Hypoxia-Responsive Cobalt Complexes in Tumor Spheroids: Laser

Aug 2, 2017 - For maximizing sensitivity and ensuring low oxide formation (ThO/Th < 0.3%), a NIST 612 Trace Element in Glass CRM was ablated. High-pur...
0 downloads 7 Views 2MB Size
Download PDF
Article pubs.acs.org/IC

Hypoxia-Responsive Cobalt Complexes in Tumor Spheroids: Laser Ablation Inductively Coupled Plasma Mass Spectrometry and Magnetic Resonance Imaging Studies Edward S. O’Neill,† Amandeep Kaur,† David P. Bishop,‡ Dmitry Shishmarev,§ Philip W. Kuchel,§ Stuart M. Grieve,∥,⊥,# Gemma A. Figtree,∇,○ Anna K. Renfrew,† Paul D. Bonnitcha,#,◆ and Elizabeth J. New†,* †

School of Chemistry, University of Sydney, Sydney, New South Wales 2006, Australia Elemental Bio-imaging Facility, University of Technology Sydney, Thomas Street, Broadway, New South Wales 2007, Australia § School of Life and Environmental Sciences, University of Sydney, Sydney, New South Wales 2006, Australia ∥ Sydney Translational Imaging Laboratory, Heart Research Institute, Charles Perkins Centre, University of Sydney, Camperdown, New South Wales 2006, Australia ⊥ Department of Radiology, Royal Prince Alfred Hospital, Camperdown, New South Wales 2050, Australia # Sydney Medical School, University of Sydney, Camperdown, New South Wales 2006, Australia ∇ Kolling Institute of Medical Research, University of Sydney, St Leonards, New South Wales 2065, Australia ○ Cardiology Department, Royal North Shore Hospital, St Leonards, New South Wales 2065, Australia ◆ Chemical Pathology Department, Royal Prince Alfred Hospital, Campderdown, Sydney, New South Wales 2050, Australia ‡

S Supporting Information *

ABSTRACT: Dense tumors are resistant to conventional chemotherapies due to the unique tumor microenvironment characterized by hypoxic regions that promote cellular dormancy. Bioreductive drugs that are activated in response to this hypoxic environment are an attractive strategy for therapy with anticipated lower harmful side effects in normoxic healthy tissue. Cobalt bioreductive pro-drugs that selectively release toxic payloads upon reduction in hypoxic cells have shown great promise as anticancer agents. However, the bioreductive response in the tumor microenvironment must be better understood, as current techniques for monitoring bioreduction to Co(II) such as X-ray absorption near-edge structure and extended X-ray absorption fine structure provide limited information on speciation and require synchrotron radiation sources. Here, we present magnetic resonance imaging (MRI) as an accessible and powerful technique to monitor bioreduction by treating the cobalt complex as an MRI contrast agent and monitoring the change in water signal induced by reduction from diamagnetic Co(III) to paramagnetic Co(II). Cobalt pro-drugs built upon the tris(2pyridylmethyl)amine ligand scaffold with varying charge were investigated for distribution and activity in a 3D tumor spheroid model by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) and MRI. In addition, paramagnetic 1H NMR spectroscopy of spheroids enabled determination of the speciation of activated Co(II)TPAx complexes. This study demonstrates the utility of MRI and associated spectroscopy techniques for understanding bioreductive cobalt pro-drugs in the tumor microenvironment and has broader implications for monitoring paramagnetic metal-based therapies.

