Re-assessing gallium-67 as a therapeutic radionuclide☆

Introduction Despite its desirable half-life and low energy Auger electrons that travel further than for other radionuclides, 67Ga has been neglected as a therapeutic radionuclide. Here, 67Ga is compared with Auger electron emitter 111In as a potential therapeutic radionuclide. Methods Plasmid pBR322 studies allowed direct comparison between 67Ga and 111In (1 MBq) in causing DNA damage, including the effect of chelators (EDTA and DTPA) and the effects of a free radical scavenger (DMSO). The cytotoxicity of internalized (by means of delivery in the form of oxine complexes) and non-internalized 67Ga and 111In was measured in DU145 prostate cancer cells after a one-hour incubation using cell viability (trypan blue) and clonogenic studies. MDA-MB-231 and HCC1954 cells were also used. Results Plasmid DNA damage was caused by 67Ga and was comparable to that caused by 111In; it was reduced in the presence of EDTA, DTPA and DMSO. The A50 values (internalized activity of oxine complexes per cell required to kill 50% of cells) as determined by trypan blue staining was 1.0 Bq/cell for both 67Ga and 111In; the A50 values determined by clonogenic assay were 0.7 Bq/cell and 0.3 Bq/cell for 111In and 67Ga respectively. At the concentrations required to achieve these uptake levels, non-internalized 67Ga and 111In caused no cellular toxicity. Qualitatively similar results were found for MDA-MB-231 and HCC1954 cells. Conclusion 67Ga causes as much damage as 111In to plasmid DNA in solution and shows similar toxicity as 111In at equivalent internalized activity per cell. 67Ga therefore deserves further evaluation for radionuclide therapy. Advances in knowledge and implications for patient care The data presented here is at the basic level of science. If future in vivo and clinical studies are successful, 67Ga could become a useful radionuclide with little healthy tissue toxicity in the arsenal of weapons for treating cancer.


Introduction
Radiopharmaceutical therapies, such as 131 I-MIBG, anti-CD20 antibodies (labeled with 90 Y or 131 I), and 177 Lu-Octreotate, have become standard in the clinic. These beta particle-emitting treatments, however, are generally not curative and can cause toxicity to healthy tissue due to the long range (up to 1 cm for 90 Y) by high beta energies. Radioisotopes emitting Auger electrons with a much shorter range (b1 μm) are now being considered for targeted radionuclide therapy and could become useful tools in targeting micrometastases that play a detrimental role in tumor recurrence.
Gamma camera imaging with 67 Ga has been used regularly in the clinic since the 1980s to image lymphoma where it was useful in disease staging, monitoring disease progression and relapse, and predicting therapy response [1]. In Hodgkin's disease, the detection sensitivity is 70 to 83% [2]. In non-Hodgkin's lymphoma the detection sensitivity depends on cell differentiation status; less differentiated cells show higher avidity for gallium [1]. Other gallium-avid cancers include lung cancer, melanoma and multiple myeloma [1,2]. In these applications 67 Ga is administered as 67 Gacitrate and 67 Ga uptake by cells is believed to be transferrin-mediated [3], although there is also evidence for transferrin-independent mechanisms [4][5][6].
A feature of 67 Ga is that besides its gamma emissions for scintigraphy and SPECT imaging, it also emits Auger electrons [7] and thus has potential as a therapeutic radionuclide. As such it could form part of a "theranostic pair" with the generator-produced positron emitter 68 Ga. Despite producing fewer Auger electrons per decay (average of 4.7) than fellow Auger electron emitter 111 In (14.7), the average total Auger electron energy released per decay of 67 Ga (6.3 keV) is similar to that of 111 In (6.8 keV) [8]. In fact, amongst Auger electron emitters, 67 Ga produces amongst the most energetic (7 to 9 KeV) and longest ranging (up to 2.4 μm in water) Auger electrons [7]. This may reduce the need for the radionuclide to be localized in specific subcellular compartments in order to be effective. 67 Ga has been explored previously, to a limited extent, as a radionuclide for therapeutic applications [9][10][11][12][13]. In vitro results were promising and showed that treatment with 67

