Iron oxide nanoparticles (IONs) were synthesized by heating a mixture of FeO(OH), oleic acid and eicosane to 360 °C for 2 hours () as described by Burdinski and coworkers 
. IONs were subsequently purified by precipitation and re-dispersion of the nanoparticles, using acetone and hexane respectively. shows a TEM image of the obtained IONs. The nanoparticles were relatively monodisperse, with an average size of 24.9 ± 1.9 nm (calculated from 400 nanoparticles; ). Occasionally, a minor fraction of much smaller IONs was observed as well (arrows; inset ). The diameter of the synthesized IONs was approximately a factor 5 larger compared with Endorem, Resovist and Sinerem (), which were used as benchmark iron oxide formulations in this study. displays a selected area electron diffraction (SAED) pattern acquired from the synthesized IONs. The measured lattice spacings based on the rings in the diffraction patterns match well with the known lattice spacings of magnetite ().
Schematic overview of the (A–B) iron oxide nanoparticles and (C–D) ION-Micelle synthesis.
Transmission electron microscopy analysis of the IONs.
Physical characteristics and relaxivities. Relaxometric measurements were performed at 1.41 T and 37 °C.
Calculated atomic lattice spacing d (Å) corresponding to diffraction pattern in compared to standard atomic spacing for bulk magnetite (Fe3O4) along with their respective hkl indices from the PDF database.
Water-soluble ION-Micelles were formed by infusing a hexane/chloroform mixture containing IONs and 1,2-Distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethyleneglycol)-2000] (PEG2000–DSPE) lipids into ultrapure water at 80 °C (). ION-Micelles were separated from micelles not containing an iron oxide core using ultracentrifugation and subsequently the ION-Micelles were redispersed in HEPES-buffered saline (HBS) at pH 7.4. The dispersion state of the ION-Micelles in HBS was studied using cryogenic-TEM (cryo-TEM) and dynamic light scattering (DLS). shows typical high-resolution cryo-TEM images of ION-Micelles. The ION-Micelles were mostly dispersed in HBS as single particles or as small aggregates of nanoparticles. Occasionally, also larger, worm-like aggregates were observed (). Other lipidic structures, such as liposomes and (empty) micelles, were not observed. shows the ION-Micelle hydrodynamic size-distribution obtained from an intensity-weighted analysis of the time correlation function measured with DLS. One dominant peak was observed at 47 nm for the ION micelles. The minor peak at larger sizes was indicative of a small fraction of aggregated nanoparticles, which matches well with the cryo-TEM findings (). This peak disappeared in the number-weighted analysis (), implying that the relative contribution of the larger sized aggregates to the overall size-distribution was negligible. The hydrodynamic diameter of the ION-Micelles obtained using number-weighted analysis was 38 nm, which corresponds well with the nanoparticles core size-measurements (25 nm, ). The apparent increase of the hydrodynamic size as determined by DLS compared to the nanoparticles core size measured by TEM is caused by the (hydrated) PEG2000-DSPE coating of the particles and the fact that a fraction of the particles contains multiple iron oxide cores. The hydrodynamic diameter of the ION-Micelles was similar to Resovist, approximately a factor two larger than Sinerem and a factor two smaller than Endorem (). The magnetic properties of the ION-Micelles were analyzed using a vibrating sample magnetometer (VSM). The ION-Micelles behaved superparamagnetically at room-temperature (RT), as the magnetization curve had no hysteresis (). The saturation magnetization of the ION-Micelles was found to be 82 Am2/kg Fe3O4, which is relatively close to the saturation magnetization of bulk magnetite (~90 Am2/kg Fe3O4 at RT).
Cryo-TEM and DLS analysis of the ION-Micelles in HEPES buffered saline (HBS).
Analysis of ION-Micelles magnetic properties.
