Murine mammary tumor virus-polyoma middle T (MMTV-PyMT) mice for the imaging studies were bred and housed at the Department of Medical Biochemistry and Biophysics, Division of Vascular Biology, Karolinska Institutet, Stockholm, Sweden. For pharmacokinetic analysis, Naval Medical Research Institute (NMRI) mice and Sprague Dawley (SD) rats (Charles River Laboratories, Germany) were housed at Adlego Biomedical AB (Solna, Sweden).

All animal experiments were carried out in accordance with the recommendations of the Swedish Animal Welfare Act. The protocol was approved by the Regional Ethical Committees for Animal Research at Stockholm and Lund, Sweden (permit ID; N96/11, N200/12 and N258/11). MRI analyses were performed under anesthesia, and efforts were made to minimize animal discomfort. Following MRI and pharmacokinetic analysis, animals were euthanized by cervical dislocation. All animals were housed under standard animal facility conditions with free access to standard rodent food and water.

Synthesis of the Spago Pix nanomaterial [12] was initiated by dissolving 12.5 mmol 1,1-bis(triethoxysilylpropyl)-1,1- bis(dimethylphosphonato)methane in 250 ml aqueous 80% ethylene glycol. The mixture was heated for 26 h at 114°C (inner temperature) whereupon mPEG 6–9 propyltriethoxy silane was added (5 mmol) and heating at 114°C was continued for 5 h. The solution of the PEGylated nanostructure was adjusted to pH 4.5 with 1 M Trizma base and 1.1 g manganese (II) chloride tetrahydrate was added. Thereafter, the solution was heated at 100°C for 112 h. The resulting solution was subjected to ultrafiltration using Millipore Centriprep centrifugal filter units. First, the solution was passed through a 100 kDa NMWC filter to remove high molecular weight nanomaterial. Secondly, the nanomaterial was filtered on a 10 kDa NMWC filter followed by repeated washing with MilliQ water as to remove unpolymerized monomer and unbound manganese. The diafiltered material was then concentrated on a 10 kDa NMWC filter and analyzed for elemental composition using ICP-AES. Concentrated nanomaterial was prepared for injection by electrolytical balancing, adjustment of pH to 7.4, osmotic equilibration using mannitol and dilution to a manganese concentration of 2 mM. The nanomaterial was further characterized by gel permeation chromatography (GPC), dynamic light scattering (DLS), zeta potential and relaxivity measurements. The size distribution and average hydrodynamic diameter were measured by GPC by comparing with protein standards of known size (bovine serum albumin and myoglobin). Hydrodynamic diameter and zeta-potential were measured by DLS on a Malvern Zetasizer Nano ZS equipped with a back-scattering detector (173°) (Malvern Instruments Ltd, Malvern, England). For zeta potential measurements the parameters were 25°C and 150 V. Longitudinal relaxivity (r₁) and transverse relaxivity (r₂) were measured using a Minispec mq60 NMR analyzer (60 MHz, Bruker Corporation, Billerica, MA) at 37°C. Samples were diluted in either MilliQ water or human blood plasma (Sigma Aldrich, St. Louis, MO).

After imaging, pieces of tumor, usually from beneath the fore limbs, and skeletal muscle from the hind limbs were excised and frozen. Tumor and skeletal muscle from one non-treated animal was included as a control. Since Mn is the paramagnetic component providing the MR contrast and Si is part of the nanomaterial matrix, the frozen tissues were digested and analyzed for content of these two elements. Elemental analysis was performed on an Optima 8300 instrument (PerkinElmer, Waltham, MA). Plasma samples were diluted in 0.1% HNO₃ prior to injection.

