Functionalized TiO2 nanoparticles labelled with 225Ac for targeted alpha radionuclide therapy

The 225Ac radioisotope exhibits very attractive nuclear properties for application in radionuclide therapy. Unfortunately, the major challenge for radioconjugates labelled with 225Ac is that traditional chelating moieties are unable to sequester the radioactive daughters in the bioconjugate which is critical to minimize toxicity to healthy, non-targeted tissues. In the present work, we propose to apply TiO2 nanoparticles (NPs) as carrier for 225Ac and its decay products. The surface of TiO2 nanoparticles with 25 nm diameter was modified with Substance P (5-11), a peptide fragment which targets NK1 receptors on the glioma cells, through the silan-PEG-NHS linker. Nanoparticles functionalized with Substance P (5-11) were synthesized with high yield in a two-step procedure, and the products were characterized by transmission electron microscopy (TEM), dynamic light scattering (DLS) and thermogravimetric analysis (TGA). The obtained results show that one TiO2-bioconjugate nanoparticle contains in average 80 peptide molecules on its surface. The synthesized TiO2-PEG-SP(5-11) conjugates were labelled with 225Ac by ion-exchange reaction on hydroxyl (OH) functional groups on the TiO2 surface. The labelled bioconjugates almost quantitatively retain 225Ac in phosphate-buffered saline (PBS), physiological salt and cerebrospinal fluid (CSF) for up to 10 days. The leaching of 221Fr, a first decay daughter of 225Ac, in an amount of 30% was observed only in CSF after 10 days. The synthesized 225Ac-TiO2-PEG-SP(5-11) has shown high cytotoxic effect in vitro in T98G glioma cells; therefore, it is a promising new radioconjugate for targeted radionuclide therapy of brain tumours.


Introduction
Despite of the predominant role of β − -particle emitters in radionuclide therapy trials (Iagaru et al. 2010), targeted radiotherapy based on α-particles is a promising alternative because they are highly cytotoxic agents, which deposit the whole of their energy within a few cell diameters (50-100 μm). Cell survival studies have shown that α-particle-induced killing is independent of oxygenation state (Wulbrand et al. 2013) or cell-cycle during irradiation and that as few as 1-3 tracks across the nucleus may result in cell death (Macklis et al. 1988;Sgouros et al. 2010). Additionally, the short range of α-particles corresponds to the dimensions of tumour micrometastases, allowing for localized irradiation of target cells with limited toxicity to surrounding healthy cells.
Recently, it has been found that in leukaemia, breast and brain cancer small subpopulations of tumour cells are able to self-renew and also to reconstitute the heterogeneous tumour-cell population (Al-Hajj et al. 2003;Singh et al. 2004). These, so called stem cancer cells, may be involved in the widespread metastatic dissemination of cancer (Wichal 2006). These findings have led to the suggestion that failure in tumour treatment may be associated with the failure to eradicate cancer stem cells (Al-Ejeh et al. 2011), especially that it has been found they are not sensitive to chemotherapy agents and to external and internal radiotherapy including β − -emitters (Al-Hajj et al. 2004). Given the properties outlined above, tumour stem cells are ideal targets for targeted α-particle therapy (TAT) (Sgouros and Song 2008).
There are hundreds of α-particle emitting radionuclides, but only a few have properties suitable for developing therapeutic radiopharmaceuticals: 212 Bi (t 1/2 = 60 min), 213 Bi (t 1/2 = 46 min), 211 At (t 1/2 = 7.2 h), 212 Pb (t 1/2 = 10.6 h), 227 Th (t 1/2 = 18.7 day), 223 Ra (t 1/2 = 11.4 day) and 225 Ac (t 1/2 = 9.9 day). However, most of the α-emitting radionuclides with optimal half-life for medical application, have one or more unstable daughter nuclides, which often emit α-particles as well. In principle such a cascade of α-particles can be very efficient in cancer treatment, but reliable delivery to the diseased site can be difficult to achieve. In the case of 225 Ac, the most perspective radionuclide for TAT (Sattiraju et al. 2017;Zhu et al. 2017), the four consecutive decay daughters (i.e. 221 Fr, 217 At, 213 Bi and 209 Pb) receive recoil energies ranging from 105 to 160 keV as a direct result of the law of conservation of momentum. The recoil energy of firstly produced 221 Fr is around 100 keV which is 1000 times greater than the energy of a chemical bond, implying that the recoiling daughter will always break free from the macrocyclic or acyclic chelating agents (Schwartz et al. 2011). Thus, produced daughter radionuclides are released from complexes, may accumulate in healthy organs and cause high toxicity which is a limiting factor in application of these radionuclides in targeted radiotherapy.
