Clustering superparamagnetic iron oxide nanoparticles produces organ-targeted high-contrast magnetic resonance images

Nicholas J Hobson1,2, Xian Weng1,2, Bernard Siow3, Catarina Veiga4, Marianne Ashford5, Nguyen TK Thanh2,6, Andreas G Schätzlein1 & Ijeoma F Uchegbu*,1 1UCL School of Pharmacy, 29–39 Brunswick Square, WC1N 1AX London 2UCL Healthcare Biomagnetic & Nanomaterials Laboratories, 21 Albemarle Street, W1S 4BS London 3Centre for Advanced Biomedical Imaging, UCL, Gower Street, WC1E 6BT London 4UCL Medical Physics & Biomedical Engineering, Malet Place, Gower Street, WC1E 6BT London 5Advanced Drug Delivery, Pharmaceutical Sciences, IMED Biotech Unit, AstraZeneca, Macclesfield, SK10 2NA Cheshire 6Biophysics Group, UCL, Gower Street, WC1E 6BT London *Author for correspondence: ijeoma.uchegbu@ucl.ac.uk

In the future, it is predicted that medicine will rely heavily on more accurate diagnosis of patients, especially as medicines become more personalized [1,2].Magnetic resonance imaging (MRI) is a powerful diagnostic technique that is able to give excellent anatomical images in a noninvasive real-time manner using nonionizing radiation [3].This said, however, it can often be challenging to distinguish between healthy and diseased sites in the same tissue and so contrast agents are routinely employed to enhance the sensitivity of the technique [4].Over the last three decades, superparamagnetic iron oxide nanoparticles (SPIONs) have gained considerable attention as negative MRI contrast agents due to their unique magnetic properties [5][6][7][8][9].In recent years however, a number of prominent SPION contrast agents have been withdrawn or discontinued [10][11][12][13].While it is difficult to ascertain the precise reasons for the withdrawal, clinical uptake was obviously a factor due presumably to the poor contrast obtained with these agents, when compared with gadolinium-based agents [14].However, there are still considerable toxicity concerns associated with the use of gadolinium contrast agents, such as brain and bone accumulation and kidney toxicity [15], a fact that underpins the requirement for safer contrast agents.Our aim is to use nanoparticle engineering to prepare contrast agents from biocompatible materials, such as SPIONs, that produce unequivocal contrast images.We have discovered that clustering SPIONs into raspberry shapes within a polymeric envelope leads to vastly improved image contrast and it is the improved image contrast that is the thrust of the current work.The self-assembling polymer used is a chitosan amphiphile: N-palmitoyl-N-monomethyl-N-N-dimethyl-N-N-N-trimethyl-6-O-glycolchitosan (GCPQ), which has previously been used to deliver challenging drug molecules [16][17][18][19].GCPQ has been through good laboratory practice standard preclinical toxicology testing with no toxicity issues observed and has shown good cell biocompatibility (IC 50 ∼1 mg ml -1 ) [20,21].Previous studies attempting to encapsulate SPIONs within self-assembling aggregates have often yielded impressive in vitro results but many have limited or no information regarding their in vivo biological stability and/or their pharmacokinetic and pharmacodynamic characteristics [22- 29].In vivo evaluations are often challenging as polymeric surfactant nanoparticles are destabilized by biological environments where there is the presence of salts and proteins.Protein adsorption in vivo is a possible destabilizing factor.The work presented here clearly demonstrates that a positively charged raspberry SPION, comprising 5 nm SPIONs clustered into a larger raspberry shape, may be used as a superior MRI negative contrast agent.Clustering, as opposed to the synthesis of larger SPIONs, means that these 4-5 nm core size SPIONs are more likely to be excreted via the urine [30,31].

Materials & methods
All chemicals and reagents were purchased from Sigma-Aldrich, UK unless otherwise specified.Dialysis membranes were purchased from Medicell International Ltd., London UK.All reagents were used without further purification.Animals were purchased from Harlan, UK.
Degraded glycol chitosan (1.0 g) and NaHCO 3 (0.75 g) were dissolved into a mixture of ethanol and water (24:76).Palmitic acid N-hydroxysuccinimide (1.58 g, Toronto Research Chemicals, Canada) was dissolved into ethanol (300 ml) and was added drop wise.The reaction was left to stir for 72 h protected from light.Ethanol was removed under vacuum and the remaining aqueous phase extracted using diethyl ether three-times.The aqueous phase was dialyzed exhaustively against water (MWCO: 3.5 kDa) and then lyophilized to obtain a white fibrous solid.Yield = 96%.
Palmitoylated glycol chitosan chitosan (300 mg) was dispersed in N-methyl-2-pyrrolidone (25 ml) and stirred vigorously overnight.Sodium hydroxide (40 mg) dissolved in ethanol (2 ml) and sodium iodide (45 mg) dissolved in ethanol (2 ml) were both added to the solution and the reaction was heated to 36 • C. Methyl iodide (0.45 ml) was added and the reaction was stirred under a nitrogen environment for 3 h.The product was precipitated using diethyl ether and left overnight.The resulting precipitate was washed three-times with diethyl ether (50 ml), dissolved in water (100 ml) and then dialyzed exhaustively against water (MWCO: 3.5 kDa).The dialysate was passed through an Amberlite IRA-96 column to remove iodide and lyophilized to obtain a white fibrous solid.Yield = 70%.