It has been recognized that the usual assumption that uniform size and shape guarantee well-defined magnetic properties is often in stark contrast with reality. For instance, Luigjes and coworkers showed that two formulations of iron oxide nanoparticles, both with a 20 nm size on TEM, displayed significantly different effective magnetic core sizes (16 and 6 nm), which were calculated from the magnetization curves 
. As the magnetic properties of the nanoparticles are likely the sole critical factor for MPI-purposes, we calculated the effective magnetic core size distribution of our ION-Micelles as described by Luigjes et al. 
. shows the obtained number-weighted magnetic core size distribution, which displayed a maximum at 16 nm. The fact that the effective magnetic core size of the ION-Micelles is already larger than the average size of an entire iron oxide core in Resovist (4–6 nm), suggests that the ION-Micelles will allow for significantly more sensitive MPI.
Magnetic particle spectrometry (MPS) measurements were performed to further evaluate the potential of the ION-Micelle nanoplatform for MPI. MPS is essentially zero-dimensional MPI and therefore allows to probe the suitability of iron oxide nanoparticles for MPI purposes 
. In addition to the ION-Micelles, also three commercially available iron oxide formulations (Endorem, Resovist and Sinerem) were measured as a benchmark. MPS measurements were performed using a dedicated magnetic particle spectrometer and each sample was acquired over thirty seconds upon application of an oscillating magnetic field with an amplitude of 10 mT at 25 kHz and RT. ION-Micelles generated increased MPS signals in comparison to the commercially available iron oxide nanoparticles over the entire frequency range (). For the lower frequencies (<0.5 MHz), the ION-Micelles generated 4–6 times more signal per gram iron than Resovist, which was the best performing benchmark formulation with respect to inducing an MPS signal. At higher frequencies (>1 MHz), the improvement in ION-Micelle MPS signal compared to that of the other preparations was even more profound: the ION-Micelles generated an MPS signal per gram iron that was more than a factor 200 higher than that of the commercially available iron oxides. Similar results were obtained for samples that were diluted in whole blood (Figure S2
). These findings support the hypothesis that the ION-Micelle nanoplatform will allow significantly more sensitive MPI than Resovist, Endorem and Sinerem.
In addition to MPI, iron oxides are frequently employed for CA-MRI purposes. To evaluate the potential of the ION-Micelle nanoplatform for CA-MRI, proton relaxometry measurements were performed. The ION-Micelles displayed a longitudinal relaxivity r1 of 6.7 mM−1 s−1 and a transversal relaxivity r2 of 253 mM−1 s−1 (). Both the transversal relaxivity and the ratio of transversal/longitudinal relaxivity of the ION-Micelles were much higher than the corresponding values found for all tested commercially available iron oxide compounds. Thus, the ION-Micelles allow for more sensitive MR imaging than the three benchmark formulations.
A proof of concept, in vitro
blood clot-targeting experiment was performed using fibrin-targeted ION-Micelles to evaluate the suitability of the ION-Micelle nanoplatform for molecular MPI and MRI. Fibrin is a major component of blood clots and plays an important role in thrombi-related pathologies such as deep venous thrombosis, pulmonary embolism and atherosclerosis. Because of the high potency of the ION-Micelle nanoplatform to generate contrast in MPI and MRI, a fibrin-targeted ION-Micelle nanoplatform was envisioned to be suitable for non-invasive detection of thrombi using MPI and/or MRI. To this aim, ION-Micelles were modified with fibrin-binding peptides (FibPeps), which were directly linked to the lipid coating of the nanoparticles. FibPep is a peptide constituted of a fibrin-binding motif (RWQPCPAESWT-Cha-CWDP) 
, which is coupled to an n-succinimidyl-s-acetylthioacetate (SATA) group via a glycine linker (Figure S1A
-B). To facilitate conjugation of FibPep to the nanoplatform, the coating of the ION-Micelles was adapted by exchanging 10% of the PEG2000-DSPE lipids with maleimide-functionalized PEG2000-DSPE lipids during the phase-transfer process. Prior to conjugation of the fibrin-binding peptides to the ION-Micelles, the SATA group of the peptides was deacetylated to provide a functional thiol group (Figure S1C
). Subsequently, the deacetylated fibrin-binding peptides were conjugated to the ION-Micelles using standard maleimide-thiol chemistry, forming a covalent thioether linkage between the lipidic coating of the nanoparticles and the fibrin-binding peptides (FibPep-ION-Micelles, ). As a negative control, a scrambled peptide with C-A substitutions (NCFibPep, Figure S1D
–F) was synthesized and coupled to the thiol-modified ION-Micelles, to obtain non-targeting NCFibPep-ION-Micelles.
