Magnetically amplified photothermal therapies and multimodal imaging with magneto- plasmonic nanodomes

Nanotherapies require new ways for controlling and improving the delivery of the therapeutic agents to the site of action to maximize their efficacy and minimize the side effects. This control is particularly relevant in photothermal treatments to reduce the required light intensity and amount of injected nanoparticles, and to minimize necrotic cell deaths. Here we present a novel concept for multifunctional nanobiomedical agents: magneto-plasmonic (MP) nanodomes for magnetically guided and amplified photothermal therapies and as contrast agents for multimodal imaging. The MP nanodomes are composed of a Fe/Au bilayer semi-shell deposited on a 100 nm diameter fluorescent polystyrene nanosphere, which gather a unique combination of straightforward functionalization, high colloidal stability, very strong ferromagnetic behavior and intense optical absorption efficiency in the near infrared. We show that the photothermal

conversion efficiency of the Fe/Au nanodomes with high Fe ratios is substantially larger than pure plasmonic Au nanodomes and the state-of-art plasmonic nanoheaters, i.e. Au nanorods and nanoshells, by merging strong optical absorption, minimized scattering and low optical anisotropy. Remarkably, the effective magnetophoretic concentration of the Fe/Au nanodomes at the illumination region enables large local increase of the optically induced temperature rise. The Fe semishell also provides very intense T2 contrast in nuclear magnetic resonance, which is at least 15-fold larger per particle than commercial iron oxide contrast agents. Moreover, the fluorescent polystyrene nanosphere and the Au semishell integrate valuable fluorescent and Xray contrasts, respectively, which we have used to assess the nanodomes internalization by cancer cells. The MP nanodomes are nontoxic to cells even in the case of magnetophoretic local enrichment with initially high particle concentration (100 g/mL). Remarkably, we demonstrate amplified local photothermal treatments by the magnetic enrichment of the nanodomes at the illumination region, which enables reaching nearly 100% reduction of cell viability with low particle concentration (10 g/mL) and mild NIR laser intensity (5 W/cm 2 ). These results highlight the high potential of MP nanodomes for magnetically guided and amplified photothermal therapies.

Introduction
Nanotherapies are playing an increasingly important role to create new therapies with higher efficacy and lower side effects than traditional chemical treatments, especially in diseases such as cancer [1][2][3][4]. In addition to efficient drug carriers, nanoparticles can also be the source of efficient physical therapies, such as local hyperthermia, which can be employed to thermally destroy the tumors [5][6][7][8][9] or as adjuvant of chemotherapies [10][11][12][13][14]. Photothermal therapies are typically based on plasmonic nanoparticles due to the amplified optical absorption associated with their localized surface plasmon resonance (LSPR) [15,16], which can be tuned to the nearinfrared (NIR), where the skin and capillaries show higher transparency [17]. However, the passive delivery of nanoparticles hampers their control and efficient concentration at the tumor.
The main obstacles that passively delivered nanoparticles encounter in cancer therapies are [18]: i) hepatic, renal or immune system clearance, which can drastically reduce the probability to target the tumor; ii) less pronounced enhanced permeability and retention effect in clinical tumors than in murine models, iii) inherent elevated interstitial hydrostatic pressure in solid tumors that inhibits nanoparticle extravasation, and iv) poor diffusion inside solid tumors due to abnormally high cross-linked extracellular matrix. An attractive way to overcome these hurdles is the external control and guidance of the nanoparticles to the site of action. The ability of magnetic nanoparticles to respond to magnetic forces has fueled the development of nanostructures to magnetically propel the nanoparticles with external magnetic fields [19][20][21][22][23][24].
Therefore, the combination of plasmonic and magnetic materials can be an interesting alternative to overcome the drawbacks associated to the passive delivery of the nanoparticles. Magnetoplasmonic nanoparticles, in addition to enhanced magneto-optic effects [25,26], could combine efficient plasmonic light absorption with magnetic manipulation [27][28][29][30].
To date, the majority of colloidal magneto-plasmonic nanostructures are based on merging plasmonic nanoparticles with small superparamagnetic iron oxide nanoparticles (SPIONs) [30][31][32][33]. However, the small size and weak magnetic moment of the SPIONs, required to minimize their magnetic interaction (which is crucial to ensure the colloidal stability), drastically limits their magnetic actuation capabilities.
