In Vitro and In Vivo Comparison of Selected Ga-68 and Zr-89 Labelled Siderophores

Purpose Some [68Ga]siderophores show promise in specific and sensitive imaging of infection. Here, we compare the in vitro and in vivo behaviour of selected Ga-68 and Zr-89 labelled siderophores. Procedures Radiolabelling was performed in HEPES or sodium acetate buffer systems. Radiochemical purity of labelled siderophores was determined using chromatography. Partition coefficients, in vitro stability and protein binding affinities were determined. Ex vivo biodistribution and animal imaging was studied in mice. Results Certain differences among studied siderophores were observed in labelling efficiency. Protein binding and stability tests showed highest stabilities and lowest protein binding affinities for Ga-68 and [89Zr]triacetylfusarinine C (TAFC). All studied Ga-68 and [89Zr]siderophores exhibited a similar biodistribution and pharmacokinetics in mice with the exception of [89Zr]ferrioxamine E (FOXE). Conclusions Zr-89 and [68Ga]siderophores showed analogous in vitro and in vivo behaviour. Tested [89Zr]siderophores could be applied for longitudinal positron emission tomography (PET) studies of fungal infections and especially TAFC for the development of novel bioconjugates. Electronic supplementary material The online version of this article (doi:10.1007/s11307-015-0897-6) contains supplementary material, which is available to authorized users.


Introduction
Siderophores are high affinity, iron-selective chelators produced by almost all bacteria, fungi and some plants [1,2]. They serve as carrier molecules transporting iron across the microbial cell membranes, thus representing an excellent source of iron for many microorganisms. Iron is an essential nutrient for all cells and plays an important role also in microbial virulence [3]. Under iron-restricted conditions, some microorganisms secrete siderophores into the environment where they collect iron from various sources and deliver it into the producing organism [4]. In some species, specialized siderophores are synthesized, which are not excreted and serve as a unique iron storage form. Siderophores and their analogues have found wide application in agriculture and in medicine [5]. In medicine, siderophores have shown potential in applications like selective drug delivery; treatment of diseases, e.g., thalassemia, malaria and sickle cell disease; cancer therapy and imaging cancer and infection [5,6].
Some Gallium-68 (Ga-68) labelled siderophores have demonstrated their use as specific agents for imaging Aspergillus infection [7]. Gallium (Ga 3 + ; ionic radius=62 pm) and iron (Fe 3+ ; ionic radius=64 pm) are structurally similar and have similar coordination properties. Ga-siderophore complexes are recognized by cellular transporters and receptors on bacterial cells allowing the complex to enter inside the microbial cell. Ga-68 is a radioactive isotope of gallium and has in recent years gained a lot of attention in the field of nuclear medicine for molecular imaging using positron emission tomography (PET) [8,9]. It is a short-lived positron emitter (half-life=67.7 min) that can be produced from a long shelf-life and cost-effective [ 68 Ge]/ [ 68 Ga] generator system. The relatively short half-life of Ga-68 can be a limitation especially in longitudinal studies. This deficit can be resolved using longer-lived positron-emitting radionuclides such as Cu-64, I-124, Y-86, Nb-90 or Zr-89 [10].
The interest in Zirconium-89 (Zr-89) has increased over the last years as it displays almost ideal properties allowing imaging of biological processes at late time-points after the tracer application [11]. Even though Zr-89 has comparably low positron abundance and due to the long half-life results in higher radiation dose, it allows long-term follow-up especially of slowly accumulating biomolecules such as antibodies, nanoparticles and other large biomolecules both for preclinical and clinical applications, thereby complementing Ga-68 with its limitations of a very short half-life. Zr-89 can be produced in a cyclotron in large quantities and at high radionuclidic purity (99.99 %) [12]. It decays with a half-life of 78.4 h via both positron emission (23 %) and electron capture (77 %). The relatively low positron energy of 396.9 keV results in high image resolution [13]. The preferred oxidation state of zirconium is +4 with an ionic radius of 72 pm, also similar to the ferric iron.
In our previous studies, we have shown that different siderophores can be labelled with Ga-68 [14]. We have also demonstrated that [ 68 Ga]triacetylfusarinine C (TAFC) and [ 68 Ga]ferrioxamine E (FOXE) are able to detect Aspergillus fumigatus infection in a rat infection model using PET imaging [7,15] and that [ 68 Ga]TAFC is highly specific to A. fumigatus in vitro [16]. Desferrioxamine B (FOXB), a hydroxamate siderophore, is the most commonly used bifunctional chelator for coordination of Zr-89 4+ . Several preclinical proof-of-principle studies have been conducted using FOXB to label antibodies with Zr-89, but the in vivo stability of this complex remains an issue [17]. Our aim in this study was to evaluate the potential of alternative siderophores for radiolabelling with Zr-89 and as imaging agents for Aspergillus infections. We also wanted to investigate any potential differences between the resulting Zr-89 complexes and the respective Ga-68 complexes; the differences in charge (Zr 4+ vs. Ga 3+ ) could impact complex stability and other properties such as lipophilicity. We have investigated this issue and report here the in vitro and in vivo properties of selected Zr-89 and Ga-68 labelled siderophores (Fig. 1).

