Daniela
Maggioni
*ae,
Paolo
Arosio
be,
Francesco
Orsini
be,
Anna M.
Ferretti
c,
Tomas
Orlando
de,
Amedea
Manfredi
a,
Elisabetta
Ranucci
ae,
Paolo
Ferruti
ae,
Giuseppe
D'Alfonso
ae and
Alessandro
Lascialfari
*be
aDipartimento di Chimica, Università degli Studi di Milano, Via Golgi 19, 20133, Milano, Italy. E-mail: daniela.maggioni@unimi.it; Fax: +39-02-50314305
bDipartimento di Fisica, Università degli Studi di Milano, Via Celoria 16, 20133, Milano, Italy. E-mail: alessandro.lascialfari@unimi.it
cLaboratorio di Nanotecnologie, CNR-Istituto di Scienze e Tecnologie Molecolari, via G. Fantoli 16/15 I, 20138, Milano, Italy
dDipartimento di Fisica, Università degli Studi di Pavia and CNISM, Via Bassi 6, 27100, Pavia, Italy
eUdR Milano Consorzio INSTM, via G. Giusti 9, 50121 Firenze, Italy
First published on 16th October 2013
Three-component nanocomposites, constituted by a superparamagnetic iron oxide core coated with a polymeric surfactant bearing tightly bound Re(CO)3 moieties, were prepared and fully characterized. The water soluble and biocompatible surfactant was a linear poly(amidoamine) copolymer (PAA), containing cysteamine pendants in the minority part (ISA23SH), able to coordinate Re(CO)3 fragments. For the synthesis of the nanocomposites two methods were compared, involving either (i) peptization of bare magnetite nanoparticles by interaction with the preformed ISA23SH-Re(CO)3 complex, or (ii) “one-pot” synthesis of iron oxide nanoparticles in the presence of the ISA23SH copolymer, followed by complexation of Re to the SPIO@ISA23SH nanocomposite. Full characterization by TEM, DLS, TGA, SQUID, and relaxometry showed that the second method gave better results. The magnetic cores had a roundish shape, with low dispersion (mean diameter ca. 6 nm) and a tendency to form larger aggregates (detected both by TEM and DLS), arising from multiple interactions of the polymeric coils. Aggregation did not affect the stability of the nano-suspension, found to be stable for many months without precipitate formation. The SPIO@PAA-Re nanoparticles (NPs) showed superparamagnetic behaviour and nuclear relaxivities similar or superior to commercial MRI contrast agents (CAs), which make them promising as MRI “negative” CAs. The possibility to encapsulate 186/188Re isotopes (γ and β emitters) gives these novel NPs the potential to behave as bimodal nanostructures devoted to theranostic applications.
Despite the wide effort of the scientific community, up to now no effective synthetic method for biocompatible NPs has been shown to optimize all the above microscopic requirements for obtaining the highest relaxivity, keeping at the same time additional functionalities suitable for e.g. γ- or β-radioemission (to be used in diagnostic SPECT and in radiotherapy, respectively), luminescence, drug delivery or MFH.1–24
The most widely used method for the synthesis of iron oxide based SPIO NPs consists of the co-precipitation of Fe3+ and Fe2+ salts in basic aqueous solution, a major drawback of this synthesis being the poorly achievable monodispersity. For this reason many authors in the last decade investigated alternative synthetic strategies,30–32 which, however, in many cases require extensive post-synthesis treatments to impart hydrophilicity to the NPs.
