Targeted Auger electron-emitter therapy: Radiochemical approaches for thallium-201 radiopharmaceuticals

Introduction Thallium-201 is a radionuclide that has previously been used clinically for myocardial perfusion scintigraphy. Although in this role it has now been largely replaced by technetium-99 m radiopharmaceuticals, thallium-201 remains attractive in the context of molecular radionuclide therapy for cancer micrometastases or single circulating tumour cells. This is due to its Auger electron (AE) emissions, which are amongst the highest in total energy and number per decay for AE-emitters. Currently, chemical platforms to achieve this potential through developing thallium-201-labelled targeted radiopharmaceuticals are not available. Here, we describe convenient methods to oxidise [201Tl]Tl(I) to chelatable [201Tl]Tl(III) and identify challenges in stable chelation of thallium to support future synthesis of effective [201Tl]-labelled radiopharmaceuticals. Methods A plasmid pBR322 assay was carried out to determine the DNA damaging properties of [201Tl]Tl(III). A range of oxidising agents (ozone, oxygen, hydrogen peroxide, chloramine-T, iodogen, iodobeads, trichloroisocyanuric acid) and conditions (acidity, temperature)were assessed using thin layer chromatography. Chelators EDTA, DTPA and DOTA were investigated for their [201Tl]Tl(III) radiolabelling efficacy and complex stability. Results Isolated plasmid studies demonstrated that [201Tl]Tl(III) can induce single and double-stranded DNA breaks. Iodo-beads, iodogen and trichloroisocyanuric acid enabled more than 95% conversion from [201Tl]Tl (I) to [201Tl]Tl(III) under conditions compatible with future biomolecule radiolabelling (mild pH, room temperature and post-oxidation removal of oxidising agent). Although chelation of [201Tl]Tl(III) was possible with EDTA, DTPA and DOTA, only radiolabeled DOTA showed good stability in serum. Conclusions Decay of [201Tl]Tl(III) in proximity to DNA causes DNA damage. Iodobeads provide a simple, mild method to convert thallium-201 from a 1+ to 3+ oxidation state and [201Tl]Tl(III) can be chelated by DOTA with moderate stability. Of the well-established chelators evaluated, DOTA is most promising for future molecular radionuclide therapy using thallium-201; nevertheless, a new generation of chelating agents offering resistance to reduction and dissociation of [201Tl]Tl(III) complexes is required.


Introduction
Since first being described by Lise Meitner and Pierre Auger in the 1920s, Auger electrons (AEs) have been investigated for use in molecular radionuclide therapy (MRT). AEs are a product of radionuclide decay, typically via electron capture or internal conversion, occurring in large numbers (4.7-36.9 per disintegration) and at low energies (<25 keV) [1]. However, this energy is deposited across a small distance (<0.5 μm), leading to higher linear energy transfer than radiotherapies involving, for example, beta particles with energies up to 2 MeV, where energy is deposited over 0.1-10 mm. For reference, alpha particle-emitters deposit their energy over 40-80 μm [2]. AE-emitters could therefore permit highly targeted therapies, capable of extreme radiotoxicity, even in micrometastases and single circulating tumour cells, but only if they can be delivered to certain targets such as the cell nucleus or membrane [3,4]. AE emissions accompany the decay of many radionuclides used in medical imaging, including 111 In, 67 Ga, 99m Tc, 64 Cu and 201 Tl, thereby allowing therapeutic radionuclides to be tracked to their biological target using single photon emission computed tomography (SPECT) or positron emission tomography (PET; in the case of 64 Cu) imaging.

Contents lists available at ScienceDirect
Nuclear Medicine and Biology sequences to amplify the effectiveness of AEs [7]. Recently, the therapeutic efficacy of an iodine-123-labelled poly(ADP-ribose) polymerase 1 (PARP1) inhibitor ([ 123 I]I-MAPi) in glioblastoma models was presented [15]. Generally, however, despite excellent preclinical results, translation of AE to the clinic has met with limited therapeutic impact due to their inability to deliver a lethal dose to the tumour. One example is [ 111 In]In-Octreotide, which so far has come closest to clinical translation [8]. Future AE-emitting MRT may be more successful if a more potent AE-emitting radionuclide was used that emitted many AEs per decay, such as thallium-201 or platinum-195 m. Thallium-201 (t 1/2 = 73 h, [ 201 Tl]Tl) has been used in medical imaging since the 1970s for myocardial perfusion SPECT imaging [16]. However, it has fallen out of favour since the development of technetium-99m-based agents, such as sestamibi and tetrofosmin, due to its long physical half-life (73 h) and consequent high absorbed radiation dose compared to technetium-99m (6 h) as well as the ready availability of technetium-99m from a generator. In 2005, it was demonstrated that clinical myocardial blood flow scans with thallium-201 led to genotoxicity in lymphocytes at day 3 after administration [17], highlighting the potential of healthy tissue toxicity from thallium-201. Similarly, intravenous injection of thallium-201 led to high testis uptake and toxicity in mice [18,19].
There are few published therapeutic studies involving thallium-201. Early studies in the 1980s highlighted toxicity of thallium-201 in V79 Chinese hamster lung fibroblasts [22]. Others have relied on in silico simulations. For example, Monte Carlo computational methods were used to accurately model the radiation dose from thallium-201 at target volumes of <1 μm in diameter by taking into account the contribution from AEs [23][24][25][26]. More recently, Geant4-DNA, another Monte Carlo simulation toolkit, demonstrated the theoretical number of single and double strand breaks that could be produced by AE-emitters on the DNA scale; thallium-201 was amongst the most effective in causing DNA damage [27].
Thallium-201 radiobiological studies have been compounded due to the difficulty of synthesising a thallium-201-labelled radiopharmaceutical. Whereas putative 201 Tl-labelled drugs like bleomycin and vancomycin have been assessed as imaging agents [28,29], a bifunctional chelator that forms a stable complex with thallium-201 still needs to be developed to accurately deliver the radionuclide to a tumour. It is expected that thallium-201 needs to be converted from its commercially available 1+ to a 3+ oxidation state, which is more amenable to complexation by multi-dentate ligands. However, oxidation methods suggested to date require harsh conditions (such as high temperature and concentrated acid) incompatible with biomolecules [30]. Moreover, reported stability studies with DTPA as chelator have been inconsistent or inconclusive [31,32]