Chemicals
All commercially available reagents were used as supplied with no further purification. Siderophores were purchased from EMC microcollections GmbH (Tubingen, Germany

In Vitro Characterization of Radiolabelled Siderophores
Partition Coefficient DeterminationRadiolabelled siderophore in 0.5 ml phosphate-buffered saline (PBS) pH=7.4 was added to 0.5 ml octanol and the mixture was vigorously vortexed for 15 min. The aqueous and organic solvents were separated by centrifugation and 50 μl aliquots of both layers were collected and measured in the automatic gamma counter (WIZARD 2 ; PerkinElmer, Waltham, USA). Log P values were calculated from obtained data (mean of n=6).

Animal Experiments
All animal experiments were conducted in accordance with regulations and guidelines of the Czech Animal Protection Act

Biodistribution in Balb/c Mice
Biodistibution of radiolabelled siderophores was studied in female Balb/c mice (8 weeks old). Ga-68 and Zr-89 labelled siderophores (1-2 MBq/mouse, corresponding to 0.1-0.2 μg of siderophore per mouse) were injected retro-orbitally (r.o.). Animals were sacrificed by cervical dislocation 30 and 90 min post-injection (p.i.). Organs and tissues (blood, spleen, pancreas, stomach, intestine, kidneys, liver, heart, lung, muscle and femur) were removed and weighed. The amount of radioactivity in the samples was measured in an automatic gamma counter. Results were expressed as percentage of injected dose per gram organ (%ID/g).
PET Imaging of Radiolabelled Siderophores in Balb/c Mice PET and computed tomography (CT) images were acquired with an Albira PET/SPECT/CT small animal imaging system (Bruker Biospin Corporation, Woodbridge, CT, USA). Female Balb/c mice were r.o. injected with radiolabelled siderophores in a dose of 5-10 MBq corresponding to 0.5-1 μg of siderophore per mouse. Animals were anaesthetized with isoflurane (FORANE, Abbott Laboratories, Abbott Park, IL, USA) (2 % flow rate) and positioned prone head first in the Albira system before the start of imaging. Static PET/CT imaging was carried out 5, 30 and 90 min p.i. for both Ga-68 and Zr-89 labelled siderophores. Animals injected with [ 89 Zr]siderophores were also imaged 4 h p.i. A 5-min PET scan (axial FOV 148 mm) was performed, followed by a CT scan (axial FOV 65 mm, 45 kVp, 400 μA, at 600 projections). Dynamic imaging was carried out immediately after the injection of radiolabelled siderophore for 90 min (5 min PET scan per frame). Scans were reconstructed with the Albira software (Bruker Biospin Corporation, Woodbridge, CT, USA) using the maximum likelihood expectation maximization (MLEM) and filtered backprojection (FBP) algorithms. After reconstruction, acquired data was viewed and analyzed with PMOD software (PMOD Technologies Ltd., Zurich, Switzerland). 3D images were obtained using VolView software (Kitware, Clifton Park, NY, USA). Animal imaging was performed for all studied siderophores except for ferrichrome A (FCHA).