Synthesis of iron oxide nanoparticles via high temperature decomposition
Iron oxide nanoparticles were synthesized in a procedure outlined by Kim et al. [33].In a typical reaction, FeCl 3 .6H 2 O (3.0 g) and sodium oleate (10.2 g, TCI Chemicals, Japan) were dissolved in a solvent mixture comprising: 17 ml water, 23 ml ethanol and 40 ml hexane.The resulting mixture was stirred for 4 h at 70 • C.After 4 h, the reaction mixture was allowed to cool to room temperature and the hexane layer extracted and washed three-times with distilled water (30 ml).The hexane was then evaporated under vacuum to yield a waxy orange oil.
The iron-oleate complex (1.8 g) from above, oleic acid (0.57 g, CRODA, UK) and oleyl alcohol (1.61 g) were dissolved in diphenyl ether (10 g).The reaction mixture was heated to 250 • C and then kept at this temperature for 20 min under a nitrogen atmosphere.The mixture was then rapidly cooled to room temperature and acetone (50 ml) was added to precipitate out the nanoparticles.Nanoparticles were collected via centrifugation (3000 rcf, 10 min) and the supernatant was discarded and replaced with acetone (40 ml) to wash the nanoparticles.The acetone was removed by centrifugation and the acetone wash was repeated three-times.The washed nanoparticles were finally dispersed and stored in hexane.To obtain dry nanoparticles for characterization, the hexane was removed under vacuum.Acetone was added and the nanoparticles gently collected with a spatula and dried overnight under vacuum.The particles were then weighed and made up to the desired concentration in hexane, chloroform or THF.

Formulation raspberry SPIONs
In a small vial, iron oxide nanoparticles in hexane (2 ml, 4 mg/ml) were added to a GCPQ dispersion (5 ml, 1 mg/ml).The dispersion was then sonicated using probe sonication (2 × 5 min, Soniprep 150, MSE, UK) at the interface between the organic and aqueous phases, creating a brown emulsion.This emulsion was left to separate over 48 h.The aqueous layer was collected and any residual hexane was allowed to evaporate overnight.The product -raspberry SPIONs -was frozen and lyophilized to obtain a dry product, which rapidly dispersed when added to water (5 ml).For the in vivo investigations, a formulation with a higher level of iron oxide nanoparticles was made by increasing the SPIONs and polymer concentration by a factor of 15: iron oxide nanoparticles in hexane (2 ml, 60 mg/ml) were added to a GCPQ dispersion (5 ml, 15 mg/ml) and processed as described above.
Fourier-transform infra-red spectroscopy Fourier-transform infra-red (FT-IR) spectra were obtained using a PerkinElmer Spectrum 100 FT-IR Spectrometer equipped with a Universal attenuated total reflectance (ATR) accessory (PerkinElmer Inc., MA, USA).The sample was mounted onto the Universal diamond ATR top-plate and the force gauge was increased to 100 N. The spectrum was recorded in a range of 4000-650 cm -1 using PerkinElmer Spectrum Express Version 1.02.00 software.

Thermogravimetric analysis
Thermogravimetric analysis (TGA) measurements (Pyris 6 TGA; PerkinElmer Inc.) were performed on samples.Samples (5-6 mg) were weighed into a ceramic crucible.Nitrogen (20 ml min -1 ) was used as a purge gas.The samples were heated from 20 to 450 • C at 10 • C min -1 .The mass loss between 20 and 100 • C was used to determine the water content of samples.The onset of degradation was obtained using the Pyris Software for Windows Version 3.81 (PerkinElmer Instruments; PerkinElmer Inc.) Ligand exchange of iron oxide nanoparticles with meso-2,3-dimercaptosuccinic acid This ligand exchange protocol was carried out according to the method previously described by Palma et al. [34].Briefly, iron oxide nanoparticles (20 mg) were precipitated from the hexane disperse phase (1 ml) by the addition of ethanol and were collected via centrifugation (9000 rcf, 10 min).The supernatant was discarded and replaced with ethanol and this procedure was repeated twice.The final precipitate was re-dispersed in toluene.Dimercaptosuccinic acid (DMSA) (36 mg) in DMSO (2 ml) was added to the toluene mixture and the reaction was bath sonicated (10 min) and left to stir at room temperature for 48 h.After 48 h, a sediment was observed and this was collected via centrifugation (9000 rcf, 10 min).The supernatant was discarded and replaced with ethanol and this procedure was repeated twice.The final precipitate was re-dispersed in water, titrated to pH 10 using NaOH and then filtered through a 0.22 μm membrane.

Transmission electron microscopy
Images were collected using an FEI CM120 BioTwin Transmission Electron Microscope (Ex.Philips, Eindhoven, The Netherlands).Digital images were captured using an AMT digital camera.A drop of the sample was placed future science group 10.2217/nnm-2018-0370 on Formvar C /Carbon Coated Grid (F196/100 3.05 mm, mesh 300, TAAB Labs Ltd, UK).Samples were stained with uranyl acetate (1%) where indicated.

Dynamic light scattering
Particle size was determined by dynamic light scattering (DLS) on a Malvern Zetasizer Nano ZS (Malvern Instruments, UK) using a laser wavelength of 633 nm.Samples were inverted several times and then left to equilibrate for 10 s before particle sizing was carried out.The accuracy of the instrument was assessed periodically using latex beads (polystyrene, mean size: 0.1 μm) in 50 mM sodium chloride dispersant phase.

Zeta potential & size measurements
The particles' zeta potentials were measured using a Malvern Zetasizer Nano ZS (Malvern Instruments) paired with an MPT-2 automatic titrator (Malvern Instruments).Prior to any titration, the pH probes were calibrated using pH 4, pH 7 and pH 10 standards.Nanoparticles were manually titrated to pH 4. The instrument then automatically titrated the sample and measured the particle size and zeta potential between pH 4 and 11.Reservoirs were filled with HCl and NaOH at concentrations of 0.05 M. Measurements were carried out in gold-plated disposable capillary cells (Malvern Instruments).The accuracy of the instrument was assessed periodically using a polystyrene zeta standard (Malvern Instruments).
X-ray diffraction X-ray diffraction patterns were collected on an x-ray diffractometer (PanAlytical, UK) using CoKα radiation λ = 1.79 Å.Samples were prepared by pressing dried powders on a zero background silicon wafer and diffraction patterns were collected within a 20-100 2θ (degrees) range.