■

INTRODUCTION

hypoxic tissues for selective targeting to tumors while minimizing the effect on normoxic healthy tissues.3 Many metal-based pro-drugs are complexes of ruthenium, iron, and cobalt.4 Transition metals incorporated into a prodrug enable deactivation of the cytotoxin in coordinated form as the inert oxidized metal. Upon reduction to a more labile reduced state, release of the cytotoxic payload occurs within reducing chemical environments such as hypoxic tissues.5 The

There is considerable therapeutic potential in anticancer prodrugs that can selectively deliver cytotoxic payloads to dense neoplastic tumors with minimal toxic side effects. The unique chemical characteristics of solid tumors typically result in resistance to conventional anticancer treatments.1 The hypoxic and acidic microenvironment of a tumor promotes chemoresistance in which dormant cancer cells in the hypoxic region are less responsive to conventional antiproliferative drugs.2 Metal-based pro-drugs exploit characteristics of abnormal © 2017 American Chemical Society

Received: May 29, 2017 Published: August 2, 2017 9860

DOI: 10.1021/acs.inorgchem.7b01368 Inorg. Chem. 2017, 56, 9860−9868

Article

Inorganic Chemistry

tagged systems.6 In the present work, we aimed to use a nontoxic bidentate ligand, acetylacetone, to explore the use of MRI and other analytical methods, including inductively coupled plasma mass spectrometry (ICP-MS) and laser ablation (LA)-ICP-MS, to gain a comprehensive picture of the distribution, oxidation state, and speciation of cobalt within a tumor spheroid.

inherent charge on transition metal ions can be exploited to bias accumulation using the acidic environments that are characteristic of hypoxic tumors.6 One cobalt-based bioreductive pro-drug with a bridged cyclam cobalt(III) scaffold showed great potential with over 200-fold greater potency under hypoxic conditions than normoxia in vitro but failed to elicit significant hypoxic cell death in an in vivo tumor xenograft model.7 Another in vivo study demonstrated enhanced anticancer activity with reduced side-effects when a clinically approved anticancer drug was appended to Co(III) in a bidentate fashion.8 This study also revealed that larger tumor volumes were required for anticancer activity, enforcing the need for greater understanding of the activity of these drugs in the tumor microenvironment. For understanding and demonstrating the efficacy of this bioreductive pro-drug strategy in the tumor microenvironment, tumor spheroids are a convenient in vitro system.5 These 3D scaffolds develop the hypoxic characteristics of a solid tumor with high cellular density and distance from the culture media limiting nutrient and drug diffusion throughout the entire spheroid. These characteristics are ideal for investigating uptake, excretion, and activity of anticancer agents across whole tumors, including normally drug-resistant cells in the hypoxic regions. For understanding the activity of bioreductive cobalt prodrugs in these hypoxic regions, previous work has included replacing the cytotoxic agent with a fluorescent payload for confocal imaging.6 X-ray absorption near-edge spectroscopy (XANES) in cellulo has also been used to show evidence of cargo release by reduction to Co(II) for some cobalt complexes.9−11 Although synchrotron techniques such as XANES reveal metal-ion oxidation state and provide some indications of ligand speciation, limited access to such equipment requires the use of other analytical techniques for detecting activatable complexes in situ in cells. In the present work, we used magnetic resonance imaging (MRI) to investigate and monitor the redox state and spatial distribution of selected metal complexes. Cobalt is a favorable candidate for such studies with the inert diamagnetic Co(III) oxidation state being activated to the paramagnetic Co(II) complex, coincident with payload release.12 Paramagnetic Co(II) is detected in MRI via its effect on the relaxation rate of the water proton signal, and it can also induce upfield and/or downfield shifts in ligand protons.13 These attributes obviate the need for a fluorescent ligand payload as the Co(II) complex can be monitored directly, allowing for wider screening of nonfluorophore anticancer drug payloads. MRI of tumor spheroids is also an approach that can be readily scaled up in the drug development pipeline with abundant preclinical MRI scanners available for in vivo studies and clinical MRI scanners ultimately used for final patient trials of anticancer drugs. We have already reported the efficacy of cobalt(III)tris((2pyridylmethyl)amine)acetylacetone (Co(III)TPAacac) as a hypoxia-responsive MRI contrast agent. Co(III)TPAacac is MRI-silent, whereas the Co(II)TPA that is generated upon reduction is MRI-active.14 Such reversible off-to-on response means that any change in the water proton MRI signal can be used to characterize the nature of the reducing environment in tissues. Previous studies of the cobalt TPA system, as a potential pro-drug system, explored optimization of the complexes by varying the number of pyridyl carboxylate groups, and hence the overall charge, and investigated the effects on fluorescence emission and lifetime of fluorophore-

