This document is the accepted manuscript version of a published work that appeared in final form in Dalton Trans., 2018, 47, 11960, DOI: 10.1039/c8dt01687a View Article Online PAPER Multimodal Prussian blue analogs as contrast agents for X-ray computed tomography Premrudee Promdet, Bárbara Rodríguez-García, Alexandria Henry, Cathie Nguyen, Thien Khuu, Jose-Ramon Galan-Mascaros and Karn Sorasaenee Prussian blue analogs (PBAs) are versatile materials with a wide range of applications. Due to their tunability, intrinsic biocompatibility, as well as low toxicity, these nanoscale coordination polymers have been successfully studied as multimodal contrast agents for multiple imaging techniques. Herein, we report the expanded biomedical application of PBAs to X-ray computed tomography (CT). In our systematic study of the series A{MnII[FeIII(CN)6]} (A = K+, Rb+, Cs+), we showed that derivatives incorporating Rb+ and Cs+ ions in the tetrahedral sites of the parent face-centered cubic cyano-bridged networks exhibited substantially increased X-ray attenuation coefficients, thus yielding significant contrast compared to the clinically approved X-ray contrast agent iohexol at the same concentrations. Additionally, our μ-CT studies revealed that these PBAs could be useful as dual-energy CT contrast agents for different biological specimens by using the lower varying scanning X-ray tube voltages. Finally, in vitro studies using U87-Luc cells treated with PBAs, including cellular CT imaging and bioluminescence cell viability assays, revealed that PBAs were taken up by the glioblastoma cells, with moderate biocompatibility at concentrations below the mM range. 1. Introduction The family of Prussian blue materials and its analogs (PBAs) have been used in multiple applications for over 300 years.1 Obtained from a combination of transition metal cations and cyanide ligands, these coordination polymers offer a unique versatility and tunability. They can be regarded as frontier materials between the solid-state chemistry of metal oxides and the molecular chemistry of transition metal complexes, as they exhibit characteristic features of both. On the one hand, they are robust, stable compounds that can accommodate multiple stoichiometries, redox states and variable compositions in their rigid structures (Fig. 1). On the other hand, a Translational Biomedical Imaging Laboratory, Department of Radiology, The Saban Research Institute, Children’s Hospital Los Angeles, Keck School of Medicine of USC, Los Angeles, CA 90027, USA. E-mail: ksorasaenee@chla.usc.edu b Institute of Chemical Research of Catalonia (ICIQ), The Barcelona Institute of Science and Technology (BIST), Av. Països Catalans 16, Tarragona, E-43007, Spain. E-mail: jrgalan@iciq.es c ICREA, Passeig Lluïs Companys 23, Barcelona, E-08010, Spain d Department of Chemistry and Beckman Institute, California Institute of Technology, Pasadena, CA 91125, USA † Dedicated to Prof. Kim R. Dunbar on the occasion of her 60 th birthday. ‡ Electronic supplementary information (ESI) available: Additional physical characterization, elemental X-ray attenuation graphs as well as detailed experimental data. See DOI: 10.1039/C8DT01687A they are crystalline, light weight, transparent, insulating coordination polymers, conveniently prepared from the corresponding molecular building blocks. This chemical variety makes them very unique materials. After decades of PBA chemistry, these materials are still at the forefront of scientific research in various fields. For example, due to (but not limited to) their redox tunability and charge transfer features,2 porosity,3 negative thermal expansion,4 or proton conductivity,5 they have been studied as room temperature magnets,6 biosensors,7 electrocatalysts,8 and electrochromic displays. 