The new alternative strategy to sequester 225 Ac and its daughters at targeted side is based on the use of nanocarriers. Such encapsulation is expected to prevent, or at least considerably reduce, the release of the recoiling daughters and by this to avoid undesired accumulation in healthy organs and tissues. Moreover, the use of nanocarriers may improve delivery of larger amounts of activity to the diseased site. There has been proposed several different types of nanostructures to retain 225 Ac as well as its daughter radionuclides. Single and multi-layer liposomes have been evaluated; however, experimental studies revealed that their retention properties were poor with less than 10% of 213 Bi retained (Sofou et al. 2004;Chang et al. 2008). Theoretical studies indicated that multi-walled polymerosomes could achieve around 80% of 213 Bi retention, but their diameter has to be more than 800 nm (Thijssen et al. 2012). It would be very difficult for such large particles to extravasate/permeate/penetrate out of vasculature in vivo and also they can be clogged in microcapillary vessels. Woodward et al. (2011) developed 3-5 nm diameter monazite (LaPO 4 ) nanoparticles as carriers for 225 Ac, but these nanoparticles only partially retained daughters. More than~50% of the 221 Fr and 213 Bi (from the decay of 221 Fr) were released from the nanoparticle lattice. Addition of two layers of LaPO 4 reduced leaching of 221 Fr to 20% (Rojas et al. 2015). Similar nanocarrier based on lanthanide phosphate nanoparticles additionally coated with gadolinium and gold layers was developed by McLaughlin et al. (2013). Although multilayered nanoparticles exhibited higher retention properties, their synthesis becomes a multi-step and timeconsuming procedure.
In the present work, for sequestration of 225 Ac and its decay products, we propose to use titanium dioxide (TiO 2 ) nanoparticles (NPs), which are inorganic cation exchangers exhibiting high affinity for 225 Ac 3+ and also 211 Fr + and 213 Bi 3+ cations. TiO 2 NPs were conjugated with a fragment of substance P (5-11) for targeting NK1 receptors overexpressed in brain carcinomas. We examined the ability of TiO 2 NPs to retain daughter products of 225 Ac and investigated the cytotoxicity of the radiobioconjugate on human glioblastoma multiforme cells (T98G cells).

Instrumentation
The shape and diameter of titanium dioxide nanoparticles was determined by scanning electron microscopy (SEM, Zeiss) and transmission electron microscopy (TEM, LEO 912AB). The hydrodynamic diameter and zeta potential (ζ) were measured by dynamic light scattering (DLS, Malvern). The presence of surface coating of nanoparticles by PEG and substance P was determined by thermogravimetric analysis (TGA, Q500, TA Instruments). About 5 mg of dried powder was placed in a TGA furnace and the analysis was conducted from room temperature to 900°C at a heating rate of 10°C per min in the presence of inert (nitrogen) atmosphere.
Gamma-ray spectra of radionuclides were measured using coaxial HPGe detector (GX 1080) for gamma spectroscopy with multichannel analyzer DSA-1000 (Canberra Packard) with Genie 2000 software; detection range 10-5000 keV. High-performance liquid chromatography (HPLC) was performed using the ELITE LaChrom (VWR-Hitachi) system with L-2310 pump coupled to L-2455 diode array detector, radiometric (GabiStar, Raytest) detector and L-2350 column oven. Hamilton column (10 μm, 10 × 250 mm) eluted at the flow rate of 2 mL/min was used for the analytical and semi-preparative chromatography. The gradient elution system consisted of deionized water (A) and acetonitrile (B), both solvents contained 0.1% (v/v) trifluoroacetic acid (TFA). The gradient conditions were as follows: 0 to 70% solvent B in 0-15 min, 70 to 95% solvent B in 15-20 min.