Actually, it should also be taken into account that for any biomedical application, the nanoparticles should possess high water solubility, biocompatibility and stability at physiological pH. Water soluble polymers, such as dextran and its derivatives, starch or polyethylene glycol, have been largely employed for SPIO stabilization.3,25,33 The use of polymer surfactants directly in the coprecipitation step has been shown to prevent aggregation and achieve a better control of the NP size, as was recently reported.34 On the other hand, poly(amidoamine)s (PAAs) are a family of water soluble polymers which possess many favourable properties for biomedical uses:35 they are biodegradable and biocompatible. Moreover, some PAAs, such as the amphoteric but predominantly anionic one named ISA23, when injected in the bloodstream, exhibit a “stealth-like” behaviour; that is, they circulate for a long time without preferentially localizing in the liver and can be used as carriers for drugs or, if properly labeled, as diagnostic probes.36 The repeating unit of ISA23 contains a carboxyl group and, therefore, holds great potential for coating and clustering SPIO cores, since the carboxyl group is one of the most widely used binding units for surface-capping iron oxide NPs.37–42
In this work we report on the synthesis, by two different routes, and the full characterization of Fe3O4–ISA23SH-Re nanocomposites constituted by a magnetic iron oxide (Fe3O4) core, covered by a copolymer strictly related to ISA23 (namely ISA23SH, Chart 1a),43 which contains a minor fraction (ca. 10%) of repeating units bearing a cysteamine pendant, able to firmly bind to a “Re(CO)3” fragment44 (the resulting stable polymer complex is ISA23SH-Re, Chart 1b).
![]() | ||
Chart 1 Schematic drawing of (a) the ISA23SH10% copolymer, in which the majority fraction (0.9) is constituted by the same repeating unit of the ISA23 homopolymer; (b) the ISA23SH-Re complex, in which the sixth coordination position around Re (marked with X) is mainly occupied by a carboxylate group of the polymer backbone.44 |
The above SPIO@PAA-Re nanocomposite models a potential diagnostic radiopharmaceutical, since: (a) rhenium has two radioisotopes, 186Re and 188Re, which are both γ and β emitters, with lifetimes and energies compatible with applications in diagnosis (γ) and therapy (β)45 (the use of Re(CO)3 units as a rhenium source is advantageous, because the preparation of the starting material [Re(CO)3(H2O)3]+ marked with 188Re is well established);46–48 (b) the SPIO core allows the NPs to act as contrast agents for MRI. As a consequence, the SPIO@PAA-Re nanocomposites herein described are potential dual probes49 for theranostic applications. It is worth mentioning that in vitro and in vivo biological studies using ISA23SH-Re containing cold rhenium isotopes revealed neither hemolytic activity, nor cytotoxic effects on HeLa cells nor apparent toxic effects after injection in mice.44
The 1H NMR relaxometry characterization (NMR-dispersion profile) was performed at physiological temperature (37 °C) and room temperature by measuring the longitudinal and transverse nuclear relaxation times T1 and T2, in the frequency range 10 kHz–240 MHz. It should be noted that the measurements at room and physiological temperatures gave the same results within 10% (corresponding to viscosity change of water). The range of frequency ν has been chosen in order to cover the most widely used clinical fields, i.e. 0.2 T (ca. 8.5 MHz), 0.5 T (ca. 21 MHz) and 1.5 T (ca. 64 MHz), and to study the mechanisms that led to the nuclear relaxation, through the analysis of the curves of r1 and r2vs. ν (NMRD profiles). The NMR signal detection and generation was obtained with a Smartracer® Fast-Field-Cycling relaxometer (Stelar, Mede, Italy) in the range 10 kHz–10 MHz, with a Stelar Spinmaster spectrometer for 10 MHz–60 MHz and with a Tecmag Apollo spectrometer equipped with an Oxford TH9/88/15 superconductor, working in the range 100–240 MHz. In the second and the third case, standard radio frequency excitation sequences were used for T1 and T2 measurements, respectively saturation-recovery for T1, Carr Purcell Meiboom Gill (CPMG) for T2. To determine the efficiency of MRI contrast agents, we calculated the longitudinal (r1) and transverse (r2) nuclear relaxivities, defined as the increase of relaxation rates of the solvent induced by 1 mmol L−1 of iron:
ri = [(1/Ti)meas − (1/Ti)dia]/CFe i = 1,2 | (1) |
In vitro MRI experiments were performed on vials containing SPIO solutions at 8.5 MHz using an Artoscan Imager by Esaote SpA. The employed pulse sequence was a high resolution spin echo sequence with TR/TE/NEX = 500 ms/18 ms/2, matrix = 256 × 192, FOV = 180 × 180 for the T1-weighted image, and with TR/TE/NEX = 1000 ms/80 ms/2, matrix = 256 × 192, FOV = 180 × 180 for the T2-weighted image. Here TE is the echo time, TR the repetition time, NEX the number of averages and FOV the field of view.