Moreover, colloidal magnetic, plasmonic and magneto-plasmonic nanoparticles are generally obtained in organic solvents by chemical synthesis, in which achieving controlled optical and magnetic properties with narrow size and shape distributions can be complex. In addition, the biomedical applications require transferring the nanostructures to an aqueous medium and their functionalization, which are cumbersome processes in nanoparticles obtained from organic-based syntheses that can lead to particle aggregation, especially when heterostructures are involved. In contrast, scalable top-down methods capable of generating macroscopic amounts of colloidal nanostructures in a simple and effective way are considerably less developed. Therefore, achieving cost-effective ways to fabricate multifunctional nanostructures with highly controlled and strong optical and magnetic properties in a large scale may lead to new high-added value biomedical agents.
Here we present novel colloidal ferromagnetic/plasmonic Fe/Au nanodomes enabling: i) excellent colloidal stability; ii) high optical heating efficiency in the NIR comparable to that of state-of-art plasmonic nanoparticles; iii) strong magnetic manipulation via magnetophoretic forces, and iv) very high contrast for fluorescence, nuclear magnetic resonance (NMR) and Xray imaging. We show that the combination of magnetophoretic manipulation and optical heating within the magneto-plasmonic (MP) nanodomes enables a near 100% reduction of cell viability in photothermal treatments in vitro under demanding conditions of low MP nanodomes concentration (10 g/mL) and mild NIR laser intensity (5 W/cm 2 ).

Fabrication of the MP nanodomes
Unlike typical chemical synthesis methods, the MP nanodomes are fabricated by a combination of colloidal nanolithography [34] and physical vapor deposition, which are scalable and cost-effective processes that enable accurate control of their magnetic and optical properties.
A schematic drawing of the different fabrication steps can be seen in Fig. S1  The surface functionalization with proteins (e.g. for cell targeting) can also be carried out directly on the nanopatterned substrate without the need of any chemical linker. In this case we exploit the very high affinity of amine and cysteine groups of the proteins towards bare Au layers. We have actually observed the formation of uniform and stable protein monolayers on the Fe/Au nanodomes by incubating the nanopatterned substrate with a solution of 10 g/mL of protein in water for just 1 h (Fig. S2).
To disperse the particles in water, the wafer together with 10 mL of water was introduced into an ultrasonic bath for 1 min. The MP nanodomes were finally concentrated and redispersed through centrifugation (4000 rpm, 5 min), followed by ultrasonication to achieve highly stable dispersions at the required concentrations. The Si wafers can be reused after cleaning for 10 min in aqua regia, which efficiently dissolves the Fe and Au layers.

Morphological, optical, magnetic and colloidal characterization
To study the size and distribution of MP nanodomes on the Si wafers, scanning electron microscopy (SEM) studies were performed using Quanta SEM 650 (Field Electron and Ion Company (FEI)) at 20 kV. The density of nanospheres coated on the wafer was quantified by using ImageJ software. Transmission electron microscopy, TEM, images and electron energy loss spectroscopy (EELS) analysis were performed in a FEI Tecnai F20 equipped with a Quantum GIF EELS spectrometer. The magnetic characterization of the MP nanodomes was performed on monolayers that were transferred to adhesive tapes to eliminate the magnetic signal from the bilayer that is deposited on the wafer surface. Magnetization loops were acquired at room temperature using a vibrating sample magnetometer (MicroSense, LOT QuantumDesign) with a maximum applied field of 20 kOe. The measurements were performed by applying the field either parallel or perpendicular to the sample, i.e., in-plane or out-of-plane conditions.

Photothermal characterization
A custom-made photothermal testing system was used to determine the photothermal conversion efficiency of MP nanodomes in water (see Stacks of images along the z-axis were obtained for a selected area using the xyz mode of the CLSM and the ImageJ (Fiji) and Bio-formats plugins were used to obtain overlapped images of all channels (nuclei, plasma membranes and MP nanodomes) and the 3D reconstructions and cross-section projections were used to confirm MP nanodomes' internalization.