Radiochemistry
The labelling conditions as well as labelling efficiency were slightly different for studied desferrisiderophores. Desferritriacetylfusarinine C (TAFC) and desferriferichrome A (FCHA) were labelled with Ga-68 in sodium acetate buffer at room temperature for 15 min with an RCP greater than 98 %. Radiolabelling of desferrioxamine E (FOXE) and desferrioxamine B (FOXB) with Ga-68 was performed in sodium acetate buffer at 80°C for 20 min with an RCP greater than 98 %. Zr-89 labelling of all studied siderophores using Na 2 CO 3 for neutralization followed by incubation with the siderophore in HEPES buffer (pH=7) required incubation times of more than 90 min at RT with the RCP greater than 98 %. Interestingly, direct incubation of [ 89 Zr]oxalate with the corresponding siderophore in sodium acetate buffer at 80°C for 20 min resulted in equivalent results with a radiochemical purity of not less than 90 % as determined by HPLC (Suppl Fig. 1) and ITLC-SG. This was considered as sufficient for this initial evaluation of [ 89 Zr]siderophores, and the methods using sodium acetate buffer for Ga-68 as well as for Zr-89 labelling of siderophores were used for further in vitro and in vivo experiments.

In Vitro Characterization
All studied siderophores showed hydrophilic properties with partition coefficients (log P) ranging from −

Ex Vivo Biodistribution in Balb/c Mice
Ga-68 and Zr-89 labelled TAFC, FCHA and FOXB showed virtually identical ex vivo biodistribution in Balb/c mice. All six compounds were rapidly excreted via the renal system and showed minimal retention in blood and the various organs. [ 68 Ga]FOXE revealed similar biodistribution with slightly higher radioactivity values in gastrointestinal tract compared to the other radiolabelled siderophores, whereas [ 89 Zr]FOXE showed significantly higher radioactivity accumulation in the intestines and liver. In all cases, [ 89 Zr]siderophores displayed slightly higher bone uptake compared to their Ga-68 labelled counterparts, but always lower than 1 %ID/g, whereas there was a trend of higher blood activity for Ga-68 vs. [ 89 Zr]compounds, both indicating the behaviour of the free radionuclides. Ex vivo biodistribution data are presented in Fig. 2.