Materials and methods
Unless stated otherwise, chemicals and solvents were purchased from commercial suppliers (Merck, Fisher Scientific, CheMatech).
[ 201 Tl]TlCl in saline was purchased from Curium Pharma, UK, and converted to [ 201 Tl]TlCl 3 by one of nine methods described below and summarised in Fig. 1  Ozone produced from medical grade oxygen via an ozone generator (1KNT-24 from Enaly, China) was bubbled through the radioactive solution via a glass pipette for 30 min.

Oxidation method 6chloramine-T
Chloramine-T (N-chlorotoluenesulfonamide; 0.1 mg -10 mg) in water was added to a minicentrifuge tube. [ 201 Tl]TlCl (5.2 MBq, 100 μL) was then added and the mixture was agitated for 10 min. Once dissolved, HCl (0.5 M, 100 μL) was added. A white solid precipitated from the solution. The solution was then agitated for 2 min, centrifuged for 30 s using a mini benchtop centrifuge to pellet the solid. The supernatant, containing [ 201 Tl]Tl(III), was then added to a clean flask. This was then used for the chelator studies.
A non-radioactive version of the reaction in method 6 was performed and the white solid precipitate was analysed using proton NMR. NMR spectra were recorded on a Bruker Ultrashield 400WB PLUS 9.4 T spectrometer ( 1 H NMR at 400 MHz). All chemical shifts were referenced to residual solvent peaks and are quoted in ppm. 1

Oxidation methods 8 and 9 -trichloroisocyanuric acid (TCCA) and iodogen
In direct comparative studies, 10 ng -0.1 mg iodogen and TCCA, both dissolved in chloroform and left in a fume hood overnight for the chloroform to evaporate, were added to clean reaction flasks. [ 201 Tl]TlCl (1 MBq, 25 μL) was then added to the pre-coated tubes, followed by HCl (0.1 M, 0.5 M or no acid added, 2.5 μL), vortexed and pipetted into a flask.