Magnetic properties
Magnetization curves were recorded in a Quantum Design hybrid Superconducting Quantum Interference Device-Vibrating Sample Magnetometer (SQUID-VSM) at 300 K, with applied fields up to 7 T. Samples were lyophilized to obtain a dry product.Weighed samples of the dry product were placed into plastic capsules and measured in a straw sample holder.

Stability studies on raspberry SPIONs
The stability of the raspberry SPIONs was investigated in: H 2 O, 0.9%w/v NaCl, phosphate-buffered saline ([PBS], pH = 7.4) and 5%w/v dextrose.Raspberry SPIONs were frozen and lyophilized to obtain a dry product and then dispersed in MilliQ water.Following this, the raspberry SPIONs (2.5 ml) were added to: H 2 O (2.5 ml), 1.8% NaCl (2.5 ml), double strength PBS (2.5 ml), 10%w/v dextrose (2.5 ml), diluting each of the buffers by a factor of 2 and were incubated at 37 • C. At various time points aliquots were removed, centrifuged (224 rcf × 10 min) and then the supernatant was analyzed for iron content via the 1,10-phenanthroline iron quantification assay.Briefly, the supernatant (200 μl) was added to HCl (4M, 200 μl) and digested at 70 • C. In a sample vial, sodium acetate (125 mg/ml, 450 μl) and hydroxylamine hydrochloride (10 mg/ml, 50 μl) were added to the supernatant (200 μl) followed by the 1,10-phenanthroline (10 mg/ml, 300 μl) turning the sample bright red if iron was present.Sample vials were stored, protected from light for 1 h and then measured using UV/VIS (ELx808 microplate reader, BioTek Instruments, UK) at 510 nm.The iron concentrations were determined based on a calibration curve made from iron(II) chloride tetrahydrate.The plasma stability of the raspberry SPIONs was measured by measuring iron content of the colloidal fraction, using ICP-AES (Varian 720, Agilent, UK), after incubation in plasma.Blood was obtained from male Wistar rats (400-450 g, Harlan), collected into microtainer tubes coated in ethylene diamine tetraacetic acid (EDTA, K3E Vacutainer, BD Biosciences, UK) and maintained on ice.Plasma was obtained as the supernatant after centrifugation of blood samples (252 rcf × 10 min, 4 • C) and stored at -20 • C until required for the experiments.Raspberry SPIONs (0.5 mg/ml Fe, 2.5 ml) were added to rat plasma (2.5 ml), diluting the plasma by a factor of 2 and were incubated at 37 • C. At various time points, aliquots were removed, centrifuged (224 rcf, 10 min) and the supernatant was analyzed for iron content using ICP-AES.Briefly, the supernatant (300 μl) was added to 70% HNO 3 (300 μl) and the particles digested at 70 • C. The digestant was diluted by a factor of 35 with MilliQ water and the diluted sample was measured using ICP-AES.The iron concentrations were determined based on a calibration curve made from an iron standard for ICP-AES (TraceCERT, Fluka, UK).MRI contrast measurements MR relaxivities were measured using a 1T bench-top MRI scanner (ICON™, Bruker, UK) at 37 • C. For r 2 relaxation, a spin echo sequence was used with the following parameters: TE = 6-100 ms, TR = 1845 ms.The r 2 relaxivities were determined by curve fitting the echo times (TE) using MATLAB (Mathworks, UK).The relaxivities were plotted against the iron concentration and the r 2 determined from the gradient of the slope.r 1 relaxivity was measured using an inversion recovery sequence with the following parameters: TI = 35-7500 ms, TR = 15000.The r 1 relaxivities were determined by curve fitting several inversion times (TI).The relaxivities were plotted against concentration and r 1 determined from the gradient of the slope.

Biodistribution studies
Female BALB/c mice were used (20-25 g, Harlan).All animal experiments were performed under a UK Home Office Animal License and in accordance with local ethics committee rules.A raspberry SPION suspension (6.5 mg Fe/ml) was prepared in 5% dextrose, titrated to pH 6 using 0.1 M NaOH and filtered through a 0.22 μm filter.The formulation was intravenously administrated to mice via the tail vein (32.5 mg Fe/kg).At various time points, animals were euthanized with Pentoject™ (pentobarbitone sodium 20% w/v, 150 mg/kg) administered via intraperitoneal injection.Blood was immediately collected via cardiac puncture and collected into microtainer tubes coated in EDTA (K2E Vacutainer, BD Biosciences, UK).Various tissues were collected (liver, spleen, brain, heart, lungs and kidneys), thoroughly rinsed in 0.9% w/v NaCl (50 ml) and then stored at -80 • C until further analyses could be performed.The tissues were lyophilized, the dry weight was determined and a yttrium standard for ICP was added (1000 ppm, 50 μl, TraceCERT).Dry tissue samples were degraded in HNO 3 (70%, 250 μl) for 1 h at 70 • C.This was followed by the addition of H 2 O 2 (30% v/v, 200 μl) and the sample further degraded for 1 h at 70 • C. The digestant was cooled, diluted 35-times with MilliQ water and centrifugation (4032 rcf, 10 min) was used to remove any fatty residue.Samples were analyzed using ICP-AES.The iron concentrations were determined based on a calibration curve made from an iron standard for ICP-AES (TraceCERT, Fluka, UK).