In vitro blood clot binding test using FibPep-ION-Micelles and negative control NCFibPep-ION-Micelles.
Blood clots were prepared by incubating a mixture of human tissue factor, calcium chloride and citrated human blood plasma for 30 minutes at 37 °C. Next, blood clots were incubated with HBS containing either FibPep-ION-Micelles or NCFibPep-ION-Micelles for 1 hour (n
4 per preparation). Subsequently, the solution (containing unbound ION-Micelles) was carefully removed and the clots were washed three times with HBS. Thereafter, clots were subjected to MRI and MPS measurements. A photograph of the clots after the incubation and washing procedure is shown in . The clots incubated with FibPep-ION-Micelles had obtained a brownish color, whereas the clots incubated with the NCFibPep-ION-Micelles had remained white. This is a strong indication that the (brownish-colored) FibPep-ION-Micelles bound specifically to the clots. MRI measurements showed clear signal voids for these clots that were incubated with FibPep-ION-Micelles, whereas those clots that were incubated with NCFibPep-ION-Micelles did not ( and Figure S3A
). The signal to noise ratio (SNR) of the clots incubated with FibPep-ION-Micelles was significantly lower than the NCFibPep-ION-Micelles incubated clots SNR (2.75 ± 0.75 and 7.43 ± 0.40, respectively; p<0.001; ). MPS measurements displayed increased signal amplitudes for the FibPep-ION-Micelles incubated clots in comparison to the NCFibPep-ION-Micelles incubated clots throughout the whole frequency spectrum up to ~750 kHz (Figure S3B
). Above ~750 kHz, the signal dropped into noise level for both the FibPep-ION-Micelles as well as the NCFibPep-ION-Micelles incubated clots. Specific analysis of the third harmonic (72.6 kHz), which is the harmonic with the highest signal amplitude, showed a seven-fold increase in signal for the FibPep-ION-Micelles incubated clots in comparison to the NCFibPep-ION-Micelles incubated clots (10.1 ± 1.8 and 1.5 ± 1.1 nAm2
at 76 kHz, respectively; p<0.001; ). The estimated iron content was calculated to be 1.48 ± 0.27 and 0.21 ± 0.17 µg Fe for FibPep-ION-Micelles and NCFibPep-ION-Micelles incubated clots, respectively, using the third harmonic MPS signal and a previously determined conversion factor of 6.87 mAm2
/g Fe for this particular batch of IONs. Thus, the targeting of the FibPep-ION-Micelles to the clots could be detected using MRI and MPS. To validate whether the FibPep-ION-Micelles indeed bound significantly more to the clots than the NCFibPep-ION-Micelles, the clots were analyzed for iron content using inductively coupled plasma atomic emission spectrometry (ICP-AES). ICP-AES measurements showed that the FibPep-ION-Micelles incubated clots contained significantly more iron than the NCFibPep-ION-Micelles (1.81 ± 0.59 and 0.12 ± 0.03 µg Fe, respectively; p<0.01; ). The ICP-AES results match well with the estimated iron content that was calculated from the MPS signal. Hence, these results show that the FibPep-ION-Micelles can selectively bind to blood clots and that this targeting can be visualized using MRI and quantified using MPS. These findings therefore underline the potential of the ION-Micelle nanoplatform for molecular MPI and MRI and encourage future assessment of the FibPep-ION-Micelle nanoplatform for non-invasive detection of thrombi in vivo
using MRI and MPI.