■

RESULTS AND DISCUSSION Tris(2-pyridylmethyl)amine ligands with additional carboxylate groups were prepared by modification of previously reported procedures (Scheme S1).15 The TPA derivatives were complexed using CoCl2·6H2O under anoxic conditions with subsequent addition of acetylacetone and triethylamine and then oxidized in air to yield the corresponding Co(III) complex, which was purified by CM Sephadex C-25 weak cation-exchange chromatography (Figure 1).

Figure 1. Co(III) complexes prepared in this study: TPA0 = [Co(III)TPAacac], TPA1 = [Co(III)TPA(COO−)acac], TPA2 = [Co(III)TPA(COO − ) 2 acac], and TPA3 = [Co(III)TPA(COO−)3acac].

The Co(III)TPAx complexes were characterized by 1H NMR spectroscopy, which revealed the existence of two isomers of TPA1 and TPA2 with the carboxylate pyridyl ring being either in the axial or equatorial position trans to the acetylacetone ligand (Scheme S1). There was no evidence of exchange between isomers, and the octahedral Co(III) structure appeared to be very rigid as evidenced by the sharp geminal 16 Hz scalar coupling (J) between the CH2 protons on the axial methyl arms seen most clearly in the TPA0 1H NMR spectrum (Figure S1). Resonances in 59Co NMR spectra were sharp for cobalt(III) complexes with line width at half-height of ∼600 Hz, and they exhibited close similarity, revealing a remarkable ligand field similarity between the complexes (Figure 2). This is in contrast to previous reports of another series of TPA complexes that were modified with an increasing number of 6-

Figure 2. 59Co NMR spectra (95.88 MHz) of TPAx complexes in D2O (6000 transients in 9 min). 9861

DOI: 10.1021/acs.inorgchem.7b01368 Inorg. Chem. 2017, 56, 9860−9868

Article

Inorganic Chemistry Table 2. Relaxivity of [Co(II)TPAx] Complexesa

methylpyridyl rings for which there were large shifts (up to 2200 ppm) across the complexes in the 59Co NMR.16 Because of the high receptivity of 59Co NMR and the very narrow resonances for these complexes (Co(III) complexes typically have greater than 14 kHz line widths17), it may be possible in the future to tune a 13C MRI coil to conduct 59Co NMR imaging of these complexes prior to their reduction. Cyclic Voltammetry. In previous work with cyclic voltammetry, TPA0 displayed a quasi-reversible redox reaction involving acetylacetone.14 However, in the sustained reducing environment that was similar to that in hypoxic tissue, acetylacetone was irreversibly lost from the complex, but the reduced complex could be reoxidized to the Co(III) form. Co(III)TPAxacac analogues, TPA1, TPA2, and TPA3 exhibited irreversible reduction during cyclic voltammetry, and one electron reduction was seen to be diffusion-controlled as with TPA0 (Figure S2). Overall, increasing numbers of carboxylate groups led to an increase in the stabilization of the Co(III) state that was manifest in more negative reduction potentials (Table 1), which was the opposite trend observed in

Epa (mV)

CoTPA0acac CoTPA1acac CoTPA2acac CoTPA3acac

−491b −560 −543 −635

r1 (mM−1 s−1)