In vivo MRI measurements
Female BALB/c mice were used (20-25 g, Harlan).Raspberry SPIONs (5 mg Fe/kg) were administered via the tail vein and images were collected pre-treatment and 1 h post administration of the contrast agent.A multi-slice spinecho pulse sequence with the following parameters was used to collect axial and coronal images: TE = 6-100 ms, TR = 1845 ms.Animals were anesthetized using isoflurane while the animal's temperature was controlled via a heated bed and its respiration was monitored.All experiments were nonrecovery and animals were euthanized via an overdose of isoflurane and death was confirmed via neck dislocation.The r 2 relaxation was determined by curve fitting the echo times (TE) of each pixel in the image using MATLAB (Mathworks, UK).Commercially available Ferucarbotran (Resovist R ) was also administrated and measured under the same conditions.

Nanoparticle synthesis & morphology
Uniform magnetic nanocrystals were synthesized via high temperature thermal decomposition of an iron-oleate complex in the presence of oleic acid, oleyl alcohol and diphenyl ether, using a method reported by Kim et al. [33].
Diphenyl ether was the most suitable high-temperature solvent tested (b.p. 258 • C) as the nanoparticles were easily purified, when using this solvent, by washing with acetone.Transmission electron microscopy (TEM) experiments confirmed the presence of nanoparticles, which had an average diameter of 4.8 ± 0.8 nm, were spherical in nature (Figure 1A & B) and could easily be dispersed in hexane, chloroform and tetrahydrofuran (THF).The prepared nanoparticles were coated in a layer of oleic acid, making them extremely hydrophobic and not suitable for biological applications.Nevertheless their hydrophobic surface made them ideal for clustering into raspberry SPIONs as they could be aggregated within an amphiphilic polymer nanoparticle, with the hydrophobic SPION surfaces interacting with the hydrophobic parts of the amphiphilic polymer.The amphiphilic polymer, N-palmitoyl-N-monomethyl-N-N-dimethyl-N-N-N-trimethyl-6-O-glycolchitosan (GCPQ), was used to achieve the clustering.The amphiphilic nature of GCPQ allows the polymer to self-assemble into nano-sized micelles of around 32 nm in diameter that are stable even at low polymer concentrations (Supplementary Figure 1) [18].The hydrophobic oleic acid coated SPIONs may be incorporated into GCPQ nanoparticles by a system of emulsification, with the aid of ultrasonication, followed by organic solvent evaporation.This method yielded raspberry SPIONs (Figure 1C-F 1F), as evidenced by the white interior spaces and dark coronas surrounding the raspberry SPIONs in the uranyl acetate stained TEM images.The uranyl acetate stain adheres to hydrophilic regions and avoids hydrophobic regions, making the hydrophilic regions black and hydrophobic regions white in the image.FTIR data also shows vibrations for the oleic acid (2924 cm -1 = CH 2 alkyl stretch) and GCPQ moieties (1647 cm -1 = C=O stretch amide and 1059 cm -1 = C-O stretch sugar moiety) in the raspberry SPIONs (Supplementary Figure 3).TGA data reveal the removal of both GCPQ and oleic acid from the raspberry SPIONs on heating (Supplementary Figure 4).

Nanoparticle size & stability
Several raspberry SPION formulations, containing different SPION, GCPQ weight ratios (1:2.5, 1:0.5 and 1:0.25) were formulated in water and sized by dynamic light scattering (DLS) over 90 days.Formulations made with a higher SPION, GCPQ ratio than 1:0.25 did not produce raspberry SPIONs, indicating that a minimum concentration of GCPQ is needed for raspberry SPIONs to form.Raspberry SPIONs, prepared from SPION, GCPQ weight ratios of 1:2.5, 1:0.5 and 1:0.25 have mean diameters (and polydispersity indices in parenthesis) of 140 nm (0.15), 190 nm (0.14) and 230 nm (0.14), respectively (Supplementary Figure 5).These sizes remain relatively unchanged throughout the 90-day period, indicating that the raspberry SPIONs exhibit good colloidal stability and it appears that the raspberry SPION particle size may be controlled by varying the SPION, polymer weight ratio (Figure 2A).It is worth mentioning that some nanoparticles did settle when the formulations were stored for prolonged periods of time (>4 weeks).These raspberry SPIONs are quite dense since they contain a high atomic number element.However, any sediment could be easily re-dispersed with gentle agitation and this sedimentation had no effect on the measured nanoparticle size.A similar phenomenon was also observed by Lin et al. with their chitosan SPION [35].Colloidal stability was also monitored in various liquids of biomedical relevance (water, [PBS, pH = 7.4], 5% w/v dextrose, 0.9% w/v NaCl) at 37 • C and raspberry SPION dispersions in these liquids were analyzed after centrifugation, to remove the noncolloidal fraction of the formulation.Analysis involved the measurement of the total iron content in the formulations using the 1,10-phenanthroline iron quantification assay.The raspberry SPIONs retain their colloidal stability over the 48-h period in all the media studied (Figure 2B).The stability of the raspberry SPIONs was also investigated in 50% rat plasma using inductively coupled plasma-atomic emission spectroscopy (ICP-AES) as the 1,10-phenanthroline iron quantification assay did not work with biological samples (Figure 2C).The raspberry SPIONs had satisfactory colloidal stability properties in plasma, retaining 70% of the iron content within the colloidal fraction after 24 h and 66% over 48 h.