9 In biomedical fields, the potential use of PBAs as molecular imaging contrast agents has been investigated in recent years.10–17 In particular, their intrinsic physical and chemical properties make them attractive candidates for multimodal molecular imaging.16,17 Multimodal agents are advantageous because multiple complementary imaging modalities allow for a more sensitive and specific diagnosis. Depending on their composition, specific derivatives have shown contrast activity for multimodal diagnostic techniques, including magnetic resonance imaging (MRI),10 optical imaging including second harmonic generation (SHG),11,18 and photoacoustic imaging.12 The latter capability was exploited to construct multimodal X-ray contrast agents (XCAs) by coating gold nanoparticles with Prussian blue, demonstrating the great potential of these materials.19 We note that the intrinsic application of these such as magnetic resonance, optical or acoustic imaging. At the same time, the incorporation of cations with a high X-ray photon attenuation capability, such as Rb or Cs, in these tetrahedral sites may transform these nanoscale multifunctional materials into competitive X-ray contrast agents (nano-XCAs). Moreover, we report how these PBAs would be appropriate as DECT contrast agents to improve differentiation among various types of biological materials and specimens in both pre-clinical and clinical settings. Our in vitro biological studies indicate good potential biocompatibility for these nano-XCAs, at least comparable to other solid nanoparticles in the same size range. We, therefore, anticipate that this work will provide a foundation for further development of PBAs as true multimodal contrast agents, incorporating computed tomography into their already large portfolio of imaging applications. 2. Experimental 2.1. Materials Millipore water was used in all biological experiments. Dulbecco’s Modified Eagle’s Medium (DMEM) and heat-inactivated fetal bovine serum (FBS) were obtained from the Fig. 1 (a) Representation of the ideal crystal structure of PBAs, show- American Type Culture Collection, ATCC. The combination casing the tetrahedral sites occupied by alkali cations (trivalent metal = antibiotic primocin was purchased from Invivogen. Phosphate yellow; divalent metal = red; alkali metal = purple; carbon = black; nitro- buffered saline (PBS) and Hoechst 33258 were purchased from gen = blue). (b) X-Ray diffraction powder patterns for materials 1–3. Thermo Fisher Scientific. The glioblastoma cell U87-Luc was obtained in-house from the Small Animal Research Imaging Center Laboratory of the Saban Research Institute of PBAs in X-ray computed tomography (CT) has not been Children’s Hospital Los Angeles. Iohexol and irinotecan were studied, to the best of our knowledge. purchased from GE Healthcare and Sigma-Aldrich, Remarkable biomedical engineering and technological respectively. advances in X-ray computed tomography, including the recent availability of high-end multi-source multi-detector CT scanners for dual-energy computed tomographic (DECT) measure- 2.2. Synthesis of Prussian blue analogs ments, have been made in the past decade.20–22 Development K0.1 Mn[Fe(CN)6]0.70 ·7H2O (1): An aqueous solution of MnCl2 of intravenous XCAs, on the other hand, has been somewhat (0.1 M) was added to a solution of K3Fe(CN)6 (0.1 M). The limited, and it is currently a major research field in radi- reaction mixture was stirred for 1 h. The solid product was ology.23,24 Conventionally, XCAs are small molecules contain- separated by centrifugation, washed 3× with 30 mL of dising iodine (Z = 53) for it exhibits high X-ray attenuation with tilled water, and dried in air. IR: 2148 (s), 659 (w), 418 (w). its K-edge energy at 33 keV.25 It was only recently that develop- Particle size: 479 ± 100 nm. Zeta potential: −19 ± 6 mV. ment of biocompatible nanoscale XCAs, such as gold (Z = 79) Rb0.5Mn[Fe(CN)6]0.84·8H2O (2): An aqueous solution of MnCl2 and bismuth (Z = 83) nanoparticles (AuNP and BiNP, respect- (0.1 M) and RbCl (1 M) was added to a solution of K3Fe(CN)6 ively), has been explored.26–29 The well-defined nature of these (0.1 M) in an ice bath. The reaction mixture was stirred for XCAs makes it difficult to improve or fine-tune their X-ray con- 1 h. The solid product was separated by centrifugation, trast features. A recent study carried out to improve synthetic washed 3× with 30 mL of distilled