The peptization of the precipitated magnetite nanoparticles was effective when performed at pH ca. 3, to favour electrostatic interactions between the COOH groups of the polymer and iron oxide NPs. Indeed, at this pH the surface charge of the NPs is highly positive,54 while the carboxyl groups of the polymer, which are strongly acidic, are deprotonated.55 The peptization process required heating and sonicating (60 °C, 2 h), and produced clear brown suspensions of the nanocomposite, labelled as SPIO@PAA-Re(1) in Scheme 1, which were purified by dialysis (3 days against water) and lyophilized. Thermogravimetric analysis (TGA, Fig. S2†) showed that ca. 20% of the weight was attributable to the iron oxide core (corresponding to ca. 14 wt% of Fe), in good agreement with the analysis by atomic absorption spectroscopy (13 wt% of Fe). This implies the presence of about 150 polymer coils (Mn = 1.4 × 104 Da) per nanoparticle (on the basis of the mean NP size indicated by TEM analysis, see below).
The light-brown lyophilized solid easily dissolved in water, affording suspensions stable over a period of months, without separation of the precipitate. The covering of the nanoparticles by the PAA was further confirmed by ζ potential measurements, which showed the isoelectric point at pH ca. 5, roughly corresponding to that of ISA23,55 but well below the value reported for bare magnetite NPs (pH 8).54 In general the trend of ζ potential values at variable pH (Fig. 1) was very similar to that measured for ISA23SH alone, which is known to spontaneously self-assemble in spherical nanosized aggregates (dHca. 10 nm).44 The ζ potential value at neutral pH was far enough from zero to prefigure stability of the suspensions under physiological conditions.
![]() | ||
Fig. 1
ζ Potential versus pH curves recorded for SPIO@PAA-Re(1) (■) and for the copolymer ISA23SH alone (![]() |
However, the physical characterization (see the next paragraph) showed that the magnetic properties were negatively affected by the relatively high polydispersity of the sample. Moreover, the time-consuming purification by dialysis of the SPIO@PAA-Re(1) nanocomposite was poorly compatible with the relatively short lifetimes of the radioactive Re isotopes (17 h 188Re, 88 h 186Re).
Therefore, an alternative synthetic approach was investigated which involved a different kind of two-step procedure with respect to method 1 (route 2 in Scheme 1). At first a SPIO@PAA nanocomposite (devoid of rhenium) was prepared, and then, after proper extensive purification, it was conjugated to rhenium by a fast reaction. The nanocomposite produced in the first step, labelled as SPIO@PAA(2) in Scheme 1, was obtained by a one-pot reaction involving NaOH addition to an acidic solution containing Fe(II), Fe(III) and ISA23SH at room temperature under vigorous mechanical stirring with no need for heating or sonicating. The resulting clear suspension showed high stability, even in the presence of the magnetic field, and favourable morphological and magnetic properties of the nanocomposites (see below). The presence of the polymer during the nanoparticle formation has therefore a beneficial effect on the morphology of the product, as previously reported using different stabilizing polymers.34
For the subsequent conjugation of rhenium, microwave irradiation was used, since it is known that this heating source can promote and accelerate many kinds of reactions.56 In spite of the high dilution, the low temperature and the very short reaction time employed (2 × 10−4 M with respect to Re, 55 °C, 0.5 h, respectively), a satisfactory amount of rhenium was captured by the nanoparticles, corresponding to about one-sixth of the cysteamine pendants of ISA23SH (on average 0.75 rhenium atom per polymer coil and then more than 100 rhenium atoms per NP, according to the ICP analysis of the rhenium content). Such a rhenium loading is adequate, since very low dosages are usually necessary for radioemitter drugs,45,57 whereas high quantities of SPIO NPs have to be administered to provide effective contrast.58 IR spectroscopy of these SPIO@PAA-Re(2) NPs confirmed the uptake of Re(CO)3 groups, showing ν(CO) absorption in the fraction retained after ultracentrifugation. The position of these bands (2019s, 1902vs, br cm−1, Fig. S3†) was different with respect to the nanocomposite SPIO@PAA-Re(1), suggesting the presence of a water molecule in the sixth coordination position of the Re(CO)3–cysteamine complex.44 Indeed, in this case the carboxylate groups of the polymer are supposed to be already interacting with the NP surface in the SPIO@PAA(2) starting material, and are then unable to coordinate to the metal.
