Chelator free gallium-68 radiolabelling of silica coated iron oxide nanorods via surface interactions †

,


Introduction
Medical imaging can be defined as the non-invasive visualisation of the intact human body for clinical analysis. Positron Emission Tomography (PET) is a nuclear medicine technique based on the detection of positron emitting radionuclides to give a high sensitivity functional image. 1-3 However, as a molecular imaging technique, it does not provide anatomical information and is, therefore, often combined with other techniques, such as computed tomography (CT) or magnetic resonance imaging (MRI). Although the combination of PET with CT is most prevalent in the clinic, recent developments in detector design have led to the creation of combined PET/MR scanners that offer advantages over PET/CT, such as increased soft tissue contrast, lower radiation dose and the potential application of MRI contrast agents. 4,5 The introduction of clinical PET/MRI scanners increases the demand for multimodal contrast agents, achieved either by using a cocktail of known imaging agents or by developing new dual modality contrast agents in which both components are linked into one construct. [6][7][8][9] Super-paramagnetic iron oxide nanoparticles (SPIONs) are used in MRI to reduce the time taken for proton nuclei to transfer their spin to neighbouring proton nuclei, resulting in a loss of transverse magnetisation (T 2 relaxation), causing a relative darkening in regions of SPION accumulation. 10,11 Ironoxide based nanospheres have been the most commonly used T 2 contrast agents clinically, [12][13][14] largely because of their low toxicity and the potential for facile modification of the surface. [15][16][17] Iron-oxide based nanorods (NRs) are less common but may offer advantages, such as improved MRI contrast [18][19][20] and improved properties as drug delivery vehicles. 21 8 Ga (t 1/2 = 68 min) is an attractive radioisotope for PET due to high proportion of positron decay (89%) and its availability from a generator based system as a decay product of a parent isotope with a long half-life ( 68 Ge, t 1/2 = 271 days), eliminating the need for an onsite cyclotron. 22,23 In aqueous solution and at physiological pH, 68 Ga is exclusively in the oxidation state +3, thus most synthetic procedures are carried out in the presence of weakly coordinating ligands such as citrate, acetate or oxalate at low pH in order to avoid the formation of insoluble Ga(OH) 3 and soluble Ga(OH) 4 − , both of which dramatically slow the kinetics of complex formation. 24,25 Gallium(III) generally forms six coordinate complexes and is classified as a hard Lewis acid, therefore, it binds to borderline and hard Lewis base donor atoms such as nitrogen and oxygen, found in routinely used chelators for biomedical imaging. 26,27 Incorporation of 68 Ga into a targeting vector is generally achieved by the use of bifunctional chelating agents (BCAs), molecules containing a chelating moiety responsible for trapping 68 Ga in a stable in vivo form, functionalised with a reactive group allowing its conjugation to a bioactive molecule, protein or peptide. 28,29 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) is the most widely employed BCA due to the relative ease of its synthesis (due to its chemical symmetry) and the ability to complex, with adequate stability, various metal ions relevant for a range of biomedical applications. 3,22,30,31 Herein we describe the development of silica coated iron oxide NRs radiolabelled with gallium-68 for PET/MR multimodal imaging in which the inclusion of a bifunctional chelating agent (such as DOTA) is not required for stable gallium(III) complexation in vivo.