water, and dried in air. IR: methodologies for elemental BiNP suggest that although not 2150 (s), 2090 (sh), 2075 (m), 528 (w), 417 (w). Particle size: trivial, a BiNP with a dense core and high bismuth payload, 160 ± 25 nm. Zeta potential: −21 ± 7 mV. Cs1.2Mn and a large X-ray contrast, thus producing a large X-ray con- [Fe(CN)6]0.93·xH2O (3): An aqueous solution of MnCl2 (0.1 M) trast, can be prepared.28 and CsCl (1 M) was added to a solution of K3Fe(CN)6 (0.1 M) In this work, we demonstrate that nanoscale Prussian-blue in an ice bath. The reaction mixture was stirred for 1 h. The analogs can also be attractive candidates for X-ray CT, particu- solid product was separated by centrifugation, washed 3× larly, Type II Prussian blue analogs AI MAII[MBIII(CN)6](x+2)/3 with 30 mL of distilled water, and dried in air. IR: 2151 (s), x incorporating a heavy alkali counter cation AI. The alkali 2077 (m), 592 (w), 527 (w), 415 (w). Particle size: 137 ± 46 nm. cation does not modify or affect the electronic properties of Zeta potential: −15 ± 7 mV. 2.3. Physical characterization 2.5. PBAs,Micro-computed tomography (μ-CT) imagingcapabilities, maintaining all other intrinsic imaging measurement and analysis: in vitro studies Size distribution and zeta potentials were determined by dynamic light scattering using a NanoSZ (Malvern). Scanning Pathogen-free U87-Luc cells were grown in a 75 mL flask in electron microscopy was carried out using a JEOL-JMS6400 DMEM in 5% CO2 at 37 °C. The cell culture medium was supenvironmental scanning electron microscope. FT-IR spectra plemented with 10% FBS and 1% primocin combination antiwere collected in transmission mode using a Bruker Optics biotic. The cell culture medium was replenished every two FT-IR Alpha spectrometer in the 4000–400 cm−1 range. Powder days and the cells were passaged once they reached 80% conX-ray diffraction (PXRD) data were collected on a D8 Advance fluence. The glioblastoma U87-Luc cells (20 000 cells per well) series 2theta/theta powder diffractometer at room were plated in an 8-chamber slide (Nunc™ Lab-Tek™, Thermo temperature. Scientific) and treated with PBAs 1–3 (25 mg mL−1) in 300 μL total well volume for 24 h after which the cells were washed 3× and detached from the culture slide. The molar concentrations 2.4. Micro-computed tomography (μCT) imaging of PBAs 1, 2 and 3 in the wells were 7.32 × 10–2, 5.82 × 10–2, measurement and analysis: phantom studies and 4.71 × 10–2 M, respectively. The cells were then transferred All PBA samples were originally weighed out, suspended in to 500 μL Eppendorf tubes and spun for 5–10 min at 12 000 1.0 mL pH 7.4 PBS, and thoroughly sonicated. Iohexol was rpm in a centrifuge to form pellets. The pelleted samples conincluded as a control. Each sample, including iohexol, was taining U87-Luc treated with PBAs 1–4 in the corresponding diluted with 2.5% agarose gel to achieve six final concenEppendorf tubes were mounted vertically onto a styrofoam trations (1.00, 6.25, 12.5, 25.0, 50.0 and 100 mM) based on the platform. All acquisition and reconstruction parameters were corresponding formula units. Each sample was prepared in similar to those described above in the phantom studies. The triplicate in 0.5 mL microcentrifuge tubes. The samples were quantification region was selected from the bottom 151 slices allowed to solidify completely prior to the CT scans. The tubes of the stack starting from the first bottom slice that showed were mounted vertically on a specimen holder 45 mm in dia- signals, and each slice of interest is the middle slice of their meter filled at the bottom with a 5 mm layer of clay. Each scan respective stacks. included a set of 5 samples of the same concentration, oriented radially with a blank tube positioned at the center of the stage. A wood handle was also placed between two of the 2.6. Cell culture and cell viability assays Pathogen-free U87-Luc cells were grown in a 75 mL flask in samples to serve as a reference point. All samples were scanned with a SkyScan 1172 micro-CT DMEM in 5% CO2 at 37 °C. The cell culture medium was supscanner (Bruker Corporation) at a standard resolution of plemented with 10% FBS and 1% primocin combination anti26.74 μm per pixel, with an image matrix of 100 × 575 pixels. biotic. The cell culture medium was replenished every two All scans were performed at a tube current of 140 mAs with days and the cells were passaged once they reached 80% convarying X-ray tube voltages (25, 30, and 40 kV). Exposure was fluence. U87-Luc cells (20 000 cells per well ) were plated in an automatically terminated when the detector reached 70% 8-chamber slide (Nunc™ Lab-Tek™, Thermo Scientific) and fill factor. No filters were used. The angular position of treated with PBAs 1–3 suspended in PBS over a range of 10–9 acquisition was set at 1° rotational steps, and 2 frames were to 10–3 M. A primary stock solution (6.3 mg PBAs 1–3 in 1 mL taken at each angular station. The field of view was pH 7.4 PBS) was prepared. The primary stock solution was increased horizontally with the offset camera mode to further diluted to prepare secondary and tertiary stock soluencompass all of the tubes within each scan. The two tions. The various amounts of stock solutions were added to frames taken at each rotational station were averaged to the 8-chamber glass slide plated with cells to give the aforeproduce the raw data set for reconstruction. Image recon- mentioned range of concentrations. The final volume of each struction was performed using NRecon Server (Version chamber is 300 μL. After treatment, the treated cells and con1.7.3.0; Bruker Inc., Billerica, MA) and CT-Analyser (Version trols were incubated in the dark in 5% CO2 at 37 °C for a 1.13; Bruker Inc., Billerica, MA) for quantification. The period of 24 h. The concentrations of each PBA based on Hounsfield unit (HU) scale (−1050 to 4500) was chosen with their formula unit are detailed in Table S1.‡ The cells were regard to the histogram displaying the data dynamic range imaged using the cooled IVIS® animal imaging system calculated by NRecon before reconstruction. The top and (Xenogen, Alameda, CA, USA) linked to a PC running with the lower ends of the scale were chosen around the region of the Living Image™ software (Version 4.0; Xenogen) along with peak and tail ends of the histogram. All data were recon- IGOR (Version 6.3; Wavemetrics, Seattle, WA, USA) under structed and analyzed within this dynamic range. The Microsoft® Windows® 2000. This system produced high attenuation value resulted from the mean brightness inten- signal-to-noise images of luciferase signals emerging from −1 sity calculated from this HU scale. The total reconstructed the cells. Before imaging, 20 μL of 5 mg mL luciferin in output was 575 slices per tube. The volume of interest (VOI) normal saline was added to each well. An integration time of was further narrowed down for quantification to the middle 1 min with binning of 5 min was used for luminescent image 201 slices, with each slice of interest being the exact middle acquisition. The signal intensity was quantified as the flux of all detected photon counts within each well using the Living slice of the VOI. triplicate. 2.7. In vitro Confocal Fluorescence Microscopy The U87-Luc cells were seeded at 20 000 cells per well on an 8-chamber slide (Nunc™ Lab-Tek™, Thermo Scientific) and allowed to grow overnight. Cells were washed with PBS and incubated with 25 mg mL−1 PBAs 1–3 in serum-free media at 300 μL total volume for 24 h at 37 °C. Molar concentrations of PBAs 1, 2, and 3 in the wells were 7.32 × 10–2, 5.82 × 10–2, and 4.71 × 10–2 M, respectively. Cells were then washed 3× with PBS and stained with Hoechst 33258. The stained cells were kept chilled on ice without fixation until just prior to imaging. Images were acquired using an LSM 710 confocal system mounted on an AxioObserver Z.1 inverted microscope equipped with a 40×/1.2 C-APOCHROMAT water-immersion lens (Carl Zeiss Microimaging, Thornwood, NY). Two visible laser lines of 405 and 488 nm were used for fluorescence excitation. The ZEN 2010 software (Zeiss) was used for hardware control. 