Nanoparticle characteristics
The surface chemistry of a nanoparticle system is an important feature, as it is this solid-liquid interface, which will interact with the environment and influence colloidal stability [36][37][38] and in vivo performance.To investigate the raspberry SPION's surface charge, the zeta potential was measured as a function of pH (Figure 3A).It was found that, like many other nanoparticles of this nature, the zeta potential of the raspberry SPIONs was dependent on pH and that at neutral pH (pH = 7.0) raspberry SPIONs had a positive surface charge (+24.4 ± 1.5 mV).The effect of pH on nanoparticle size was also measured (Figure 3B).It was found that the size of the raspberry SPIONs increased at basic pH (pH > 9), indicating that the raspberry SPIONs aggregate at basic pH.The solubility of GCPQ is reduced at basic pH [39] and this would reduce the hydrophilic head group area of the polymer and in turn the raspberry SPIONs' surface coverage by the polymer, resulting, ultimately in aggregation.
x-ray diffraction (XRD) patterns were collected on the raspberry SPIONs and the oleic acid coated SPIONs (Figure 3C).The diffraction patterns are similar to the patterns previously reported for SPIONs made via hightemperature thermal decomposition [40][41][42].One noticeable characteristic observed was that diffraction peaks obtained were weak in intensity and broad in nature and this is due to the small nanoparticle crystal size.The average crystal size of the oleic acid SPIONs determined by the XRD data and calculated by the Scherrer equation was (3.6 ± 0.1) nm.The small decrease in measured size from that found using the TEM images could be due to the spin canting effect, a small, noncrystalline layer often found on the surface of SPIONs [43], which would be unaccounted for in the XRD assay.
The magnetic behavior of the raspberry SPIONs was investigated by collecting field-dependant magnetization plots (M vs H plots) using a super quantum interference device (SQUID) (Figure 3D).Both the oleic acid coated SPIONs and the raspberry SPIONs display superparamagnetic behavior with no observable remnant magnetization.The magnetic saturation of the oleic acid coated SPIONs and raspberry SPIONs was found to be 29.6 ± 3.7 emu/g Fe and 33.8 ± 1.8 emu/g Fe, respectively.The raspberry SPIONs do exhibit a small increase in No significant difference to the formulation on day 1 was observed in any of the formulations studied over the time period, suggesting that the raspberry SPIONs are colloidally stable in aqueous conditions.Data are means ± SD, n = 3 (p > 0.05, one-way ANOVA, Games-Howell post hoc test).(B) Raspberry SPION colloidal stability was investigated in various biomedically relevant liquids at 37 • C (H 2 O, PBS, 5% w/v dextrose, 0.9% w/v NaCl).The iron concentration was monitored using the 1,10-iron phenanthroline assay.No significant differences in iron content, over the time period, were observed in any of the media tested suggesting that the raspberry SPIONs were stable in these media over the time period.Data are means ± SD, n = 3 (p > 0.05, one-way ANOVA, Tukey post hoc test).(C) Raspberry SPION stability was investigated in 50% rat plasma at 37 • C. The iron concentration was monitored using ICP-AES.A statistically significant change in iron content was observed over the time period indicating a slight nanoparticle instability in plasma; however, colloidal stability was satisfactory as 70% of the iron content was retained over 24 h.Results suggest that while the raspberry SPIONs change slightly on incubation in plasma, their stability is satisfactory for their imaging application.Data are means ± SD; n = 3 (p < 0.05, one-way ANOVA, Tukey post hoc test).magnetic saturation, when compared with the oleic acid coated SPIONs; however, this result was not statistically significant (p-value > 0.05), suggesting that encapsulation within the polymer nanoparticles has little effect on the magnetic behavior of the SPIONs.
To investigate the MRI contrast behavior of the raspberry SPIONs, their spin-lattice relaxation (r 1 ) and their spin-spin relaxation (r 2 ) were measured as a function of iron concentration using a 1T bench-top MRI scanner (ICON; Figure 3E).The raspberry SPIONs produced an r 1 value of 0.66 mM -1 s -1 and an r 2 value of 51.8 mM -1 s -1 , giving an r 2 /r 1 of 79.1.This high r 2 /r 1 value for SPIONs with a core size of 5 nm, indicates that a highly effective negative contrast agent has potentially been created [44,45].M versus H plot at 300 • K comparing the magnetic iron content of the raspberry SPIONs and the oleic acid coated SPIONs.Samples of oleic acid coated SPIONs and raspberry SPIONs were measured for iron content and then plotted against their magnetization profile.This yielded a magnetic saturation value of (29.6 ± 3.7) emu/g Fe and (33.8 ± 1.8) emu/g Fe for the oleic acid coated SPIONs and raspberry SPIONs, respectively.There is no significant change in magnetization between the two curves, showing that the clustering of SPIONs has little to no effect on the SPION's magnetization.Data are means ± SD n = 3 (p < 0.05, Student's t-test); (E) Plot of 1/T 1 (s -1 ) and 1/T 2 (s -1 ) against the concentration of Fe in raspberry SPIONs; (F) Plot of 1/T 1 (s -1 ) and 1/T 2 (s -1 ) against the concentration of Fe in dimercaptosuccinic acid SPIONs (mM).The relaxivities (r 1 and r 2 ) can be determined from the slope of the plots.SPION: Superparamagnetic iron oxide nanoparticle.
To investigate the effect of clustering the SPIONs within polymer nanoparticles, the relaxivity of a single crystal (nonclustered) SPION dispersion was measured.The single crystal SPION was achieved by coating the surface of the oleic acid coated SPIONs with DMSA using the ligand exchange procedure outlined by Palma et al. [34].The DMSA SPIONs were colloidally stable in aqueous conditions and had an average hydrodynamic diameter of 9 nm (Supplementary Figure 6).The relaxivity values of the DMSA SPIONs were r 1 = 1.33 μM -1 s -1 and r 2 = 3.99 μM -1 s -1 , giving an r 2 /r 1 value of 3.0 (Figure 3F).This large reduction in the r 2 /r 1 value suggests that the DMSA SPIONs are a much weaker negative contrast agent and possibly a more favorable positive contrast agent [33,45].The data suggest that encapsulating the SPIONs within GCPQ nanoparticles, reduces their spin-lattice relaxation and increases their spin-spin relaxation.