CoTPA0acac CoTPA1acac CoTPA2acac CoTPA3acac

± ± ± ±

0.0607 0.0362 0.0544 0.0474

0.0009 0.0004 0.0006 0.0007

b

r2 (mM−1 s−1)

r2/r1

± ± ± ±

4.0 6.7 7.0 8.7

0.24 0.24 0.38 0.41

0.03b 0.007 0.010 0.007

a

Relaxivity values were calculated by linear regression of 1/T1,2 experimental values from Figures S2 and S3. Measured at 310 K at 9.4 T in deoxygenated 10% D2O in H2O, 0.15 M NaCl, 0.1 M tris buffer pH 7.4 (± standard error). bData previously reported.14

relaxation pathway,22 thus implying that changes in water residence time, and exchange kinetics, have minimal effect across the series of cobalt complexes. These cobalt complexes exhibited relatively small relaxivity values compared with gadolinium-based complexes, but because relaxivity is a concentration-dependent parameter, at larger millimolar concentrations an appreciable relaxation effect could still be expected for Co(II) complexes. Despite the small relaxivity values, there was a definite trend of increasing r2/r1 values with additional carboxylate groups. Cytotoxicity Screening. Before performing cellular studies, we assessed the stability of the series of complexes in cell culture media and found that, after 8 d at 37 °C, both the TPA and acac ligands remained coordinated to all complexes (Figure S5). The in vitro cytotoxicity values of the TPAx series was evaluated by the alamarBlue assay with DLD-1 colorectal cancer cells. A maximal complex concentration of 4 mM was tested; this was done to coincide with the concentrations required in MRI screening. Even at the highest concentrations tested, greater than 60% cell viability after 24 h was observed (Figure 3), so the complexes can be considered nontoxic. A 2

Table 1. Estimated Anodic Peak Potentials (Epa) for the Potential of a Half-Wave of CoTPAxacac Complexesa complex

complex

a

In PBS buffer (137 mM NaCl, 2.7 mM KCl, and 10 mM phosphate buffer solution, pH 7.4) as recorded with cyclic voltammetry (5 mM, scan rate 100 mVs−1) normalized to NHE using Ag/AgCl (saturated) reference electrode potential of 198 mV. bData previously reported.14

a previous study,6 where a bulkier ligand than acac resulted in steric-directed promotion of Co(II) reduction with increasing numbers of carboxylate groups. It is possible that because acac is a small ligand, the electronic effects of carboxylate groups direct the reduction potentials. The COO− group induced an increase in electron density on the pyridyl nitrogen in the meta position, resulting in stronger σ-bonding to the cobalt.18 The pseudo-octahedral geometry of the complex increases Co−N bond strength with favorable σ-orbital overlap with the occupied low-spin t2g Co(III) d-orbitals. This increased σorbital overlap will induce a small trans destabilization of the acac Co−O bond when the carboxylate pyridyl group is trans to the acac group.19 This accounts for the slightly less negative reduction potential of TPA2 compared to that of TPA1, as it is more likely to have a carboxylate pyridyl group in the trans position. However, the effect is still less than the overall σbonding stabilization of the complex because TPA3 has the lowest reduction potential. Despite the variations in reduction potential, all values lie within the range of reduction potentials observed for hypoxic tissue and the hypoxia responsiveness of metal-based complexes.20 Relaxivity. The longitudinal (r1) (Figure S3) and transverse (r2) (Figure S4) relaxivities revealed an off-to-on response upon reduction from Co(III) to Co(II) using the chemical reductant sodium dithionite (Table 2). There was no relaxivity trend seen with the Co(II) paramagnetic complexes as might have been expected according to the Solomon−Bloembergen−Morgan (SBM) theory.21 This was surmised to be due to the fact that very short electronic relaxation time ∼10−12 s is the dominant

Figure 3. Cytotoxicity of TPAx complexes measured by alamarBlue assay (DLD-1 cells, 24 h incubation with TPAx, followed by 4 h incubation with alamarBlue). Cell viability normalized to controls incubated with PBS vehicle alone.

mM concentration of TPAx was chosen as a conservative concentration for further in vitro studies of toxicity with incubation times of no longer than 24 h for all subsequent experiments. Cobalt pro-drug complexes typically show low toxicity in vitro and in vivo, so it would not be unreasonable to test future cobalt pro-drug complexes at these concentrations. 9862