3. Results and discussion We selected the AxMn[Fe(CN)6] Prussian blue analogs for our studies. These series present very good stability in solution, thanks to the [Fe(CN)6]3− building blocks, and have been shown to be multimodal contrast agents, thanks to their paramagnetic nature and electronic properties.11 As blank material we prepared K0.1Mn[Fe(CN)6]0.70 (1) by direct synthesis from the precursor potassium hexacyanoferrate and manganese dichloride. In order to determine the positive (if any) X-ray contrast effect of heavy alkali metals embedded in the tetragonal sites of this Prussian blue derivative, we prepared Rb0.5Mn[Fe(CN)6]0.84 (2), and Cs1.2Mn[Fe(CN)6]0.93 (3) following the same experimental protocol but in a concentrated solution of the desired alkali cation in water. Metal stoichiometry was determined by microanalysis (EDX). The same reaction conditions yielded materials with variable alkali cation contents, due to the higher affinity of larger alkalis to be trapped in the Prussian blue network. X-Ray powder diffraction confirmed the isostructural face center cubic structures for PBAs 1 and 2 where some reflections are missing. This is typical of PBAs with high occupancy of the tetrahedral sites, which reduces the symmetry of the cell.18,30–34 Particle size was determined by dynamic light scattering (DLS) analysis and confirmed by electron microscopy (Fig. S1 and S2‡), yielding 479 ± 100 nm (1), 95 ± 22 nm (2) and 137 ± 46 nm (3). The IR spectra (Fig. S3‡) showed one CN stretching band at 2148–2152 cm−1 (strong) that can be assigned to the FeIII–CN–MnII. As the alkali content increased, a second band appeared at 2075–2077 cm−1 (medium) corresponding to the FeII–CN–MnII pair. This has been previously assigned to partial reduction of the FeIII centers, imposing the need for extra cations to maintain electroneutrality in these non-stoichiometric materials.35 All samples exhibited moderate negative zeta potentials. This is helpful to prevent aggregation in solution. Indeed, suspensions of these materials in water are relatively transparent as was confirmed by UV-Vis spectra of water suspensions (Fig. S4‡). Nanoparticles of these materials were suspended in 2.5% agarose gel at various concentrations based on their formula unit (1.00, 6.25, 12.5, 25.0, 50.0, and 100 mM). X-Ray attenuation measurements were performed using a single-slice or whole-volume quantification at 40 kV. The response from 1 can be taken as that corresponding exclusively to the PBA framework, since almost all tetrahedral sites are vacant (over 90%). As expected, the incorporation of heavy alkali metals clearly enhances the contrast response (Fig. 2 and S5‡) per molecular weight. In all cases, a direct linear relationship is observed for the attenuation coefficients (AC) with respect to the PBA concentration (Fig. 2), with the substituted derivatives showing several times the response of the parent compound. 3 shows the highest AC, but it also brings a higher heavy alkaline content. If we normalize the response taking into account the stoichiometry of the heavy alkaline metal (Fig. 2b), 2 promotes a better contrast, which was noted to correlate with the corresponding mass attenuation coefficients reported (Fig. S5‡).36 Additionally, the low occupancy reached in 2 also suggests that AC may be improved, increasing Rb+ occupancy in the tetrahedral sites. The samples were also scanned at two additional X-ray tube voltages, namely 25 and 30 kV, available on our high-resolu- Fig. 2 X-ray attenuation for 1–3 and the model XCA iohexol at 40 kV (a, b) and at 30 and 25 kV (c, d). The attenuation is plotted as a function of molar concentration, C (a, c), and as a function of the heavy alkali metal cation concentration, [A] (b, d). Table 1 X-Ray attenuation coefficients (AC) in HU were performed in Image™ software package. All experiments mM−1 for Prussianblue analogs 1–3 of formula AxMn[Fe(CN)6] and the standard iohexol Table 2 X-Ray attenuation coefficients slightly mg−1 for Prussian(Fig. 1b). Compound 3 revealed a (AC) in HUdifferent pattern, blue analogs 1–3 and the standard iohexol AC (HU mM−1)a AC (HU L mg−1)a per mmol per mmol of A X-ray tube voltage (kV) X-ray tube voltage (kV) XCA 1 2 3 Iohexol a 25 3.70 7.92 8.88 7.12 30 2.82 6.53 8.78 5.69 40 1.94 4.91 7.38 4.48 XCA 25 — 15.85 7.40 — 30 — 13.05 7.32 — 40 — 9.83 6.16 — AC@25 kV AC@30 kV AC@40 kV 1 2 3 Iohexol 22 37 34 17 17 30 33 14 11 23 28 11 a Extracted from the slope of the AC vs. C plots. Extracted from the slope of the AC vs. C plots. tion small-animal X-ray scanner despite their non-clinical X-ray voltage range (Fig. 2c and d). High-resolution X-ray contrast images were also obtained in triplicate at various scanning voltages for the PBAs studied. An example of such images is shown in Fig. S6.