Contents lists available at ScienceDirect
Nuclear Medicine and Biology in human U937 lymphoma cells [9] and in myeloid leukemic blasts from acute myeloid leukemia patients [10]. A feasibility study in eight patients with relapsed acute leukemia was less successful due to the low cell uptake of 67 Ga-citrate [11], which might have arisen in part from using higher citrate concentrations in the clinical preparations than used in the in vitro work. Others have explored the therapeutic potential of 67 Ga in lymphoma when coupled to anti-CD74 antibodies [12,14] or anti-LL1 antibodies [13]. Michel et al. showed that 67 Ga was two to three times more potent than 111 In when coupled to anti-CD74 antibodies on the basis of equivalent total disintegrations in the medium per viable cell [12]. Low specific activities and lack of purposedesigned gallium chelators and conjugates at that time, however, led to a lack of further enquiry in this field.
In recent years the development of peptides and proteins labeled with 68 Ga, arising from the growing popularity of the 68 Ga generator for positron emission tomography applications, has led to a new generation of effective bifunctional chelators for gallium [15][16][17][18][19][20][21][22][23][24][25]. For example, the trishydroxypyridinone chelator allows radiolabelling of molecular targeting agents with gallium using fast and simple onestep procedures [15,20]. The resulting enhanced versatility and range of potentially useful targeting agents now presents an opportunity to reconsider 67 Ga as a targeted therapeutic radionuclide.
Here a comparison is made between 67 Ga and the wellcharacterized and clinically evaluated radionuclide 111 In, which has been successfully tested preclinically as a therapeutic radionuclide attached to cell surface or intracellular targets [26][27][28]. 67 Ga and 111 In have similar half-lives (78 and 67 h, respectively) and both produce gamma rays. We used the cell-free pBR322 plasmid assay to directly compare DNA damage induced by the radionuclides without complications due to cellular and subcellular barriers, DNA repair mechanisms and other cellular responses. For the first time, the levels of activity per cell required to achieve significant cytotoxic effects was calculated from viability and clonogenic assays using radiolabelled lipophilic complexes in three prostate and breast cancer cell lines, selected for their possible future use in in vivo models for radionuclide therapy of circulating tumor cells and micrometastases using Auger electron emitters. In-chloride ( 111 InCl 3 ) (Mallinckrodt, Netherlands) was supplied at 111 MBq in 0.3 mL 0.05 M HCl. 67 Ga-citrate (6.46 mM citrate, Mallinckrodt, Netherlands) was converted to 67 Ga-chloride ( 67 GaCl 3 ) [29]. Briefly, 67 Ga-citrate (82-160 MBq in 2.2 mL) was diluted to 5 mL with dH 2 O and passed through a Silica Light SEP-PAK column (Waters, US) at 1 mL/min. After washing with 5 mL dH 2 O, trapped 67   Plasmids were co-incubated with 14 mM dimethylsulfoxide (DMSO), excess EDTA (5 mM) or diethylenetriamine pentaacetic acid (DTPA; 5 mM). Controls included untreated plasmid (no radionuclide), external irradiation (where radionuclides in a 50 mL tube were physically separated from plasmid in 1.5 mL microcentrifuge tube inside the 50 mL tube), and equivalent amounts of non-radioactive gallium-(0.69 pmol) and indium-chloride (0.58 pmol) at molar concentrations equivalent to 1 MBq radionuclide.

Analysis of gel electrophoresis images
Images were analyzed by densitometry of each plasmid band (Figs. 1-2, S1-3; supercoiled, circular and linear; Image J 1.48, NIH, USA). Background was measured and subtracted from band intensity. The fraction of supercoiled plasmid (undamaged) of total plasmid represents undamaged plasmid.
Radiochemical yield of radiometal complexes was measured in a dose calibrator.

Cellular uptake and retention of radionuclide oxine complexes
Cells (10 6 ) in suspension were incubated with 0.1 MBq 67  In-oxine in 1 mL PBS for 1 h at 37°C and 5% CO 2, then pelleted and washed twice with PBS. Cell-bound (pellet) and free (supernatant) activity was measured by gamma counter.
For cellular retention studies, cells were treated and washed as above and plated in a 6-well plate for three days. At different times, medium was collected, cells washed, and the amount of cell-bound versus free activity measured. The percentage of cell-bound activity retained within the cell at time points after 1 h (set at 100%) was measured.
Following incubation, cells were centrifuged, washed and seeded in medium in 6-well plates for 3 days at 37°C in 5% CO 2 . Cells were then washed, trypsinised and counted for viability by trypan blue exclusion.

Clonogenic survival
Cells were treated as in the viability assay, but after treatment and washing, 800-2500 cells were seeded in 6-well plates at 37°C in 5% CO 2 for 10-14 days. Medium was replaced every 3 days. Colonies (N50 cells) were fixed, stained with methanol/1% crystal violet (Sigma, 1:1) and counted. The results were plotted as the surviving percentage relative to untreated values, with the latter set at 100%.

Statistical analysis
For plasmid studies, data were analyzed by 2-way ANOVA. Student and paired t-tests were used to compare preparations at any one particular time point or the results from the oxine studies, respectively. Statistical analyses were carried out with GraphPad Prism 5 (version 5.04, GraphPad Software Inc., USA).