DOI: 10.1021/acs.inorgchem.7b01368 Inorg. Chem. 2017, 56, 9860−9868

Article

Inorganic Chemistry Cobalt Uptake Profiles. We analyzed the uptake of different complexes into spheroids. DLD-1 spheroids of various cell numbers (5 × 103 to 2 × 104), and hence, different sizes were grown over 5 d and dosed with TPAx. The uptake of cobalt into these spheroids was measured by ICP-MS. This revealed similar uptake profiles for each complex with relatively less uptake into the larger spheroids (Figure 4). The decrease

Figure 5. Representative cobalt LA-ICP-MS images showing cyrosectioned slices of DLD-1 spheroids (15,000 cells) incubated with TPAx (2 mM) for 24 h; scale bar 500 μm.

region in spheroids of a similar size.31 TPA0 also showed general uptake across the whole spheroid with no discernible pattern of heterogeneity. TPA1 and TPA3 demonstrated greater uptake into the inner hypoxic layer of the spheroids with TPA3 having the most pronounced accumulation in the center. The accumulation of TPA3 was attributed to the electroneutrality achieved upon protonation of one carboxylate group in the hypoxic extracellular region, which is known to have a lower pH.32 It is possible that the accumulation of TPA2 in the center of the spheroid could point to a possible excretion pathway from cells not predominant for the other complexes. Except for that of TPA2, the uptake profiles match those previously seen with confocal microscopy for the fluorophorelabeled cobalt complexes.6 This confirmed the utility of LAICP-MS as a technique for measuring the tissue distribution of cobalt complexes. Magnetic Resonance Imaging. Tumor spheroids, grown and treated as for LA-ICP-MS testing, were embedded in Matrigel and imaged using the rapid acquisition with relaxation enhancement (RARE) pulse sequence (Figure 6). Straight line

Figure 4. Tumor spheroid uptake of CoTPAxacac complexes measured by ICP-MS relative to protein content. Spheroids (5 × 103 to 2 × 104 cells incubated for 2 days, n = 16 per condition) were dosed with TPAx (2 mM, 24 h incubation), washed with PBS, split into groups of eight spheroids and lysed (20 μL of lysis buffer, 10 min). Ten microliters of the lysate was digested with 10 M HNO3 at 40 °C overnight, and protein content was normalized by standard Bradford assay25 of 1 μL of original lysate.

was due to the decrease in surface area to volume ratio and also the different cellular architectures that resulted from a relatively larger necrotic core in which there was protein but no intact cells.23,24 A more complete image of cobalt accumulation and distribution required the use of an imaging technique, and for this, we turned to LA-ICP-MS. LA-ICP-MS has previously been used to examine metal distribution in tissue and small organisms. Specifically, it has been used to study platinum in tissue sections,26 lanthanidebased MRI agent accumulation,27 and cobalt mouse models of obstructive sleep apnea.28 LA-ICP-MS has lower resolution than synchrotron X-ray imaging techniques, yet it is more sensitive29 and more readily accessible. We therefore sought to determine if LA-ICP-MS would be a suitable technique for observing cobalt levels within tumor spheroids. Spheroids (1.5 × 104 cells) were cryosectioned into 20 μm sections and analyzed by LA-ICP-MS. All CoTPAxacac complexes, regardless of overall complex charge, entered the spheroids with considerable diffusion through the dense cellular architecture rather than exclusive uptake by the outer cells, as is the case for many drugs (Figure 5).30 TPA2 exhibited the greatest uptake across the whole spheroid despite the slightly lower overall uptake by ICP-MS with most of the uptake in the outermost cells compared to the other TPAx complexes (Figure 4). This outcome was attributed to the neutral charge expected on TPA2 in both the Co(III) TPA2 and Co(II) TPA2 forms, enhancing accumulation into both inner and outer regions of the spheroid. The region in which cobalt uptake appeared greatest matches with that previously identified as the hypoxic

Figure 6. Representative MR images of tumor spheroids previously incubated with TPAx (2 mM, n = 48 and 24 h) acquired using the RARE pulse sequence at 9.4 T (NA = 4, TE = 24.35 ms, TR = 400 ms, matrix size: 192 × 192 × 192, resolution 80 × 80 × 80 μm); scale bar 1 mm.