The in vivo part of the pharmacokinetics studies was performed by Adlego Biomedical AB. Male SD rats were administered a dose of Spago Pix corresponding to 20 µmol Mn/kg by injection in the tail vein. Blood samples were serially collected into heparin tubes at 5, 20, 60 and 120 minutes, centrifuged, and the resulting plasma was frozen at −20°C until analysis. Mn and Si concentrations as biomarkers for the nanomaterial were analyzed by ICP-AES. A smaller study using NMRI mice with blood sampling at 5 and 15 minutes was performed to verify that there were no major species differences in circulation time. For analysis of the biodistribution, MMTV-PyMT mice were euthanized after MR imaging and tumors and femoral skeletal muscles were excised and weighed. One half of muscle and tumor tissues were fixed in 4% paraformaldehyde for immunohistochemistry and the other half was frozen at −20°C for digestion and elemental analysis. For digestion, tumor- and muscle tissues were placed in Teflon containers. Five ml of concentrated HNO₃ and 5 ml H₂O were added to each container after which they were heat-treated in a microwave oven at 185°C for 30 minutes. The samples were adjusted to 25 ml with MilliQH₂O and analyzed for Mn and Si content by ICP-AES.

Two separate MR imaging studies were done with a clinical 3 T MRI system (Siemens Magnetom Trio, Siemens Healthcare, Erlangen, Germany) using a clinical wrist coil. In the first study, a fat-saturated turbo spin-echo sequence was used with an echo train length of 3, a repetition time of 350 ms, an echo time of 9.4 ms, and a scan time of 1 min 31 sec. Two-mm slices were imaged with 2 averages and 2.2 mm spacing in a coronal orientation. The receiver bandwidth was 300 Hz/pixel. The acquisition matrix was 256×204 (read × phase) with a field of view of 80×63 mm². In the second study, the spatial resolution and signal-to-noise were increased (acquisition matrix 256×256, field of view of 65×65 mm², 11 averages), and the repetition time was slightly reduced to 320 ms. The scan time was 5 min 30 sec. Anesthesia was given prior to imaging as an intraperitoneal injection of 10 ml/kg of 12.5 mg/ml Ketalar (Pfizer, New York, NY) and 4 mg/ml Rompun (Bayer Animal Health, Leverkusen, Germany). Anesthetized animals were immobilized in a modified 50 ml syringe on top of a 37°C heating pad and placed supine in the wrist coil. After a pre-injection scan, the animals were administered a single tail vein intravenous injection of 10 or 20 µmol Mn/kg of a formulation of 2 mM Mn Spago Pix. Coronal T₁-weighted MR images, covering the whole animal, were acquired before contrast injection and at 10–270 minutes. In cases where the animals were to be re-scanned more than 30 minutes post-dose, they were kept in cages with free access to water and food between imaging sessions.

MR images were exported as DICOM files and analyzed with the OsiriX software (32-bit Open Source, http://www.osirix-viewer.com). For quantitative analysis of the coronal imaging planes, one region of interest (ROI) was placed in air in the upper left quadrant outside the animal and defined as background. Depending on the availability of well-defined muscle tissue, one or two ROIs were placed separately on either fore- or hind limb muscle tissue. One ROI was placed on each tumor present in the determined image plane. The standard deviation, σ, of the background signal and the mean signal intensity of the muscle and tumor ROIs were determined by the software. The signal-to-noise ratio (SNR) was calculated by dividing the mean signal intensity by σ, and the tumor versus muscle contrast-to-noise ratio (CNR) was calculated by subtracting SNR_muscle from SNR_tumor. Images for visual presentation were treated with global automatic levels settings with GIMP (GNU Image Manipulation Program, version 2.8.2).

Paraformaldehyde fixed muscle and tumor tissues collected before injection and at 30 minutes and 2–4 h post-dose were paraffin embedded and thereafter cut (4 µm) and positioned on glass tissue slides. Prior to staining, the slides were placed in an antigen-retrieval solution using an automated pre-treatment module (PT-Link; Dako, Glostrup, Denmark). Slides were stained for PEG in an automated immunohistochemistry robot (Autostainer; Dako) using a primary rabbit anti-PEG-B-47 antibody (Abcam, Cambride, MA) at a concentration of 0.6 µg/ml and a secondary goat-anti rabbit antibody (Dako) at 1∶2000 dilution. Slides were counterstained with hematoxylin, dehydrated and mounted. The stained slides were digitized and total PEG-staining was quantified with the Visomorph software (Visiopharm, Hoersholm, Denmark). Findings in staining patterns were confirmed by a trained pathologist.

