HAuNSs nanoassembly, however, HAuNSs were fixed within pSi with a domain of less than 1 μm (Figure 1). It has been reported that surface plasmons penetrate dielectric media up to hundreds of nanometers away from metal surface39, 40 and that the specific orientation of HAuNSs can transfer energy over hundreds of nanometers.41 Thus, HAuNS particles within the nanoassembly should become electromagnetically coupled mostly through dipole–dipole interactions42 and can function as waveguide-like structures that increase energy transfer and heat production, as illustrated in Figure 5. To support our finding, increased heat production in HAuNS structures through varying the angle of an incident photon beam has already been observed.43 Furthermore, bringing HAuNSs closer to clusters can cause a red shift of the surface-polaron resonance44 that could be larger than 100 nm.40 The peak around 950 nm found in pSi/HAuNSs, which was not observed in free HAuNSs suspension, indicates a red shift of 200 nm. The increased wavelength in the NIR region could be used for deeper penetration into tissues, where HAuNSs are arranged in a collective nanoscale structure. Potentially, an enhanced heating effect can also be supplemented by scattering, which increases the light path.
Since convective flows inside pSi are negligible, thermal diffusion is the most important aspect. Characteristic thermal diffusion lengths of water, expressed as over given time t and the thermal diffusivity D = 1.4 × 10−3 cm2 s−1, are 0.7 μm for 1 μs and 23.4 μm for 1 ms pulses. Therefore, using sub-microsecond-range NIR pulses, thermal expansion domains around individual HAuNSs will start to overlap, while NIR pulses over 1 μs will make pSi/HAuNSs a continuous thermally excited domain (thermal spot-source). Silicon has athermal diffusivity that is almost six times as high as that of water. At 80% porosity in pSi, the effective thermal diffusivity is 3.3 × 10−3 cm2 s−1. It has already been shown that thermal dissipation of AuNPs can be increased by wrapping the gold core with a silica shell.45 Therefore, LT of pSi increases by almost 50%, which makes heat dissipation even more efficient within premises of pSi, suggesting that the induced photothermal effects can be enhanced by thermal properties of the HAuNS organizing materials.
Our results have clearly shown the superior photothermal ablation effect from pSi/HAuNSs. We have demonstrated that the pSi/HAuNSs nanoassembly is much more efficient at heat generation than free AuNPs. Moreover, the nanoassembly offers the benefit of a red shift to access deeper tissues and more efficient energy-to-heat conversion. These properties are enabled through exploiting nanoscale organization features of HAuNSs and are demonstrated for the first time with therapeutic purposes.