‡ The X-ray attenuation coefficients as a function of molar concentration (C), obtained from the AC vs. C plots (Table 1), appeared to be higher at a lower set voltage. For instance, the results for the samples scanned at 25 kV voltage were the most opaque compared to the other two set tube voltages. This is in contrast with recent reports for other metal-containing XCAs. Cormode and co-workers suggested steady trends in attenuation over higher scanning tube voltages (80–140 kV), employed in clinical settings, for bismuth nanoparticles and ions.28 We found the opposite trend, except for PBA 3 whose response is almost voltage-independent. Compared to the iodine standard iohexol,37 our PBAs exhibited substantially higher attenuation coefficients per mol in all of the range of X-ray tube voyages employed (25, 30, and 40 kV), a typical set for small animal studies. The difference is even more remarkable per gram, due to the lower molecular weight of PBAs (Fig. 3) (Table 2). Our observation of tunable radio-opaqueness over a range of low X-ray tube voltages could be useful for the development of nanoscale dual-energy XCAs for small-animal studies. We further inspected the potential development of these PBAs as DECT XCAs to allow for improved differentiation of image contrasts in different types of biological specimens. As the techno- logy emerges, DECT is a relatively new imaging technique to enhance the visualization of different organ systems, relying on the difference between the dual energy ratio (DER) of the materials exposed to the X-ray beams determined by the separation of the high- and low-energy spectra given by eqn (1):38 DER ¼ AClowV AChighV ð1Þ High DER values could implicitly be indicative of improved contrast differentiation for given materials. In addition to the difference in X-ray scanning voltages,39 particularly for contrast-enhanced DECT imaging, the contrast also depends on the difference between the DER of the contrast and that of the biological specimens. In this work, using the aforementioned three scanning voltages, we evaluated the attenuation ratios at two DER settings, namely, at 25 kV/30 kV and at 25 kV/40 kV (Table 3). A previous report by Krissak and co-workers examined DER at the scanning X-ray tube voltages of 80, 100, 120, and 140 kV, applicable in clinical settings, for compact bone and muscle.38 The corresponding DERs with the range of 1.20–1.45 and 1.01–1.08 for compact bone and muscle, respectively, were relatively constant. By comparing these DERs to those of gold and iodine XCAs, it was determined that the iodine XCA was a better DECT contrast for muscle due to a larger difference in their DERs than the gold XCA, which was a better DECT contrast for compact bone due to a similar Table 3 Dual-energy ratios (DER)a for PBAs 1–3 and various tissue samplesb Sample DER(25/30) DER(25/40) 1 2 3 Iohexol Musclec Soft tissuec Bonec 1.31 1.21 1.04 1.25 1.11 1.12 1.47 1.91 1.61 1.24 1.59 1.19 1.22 2.29 a Fig. 3 X-ray attenuation for 1–3 and the model XCA iohexol at 40 kV (a) and at 25 and 30 kV (b) normalized to XCA mass content. DER denotes the ratio between the AC value at a low X-ray voltage and the AC value at a higher X-ray voltage. b The tissue values were calculated based on the mass X-ray attenuation coefficient (μ/ρ) reported in the literature.40 c The low voltage value for DER estimation for tissue samples was 20 kV, instead of 25 kV. reason. Following the same rationale, our results at low scanning voltages for