Synthesis and surface characterisation
Iron-oxide nanorods, coated with various ratios of a siloxane terminated tetraazamacrocycle (siloxane-DO3A) and a siloxane polyethylene glycol (PEG) derivative, were synthesised using a modified method, similar to that previously reported by our group. 32 Briefly, uncoated iron oxide nanorods and the selected ratio of functionalised siloxanes were stirred in a basic 60% ethanol/water solution, as shown in Scheme 1. The preparation includes a rigorous washing procedure followed by resuspension of coated NRs in water and removal of macro aggregates by filtration through a 0.22 µm filter. Macrocycle coated nanoparticles were subsequently subjected to trifluoroacetic acid (TFA) hydrolysis to unmask the protected acetate arms to give nanoparticles coated using 100% macrocycle (1) and 50% macrocycle, 50% PEG (2). The 100% PEG coated NRs (3) do not require a hydrolysis step. Transmission electron microscopy (TEM) analysis was carried out on the final constructs (1-3) to ensure that synthetic coating protocols do not affect the core iron-oxide structure, see Figure 1. All constructs were found to have a similar size distribution, with rods of length 80-130 nm. Fourier transform infra-red (FTIR) spectroscopy was carried out to validate the presence of the surface coating, see Figure 2. All coating types had the expected stretches of Fe-O, Si-O-C and Si-O-Fe between 600 and 1000 cm -1 . 33 Surface coatings can be differentiated due to the independent peak characteristics of each type. In case of 1, characteristic macrocycle-derived C-N stretches can be assigned at ca. 1200 cm -1 . In case of 3, C-O-C linkage, characteristic to PEG, can be assigned at 1117 cm -1 . 2 was designed to contain a mixture of coating component and was produced using a 1 : 1 ratio of the two coating components, with the FTIR showing a broad peak in the region overlapping both regions. The chemical composition of surface coating is corroborated by combustion analysis of coated NRs nitrogen content (0.79, 0.29 and 0.00% for 1, 2 and 3, respectively) as the only inclusion of nitrogen is from the tetraazamacrocycle coating, decreasing for 2 relative to 1 as expected, see Figure  S1.

Nanorod construct analysis
The surface charge of nanoparticles gives an indication of their aqueous suspension stability, an essential requirement for translation to in vivo studies. A large negative or positive value would suggest an increased electrostatic repulsion between individual nanoparticles, decreasing the tendency for aggregation. [34][35][36] The zeta potential of the constructs (1-3) was measured at a constant concentration (7.5 mM Fe) at pH 7 in PBS, see Figure S2. All constructs have an overall negative charge between -10 and -20 mV. Analogous PEG methyl ether terminating nanoparticles have a similar surface charge 37 and DOTA based systems have deprotonated carboxylic acid groups at pH 7. 38, 39 All constructs have a relatively similar zeta potential, with 2 having the highest negative potential at -17.9 mV. The high negative surface charge of all constructs translates to high stability in aqueous solution, with no visible precipitation observed for any of the samples after 2 months.
The size of nanoparticles is an important characteristic governing their in vivo biodistribution pathways and, therefore, determining their potential for application as MRI contrast agents. Very small (<5 nm) nanoparticles are generally eliminated via the renal system and large (>200 nm) nanoparticles are excreted via the spleen and liver. Either of these elimination pathways results in short blood circulation times. 10,40,41 Samples of all three constructs were suspended in water (at a concentration of 0.1 mM Fe) and their hydrodynamic size was measured by nanoparticle tracking analysis (NTA) in which mean and modal sizes were determined, see Figure S3. The constructs had mean sizes in the range of 99-155 nm, with an increase in size correlating with the addition of macrocycle to the surface coating, all are an acceptable size for in vivo applications. There are challenges with NTA analysis and the use of rods rather than spheres: (a) how do these numbers relate to the different nanorod axes?; and (b) if the in vivo excretion pathways of nanorods are similar to that of nanospheres and, if not, how they are affected by length?
Prior to radiolabelling and in vivo studies, the magnetic behaviour of the NR constructs was investigated in vitro at 3 T, a magnetic field strength correlating to common clinical MRI scanners. The r 1 and r 2 relaxation rates were calculated for all three constructs (1-3), see Figure 3 and Figure 4. All constructs are relatively weak T 1 contrast agents. As T 2 contrast agents, all constructs showed fast relaxivity with improvement correlated to the increasing PEG content of the NR coating.
The large variability in both r 1 and r 2 values, as a function of chemical composition of surface coating, could be explained by a combination of factors, such as variation of hydrodynamic size, hydrophilicity and permeability of the different coatings to water molecules. 42 The increase in hydrodynamic size inversely correlates with both r 1 and r 2 values. Although superparamagnetic contrast agents generate a long-range magnetic field for the spin-spin relaxation (T 2 ) process, this effect decreases as distance increases. Similarly, as the mechanism for transverse (T 1 ) relaxation is proximity based, an increase in coating thickness leads to a decrease in r 1 values. 43 The hydrophilicity and permeability of the surface coating also has an effect on relaxivity rates, 44 although more relevant for T 1 relaxation, varying the ease of water molecule access to the iron surface. The more easily that water molecules can transfer into close proximity with the iron surface, and exchange rapidly, the faster the relaxivity rates.