Conjugation of substance P (5-11) with PEG molecules
An equal molar amount of substance P (5-11) (SP(5-11)) w a s a d d e d t o m e t h o x y l s i l a n e a n d Nhydroxysuccinimide functionalized polyethylene glycol (silane-PEG-NHS) and dissolved in 250-300 μL DMF followed by 3 μL of TEA addition. Mixture was stirred for 72 h in an inert (nitrogen) gas atmosphere. After completion of the reaction (checked by HPLC), the solvent was removed under vacuum and the product (silan-PEG-SP(5-11)) was separated from unreacted substrates using above described HPLC system and conditions, and finally lyophilized.
Functionalization of TiO 2 surface with silane-PEG-SP (5-11) The TiO 2 NPs were functionalized with earlier synthesized conjugate silane-PEG-SP(5-11) according to the procedure described in literature (Hermanson 2008). The acetic acid was added to 50 mL of ethanol (96%) to adjust pH to 4.5-5.5. Then about 5 mg of silan-PEG-SP(5-11) conjugate was dissolved and solution was stirred for 5 min. Next, 10 mg of TiO 2 NPs was added and the mixture was sonicated for 10 min followed by 2 h stirring at room temperature (RT). At the end, obtained product was washed several times with ethanol to remove unreacted silane-PEG-SP(5-11) and dried at 100°C for 30 min.
Labelling of TiO 2 -silane-PEG-SP(5-11) with 225 Ac 225 Ac 3+ cations are adsorbed on the surface of TiO 2 NPs through ion-exchange reaction on hydroxyl groups. For a typical labelling,~1 mg of nanoparticles either bare TiO 2 NPs or TiO 2 -silane-PEG-SP(5-11) NPs was suspended in 2 mL of 225 AcCl 3 (~100 kBq). The mixture was sonicated in an ultrasonic bath for 10 min and gently shaken on a circular stirrer for 1 h. After that time labelled NPs were centrifuged at 13000 rpm for 10 min, separated from supernatant and washed several times with deionized water. Finally, the product was suspended in 1 mL phosphate buffer (PB) and its activity measured on a γ-spectrometer to determine the labelling yield.
Stability studies of labelled TiO 2 -silane-PEG-SP(5-11) NPs In order to study leaching of 225 Ac and its decay daughters, a portion of labelled TiO 2 -silane-PEG-SP(5-11) NPs was dialyzed against 20 mL of 0.1 M PBS or saline (0.9% NaCl) for 10 days. Every day a 1-mL aliquot was withdrawn and the percentage of liberated activity from each daughter radionuclide was determined by its characteristic γ-ray peak. The stability study of radiolabelled nanoparticles in cerebrospinal fluid (CSF) was conducted using a centrifugal method. A 10 μL sample of labelled TiO 2 -silane-PEG-SP(5-11) NPs was added to 200 μL of CSF, vortexed and incubated at 37°C for 10 days. Every day, the probes with added nanoparticles were centrifuged and 100 μL of CSF was tested for activity of 225 Ac and its decay daughters.

Cytotoxicity evaluation
The impact of TiO 2 nanoparticles conjugated with substance P(5-11) fragment and labelled with 225 Ac on metabolic activity and proliferation of T98G cells was measured with 3-(4,5-dimetyl-2-thiazolyl)-2,5diphenyl-2H-tetrazolium bromide (MTT, Sigma Aldrich) assay. The assay was performed as described by Lankoff et al. (2012). In brief, 1 day before the experiment, T98G cells were seeded in 96-well plates (TPP) at a density of 10 3 cells/well in 100 μL of culture medium. Cells were treated for 48, 72 and 96 h with increasing concentrations of non-radiolabelled TiO 2 nanoparticles (0.0625-1 mg/mL) or the nanoparticles labelled with 225 Ac (1.25-20 kBq/mL). At least three independent experiments in six replicate wells were conducted. After the described treatment, cell culture medium was removed and 100 μL of 3 mg/mL MTT solution was added to each well. After 3 h incubation at 37°C the MTT solution was removed. Remaining insoluble formazan crystals were dissolved in 100 μL dimethylsulfoxide (DMSO, Sigma Aldrich) and absorbance of the solution was measured at 570 nm in plate reader spectrophotometer Infinite M200 (Tecan).