cross sections of the spheroids were measured using ImageJ at their maximum width for slices in the MRI that exhibited the greatest spheroid diameter (Figure 7a). The intensity values were then normalized to the surrounding matrix in each tube for each slice to minimize the effect of “hotspots” in the MRI images due to proximity birdcage coils of the probe. These normalized intensity values could then be compared to control spheroid values to reveal the effect of TPAx on the relaxation of the water signal (Figure 7b). TPA3 showed an increase in signal intensity of the inner regions but did not affect the outer 9863

DOI: 10.1021/acs.inorgchem.7b01368 Inorg. Chem. 2017, 56, 9860−9868

Article

Inorganic Chemistry

Figure 7. MRI spheroid cross-section analysis of the TPAx series. Intensities were first normalized to background matrix signal and TPAx incubated spheroids (2 mM, n = 48 and 24 h). (a) Intensities of each pixel along the cross section, and (b) mean voxel intensities of the most intense inner voxel and the outer region’s least intense voxel compared to the corresponding intensities in the control spheroids.

regions. These are encouraging results that are consistent with reduction to Co(II) TPA3 in regions previously associated with hypoxia.6 The LA-ICP-MS results, which show higher accumulation of Co in these inner hypoxia regions, further support these results (Figure 5). TPA0 and TPA1 did not induce a large difference in the intensity, which indicates a low amount of complex in the reduced Co(II) state, although ICPMS and LA-ICP-MS indicate that there should be sufficient accumulation of complexes within the spheroid for detection of the Co(II) forms. Spheroids treated with TPA2 demonstrated a decrease in signal intensity, a negative contrast compared to the other complexes. This inversion of contrast has previously been observed in endosomes and liposome constructs containing normally “positive” gadolinium-based contrast agents33 and therefore could indicate the formation of microvesicles from direct blebbing of the TPA2-embedded plasma membrane or exosomes as possible excretion pathways. The production of exosomes are prevalent in intestinal epithelial cells such as the DLD-1 cell line studied here.34,35 This is also a possible explanation for the seemingly contradictory findings of low uptake in the aggregate ICP-MS results and high values observed in the LA-ICP-MS results except in the larger spheroids (Figure 4; 20,000 cells), which are expected to have a larger necrotic center to assist in trapping these possible exosomes. Tumor hypoxia is known to actively facilitate the formation of exosomes to promote survival and invasion.34,36−38 The LA-ICP-MS results support this excretion pathway, as the hypoxic cell layer appeared to have the lowest concentration of cobalt with the highest levels being evident on the outermost and necrotic center. This was opposite to the distribution for the hypoxia-selective TPA3 complex. NMR Spectroscopy. For establishing the integrity of the TPA complexes when activated, 1H NMR spectroscopy was used to detect the resonances of complexes (Figure 8) as has been previously reported.39 If the source of the paramagnetic enhancement was simply due to the breakdown of the complex when incorporated into the cell as free Co2+, then we should not have observed the Co(II) TPAx 1H NMR resonances. Existence of the Co(II) TPAx 1H NMR resonances provide evidence of Co(II) TPAx integrity despite Co(II) typically being the most kinetically active oxidation state. This 1H NMR technique therefore provides more information than can be obtained from synchrotron XANES or extended X-ray

Figure 8. 1H (500 MHz) NMR spectra of Co(II) TPAx complexes incubated with tumor spheroids showing resonances in the “paramagnetic” region of the chemical shift domain. Tumor spheroids (n = 25) suspended in advanced DMEM media dosed with 5% D2O prior to acquisition of 1H NMR spectra using H2O presaturation pulses in a 1.7 mm capillary tube.

absorption fine structure (EXAFS) techniques, for which ligand speciation information is limited to the coordinating atoms only: for example, if the nitrogen ligand is replaced with a different coordinated nitrogen ligand, the difference cannot be detected.11