Gallium-68 radiolabelling and stability
In aqueous solutions, gallium is almost exclusively found in the oxidation state +3, and is highly influenced by pH, with the optimal coordination conditions situated between pH 3 and 5, since a more acidic environment may inhibit the complex formation due to protonation of donor atoms, and a neutral or basic environment causes unreactive hydroxides to form. The constructs described in this study were prepared and radiolabelled with 68 GaCl 3 following a standard procedure. 32 Briefly, to a dried aliquot of 68 GaCl 3 was added a sample of NRs in 0.2 M sodium acetate solution and the reaction mixture was heated to 90 ο C for 15 minutes, whilst monitoring by radiothin layer radio-chromatography (radio-TLC). In accordance with our previous results, the macrocyclic chelator was not required for the radiochemical incorporation of gallium-68 and quantitative radiolabelling was achieved in a 15 minute reaction time. 32 In order to simulate conditions of possible gallium(III) ion release in vivo, two assays were performed, see Figure 5. Firstly, the radiolabelled constructs were incubated in human serum. All three constructs (1-3) showed high stability (>95%) after 3 hours at 37 ο C, a reasonable time point considering the short half-life of gallium-68 (68 min). As gallium(III) has a similar atomic radius to iron(II), the primary mechanism for gallium(III) bioprocessing in vivo is via iron ion pathways, such as complexation with apo-trasnferrin in blood plasma. 45 In order to probe this property, the radiolabelled constructs (1-3) were incubated with an excess of apo-transferrin (0.5 mg) at 37 ο C. Again, after 3 hours >95% of gallium-68 remained attached to the constructs, with no significant difference observed between differently coated NRs in either assay.
Importantly, the 68 Ga radiolabelling results indicate that a bifunctional chelating agent is not required for rapid radiolabelling or to stably attach the metal ion in vitro, as demonstrated in chelator competition assays, with 3 showing analogous labelling stability to 1 and 2. Experiments on comparable non-PEG silica coatings using triethoxy(ethyl)silane (data not shown) yielded equivalent radiochemical results, indicating that the effect is not solely a property of the PEG chains. Spectroscopic analysis of nonradioactive 69/71 Ga isotope analogues failed to provide data on the existence of Ga-O-Si or Ga-O bonds as the peaks were masked by the intense Fe-O-Si and Fe-O peaks. Due to the polymeric nature of the surface coating and the high density, determined to be 37% w/w or 5.62 ligands per nm 2 of iron oxide for 3 (see Figure S4), 46 we assume that the attachment of gallium(III) to the core is unlikely. In order to assess the hypothesis of the formation of Ga-O-Si bonds, we attempted the 68 Ga radiochemistry at a lower pH, which would cause siloxy protonation and hence would not be able to react with the gallium(III). Reactions performed at pH 2 with 3 using the same temperature and reaction time yielded only 4.4% ± 0.9 (n=3) RCY. Also, by modifying the pH of the same reactions back up to 5 and heating again for 15 minutes, the RCY increased to 96.7% ± 1.5 (n=3), supporting this hypothesis.