Statistical analysis
At least three independent experiments in six replicate wells were conducted for each toxicity point. Difference between samples and control were evaluated using GraphPad Prism 5.0 software (GraphPad Software Inc., USA). Toxicological data were evaluated by Kruskal-Wallis One Way Analysis of Variance on Ranks (ANOVA) followed by post hoc Dunnet's method. Differences were considered statistically significant when the p value was less than < 0.05.

Results and discussion
Currently 225 Ac is the most perspective radionuclide for targeted α-therapy in nuclear medicine. However, one significant factor limiting its use is the difficulty in sequestration of its decay products 221 Fr and 213 Bi at the targeted site. In our approach, we propose to use TiO 2 nanoparticles as carriers for 225 Ac and its decay daughters. Previous studies onTiO 2 NPs properties indicated their high sorption affinity for 3+ metal cations (Metwally and Rizk 2014) and heavy alkali metal cations (Filipowicz et al. 2014) and facile functionalization and modification of their surface through hydroxyl groups (Bogdan et al. 2017). Therefore, these reports provided a compelling rationale to evaluate TiO 2 NPs as potential carriers for stable retention of 225 Ac and its decay daughters.
The morphology, shape and diameter of bare TiO 2 NPs were characterized by SEM and TEM microscopy. The TiO 2 NPs exhibit nearly spherical morphology and their diameter ranges from 12 to 25 nm, as depicted on Fig. 1.
In order to synthesize a potential radiopharmaceutical exhibiting affinity to NK1 receptors overexpressed on glioma cells, attachment of an active biomolecule is required, in our case, an undecapeptide regulatory neuropeptide, substance P (SP). However, the native SP is metabolized in vivo due to enzymatic cleavages. The fragment SP(5-11) is the most abundant metabolite of native SP and it is still biologically active as the binding site with NK1 receptors is localized in the region of 7-11 amino acids. We have previously shown that this fragment radiolabelled with different radionuclides exhibit high affinity (IC 50 = 38 nM) and interaction with NK1 receptors on T98G cells (Piotrowska et al. 2017;Lyczko et al. 2017). Additionally, fragment SP(5-11) contains L-glutamine in position 5 what allows coupling of silane-PEG-NHS via NHS-active ester to form bioconjugate that further is used to functionalize the TiO 2 NPs surface. Figure 2 shows the procedure for TiO 2 NPs surface functionalization. In a first step, the silane-PEG-NHS linker was coupled to the SP(5-11) fragment and the obtained bioconjugate silane-PEG-SP(5-11) was attached to the surface of TiO 2 NPs by siloxane bond formation. Since the OH groups on TiO 2 surface are active sites for both 225 Ac adsorption and also for conjugation through silanization process, the number of silane-PEG-SP(5-11) molecules must be low enough to remain sufficient number of OH groups for adsorption of 225 Ac and its decay products.
TGA provided experimental evidence for the presence of silane-PEG-SP(5-11) on the TiO 2 NPs surface. Figure 3 shows comparison of TGA thermograms of bare TiO 2 and TiO 2 functionalized with silane-PEG-SP(5-11). Both thermograms indicate a loss of mass1 % in the range 25-100°C and additional 0.8% in the range 100-300°C, which can be attributed to desorption of physically adsorbed and intercrystaline water, respectively. The two additional mass losses can be noticed on the thermogram of functionalized TiO 2 . The first mass loss observed between 270 and 400°C (1.5%) is due to peptide molecule degradation and the second between 375 and 450°C (1%) is related to the silane bond degradation.
On the basis of obtained TGA results the number of silane-PEG-SP(5-11) coating molecules per one TiO 2 nanoparticle can be estimated. The calculation was performed under assumption that the nanoparticle is spherical with medium diameter of 20 nm, as measured by TEM, and that the density of material is 3.8 g cm −3 . The obtained results show that in average 80 silane-PEG-SP(5-11) molecules are bound to the surface of one TiO 2 nanoparticle. Taking into account that the ion-exchange capacity of TiO 2 nanoparticles is about 1 meq g −1 (Filipowicz et al. 2014), the number of available OH groups on one TiO 2 nanoparticle should exceed 3.8 × 10 4 . This indicates that even after functionalization of the TiO 2 nanoparticle surface with silane-PEG-SP(5-11) molecules, still the great majority of hydroxyl groups remain capable for binding of 225 Ac 3+ and formed in the decay process 221 Fr + and 213 Bi 3+ cations.