Nanoparticle in vivo pharmacokinetics
To determine the in vivo biodistribution of the raspberry SPIONs, the raspberry SPIONs were injected intravenously (32.5 mg Fe/kg), tissues (blood, liver, spleen, brain, lungs, kidney and heart) were collected at various time points and the iron content was measured using ICP-AES.A relatively large dose was administered in order to distinguish SPION iron levels from the endogenous iron levels found in the blood and organs.Blood samples showed a sharp increase in iron concentration over background levels and by the first 15 min time point, 61% of the injected dose was present in the blood (Figure 4A).After this point, the iron concentration continued to decay and by the 1-h time point 43% of the injected dose was present in the blood and iron concentration in the blood had reached background levels by the 8-h time point.The raspberry SPIONs have a blood half-life (t 1/2 ) of 28.3 min.Our data show that the raspberry SPIONs accumulate in the liver and spleen (Figure 4B & C).By the first 15-min time point, 23% of the injected dose is present in the liver and 11% of the injected dose is present in the spleen.By the 1-h time point, this increases to 41% in the liver and 15% in the spleen.There was no increase in iron concentration in the brain (Figure 4D), due presumably to the blood-brain barrier, a naturally occurring defence system which is able to exclude foreign particles from entering the brain [46].A considerable iron increase was detected in the lungs; however, this was cleared after 4 h.The heart and kidneys also exhibited an increase in iron concentration in the early time points (Figure 4F & G).No adverse effects were observed in any of the animals in this experiment, even, when the administered dose was several fold higher than that needed for MRI.

In vivo MRI studies
To evaluate the imaging potential of the raspberry SPIONs in vivo, MRI was conducted using a 1T bench-top MRI scanner (ICON).Raspberry SPIONs (5 mg Fe/kg) were injected into the tail vein of female BALB/c mice.A dose of 5 mg Fe/kg has been reported by others previously [47,48].Pre-treatment and 1 h after injection T2 maps were collected (Figure 5A & B).Noticeable hypointense contrast was observed in the liver parenchyma, to the extent that hepatic liver vessels became noticeably highlighted from the surrounding tissue, showing extravasation of the particles from the liver vasculature.The portal vein also was clearly visible in post contrast images.The benefit of having high levels of contrast in the liver region is that many of the surrounding organs became easy to identify, most notably the kidneys, which were often hard to distinguish from the liver in the pre-treatment images.Negative contrast was also observed in the spleen.The data (Figure 5) correlate well with the pharmacokinetics data (Figure 4), with most of the image darkening being observed in the liver and spleen (Supplementary Figure 8).To assess the raspberry SPIONs' potential clinical relevance, a commercially available SPION negative contrast agent (Ferucarbotran R ) was also tested (Figure 5C).As predicted, darkening of the liver was also observed, however many of the fine hepatic liver vessels observed with the raspberry SPIONs were not seen with Ferucarbotran.
To be able to quantify the effect that the raspberry SPIONs had on the T2 relaxivity of the various tissues, T2 relaxivity maps were collected and compared with experiments using no contrast and Ferucarbotran (Supplementary Figure 9).Maps were collected 1 h post administration of the contrast agent.The T2 relaxation was determined by curve fitting the echo times (TE) of each pixel in the image using MATLAB (Mathworks, UK).A summary of the T2 relaxation times of specific regions of interest (lungs, liver, kidneys and spleen) can be seen in Figure 6.The heart could not be mapped accurately as it gave large variations in the data collected.This was attributed to movement from the beating heart during the experiment and the data taking several seconds to acquire for each slice.Respiration was accounted for by gating the experimental procedure and the raspberry SPIONs had no significant effect on relaxation time in the lungs unlike Ferucarbotran, which did display a significant effect on relaxation time in the lungs.The same observation was made in the kidneys, with no change in relaxation times seen with the raspberry SPIONs.Both the raspberry SPIONs and Ferucarbotran produced significant effects Maps were collected 1 h post administration with the contrast agents.The heart was unable to be mapped most likely due to movement from the beating heart during the experiment.Respiration was accounted for by gating the experimental procedure and as a result, the data show that the raspberry SPIONs had no significant effect on T2 relaxation in the lungs unlike Ferucarbotran, which did display significant T2 relaxation reduction in the lung.The same observation was made in the kidneys.Both the raspberry SPIONs and Ferucarbotran produced significant signal reduction in the liver and spleen and to a similar extent.Data are means of pixels of interest ± SD (n > 30, significance = p < 0.05, 1-way ANOVA, Games-Howell post hoc test).
10.2217/nnm-2018-0370 Nanomedicine (Lond.)(Epub ahead of print) future science group on relaxivity in the liver and spleen and to a similar extent.Overall, these data show a difference between the raspberry SPIONs and the commercially available Ferucarbotran; at the 1-h time point both species may be used to give hypointense MRI contrast in healthy liver and spleen tissue; however, the raspberry SPIONs do not affect other organs such as the lungs and kidneys making them selective in their organ imaging.In particular, the unique properties of the raspberry SPION, give far superior resolution of the structural features of the liver, such as the liver vasculature, when compared with the commercial product.