small-animal CT studies suggest that PBA 1 could be a better XCA for muscle, while PBA 3 could be a more suitable XCA for compact bone. We, however, note that there were appreciable differences in DER values for 3 and iohexol in both muscle and bone, suggesting that they may not be useful as DECT contrasts under these experimental settings. Additionally, we evaluated in vitro imaging of the glioblastoma U87-Luc labeled with our PBAs, using X-ray μ-CT to demonstrate cancer cell labeling with the nanoscale coordination polymers. Fig. 3a shows the μ-CT images of cells treated with and without PBAs scanned at 25 kV. The U87-Luc cells were chosen as the model cell line in this study because they have been well studied in our laboratory in the past. These cells were dispersed in culture media and treated with highly concentrated suspensions of 1–3 (25 mg mL−1) at 37 °C for 24 h. Although this concentration is toxic for this or any other nanoparticle-based drug,41–43 these experiments were aimed at confirming the cellular PBA uptake. X-Ray CT images of the PBA-treated collected cells showed considerable contrasts when compared to the control. As expected, based on the phantom studies, cells labeled with PBA 2 and 3 exhibited the highest enhanced contrast, particularly at lower scanning X-ray tube voltages (25 kV). Internalization of the PBAs was also confirmed by the corresponding bright-field image of the U87-Luc treated PBA 2 (Fig. 4). The arrows in Fig. 4b indicate the location of 3 nanoparticles and the blue domains designate the nuclei stained with Hoechst 33342 (see also Fig. S7‡). These images obtained from transmitted light microscopy correlate well with the findings of the in vitro μ-CT images, suggesting potential labeling of the U87-Luc cells with PBAs and their probable cellular uptake. It is important to highlight the non-toxic nature of Prussian blue and most of its derivatives.44,45 Prussian blue treatments of rat showed low oral toxicity and those of rabbit exhibited no skin or eye irritation. Treatment of bacteria using the parent Fe4[Fe(CN)6]3 provided no evidence of mutagenicity. Additionally, repeated oral administration in high doses in humans did not result in harmful adverse effects. In humans oral doses of Prussian blue up to 10 g per day are well tolerated. Indeed, the FDA has approved Prussian blue as an antidote against non-radioactive and radioactive thallium as well as radioactive cesium poisoning.46 However, all these medical applications do not require cell uptake. Because of this, we carried out in vitro viability assays on U87-Luc cells treated with various PBA concentrations (up to 10–3 M). The cells were incubated in an 8-well culture plate at 37 °C for 24 h, after which their viability assays were evaluated using the bioluminescence technique in triplicate. Quantification of the bioluminescence signals from the cells treated with PBAs could be correlated to cell viability.47 These assays revealed that all PBAs were relatively non-toxic compared to the clinically approved anticancer drug irinotecan, which yielded a viability of less than 50% at all the concentrations studied (Fig. 5). In the case of PBAs, the cytotoxic effect on U87-Luc cells became drastically apparent (cell viability below 30%) at Fig. 4 Uptake of Prussian blue XCAs by U87-Luc studied by μ-CT imaging. U87-Luc cells were treated with PBAs 1–3 for 24 h, after which the cells were detached, washed, and analyzed using the μ-CT scanner. Blank images correspond to those of cells without PBA treatment. (a) μ-CT images of PBAs scanned at 25 keV and 5 μm resolution. (b) Transmitted light microscopy image showing the uptake and internalization of PBA in U87-Luc cells. The blue areas are nuclei stained with the nuclear dye Hoechst 33342. Fig. 5 Percent viability assays of the glioblastoma U87-Luc after a 24 h treatment with PBAs 1–3 as well as the anticancer drug irinotecan measured by the bioluminescence technique. the mM concentration range. Although our current results suggest promising biocompatibility of these PBAs with U87Luc cells at very low concentrations (