Soft Transmission X-Ray microscopy
Cells were seeded onto gold grids covered with FORMVAR and carbon foil at a density of Cyototoxicity assay (Life Technologies) following manufacturer's guidelines. Images were acquired using the inverted fluorescence microscope Olympus IX71 (Olympus) and processed through ImageJ (Fiji). Three independent experiments were performed. The MP nanodomes photothermal effect was analyzed using Fisher's exact tests, using Graphpad Prism ® 7.0a software (Graphpad Software). Statistical significance was considered when P < 0.05.

Results and discussion
The MP nanodomes are composed of a 100 nm diameter polystyrene core that is partially coated by a Fe/Au bilayer semi-shell. To show their tunable magnetic and optical properties, in this study we vary the relative thickness of the Fe and Au layers, but we keep a total bilayer thickness of 40 nm (Fig. 1a). The fabrication process yields a monolayer of well-separated nanodomes with a short-range order distribution and homogeneous density of 1.6·10 9 nanodomes/cm 2 (Fig. 1b), which is equivalent to ca. 1.3·10 11 particles per wafer.  Fig. 2b). In the longitudinal configuration there are two main resonances of magnetic dipolar and quadrupolar character, located in the NIR and in the red part of the spectrum, respectively (Fig. S4). In contrast, two electric dipolar and quadrupolar resonances are observed in the transversal configuration that are blue shifted with respect to those in the longitudinal orientation (Fig. S4). Since the dispersed nanodomes are randomly distributed, the experimental spectrum is the convolution of these resonances averaged over all the possible orientations. The resonance band in the NIR region suites perfectly within the spectral region with higher penetration in physiological tissues (biological window) and, therefore, has high potential for photothermal applications. This resonance can be red-shifted even further by reducing the Au thickness (Fig. S5).  Interestingly, the MP nanodomes also offer tunable ferromagnetic properties by modifying the Fe thickness. Vibrating sample magnetometer measurements show that the MP nanodomes exhibit a ferromagnetic behavior at room temperature in all the studied Fe thicknesses (Fig. 2c).
Nanodomes with 5 nm Fe thickness present single domain-like hysteresis loop with small coercitivity, which substantially increases for the 10 nm Fe thickness. Interestingly, a magnetic vortex is formed in the nanodomes with 20 nm and 30 nm Fe thickness showing near zero remanence hysteresis loops, as expected from the size and thickness of the Fe layer [35,36]. This magnetic behavior explains the observed high colloidal stability in all the MP nanodomes, even for high Fe content. Namely, when the magnetic vortex is formed, the magnetostatic interaction in the absence of external magnetic field is negligible. On the other hand, the magnetic dipoledipole interactions of the nanodomes with 5 nm and 10 nm Fe thickness are drastically reduced due to the large thickness of the Au layer and the strong electrostatic repulsion between particles.
We have analyzed the photothermal response of MP nanodomes for the different Fe/Au ratios, with concentrations ranging from 3·10 9 up to 1.2·10 11 nanodomes/mL (Fig. 3b). A typical heating curve is shown in Fig. 3a, which displays the temperature rise in the nanodomes suspension when the laser is switched on, until the thermal equilibrium, due to the equal balance of absorbed and dissipated energy by the sample, is reached. The suspension slowly recovers the initial room temperature level once the laser is switched off.  As expected, the temperature increase follows a linear dependence with the colloidal concentration for low concentration levels. However, the temperature increase saturates at a concentration of 3·10 10 nanoparticles/mL due to the complete absorption of the laser light along the 1 cm thick cuvette for higher nanodomes concentrations. Nevertheless, the most remarkable result is the almost identical optical heating efficiency for all nanodome configurations, regardless of the Fe/Au ratio (Fig. 3b). The reason behind this striking behavior can be inferred Interestingly, the decrease in the absorption efficiency of Fe/Au nanodomes with respect to Au nanodomes in the longitudinal configuration is partially compensated by a higher efficiency in the transversal configuration. In addition, the heating efficiency of nanodomes with high Fe content benefits from their low scattering cross section (Fig. S4), thereby minimizing the backscattered radiation that does not contribute to nanodomes heating. As a result, all the MP nanodomes achieve similar temperature increments for a given laser power and particle concentration. Interestingly, even at rather low particle concentrations (in the 10 9 nanodomes/mL range), temperature increments for therapeutic applications (from 5 ºC to 8 ºC) can be easily where ℎ is the heat transfer coefficient, is the laser irradiating area, Tmax is the optically induced temperature change when the thermal equilibrium is reached, is the heat dissipation from the experimental set-up, I is incident laser power (166 mW), and 808 is the absorbance of the nanoparticles at 808 nm. The value hS is given by: As can be observed in Table 1, the Fe/Au nanodomes efficiency is clearly higher than that of Au nanodomes and is equal to that of the Au nanorods, which are the most efficient plasmonic nanoheaters. Such a high efficiency in the Fe/Au nanodomes is due their minimized scattering cross section of the Fe/Au nanodomes and low optical anisotropy, as it was discussed below. The combination of both effects enables a deeper light penetration and a more uniform heating of the colloidal dispersion in the case of Fe/Au nanodomes. In addition, the heating efficiency is much larger than that of nanoshells, which are nanostructures that exhibit a large scattering cross section, and lower absorption cross section in the near infrared.