The thermal profile by TGA analysis (Fig. S2†) was slightly different with respect to that of NPs obtained by peptization, but the amount of iron oxide was strictly comparable (23%). Moreover, no significant change after rhenium addition was observed.
It is worth pointing out that magnetite cores are produced by the coprecipitation method, but partial subsequent oxidation (in spite of the use of a nitrogen atmosphere for NP handling) might produce different iron oxide nanophases. Among these, maghemite NPs (γ-Fe2O3) have magnetic properties comparable with magnetite and can hardly be differentiated from magnetite NPs by diffraction experiments. The electron diffraction pattern reported below showed a pattern compatible with these two phases only, which throughout this paper will be referred to as iron oxide.
![]() | ||
Fig. 2 (a) TEM images of SPIO@PAA-Re(1) prepared by peptization (the white arrow shows the aggregates), (b) histogram of size distribution and (c) electron diffraction pattern. |
At variance with this, the TEM images of the SPIO@PAA(2) sample (Fig. 3a) revealed that the magnetic cores had a roundish shape with a low dispersion, mean diameter = 6.6 ± 1.8 nm and distribution median = 6.6 nm (Fig. 3b), which stress the sharper and more symmetric distribution of NP dimensions, with respect to SPIO@PAA-Re(1). It is worth noting the presence of small and relatively regular aggregates, having a length of 40–60 nm and a width of 10–30 nm, surrounded by a thin layer of PAA, well visible like a halo in TEM image (Fig. 4). This TEM information is qualitative, but well agrees with DLS data.59 Indeed the analysis of the dialyzed suspension of SPIO@PAA(2) showed a size distribution (numbers) centred at ca. 70 nm (Fig. S4†), which remained constant over several months. The analysis of the correlograms by a non-linear bi-exponential fitting estimated two populations centred at ca. 70 and 170 nm, with the first one being largely dominant (ratio 1000:
1), in line with the numbers distribution size analysis.
![]() | ||
Fig. 3 (a) TEM images of SPIO@PAA(2) prepared by the “one-pot” synthesis, (b) histogram of size distribution and (c) electron diffraction pattern. |
![]() | ||
Fig. 4 Magnification of a TEM image of SPIO@PAA(2) NPs. The black arrow indicates the polymer presence. |
The electron diffraction (ED) analysis (Fig. 2c and 3c) revealed the typical diffraction pattern of magnetite/maghemite crystals. The rings were easily assigned to the low index crystal planes. They were sharper for SPIO@PAA(2) than for SPIO@PAA-Re(1), because of the smaller size distribution, and were slightly blurry, because of the presence of PAA surrounding nanocrystals, as the TEM images pointed out.
The comparison of TEM images acquired before and after the complexation of fac-Re(CO)3 moieties on SPIO@PAA(2) (Fig. S5†) showed that shape, size and size distributions were conserved after microwave irradiation at 50 °C for 30 min for rhenium conjugation. Moreover, the powder diffraction pattern confirmed that the spinel crystal structure of iron oxide nanoparticles was preserved. The only significant effect was a modification in the aggregation of the nanocomposites, since after the complexion reaction, aggregates were no longer clearly detectable in TEM images. In agreement with this, DLS showed a decrease in the size, which after MW irradiation was centred at ca. 25 nm (according to distribution by numbers, Fig. S6†). However, this did not affect the magnetic properties (see the next paragraph).
The formation of aggregates, revealed both by DLS and TEM, likely results not only from non-bonding interactions between polymer coils coating different cores, but also from interactions of single polymer coils with the surfaces of different iron oxide NPs (each coil bearing dozens of carboxylate groups regularly distributed along its chain). Such aggregation does not compromise the stability of the nano-suspensions, since no precipitate formation was observed for very long times.