DNA damage assessment
For plasmid DNA incubated with 0.5 MBq [ 201 Tl]Tl(III), increasing the incubation time decreased the percentage of supercoiled DNA from 88 ± 1% to 51 ± 2% at 1 and 24 h, respectively (Figs. 2, S9). The presence of relaxed DNA increased from 12 ± 1% at 1 h to 49 ± 2% at 24 h, whereas linear DNA was first detectable (6.27 ± 0.15%) at 144 h. In all studies, negative controls consisting of the addition of PBS or non-radioactive [ nat Tl]Tl(III) to the plasmid did not show evidence of damage over the corresponding timeframe within the errors associated with the measurement (Fig. 2; p = 0.22).  3; Fig. 3B).
In order to develop bioconjugates of 201 Tl to evaluate their potential use in MRT, a platform for stable chelation of thallium must first be established. To date, none of the conventional radiometal chelators widely used in nuclear medicine has been adequately evaluated for chelation of radiothallium. Thallium is most stable under ambient conditions in oxidation states (I) and (III). In oxidation state (I), thallium is known to be strongly hydrated and behaves biologically much like the heavier alkali metals; for example, like potassium, it is a substrate for the sodium-potassium ATPase pump. Moreover, its electronic structure features a sterically active lone pair of electrons. With these properties, it is hard to conceive a likely kinetically stable thallium(I) chelate complex. On the other hand, thallium(III) is electronically analogous to indium(III) for which a range of highly stable chelates is known with well-established uses in nuclear medicine. Based on these considerations, thallium(III) would appear to be the more attractive option for developing a suitable chelation system. A prerequisite for developing such a platform is to find an efficient and convenient method to oxidise thallium(I) chloride, the form in which 201 Tl is manufactured and supplied, to thallium(III). Such a method would need to be sufficiently mild to be used in the context of labelling sensitive biomolecules. The Tl(I)/Tl(III) redox couple has a standard redox potential of +0.77 V, suggesting that unless the metal ion can be stabilised by a chelator, it may be reduced back to Tl(I).
Published methods to oxidise [ 201 Tl]Tl(I) to [ 201 Tl]Tl(III) included ozone, hydrogen peroxide, HCl or a combination of oxidising agents and high temperatures (95°C) [32,45,46]. In our hands, using iTLC-SG plates and acetone as an effective and reliable method to distinguish Tl (I) from Tl(III) [46], published oxidation methods (methods 1 -4 Table 2 Conversion yields of [ 201 Tl]Tl(I) to [ 201 Tl]Tl(III) using oxidation methods 1-9. Values are average ± standard deviation (n = 3). Also shown are characteristics of the nine oxidation methods in terms of simple set-up, ability to remove the oxidising reagent after the reaction, and whether the oxidation process is compatible with radiolabelling biomolecules such as antibodies. here) did not always prove successful when evaluated. For example, in method 2, the conversion yield was between 3 and 12%. Although oxidation method 1 was reproducible, heating at 95°C is not biocompatible ( Table 2). Oxidation methods 3-5 used ozone or oxygen as oxidants and avoided the need for high temperatures. Comparing conversion yields obtained from either method 2 or 3 showed the importance of hydrogen peroxide for the oxidation using ozone, although this appeared less important for oxygen (method 5), despite the decrease in oxidation potential from +2.07 V to +1.78 V [47]. Although the solution could be neutralised, this would dilute the radionuclide and increase the complexity of the labelling procedure. Equally, the practical set-up of bubbling oxygen through a reaction vessel using a large cylinder of compressed oxygen adds undesirable complexity and hazard (Table 2). Therefore, alternative, safer methods of oxidation were investigated. A range of biocompatible oxidising agents has been available for many years, developed for the purpose of radiolabelling biomolecules with radioiodine. Chloramine-T, first used by Greenwood et al. in the 1960s [48], is still popular in this field. In our experiments with oxidation of 201 Tl, we found that conversion yields for chloramine-T were 99%, even at the low amounts previously used to synthesise [ 123 I]diiodotyrosyl-salmon calcitonin (0.1 mg) [49]. Although chloramine-T is relatively biocompatible, it is known to cause protein damage in some cases [34], and its presence could lead to misleading stability results or damaged cells during in vitro uptake and stability studies described below. It should, therefore, ideally be quenched or removed from the reaction solution prior to introducing the biomolecules. This step is not always simple due to its water solubility. Chloramine-T also has an oxidation potential of +1.14 V under acidic conditions, so marginally lower than that of oxygen. [50] We therefore evaluated a range of solid-phase oxidants that could easily be removed after oxidation is complete. Iodo-beads (method 7), for example, consist of a chloramine-T analogue covalently bound to a solid polystyrene bead, allowing the supernatant containing [ 201 Tl]Tl (III) to be easily be removed from the vessel; this is also an advantage when using iodogen or TCCA (methods 8-9; Table 2). All three oxidants gave a good conversion yield (99%) from [ 201 Tl]Tl(I) to [ 201 Tl]Tl(III). An extra advantage of iodogen and TCCA over chloramine T is their solubility in organic solvents and low solubility in water; this enables precoated tubes to be created with the volatile solvent evaporating during the process. Additionally, TCCA has an oxidation potential of +4.84 V which is far higher than ozone and oxygen. [ (Fig. 4). As DTPA and EDTA are both acyclic chelators, with 6 and 8 donor atoms, respectively, this instability is likely due to low free energy barriers to conformational changes required to dissociate. The complexes are thermodynamically favourable and quick to form but not kinetically stable. These results conflict with claims in previous studies that [ 201 Tl]Tl(III)-DTPA-HIgG, is stable in human serum for more than 24 h [45]. [ 201 Tl]Tl(III)-DOTA, on the other hand required longer for the complex to initially form than [ 201 Tl]Tl(III)-EDTA and [ 201 Tl]Tl(III)-DTPA. The crystal structure of the complex showed the thallium ion directly coordinated to all eight donor atoms in a twisted square antiprismatic coordination and previous work has indicated that DOTA does indeed enable more stable chelation of [ 201 Tl]Tl(III) than DTPA, at least in vitro [32,52]. A crystal structure of [ nat Tl]Tl(III)-DOTA obtained by Fodor et al. shows the metal sitting above the plane of the cyclen ring [52]. As such, DOTA looks a more promising chelator of [ 201 Tl]Tl(III) for MRT than DTPA or EDTA, but it still not an ideal candidate, unless perhaps for a small targeting molecule with a fast biological half-life.

Conclusion
We have described simple, convenient and mild reactions, using iodo-beads, TCCA or iodogen, to convert DNA-damaging thallium-201 from Tl(I) to Tl(III), and evaluated a range of conventional chelators for their potential to serve as bifunctional chelators for thallium(III). EDTA and DTPA have inadequate stability for use in bioconjugates for MRT. DOTA shows greater kinetic stability which may suffice for some applications but will unlikely meet the need for a generally applicable thallium bifunctional chelator. This justifies further research into alternative chelators for [ 201 Tl]Tl(III).