Bare TiO 2 NPs and TiO 2 -silane-PEG-SP(5-11) were also characterized by dynamic light scattering (DLS). The obtained values of hydrodynamic diameters and zeta (ζ) potentials are presented in Table 1. Due to the hydration layer formed on bare nanoparticles, the hydrodynamic diameter measured by DLS is usually greater than that measured by TEM. The observed increase of the hydrodynamic diameter for TiO 2 NPs after functionalization process is another confirmation that silanization reaction proceeded and silane-PEG-SP(5-11) molecules were indeed coupled to the surface of TiO 2 NPs. The measured negative zeta potential value of − 28.5 mV for TiO 2 -silane-PEG-SP(5-11) indicates that particles repel each other and do not aggregate what was confirmed by monitoring changes in hydrodynamic diameter over many days.
Bare TiO 2 and TiO 2 -silane-PEG-SP(5-11) NPs were radiolabelled with 225 Ac (100 kBq) at pH = 6.0 with a high yield of 99.8 ± 2.1% (n = 13) and 98.2 ± 1.3% (n = 10), respectively. The obtained results indicate that there is no difference in labelling yield between both nanoparticles and that surface functionalization did not have any negative impact on the radiolabelling process. It was expected, as previous calculations confirmed that the number of attached silane-PEG-SP(5-11) molecules is significantly much lower than the number of available OH functional groups responsible for ion-exchange approach with 225 Ac.
However, during our studies, it was found that due to low acidic character of the OH functional groups, the radiolabelling process is pH dependent and in acidic environment (pH = 3) the yield decreases even to only 27% for bare TiO 2 NPs. Therefore, the optimal pH for radiolabelling reaction is 6.0 and it was used in further studies.
The stability of TiO 2 -silane-PEG-SP(5-11) NPs labelled with 225 Ac was examined in 0.02 M PBS (pH = 7.4), physiological salt (0.9% NaCl) and cerebrospinal fluid (CSF) for 4 days ( Table 2). The leaching of mother radionuclide 225 Ac and its first decay product 221 Fr was measured on γ-spectrometer. In PBS and physiological salt NPs retained more than 95% of 225 Ac and 221 Fr over 10 days. In CSF, the fraction of 225 Ac released from NPs was still less than 5%, but the retention of 221 Fr was decreasing for the first 2 days and later stabilized and remained constant up to 4 days. The high stability in CSF is very important, as treatment of glioblastoma multiforme brain tumour is mostly done through intratumoral or intracavitary injection of radiopharmaceutical, which than can readily penetrate brain parenchyma and target widely disseminated cancerous cells.
The obtained retention results might be interpreted on the base of recoil energy in the 225 Ac → 221 Fr reaction and subsequent α decays. Calculated, using the SRIM program (SRIM 2010), recoil range of 221 Fr in water exceeds 84 nm, but in the material with density close to TiO 2 -anatase shows only about 44 nm. When the decay series is taken into consideration, we can assume that the radionuclides 217 At, 213 Bi and 209 Pb (daughters of 221 Fr) receive similar recoil energy. The escape of the radioactive isotopes from recoiled daughter and targeting agents allows them to freely migrate in the body, causing toxicity to healthy tissues and decreasing the therapeutic dose delivered to the diseased site. The renal toxicity induced by 213 Bi is considered to be the major constraint to apply 225 Ac in a large number of clinical trials (Jaggi et al. 2005;Schwartz et al. 2011). Based on the literature data, one would expect that higher amount of 221 Fr and 213 Bi nuclides would be ejected from a 20nm-diameter TiO 2 nanoparticle than observed in our studies. The observed smaller release of the daughter radionuclide 221 Fr (< 5% in PBS and saline solution) suggests that a fraction of the recoil energy takes part in translation of the whole NP rather than being entirely dissipated through atomic displacements resulting, as it occurs in the bulk, in shorter recoil range. Additionally, in the case of TiO 2 which is an effective cation exchanger, we observe subsequent reloading of escaped recoils 221 Fr + and 213 Bi 3+ as it contains hydroxyl surface groups with high affinity for these cations (Filipowicz et al. 2014;Bogdan et al. 2017;Perekhozheva et al. 1985;Bilewicz et al. 1991). This process can also explain slightly lower stability of radiolabelled NPs in CSF, as in this medium hydroxyl surface groups of NPs are blocked by the proteins and peptides present in CSF what prevents rebinding of released daughter radionuclides.