On the other hand, magnetic measurements on a sample of SPIO@PAA(2) confirmed the higher monodispersity of the magnetic core, with respect to SPIO@PAA-Re(1), as indicated by TEM micrographs. Indeed the ZFC–FC magnetization curves reported in Fig. 5c show that their divergence occurs at a temperature close to the blocking temperature TB (T = 50 K). In addition a lower TB is indicative of a smaller nanoparticle size as observed in TEM, while the increase of magnetization in the FC curve at low temperature is probably due to paramagnetic impurities. The field dependent magnetization curve for SPIO@PAA(2) confirms that the sample is superparamagnetic at room temperature with zero coercivity, whereas at 2 K an open hysteresis loop is observed, Fig. 5d, with a coercive field Hc = 270 Oe. The hysteresis curve is saturated at 4 T.
SPIO@PAA-Re(1) NPs were characterized by NMR relaxometry and their r2 relaxivity, compared to the negative commercial contrast agent Endorem®, is reported in Fig. 6a. r2(ν) Values were higher than commercial compounds in the whole frequency range and interestingly showed a continuous increase at high frequencies (ν ≥ 10 MHz). The longitudinal nuclear magnetic relaxation dispersion (NMRD) curve (r1vs. ν) obtained for the SPIO@PAA-Re(1) NPs (see Fig. S7†) displayed a typical behaviour for a superparamagnetic CA: it stays flat for low frequencies, and then reaches a maximum and finally decreases rapidly at higher frequencies, showing a profile qualitatively similar to Endorem®, except for a shift towards lower frequencies and an increased height, of the peak. In the case of SPIO@PAA(2) the NMRD longitudinal relaxivity curve (r1vs. ν), reported in Fig. 6b, follows the profile of Endorem®, except for the height of the peak. In general the shape of the r1(ν) NMRD curve can be explained through the known mechanisms of nuclear longitudinal relaxation in superparamagnetic particles.2,3,60,61 The mechanism dominating at low frequency is the Néel relaxation, induced by the reversal of the magnetic moment through the anisotropy energy barrier. The other mechanism, predominating at higher frequencies, is the Curie relaxation, ascribed to the progressive orientation of the magnetic moment with increasing the field. According to Langevin's law, this mechanism depends on the core size and magnetization of the nanoparticles and on the temperature. Curie's mechanism is responsible for the maximum of r1(ν) at higher frequencies and the frequency shift of the peak is directly explained by the change in the nanoparticles size. Therefore the different position of r1 maxima for SPIO@PAA-Re(1) and SPIO@PAA(2), see Fig. S7† and Fig. 6b, is correlated to the different NP sizes, as also evidenced by TEM and magnetic measurements. Measurements on SPIO@PAA-Re(2) NPs (Fig. 6b) at selected frequencies (in the range 1–42 MHz, of interest for the chemico-physical properties of the sample and clinical applications) demonstrated that the subsequent conjugation of rhenium on SPIO@PAA(2) does not affect the mechanisms of longitudinal nuclear relaxation in the system. In Fig. 6c the NMRD transverse relaxivity curve (r2vs. ν) of SPIO@PAA(2) is reported, together with the data for SPIO@PAA-Re(2). Values similar to Endorem were found in the low frequency region, while for ν > 10 MHz a clear enhancement of r2 values was observed. The transverse nuclear relaxation rate (1/T2) is partially described by an approximate heuristic model in the framework of the so-called motional averaging regime (MAR), in which the protons of water molecules surrounding the material are sensitive to the magnetic dipolar field created by the electronic magnetic moments.2,3 The model in the MAR regime (valid when the Redfield condition is verified, i.e. ΔωτD < 1) predicts that the transverse relaxation rate at high field is described by the equation: R2 = 1/T2 = (16/45)fτD(Δω)2, where f is the volume fraction occupied by the nanoparticles in the suspension, Δω = γμ0MV/3 is the angular frequency shift experienced by a proton at the equator of the particle, γ is the gyromagnetic factor of the proton, μ0 is the magnetic permeability of vacuum, MV is the saturation magnetization divided by the particle volume and τD = d2/4D is the translational diffusion time of the protons in the magnetic