Discussion
Until recently, the majority of preclinical and clinical research has focused on the development of PET radiopharmaceuticals using conventional PET radionuclides such as C-11, N-13, O-15 and F-18 [18]. This has changed with the increased availability and production of non-conventional positron-emitting radionuclides such as Cu-64, Ga-68, Y-86, Zr-89 and I-124. Recently, Ga-68 and Zr-89 have generated significant interest in the nuclear medicine community with a concomitant increase in the use of Ga-68 labelled peptides for PET imaging of neuroendocrine tumours and Zr-89 imunoPET [8,19]. The radiolabelling of biomolecules with Zr-89 is usually performed by the desferrioxamine B (FOXB) chelator, commercially available as Desferal. FOXB is a hexadentate siderophore containing three hydroxamate groups for chelating metals and a primary amine functional group for conjugation to a biomolecule. It is a chelating agent for several metal ions besides Zr-89 [20]. Zr-89 radiolabelling of biomolecules using the FOXB chelator has shown that the in vivo stability of this complex remains an issue. This instability has been observed in several preclinical studies with uptake of Zr-89 in bones [21,22]. While many investigators have focused on the modification of the linkage between the FOXB and the biomolecule [23,24], others have attempted to improve the chelator itself [25][26][27][28][29]. Despite these attempts, a novel high-stability Zr 4+ ligand which would minimize the uptake of liberated Zr 4+ in the bone and other non-targeted tissue is warranted.
In this study, we have investigated the in vitro and in vivo behaviour of selected hydroxamate siderophores labelled with Zr-89 to evaluate their potential to be used as novel Zr-89 chelators and to compare their in vitro and in vivo properties with their Ga-68 labelled counterparts. We note that studied siderophores are only hexadentate ligands and cannot coordinatively saturate Zr 4+ that can bind up to eight ligand atoms. Therefore, it is to be expected that the additional positive charge of Zr 4+ will have a pronounced effect on chemical and biological properties of small chelators such as siderophores. However, the complexation of Zr 4+ is affected not only by number of coordinating atoms but also by linear vs. cyclic character, optimal ring size and appropriate geometry of the ligand [26].
Triacetylfusarinine C (TAFC) and ferrioxamine E (FOXE) are macrocyclic hydroxamate siderophores with the ring size of 35 (TAFC) and 33 (FOXE) atoms, which is deemed to be in the appropriate range for Zr 4+ chelation to preserve the optimal spatial orientation [25]. These siderophores were chosen due to their successful application as Ga-68 labelled tracers in the imaging of Aspergillus infection in rats [7]. We aimed to further test the ability of Zr-89 to label these promising  All siderophores that were tested could be labelled to near-completeness with both Ga-68 and Zr-89 using established labelling protocols. Both Ga-68 and Zr-89 labelled siderophores showed hydrophilic properties with [ 89 Zr]siderophores having a tendency to be slightly more lipophilic. This is rather surprising, as the binding of Ga 3+ results in a neutral molecule, whereas Zr 4+ introduces an additional positive charge. Significant differences among studied siderophores were observed mainly in protein binding and in vitro stability. injected into Balb/c mice displayed similar pharmacokinetics and minimal accumulation of radioactivity in blood and other organs and tissues with the exception of [ 89 Zr]FOXE, which caused significant retention of radioactivity in the gastrointestinal tract. In all cases, [ 89 Zr]siderophores displayed slightly higher (90 min p.i.), but still very low, retention of radioactivity in bones, indicating favourable in vivo stability in the studied time intervals. The differences in the behaviour of Ga-68 and the corresponding [ 89 Zr]siderophores could be explained by the different charge of the two radio-metal ions, resulting in a positively charged complex for hydroxamate coordinations for Zr in contrast to a neutral overall charge in the case of Ga. Even though there was no obvious instability of Zr-89 compounds, further investigations are required especially in the view of a higher stability of Zr-complexes by using octacoordinating ligands [25,28].
The overall comparable properties of Ga-68 and Zr-89 labelled siderophores (despite the differences of Ga and Zr in terms of charge, coordination chemistry and ionic radius) warrant further investigation of the targeting behaviour of Zr-89 labelled siderophores. This is of high interest in the application of the radiolabelled siderophores in the imaging of infection, in particular, infection by invasive fungal pathogens. The preclinical monitoring of therapy could benefit from a radionuclide that has a longer half-life, as it would enable longer imaging times (from minutes to days). However, there is no data available so far that proves that [ 89 Zr]siderophores such as TAFC are recognized by the specific transporters of microorganisms in the same way as their Ga-68 labelled counterparts [7,15]. Further studies in this respect are ongoing. The other application could lie in the development of novel, more stable [ 89 Zr]bioconjugates, coupling siderophores to, e.g., receptor of antigen targeting sequences. Here, the difference of Zr-89 vs. Ga-68 labelling will not result in a difference in targeting properties, but may result in different pharmacokinetics due to changes in charge and resulting lipophilicity. In this regard, preliminary studies of fusarinine C (FSC)-based bioconjugates that have been described recently for Ga-68 labelling [30] and Zr-89 labelling [31] showed excellent in vivo stabilities and tumour targeting.

Conclusion
We have shown in this study that a variety of siderophores can be labelled with Ga-68 and Zr-89. Ga-68 and [ 89 Zr]siderophores displayed analogous, but not fully identical in vitro and in vivo behaviour. Ga-68 3+ seems to be the more suitable radionuclide for labelling of hexadentate hydroxamate siderophores. Nevertheless, [ 89 Zr]TAFC showed favourable properties for potential longitudinal Aspergillus infection imaging using the animal model from our previous studies, and also availability as a model structure for preparation of novel bioconjugates.