In addition to a higher photothermal conversion efficiency, the Fe layers confer the nanodomes with the unique capacity to magnetically control and amplify photothermal therapies.
To experimentally assess such ability, we have first analyzed their magnetic trapping efficiency via magnetophoretic forces, by attaching a cylindrical FeNdB magnet (6 mm diameter, 10 mm length, with a 2.5 kOe field at the surface) at the lateral side of the cuvette that is parallel to the light path (see Fig. S2). To compare the magnetophoretic forces among nanodomes, we have quantified the time that is required to achieve the 95% of the transmitted laser power (taking 100% as a water sample without particles), as a method to determine when the majority of the nanodomes are magnetically trapped at the cuvette wall. Magnetic trapping takes more than 1 h for nanodomes with only 5 nm of Fe. In contrast, the trapping time is reduced to 6 min for 10 nm Fe nanodomes, and less than 2 min for 20 nm and 30 nm Fe nanodomes (Fig. 4a). The large trapping time for 5 nm Fe nanodomes is due to their weak magnetic dipole moment and large mass given by the thick of Au layer. Increasing the Fe content to 10 nm, induces a 9-fold enhancement of the nanodomes magnetization (Fig. S6), probably due to a reduced magnetization at the Fe/Au interface [38,39], whose net magnetic effect is more pronounced for The only slight reduction of the trapping time in the 30 nm Fe nanodomes compared to that in the 20 nm Fe nanodomes is due to the more tilted hysteresis loop in the former ones (see Fig.   2c), which compensates their higher Fe content and lower mass. nanodomes in the laser path. The initial particle concentration is 2.4·10 9 particles/mL.
To highlight the strength of the magnetic manipulation it is worth comparing the magnetic dipole (m) that can be generated in the nanodomes and in standard colloidally stable SPIONs.
The maximum magnetic dipole that can be generated in the nanostructures is given by m = MS VP, where MS is the saturation magnetization and VP is the volume of the magnetic element. In the case of nanodomes with 20 nm Fe thickness, the maximum magnetic dipole is ca. 630-fold larger than that of FDA (US Food and Drug Administration)-approved SPIONs with 12 nm diameter, and almost three orders of magnitude larger for Fe 30 nm nanodomes. This huge difference is due to the 3-fold higher MS of metallic iron compared to that of iron oxide (i.e., 1716 emu/cm 3 for Fe versus 476 emu/cm 3 for Fe3O4) and the much larger volume of the Fe layer in the nanodome. As a result, the nanodomes can act as very strong nanomagnets in the presence of a magnetic field (thus, reacting quickly to them), although they can keep high colloidal stability in the absence of magnetic fields due to their near zero remanence (given by their magnetic vortex state). Actually, mild sonication can easily disperse the magnetically trapped nanodomes and fully recover the homogeneous colloidal dispersion once the magnet is removed.
Interestingly, the efficient magnetophoretic manipulation can be used to locally amplify the optical heating efficiency by increasing the particle concentration at the illumination region. This effect is demonstrated for Fe 20 nm / Au 20 nm nanodomes in Fig. 4b, in which an 85% enhancement in the temperature rise is observed when particles are magnetically concentrated at the region that blocks the laser path in the cuvette wall (see Fig. S2). As we show below, this effect is especially valuable to locally enhance the photothermal treatments for efficiently killing tumor cells.