Differences between indicated groups were analyzed with unpaired t-tests with the statistical package R (version 3.0.2). For MR CNR quantification, all tumors were treated separately and pooled data from the two studies was analyzed. A p-value of <0.05 was considered statistically significant.

The Spago Pix nanomaterial consists of an organophosphosilane hydrogel with strongly chelated manganese (II) ions and a covalently attached PEG surface layer. It has a globular shape with an average hydrodynamic diameter of 5 nm, as determined both by GPC and DLS measurements. The zeta potential is essentially neutral. The longitudinal relaxivity (r₁) is 30 mM Mn ⁻¹ s⁻¹ (60 MHz) in both water and human blood plasma. The transverse relaxivity (r₂) in human blood plasma is 55 mM Mn ⁻¹ s⁻¹ (60 MHz). The solutions of Spago Pix used in the in vivo experiments had a manganese concentration of 2 mM and a silicon concentration of 62 mM.

Over the first 60 minutes after injection of Spago Pix, the plasma concentration of both Mn and Si declined by about 90%, with Mn dropping slightly faster than Si. From 60 to 120 minutes both Mn and Si concentrations continued to fall but at a lower rate (Figure 1). Results from a smaller study in mice showed a similar concentration profile (data not shown). Thus, Spago Pix nanoparticles are present in the circulation for more than one hour, defining the time frame in which extravasation to the tumor can occur.

Between one and nine mammary tissue tumors were detected in all animals included in the two studies. In PyMT animals of the age used in the study (10–12 weeks) tumors occur mostly in the second and third mammary glands (Figure 2). The time frame for when the tumor contrast develops in animals injected with Spago Pix is shown in Figure 3. The two MR imaging studies were performed with slightly different imaging parameters, but the mean CNR of tumors prior to or 30 min and 4.5 h after injection of Spago Pix did not differ significantly between the studies (data not shown). Therefore, CNR data from both studies could be pooled for further analysis. The mean CNR of the tumors prior to administration of Spago Pix was 13.4 (±4.5, n = 24 tumors in 7 animals). Tumor contrast increased within 10 minutes after administration of Spago Pix and the mean CNR 30 min post administration of Spago Pix was 24.4 (±7.9, n = 29 tumors in 7 animals). This corresponds to an 82% increase in CNR. After 2–4.5 h, the mean CNR had further increased to 31.4 (±11.2, n = 31 tumors in 6 animals) corresponding to a 134% CNR increase relative to the pre-dose imaging. A boxplot of CNR values is shown in Figure 4. In addition to an increased tumor CNR, the liver and gall bladder also exhibited increased MR signal intensities.

Tumors from MMTV-PyMT mice contained approximately 30 nmol Mn and 800 nmol Si per g tissue, which was 6- and 2-fold higher than in muscle tissue, respectively (Figure 5). Muscle tissues had approximately the same Mn and Si levels as tissues from non-injected animals.

The area of PEG (a coating constituent of Spago Pix) staining was large in tumoral areas, especially in stromal tissues between tumor foci. Specifically in these areas, vessels and fibroblasts where strongly stained. Several tumor foci also showed PEG-positive signals, and the area of those foci was larger at the later time points. In contrast, staining in muscle tissue was limited to the area inside vessels between muscle structures (Figure 6). Tumor vessels frequently showed signs of extravasation of Spago Pix into surrounding tumor tissue (Figure 7) in contrast to vessels in muscle tissue in which Spago Pix was contained within the vessels (Figure 6, panels A4 and B4). The PEG staining area was larger at the later time points (2–4 h) compared to 30 minutes after injection as measured by image analysis (Figure 8). In contrast, muscles showed only a weak increase in PEG staining. The appearance of the staining in tumors was more focal at the early time point and more widespread throughout the tumor at the later time points.