Discussion
We present here raspberry SPIONs which have been prepared by encapsulating oleic acid coated SPIONs within a chitosan amphiphile: GCPQ (Figure 1).The driving force behind this clustering is thought to be due to the removal of the energetic penalty at the oleic acid-water interface, where the water molecules are at an entropy deficit due to their inability to hydrogen bond in all directions and thus is similar to the driving force behind most aqueous self-assemblies [49].The entropy gain of the water molecules, achieved once the hydrophobic oleic acid coated SPIONs are encapsulated, drives the self-assembly.The eventual result is that the hydrophilic face of the polymer is in contact with the aqueous disperse phase and the hydrophobic portions of the polymer and the oleic acid coated SPIONs are removed from the aqueous phase and buried within the nanoparticle.Others have reported clustering with polymer amphiphiles [50] and low molecular weight amphiphiles [51] and the novelty of the current work lies with the superior MRI contrast images obtained with the clustered SPIONs (Figure 5).The raspberry SPIONs are remarkably stable with no significant change in their colloidal properties for up to 90 days (Figure 2) and exhibited satisfactory stability in plasma, despite the slight loss of iron recorded over the 24-h period.It is known that a protein coating forms on nanoparticles when in contact with plasma [52] and this protein coating may be implicated in the slight drop in iron concentrations observed.Examining all of the stability data we conclude that the raspberry SPIONs are ideal for the imaging application for which they were intended [36,53] and we proceeded to test them.The raspberry SPIONs produced an r 1 value of 0.66 mM -1 s -1 and an r 2 value of 51.8 mM -1 s -1 , giving an r 2 /r 1 of 79.1.This high r 2 /r 1 value for SPIONs with a core size of 5 nm, indicates that a highly effective negative contrast agent has potentially been created [44,45].Others have reported higher r 2 values of 128 mM -1 s -1 on a 0.5 T instrument with SPIONs of 18 nm in diameter; SPIONs that are over three-times the diameter of the current SPIONs [54].The relatively large SPIONs reported by Yang and colleagues had r 1 values of 19 mM -1 s -1 , with a corresponding r 2 /r 1 value of 6.6, making them likely to be less efficient contrast agents than the current SPIONs.The relaxivity values of the nonclustered DMSA SPIONs were r 1 = 1.33 μM -1 s -1 and r 2 = 3.99 μM -1 s -1 , giving an r 2 /r 1 value of 3.0 (Figure 3F).This large reduction in the r 2 /r 1 value suggests that the DMSA SPIONs are a much weaker negative contrast agent [33,45].
The data suggest that encapsulating the SPIONs within GCPQ nanoparticles reduces their spin-lattice relaxation and increases their spin-spin relaxation.These observations can be explained as follows: particles encapsulated within the polymeric nanoparticle undergo the clustering effect, a well-documented phenomenon in which large clusters of packed SPIONs are much more effective at generating magnetic field inhomogeneities and therefore, increase the spin-spin relaxation of the neighboring water protons [26,55,56].The SPION's ability to affect the spin-lattice relaxation is a less well understood mechanism, although a compelling hypothesis is that encapsulation within the hydrophobic core of the polymer nanoparticle, inhibits the water molecule's dipolar interactions with the magnetic iron cores, a key interaction underpinning the mechanism of spin-lattice relaxation, with the overall result of reducing the spin-lattice relaxation [57,58].Our data show that encapsulation of SPIONs within polymeric nanoparticles enhances the SPIONs ability to act as potential negative contrast agents.
In order to evaluate these raspberry SPIONs as imaging agents, we examined their biodistribution (Figure 4).The raspberry SPIONs have a blood half-life (t 1/2 ) of 28.3 min.Literature findings report considerably varied SPION blood half-lives, typically ranging from several minutes to several hours [53].This high variation in SPION half-life is often attributed to the variation in size, charge and surface coating of each SPION system under examination.A comprehensive summary of SPION blood half-lives can be found in a review by Arami et al. [53].Our data show that the raspberry SPIONs accumulate in the liver and spleen (Figure 4B & C).Accumulation in the liver and spleen is a common elimination pathway with SPION systems and is most likely the result of the nanoparticles accumulating in the macrophage-rich tissues of the reticuloendothelial system [53,[59][60][61].Raspberry SPIONs were also delivered to the lungs; however this was cleared after 4 h.SPIONs have been found in the lungs in other reports and this is often attributed to the lung's large vasculature combined with the presence of macrophage-rich future group 10.2217/nnm-2018-0370 tissue in the alveoli (Figure 4E) [62,63].In this case, it appears as if the SPIONs present in the blood pool in the lungs are responsible for the lung iron levels seen.It is also possible that the limited lung tissue accumulation could have been due to the short plasma half-life of the SPIONs.The heart and kidneys also exhibited an increase in iron concentration in the early time points and this is probably also due to the raspberry SPIONs present in the blood pool (Figure 4F & G).It is likely that the iron found in the lung, heart and kidney is due to the blood pool in these organs as the decay kinetics mirror the blood decay kinetics.Ultimately the value of these raspberry SPIONs resides in their ability to be used for MRI imaging.From the findings presented here, this novel contrast agent could be used with shorter MRI scan times to image whole body pathological conditions whereas at longer scan times, it could be used for imaging organs of the reticuloendothelial system such as the liver and spleen.On intravenous administration of the raspberry SPIONs, noticeable hypointense contrast was observed in the liver parenchyma, to the extent that hepatic liver vessels became noticeably highlighted from the surrounding tissue, showing extravasation of the particles from the liver vasculature.The portal vein also was clearly visible in post contrast images.Negative contrast was also observed in the spleen.The data (Figure 5) correlate well with the pharmacokinetics data (Figure 4) and the T2 relaxation data (Figure 6), with most of the image darkening being observed in the liver and spleen (Supplementary Figure 8).The T2 relaxation data with the raspberry SPIONs was localized in the liver and spleen (Figure 6).Whilst with the commercially available SPION negative contrast agent (Ferucarbotran), T2 relaxation was seen in all the organs examined (Figure 6), the raspberry SPIONs target the image contrast to the liver and spleen, making them selective for imaging these organs.
The hepatic liver vessels were not visualised with Ferucarbotran (Figure 5C).The reason why these hepatic liver vessels are visualized with the raspberry SPIONs and not with the commercially available Ferucarbotran is hypothesized to be due predominantly to the surface chemistry and size differences of the nanoparticles under investigation.The raspberry SPIONs and Ferucarbotran have average hydrodynamic sizes of 190 and 50 nm, respectively.After injection into the blood, the smaller sized Ferucarbotran nanoparticles have the ability to pass from the blood to the nonphagocytic liver sinusoidal endothelial cells [64] that line the liver sinusoids; via clathrinmediated endocytosis or to pass through their many fenestrations (which allow bidirectional flow of solutes and small particles less than 100-150 nm from the blood to the liver parenchyma and vice versa).In contrast, the larger raspberry SPIONs are liable to be phagocytosed by resident liver macrophages located in the luminal side of the lining of the liver sinusoid taking the hypointense contrast away from the lumen of the blood vessels.In essence, clustering enabled an extravascular localization in the liver and a more intense MRI contrast in extravascular region, which enabled imaging of the vasculature.Accumulation of negatively charged 105 nm latex beads has been reported in the liver Kupffer cells by Tanimoto and colleagues [65].Considerably more darkening was noticed in the lungs when using Ferucarbotran than was observed with the raspberry nanoparticles (Supplementary Figure 8).The reason for this is that the raspberry SPIONs do not accumulate in the lung tissue (Figure 4E & Figure 6).
The unique properties of the raspberry SPIONs, give far superior resolution of the structural features of the liver, such as the liver vasculature, when compared with the commercial product.