In addition to the significant fabrication cost reduction with respect to pure plasmonic nanoparticles and the magnetically enhanced photothermal effects, the Fe layer in the MP nanodomes provide an intense NMR contrast for imaging. To maximize the NMR signal we focus the NMR analysis on the MP nanodomes with the largest magnetic dipole, i.e. 30 nm of Fe. We have studied the relaxation rates (R1 = 1/T1 and R2 = 1/T2) as a function of the Fe molar concentration in samples of MP nanoparticles dispersed in agar gel (Fig. 5a). While we do not observe any significant T1 contrast, which is a consequence of the presence of the diamagnetic Au layer in contact to water molecules, we detect very high T2 contrast. The slope of the R2 curve as a function of Fe concentration reveals that the relaxivity r2 is 26627 mM -1 s -1 . This relaxivity is between 1.5-and 2-fold larger than that of superparamagnetic iron oxide nanoparticles that have been commercially available as T2 contrast agents, like Feridex (105 mM -range (520 eV). By employing cryogenic conditions, the soft X-ray microscope enables operating in an environment close to the hydrated physiological conditions. Thus, soft X-ray microscopy can also yield 3D structural information of the entire cell without the need of fixation, dehydration, embedding and sectioning of the samples. The resolution of the soft X-ray microscope is about 40 nm, which is sufficient to visualize individual MP nanodomes at different cellular planes and to map their interaction with the cellular compartments. The X-ray images of showing similar morphology and X-ray absorption. These results highlight the potential of the MP nanodomes as contrast agents in X-ray computed tomography. can be seen in the orthogonal cross-sections (Fig. 5d). Discrete red spots, corresponding to MP nanodomes trapped in endosomes/lysosomes can be observed inside the cytoplasm, but not inside the nucleus due to their large size (100 nm in diameter), which prevents crossing the nuclear porous complex [41]. Probably, MP nanodomes are internalized via pinocytosis, forming endosomes that can be visualized as discrete points.
To assess the biomedical potential of the MP nanodomes, we finally tested the effect of photothermal treatments in HeLa cells in challenging conditions of low particle concentration and mild laser intensity (Fig. 6). We used glass bottom dishes with a thickness of 0.17 mm to facilitate light irradiation, where cells were seeded only in the glass region. Cells were incubated for 3 h with cell culture medium containing a concentration of either 10 μg/mL or 100 μg/mL of Fe 20 nm/ Au 20 nm nanodomes. We have compared the photothermal treatment efficiency either with or without magnetic field concentration of the nanodomes at the illumination region. After the incubation time, the cell culture medium with nanoparticles was replaced by 1 mL of fresh medium to leave only the particles in contact or inside the cells. The cell monolayer was irradiated with a laser at 808 nm emission wavelength during 30 min with an incident intensity of 5 W/cm 2 . Due to the limited irradiated area (laser spot diameter of 6 mm), we analyzed cell viability using a colorimetric assay (LIVE/DEAD ® ), which enables quantifying live (green) and dead (red) cells according to their esterase activity and membrane integrity, respectively (Fig. 6 a-c). In these assays we compared the light treatment effect in control samples without nanodomes and in samples with two different initial concentrations of nanodomes (10 g/mL and 100 g/mL), either in the absence of magnetic field or with magnetic concentration at the illumination region by a spherical FeNdB magnet (12 mm in diameter, 2.5 kOe at the surface) (see Fig. S8). Firstly, the results show that irradiated cell cultures without MP nanodomes do not exhibit statistically significant differences with the non-irradiated control (Fig. 6a) under magnetic concentration conditions (Fig. 6a). Finally, we assessed the effect of light irradiation in cells containing MP nanodomes. The results show a minimal decrease in the cell viability without magnetic concentration for a nanodomes concentration of 10 g/mL. The concentration must be increased up to 100 g/mL to observe a significant viability reduction in the absence of magnetic concentration. In contrast, a viability reduction of nearly 100% after the light treatment is observed in the case of magnetic concentration at the illumination region for both initial nanodomes concentrations of 10 g/mL and 100 g/mL (Fig. 6a). Interestingly, such drastic viability reduction is observed 24 h after the treatment for the initial 100 g/mL concentration, whereas it takes 48 h to get the near 100% reduction when the initial concentration is 10 g/mL. The different light effect on the cancer cells can be clearly observed in the scanning electron microscopy images of Fig. 6 d-f. For magnetically concentrated samples with initial 100 g/mL concentration, a clear loss of cell membrane integrity is observed, thereby reflecting the unfavorable fast necrotic cell death that is caused by the very intense photothermal actuation under these conditions. On the contrary, for the initial concentration of 10 g/mL, the cell membrane disruption is not perceived, although there is a clear morphological change in the cancer cells. The longer time to generate the viability reduction suggests a more controlled cell death pathway, which is desirable for the photothermal treatments. Importantly, the magnetic concentration at the light treatment region enables at least a 10-fold reduction of the quantity of injected particles to achieve the nearly 100% therapeutic efficacy, which can be highly relevant to reduce the therapy cost and the bioaccumulation of the nanoparticles in other organs.