In vivo imaging biodistribution and stability
All three radiolabelled NRs demonstrate comparable in vitro stability. The most interesting construct is 3 due to its lack of a chelator moiety. Thus, PET-CT and MR imaging studies were carried out in mice using gallium-68 labelled 3 to assess the co-localisation of PET and MRI signals and evaluate the in vivo stability of the construct.
Iron oxide nanoparticles of a similar size to ones used in this study are quickly taken up by the reticuloendothelial system (RES) and accumulate in the liver and spleen. 12,37,47 PET scanning revealed that the 68 Ga-3 construct rapidly (<5 min) accumulated in the liver ( Figure 6). Some limited uptake in the lung was noted in some scans; this has been observed previously with iron oxide nanoparticles and has been attributed to possible embolisation caused by post-injection aggregation. 48,49 After the PET-CT study the mouse was transferred to the MR scanner and both T 1 and T 2 MR images were acquired (see Figure 6 and Figure S6). As expected the liver T 1 signal was not significantly affected by the 68 Ga-3 construct. Whereas T 2 weighted images revealed hypointense liver signal intensity, indicating localisation of the iron oxide NRs in the liver.
As a comparison, 68 Ga 3+ was administered as gallium(III) citrate and its biodistributon assessed in order to observe the biodistribution that may occur if the 68 Ga were released from the nanorod in vivo. PET scans show that the 'free' 68 Ga 3+ is distributed more widely throughout multiple organs and is cleared much more slowly from the blood pool, see Figure S5. A significantly lower amount of radioactivity is noted in the liver, which is expected due to 68 Ga coordination to transferrin.
The presence of 68 Ga on the nanoparticle was also confirmed by ex vivo analysis, see Figure S7. After sacrifice (90 min post-injection), the liver was harvested and processed to separate the nanoparticles from the cell pellet, after which radio-TLC was used to determine if the radioactivity was still ARTICLE Nanoscale 4 | J. Name., 2012, 00, 1-3 This journal is © The Royal Society of Chemistry 2012 attached to the nanoparticle. As a comparison, the same experiment was carried out after independent administration of 68 Ga-citrate and 68 Ga-transferrin. It can be seen that after administration of 68 Ga 3+ or 68 Ga-transferrin, the majority of the radioactivity is attached to transferrin. When 68 Ga-3 is administered, >90% of the radioactivity is still attached to the nanorod even after analysis 90 min post-injection, which is a relatively late in vivo time point for gallium-68 imaging given its short half-life.
Debate is ongoing as to the best design parameters and applications for combined PET/MR imaging agents. 50 Ironoxide nanoparticles of this size localise in the liver via phagocytosis into the Kupfer cells and malignant liver legions often lack normal Kupfer cells and remain "bright" in an MRI scan upon SPION administration. 51,52 However, the lower MRI sensitivity means that differentiation between regions with and without SPION may be challenging and may limit widespread clinical utility. Clinical utility could be increased by using PET radioisotope labelled SPIONs such as those developed in this study due to the increased sensitivity of PET and subsequent clarity of low uptake regions.

Conclusion
In this study, we have demonstrated that the positron emitting radioisotope 68 Ga can be used to radiolabel silica coated ironoxide nanorods via surface interactions in the absence of a bifunctional chelating agent. Studies of 68 Ga-3 (NRs coated with 100% PEG) confirm that the construct is stable in vivo and demonstrate its applicability as a PET/MR multimodal imaging agent. This is particularly interesting, because coated NRs can be easily obtained in one step from commercial materials, removing challenging and complex organic synthesis steps. Thus, this study establishes a novel straight-forward method for the preparation of PET/MR multimodal imaging agents which have the potential to combine the high sensitivity of PET with MR contrast for malignant liver lesion imaging and also offer the potential for facile surface labelling methods that can be applied to currently licensed iron oxide MRI contrast agents for clinical use. This would facilitate more rapid progress through patient trials to regulatory approval.