field inhomogeneities created by the nanoparticles (D being the water translational diffusion constant and d the particle diameter). Recently Gossuin, Sandre et al.62 proposed a universal scaling law to predict the efficiency of magnetic nanoparticles as T2-contrast agents revising the model in the MAR regime. The authors proposed to modify the r2 relaxivity, obtained within the MAR regime, by means of the intra-aggregate volume fraction of magnetic materials ϕintra to derive a corrected relaxivity r2′. In this way the authors state that it is possible to properly compare the obtained relaxation data of different kinds of nanoparticles “as if they were filling the same volume fraction of suspensions of single USPIO nanoparticles”. Our data at a fixed frequency (20 and 60 MHz, near clinical ones) follow this model which, on the other hand, cannot describe the whole transverse NMRD curve (r2vs. ν) or explain the relaxation mechanisms as a function of the applied magnetic field. Nevertheless being the transverse relaxivity r2 the fundamental parameter to test the MRI efficiency for a superparamagnetic material, the reported experimental data demonstrate that our nanoparticles act as promising negative contrast agents. The lowering of r2 values for SPIO@PAA(2) (30–40%) with respect to SPIO@PAA-Re(1) is probably due to the different core size and the different aggregation of the two samples.
To get a confirmation of the contrast enhancement ability of our SPIO@PAA(2) samples, MRI in vitro experiments have been performed on a phantom composed of four vials containing: a suspension of the SPIO@PAA(2) sample, Magnevist® (positive or T1 relaxing commercial CA), Endorem® (negative or T2 relaxing commercial CA) and ultrapure water. A standard high resolution spin echo sequence was used to verify the MRI efficiency of SPIO@PAA(2) both as T1 and T2 relaxing agents. In Fig. 7a the T1-weighted image is shown, where a mild ability of the sample to contrast the image as a positive CA is evident. The T2-weighted image, reported in Fig. 7b, is in good agreement with the measured r2 relaxivities at approximately 8 MHz, corresponding to our MRI tomograph operating field (0.2 T). As can be seen from the image, the solution containing SPIO@PAA(2) contrasts with the image at the same quality level of the commercial CA, confirming the efficacy of our material.
The synthesized hybrid organic–inorganic nanocomposites show superparamagnetic behaviour at room temperature. Remarkably, the samples have demonstrated the capacity to behave as MRI contrast agents similar or superior to a well-known commercial product, i.e. Endorem®, as demonstrated both by transverse relaxation measurements and in vitro MRI experiments. The presented nanocomposites have thus proven to be very interesting systems, since the capability to encapsulate rhenium might allow the NPs to behave as bifunctional imaging probes, useful for both MRI and SPECT, if hot 186/188Re isotopes were used. On the other hand, the potential of the nanocomposites to behave as theranostic agents must be noted, since Re-containing hybrid materials could be employed for radiotherapy, owing to the β-emission of the hot rhenium isotopes.45 This makes fundamental the achievement of selectivity in cancer tissues addressing. The spontaneous tendency of NPs to accumulate in the tumour tissues (by enhanced permeability and retention effects)63 and the magnetic transport64 could be exploited, but functionalization of the nanocomposites with actively targeting moieties would also be highly advisable.65 Further investigation in this direction is under way.
In the future, efforts will be devoted also to the preparation of SPIO NPs with a suitable size for hyperthermia treatments,22,66,67 thus enhancing the multifunctionality of the hybrid organic–inorganic nanocomposite. A further possible development towards multifunctional magnetic/luminescent/γ-emitter materials could be provided by using, for SPIO NPs stabilization, the recently reported ISA23-based copolymer bearing a phenanthroline pendant in its minority part, able to rapidly bind to Re(CO)3 fragments, affording luminescent polymeric complexes.68
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c3dt52377b |
This journal is © The Royal Society of Chemistry 2014 |