The comparison of our results with other nanoparticles for photothermal therapies is not straightforward, since different composition, surface functionalization and concentration of particles have been employed, in addition to different irradiation conditions, and cell types [42].
Nevertheless, some trends can be found regarding particle concentration and exposure conditions. Studies that use high nanoparticle concentration (from 100 to 250 μg/mL) apply low or mild exposure conditions (from 0.1 to 10 W/cm 2 ), while studies using low nanoparticles concentration (from 6.6 to 36.5 μg/mL) typically need higher exposure conditions (from 15.3 to 250 W/cm 2 ) to obtain good efficiencies on cell death. In the present work, MP nanodomes have demonstrated to be a good candidate for photothermal treatments using low initial concentration (10 μg/mL) and mild incident intensity (5 W/cm 2 ) compared to other published studies [43]. This laser intensity is higher than the recommended clinical safety values (0.3 W/cm 2 ), but our in vitro assays with cell monolayers are much more demanding from the optical perspective than in vivo conditions, in which light can be efficiently absorbed along several millimeters. In contrast, the light absorption region is thinner than 10 m in a 2D cell culture with nanoparticles only in the cell monolayer. The generated heat in the cell monolayer rapidly diffuses towards the cell medium and the glass substrate due to the generated temperature difference between the cell monolayer and both the cell media and the substrate, thereby forcing the need of higher light intensity to achieve temperature increments with sufficient therapeutic effect.
Finally, it is worth mentioning that, although the magnetic manipulation is a powerful tool to locally control the photothermal effects, the nanodomes could offer additional cell targeting by exploiting their straightforward functionalization with specific antibodies, proteins such as transferrin (see Fig. S2), and molecules such as folate [44], musuin7 [41], RGD peptide [42] or other specific molecules capable of preferentially binding the MP nanodomes to cancer cells overexpressing particular membrane receptors.

Conclusion
We have shown that MP nanodomes can have high potential for therapeutic and diagnostic applications. Compared to other magnetic, plasmonic or magneto-plasmonic nanoparticles fabricated by chemical synthesis, our fabrication process enables easier control in the magnetic and optical properties. The capacity to functionalize the MP nanodomes on the substrate and their direct dispersion in water or buffer represent also significant advantages with respect to chemically synthesized nanoparticles, yielding highly stable colloidal dispersions during months.
Compared to state-of-the-art plasmonic nanoparticles for photothermal therapies, MP nanodomes offer higher heating efficiency at lower cost given by their strong manipulation via magnetophoretic force, very high contrast for NMR and X-ray imaging, and easy incorporation of fluorophores in the polymer core. Compared to iron oxide nanoparticles used in magnetic hyperthermia, the high heating efficiency of MP nanodomes enables local thermal treatments at much lower particle concentration [45]. Moreover, MP nanodomes show much higher T2 contrast per particle for NMR imaging than that of commercial iron oxide nanoparticles. Demonstration of both low cytotoxicity and magnetically enhanced efficiency for photothermal therapy at low particle concentrations and mild light intensity encourages the transfer of this nanotechnology to in vivo therapies. The high optical heating efficiency of the nanodomes could be also applied to develop temperature responsive drug delivery systems that could be magnetically controlled and visualized via computed X-ray tomography, NMR imaging, or fluorescence. The optical anisotropy in nanodomes with low Fe ratio and their capacity to efficiently rotate in the liquid under an alternating magnetic field can also provide interesting tools for the development of nanobiosensors and nanothermometers [46].