Synthesis of coated iron-oxide nanorods
General procedure for coating nanorods with siloxanes: Freshly prepared nanorods were coated with siloxane-DO3A and 2methoxy(polyethyleneoxy) propyltrimethoxysilane (Siloxane-PEG) in different ratios. In each reaction, bare NRs (0.3 g, 1.3 mmol) were suspended under nitrogen in 60% ethanol (100 ml) by extensive sonication. Ammonium hydroxide (20 ml, 35%) was added drop-wise and the solution was stirred for 30 min. The designated ratio of siloxane-DO3A and siloxane-PEG (total 0.34 mmol) was dissolved in ethanol (10 ml) and added over 10 mins, the reaction was then stirred for 48 hours at room temperature. The solid precipitate formed was attracted to the bottom of the flask using an external rare earth magnet and the clear aqueous supernatant liquid was decanted. The washing and decanting procedure was repeated with 60% ethanol (2 x 100 ml), ethanol (2 x 100 ml), methanol (2 x 100 ml) and diethyl ether (100 ml) and dried. The crude sample was resuspended in water (20 ml) filtered through a 5 µm filter followed by a 0.45 µm filter (Merck®) and dried under high vacuum to give coated NRs as a black/brown powder. Nanorods functionalised with 100% siloxane-DO3A or with 50% siloxane-DO3A were suspended in DCM (20 ml), TFA (2 ml) was added and the reaction was shaken at room temperature for 24 hours. The solid precipitate was attracted to the bottom of the flask using an external rare earth magnet and the clear aqueous supernatant liquid was decanted. The washing and decanting procedure was repeated with ethanol (2 x 20 ml), methanol (20 ml) and diethyl ether (100 ml) and dried. As previously, the crude sample was re-suspended in water (20 ml) filtered through a 5 µm filter followed by a 0.45 µm filter (Merck®) and dried under high vacuum to give coated NRs as a black/brown powder. (1 = 124 mg, 2 = 154 mg, 3 = 97 mg)

MRI relaxivity studies
Relaxivity measurements were carried out on a clinical 3.0 T Discovery MR750 General Electric scanner. Longitudinal (T 1 ) and transverse (T 2 ) relaxation times were determined at a range of iron concentrations: 1, 0.5, 0.25 and 0.125 mM in water. A linear plot of concentration versus 1/relaxation time gives r 1 and r 2 as the plot gradient.

PET/CT image acquisition and image analysis
CD1 nude female mice (Charles River, UK) were anaesthetised with 1-2% Isoflurane, the tail vein was catheterised and they were placed in the Minerve Small-Animal Imaging Cell (Minerve, France) and transferred to the SuperArgus 2R preclinical PET/CT scanner (Sedecal, Spain). At the start of the PET acquisition, animals were injected with ~10 MBq, 50 µg Fe of 68 Ga-3 intravenously (IV) via the tail vein and a 90minute whole-body imaging sequence acquired (2 bed positions of 2 x 60s, 4 x 120s and 7 x 300s). CT images were acquired via 360 projections/8 shots at a tube voltage of 45 kV and current of 140 µA. Respiration and temperature were monitored throughout using the SA Instruments 1025T monitoring system (SA Instruments, CA, USA). All procedures were carried out in accordance with the Animals in Scientific Procedures Act 1986 and UKCCCR Guideline 2010 53 by approved protocols following institutional guidelines (Home Office Project License number 60/4549 held by Dr. Cawthorne).

MR image acquisition
Magnetic resonance imaging was performed on a wide bore vertical 11.7 T MR scanner (Bruker, Germany). With a 4 cm vertical field of view with slide thickness' of 1.0 mm and interslice distances of 1.5 mm. T 1 images were acquired using a RARE inverse recovery sequence in which TE = 7.5 ms, TR = 3500 ms, flip angle = 180 ο , SW = 79365.1 Hz. T 2 images were acquired using a T 2 weighted FLASH sequence in which TE = 6.9 ms, TR = 350 ms, flip angle = 30 ο , SW = 50505.1 Hz. 68 Ga-transferrin synthesis: Processed and dried 68 GaCl 3 was added to apo-transferrin (0.5 mg) in water (200 µl) and incubated at 37 o C for 30 minutes. Radio-TLC indicated the synthesis of 68 Ga-transferrin in >95% radiochemical purity.

68 Ga-3 in vivo liver stability
Ex vivo liver processing: The liver was removed after sacrifice and ground in liquid nitrogen. The liver samples (ca. 80 mg) were added to lysis buffer (10x w/w), sonicated for 10 minutes then centrifuged at 20,000 RPM for 10 minutes. The supernatant was separated from the cell pellet. Radio-TLC was carried out on the supernatant to determine gallium attachment