Conclusion
In conclusion, highly crystalline 5 nm SPIONs were synthesized via high temperature thermal decomposition and clustered into raspberry shapes by a polymeric amphiphile using physical means and without the need for cross-linking.These raspberry SPIONs are colloidally stable in a range of biomedical liquids.MRI experiments demonstrate that this clustering of SPIONs greatly enhances the r 2 relaxivity and decreases the r 1 relaxivity compared with the single crystals alone, giving an effective negative contrast agent.The blood half-life of the raspberry SPIONs is 28.3 min in rodents and the raspberry SPIONs accumulate in the liver and spleen.Finally, in vivo MRI studies show that these new raspberry SPIONs act as effective MRI contrast agents in the liver and spleen, localizing in the liver extravascular space and allowing the unique visualization of the liver vasculature, including the portal vein, when compared with commercially available SPIONs.

Future perspective
It is predicted that the future of medicine will be heavily focused on more effective diagnostics and as such a diagnostic nanoparticle has been created that could be used to better understand liver disease pathology.It is possible that the knowledge gained in this study will inform the development of more efficient contrast agents.

Figure 2 .
Figure2.Superparamagnetic iron oxide nanoparticle stability.Raspberry superparamagnetic iron oxide nanoparticle (SPION) colloidal stability was investigated by DLS in water over 90 day at 5 • C. Polydispersity values were given in labels above each column.No significant difference to the formulation on day 1 was observed in any of the formulations studied over the time period, suggesting that the raspberry SPIONs are colloidally stable in aqueous conditions.Data are means ± SD, n = 3 (p > 0.05, one-way ANOVA, Games-Howell post hoc test).(B) Raspberry SPION colloidal stability was investigated in various biomedically relevant liquids at 37 • C (H 2 O, PBS, 5% w/v dextrose, 0.9% w/v NaCl).The iron concentration was monitored using the 1,10-iron phenanthroline assay.No significant differences in iron content, over the time period, were observed in any of the media tested suggesting that the raspberry SPIONs were stable in these media over the time period.Data are means ± SD, n = 3 (p > 0.05, one-way ANOVA, Tukey post hoc test).(C) Raspberry SPION stability was investigated in 50% rat plasma at 37 • C. The iron concentration was monitored using ICP-AES.A statistically significant change in iron content was observed over the time period indicating a slight nanoparticle instability in plasma; however, colloidal stability was satisfactory as 70% of the iron content was retained over 24 h.Results suggest that while the raspberry SPIONs change slightly on incubation in plasma, their stability is satisfactory for their imaging application.Data are means ± SD; n = 3 (p < 0.05, one-way ANOVA, Tukey post hoc test).

Figure 3 .
Figure 3. Characterisation of superparamagnetic iron oxide nanoparticles.(A) Zeta potential titration to study the effect of pH on raspberry SPION surface charge.Raspberry SPIONs have a positive (+24.4 ± 1.5 mV) zeta potential at neutral pH (pH 7.0).Data are means ± SD n = 3; (B) Dynamic light scattering titration to study the effect of pH on nanoparticle size.In strongly basic conditions (pH > 9), aggregation is observed with the raspberry SPIONs, a strong indicator of colloidal instability at alkaline pH.Data are means ± SD n = 3; (C) x-ray diffraction patterns of the (i) raspberry SPIONs and (ii) oleic acid coated SPIONs (5 nm).Both patterns indicate the presence of magnetic iron oxide.(D)M versus H plot at 300 • K comparing the magnetic iron content of the raspberry SPIONs and the oleic acid coated SPIONs.Samples of oleic acid coated SPIONs and raspberry SPIONs were measured for iron content and then plotted against their magnetization profile.This yielded a magnetic saturation value of (29.6 ± 3.7) emu/g Fe and (33.8 ± 1.8) emu/g Fe for the oleic acid coated SPIONs and raspberry SPIONs, respectively.There is no significant change in magnetization between the two curves, showing that the clustering of SPIONs has little to no effect on the SPION's magnetization.Data are means ± SD n = 3 (p < 0.05, Student's t-test); (E) Plot of 1/T 1 (s -1 ) and 1/T 2 (s -1 ) against the concentration of Fe in raspberry SPIONs; (F) Plot of 1/T 1 (s -1 ) and 1/T 2 (s -1 ) against the concentration of Fe in dimercaptosuccinic acid SPIONs (mM).The relaxivities (r 1 and r 2 ) can be determined from the slope of the plots.SPION: Superparamagnetic iron oxide nanoparticle.