■

CONCLUSIONS The combined study of a series of cobalt TPAx complexes using LA-ICP-MS and MRI enabled the evaluation of their diagnostic utility for cobalt pro-drugs in tumor spheroids. TPA3 was selective for hypoxic regions of cells both in accumulation and activation. Future modification of the acetylacetone group on TPA3 with neutral cytotoxic agents would be a suitable pairing for hypoxia-responsive anticancer therapies. The results from TPA2 warrant further investigation for the formation of potential exosomes as a mechanism of excretion from cells and the existence of a negative MR contrast by other metal complexes and similar accumulation patterns by LA-ICP-MS could serve as a potential warning sign for hypoxic resistance mechanisms to metal-based therapies. 1H NMR spectroscopy of paramagnetically shifted ligand protons offers a valuable alternative to XANES and EXAFS synchrotron techniques for monitoring Co(II) speciation. The use of MRI for the study of cobalt pro-drugs is therefore of great utility during all stages of the drug development pipeline. Both the 9864

DOI: 10.1021/acs.inorgchem.7b01368 Inorg. Chem. 2017, 56, 9860−9868

Article

Inorganic Chemistry

laser emitting nanosecond laser pulses at a wavelength of 193 nm. Samples and standards were ablated with 25 μm spot-size scanning at 100 μm s−1. S-lenses with a Pt sampler, and skimmer cones, were used in the ICP-MS. For maximizing sensitivity and ensuring low oxide formation (ThO/Th < 0.3%), a NIST 612 Trace Element in Glass CRM was ablated. High-purity liquid Ar boil-off was used (Ace Cryogenics, Castle Hill, NSW, Australia) as the carrier gas. Ultrahigh purity H2 (99.999%) was used as the reaction gas (BOC, North Ryde, NSW, Australia). Preparation of Tissue Standards. Tissue standards were prepared as previously described:40 briefly, sheep brains were homogenized on ice using a tissue homogenizer (Omni Scientific, Kennesaw GA, USA) and fitted with disposable 7 mm × 110 mm polycarbonate hard tissue OmniTips. The tissue was then spiked with standard metal-ion solutions prepared in 1% (v/v) HNO3 from a cobalt nitrate salt with a minimum purity of 99.995% (Sigma-Aldrich, Castle Hill, NSW, Australia). The salt was distributed through the tissue by a second homogenization step. Six ∼50 mg aliquots of each standard were digested in 4 mL of 69% HNO3 and 1 mL of 30% H2O2 in a Milestone MLS 1200 microwave digester (In Vitro Scientific, Noble Park North, VIC, Australia) diluted to ∼50 g (measured using an analytical balance) and analyzed with an Agilent 7500cx ICP-MS (Agilent) to confirm the concentration and homogeneity of each element in the tissue standards. BCR185R (bovine liver reference standard) was digested and analyzed to measure the digestion recovery. Frozen sections of tissue standards were cut at 30 μm on a Leica crytome (Leica, Solms, Germany) at −16 °C and mounted on Starfrost microscope slides (Waldemar Knittel Glasbearbeitungs, Wildhagen, Germany). Sections were air-dried prior to use. Calibration and Specifications. Four 2 mm lines of each standard were ablated to construct the calibration curves (see Table S1). The data was averaged for each calibration level. Calibration curves and construction of images were performed using in-house developed imaging software with ParaView used for visualization. Calibration equations were used to convert the signal intensities of every pixel in each image to concentrations (mg kg−1). Tumor Spheroid MRI. MR images were acquired using an Avance III DRX NMR spectrometer with an Oxford Instruments vertical-bore 9.4 T magnet (Bruker, Etttlingen, Germany) and a mini 0.5 small animal imaging probe. The total scanning time was 3 h 15 min. Spheroids (n = 48 per condition) were grown as above and suspended in ECM-gel recovered from Engelbreth−Holm−Swarm (EHS) mouse sarcoma cells (Matrigel) at