Cytotoxicity studies were performed on human glioma T98G cells with NK1 receptor overexpression. Cells were exposed to different radioactivities of 225 Ac-silane-TiO 2 -PEG and 225 Ac-silane-TiO 2 -PEG-SP(5-11) as well as different concentrations of non-radioactive TiO 2 -silane-PEG and TiO 2 -silane-PEG-SP(5-11) for up to 96 h. The effect of all studied compounds was assessed by the MTT assay. The metabolic activity of cells exposed to 225 Ac-silane-TiO 2 -PEG-SP(5-11) decreased rapidly with increasing radioactive doses and incubation time (Fig. 4b). On the contrary, 225 Ac-silane-TiO 2 -PEG NPs control, without targeting biomolecule, reduced cells viability, although the degree was not so much significant. Thus, the obtained data clearly indicate that observed cytotoxicity was specific to NK1 receptors and directly correlated with its expression level on tumour cells. The non-radiolabelled TiO 2 -silane-PEG and TiO 2 -silane-PEG-SP(5-11) NPs were relatively non-toxic as decrease in metabolic activity was observed only for the highest concentration (1 mg/mL) (Fig. 5). This concentration reduced the metabolic activity of T98G cells after 96 h incubation to 75 and 73% of control value for TiO 2 -silane-PEG and TiO 2 -silane-PEG-SP(5-11) NPs, respectively. These results reveal that indeed 225 Ac attached to NPs is a source of toxicity. The cytotoxicity of 225 Ac-silane-TiO 2 -PEG-SP(5-11) is very high in comparison to other radiopharmaceuticals including conjugates radiolabelled with α-emitters like 213 Bi or 211 At (Gadbois et al. 1996;Dziawer et al. 2017), however, it is comparable with other radiobioconjugates of 223 Ra and 225 Ac (Borchardt et al. 2003;Piotrowska et al. 2017). The decay processes of 225 Ac and 223 Ra include four α and two β − emissions to a stable 209 Bi and 207 Pb daughters. Because a very large amount of energy (~28 MeV) is released during these decay processes, much smaller amount of radionuclides activity is required to produce the desired effect and 225 Ac and 223 Ra radionuclides demonstrate extreme cytotoxicity. Comparison of obtained cytotoxicity results of 225 Ac-silane-TiO 2 -PEG-SP(5-11) NPs with literature data for radiopharmaceuticals labelled with 213 Bi (Friesen et al. 2007) demonstrate that several logs less 225 Ac radioactivity is required to reach similar LD 50 , value, because of the longer half-life and multiple α-emissions.

Conclusion
We have shown that TiO 2 NPs functionalized with substance P (5-11) can be used to deliver and retain 225 Ac and its daughter radioisotopes at a target site; thereby reducing the absorbed dose to non-targeted organs. The TiO 2 NPs successfully retain a large fraction of the 225 Ac decay products without compromising the tumoricidal properties of the α radiation. Because of the reloading of the 225 Ac decay products the retention efficiencies of TiO 2 NPs are comparable to the radiolabelled LnPO 4 -Au core shell Fig. 4 Metabolic activity (MTT assay) of T98G cells treated with different radioactivities of 225 Ac-TiO 2 -silane-PEG (a) and 225 Ac-TiO 2 -silane-PEG-SP(5-11) (b) NPs for 48, 72 and 96 h. Data are expressed as percent of control and the mean ± SD from three independent experiments nanostructures; however, the synthesis process is much simpler and less time-consuming. Furthermore, it has been shown that 225 Ac-TiO 2 -silane-PEG-SP(5-11) exhibits satisfactory stability in CSF what is very important as it is intended for direct intratumoral or post-resection application. The intravenous injection of the 225 Ac-TiO 2 -silane-PEG-SP(5-11) is excluded due to the relatively large size and high hydrophilicity which prevents crossing the blood-brain barrier.