Cell free plasmid DNA damage by 111 In and 67 Ga
Incubation of pBR322 supercoiled DNA with 111 In and 67 Ga (0.1-1 MBq) led to single and double DNA strand breaks, i.e. conversion of supercoiled plasmid to either relaxed or linear plasmid, respectively. Plasmid integrity (i.e. the fraction of plasmid remaining in the supercoiled form) decreased as radioactivity increased; Fig. 1A; Fig. S1). As this activity produced significant damage without assay saturation, it was deemed suitable for comparative studies. DNA damage was detected as early as 4 h post incubation (Fig. 1B-C) and after 24 h of incubation, the supercoiled DNA fractions were reduced to 0.001 ± 0.002 and 0.06 ± 0.01 for 67 Ga and 111 In respectively (p = 0.002 compared to untreated). In contrast, untreated controls (0.76 to 0.90) and nonradioactive In-chloride (5.8 nM; 0.89 ± 0.002) or Ga-chloride (6.9 nM; 0.86 ± 0.04) controls produced little DNA damage.
Plasmid DNA was partially protected against 111 In-induced damage, and less so from 67 Ga, by co-incubation with DMSO ( Figs. 1 and S2). At 24 h, DMSO co-incubation led to supercoiled fractions of 0.47 ± 0.13 and 0.20 ± 0.04, for 1 MBq 111 In and 67 Ga, respectively.
External irradiation (i.e. radionuclide separated from the plasmid by the walls of a plastic tube; so that only gamma photons were incident upon the plasmid-containing solution) produced relatively little DNA damage for 67 Ga (p N 0.05 at 48 h compared to untreated controls). External 111 In produced significantly more DNA damage than untreated controls (p b 0.05 at 48 h), but significantly less than internal 111 Inchloride with and without DMSO (p b 0.001 and p b 0.001 at 48 h, respectively).

Radionuclide oxine synthesis
Radiolabelling yields for 67 Ga-oxine, -tropolone, and -MPO were 92%, 80%, and 25%, respectively, and 98% for 111 In-oxine. 67 Ga-oxine gave the highest cell binding (Fig. S4); all subsequent studies were carried out with the oxine complex. In DU145 cells, a one-hour incubation with 111 In-oxine or 67 Ga-oxine allowed radionuclide binding at 60.6 ± 8.8% or 7.5 ± 1.3%, respectively (Fig. 3A). This decreased with time with only 31.2 ± 1.4% and 38.8% ± 0.7% of the initially-bound 111 In and 67 Ga, respectively, retained 72 h after a onehour incubation period (Fig. 3B). Similar results were found in cell lines MDA-MB-231 and HCC1954 (Figs. S5, S6). The different cell labeling efficiencies of 111 In-oxine and 67 Ga-oxine raise the issue of whether to discuss cellular toxicity in relation to the radioactivity to which the cells are exposed in total (i.e. the radioactivity added to the culture) or the radioactivity that is accumulated in the cells (referred to as "cellbound" activity henceforth). Both are discussed together in the following paragraphs.

Trypan blue viability assay
A 50 is defined as the cell-bound activity causing 50% reduction in viability relative to untreated cells (100%). Three days after an initial onehour incubation period with 67 Ga-oxine, the A 50 was approximately 1 Bq/cell (Table 1, Fig. 4A). Cell-bound 67 Ga activity required to reduce viability to 17.4 ± 6.6% was approximately 1.5 Bq/cell; this required incubation with 15 MBq/mL 67 Ga-oxine (Fig. 4B). At this concentration, 67 Ga-citrate, which was not taken up significantly in cells, caused only 53% loss in viability (Fig. 4B). A similar loss in viability occurred in the control sample incubated with decayed 67 Ga-oxine.  Qualitatively similar results were obtained with 111 In; the A 50 was approximately 1 Bq/cell. However, even at cell-bound activities up to 19 Bq/cell, viability did not drop below 20%. As for 67 Ga, the controls showed a significant level of toxicity caused by decayed 111 In-oxine similar to that caused by 111 In-chloride, which did not bind significantly to cells. Interestingly, 67 Ga-oxine-induced toxicity at 15 MBq/mL was the same as that caused by the same concentration of 111 In-oxine, despite this concentration of 111 In-oxine yielding almost 10-fold higher activity per cell. Non-cell bound 111 In-chloride caused toxicity (viability around 50%). A similar level of toxicity resulted from the purely chemical effect of decayed 111 In-oxine.
Qualitatively similar results for both 67 Ga and 111 In were found in cell lines MDA-MB-231 and HCC1954 (Figs. S7 and S8 and Table S1).

Clonogenic survival assay
A one-hour incubation period with 67 Ga-oxine (15 MBq/mL) with cellular uptake as little as 1.1 Bq/cell was enough to diminish clonogenic survival to 4.4% ± 3.1% compared to untreated controls (Fig. 5A). Replacing 67 Ga-oxine with 67 Ga-citrate at this same concentration, with minimal cellular uptake, caused no significant loss in clonogenicity compared to untreated controls (Fig. 5B). Qualitatively similar results were obtained for 111 In demonstrating that neither radionuclide affected clonogenicity significantly unless bound to the cell (Fig. 5B). Fully decayed radioactive 67 Ga-oxine and 111 In-oxine added to the incubation medium led to a significant decrease in relative clonogenic survival (to 74 ± 17% and 69 ± 20% for decayed 67   In-oxine, respectively, see Fig. 5B) compared to untreated controls. However this chemical toxicity was much less than the toxicity of their non-decayed counterparts, indicating that the radioactivity was by far the major contributor to the observed toxic effect. Qualitatively similar results were found in cell lines MDA-MB-231 and HCC1954 (Figs. S9 and S10 and Table S2).

Discussion
The plasmid data presented here suggest the involvement of different mechanisms of DNA damage. These include ionization and formation of free radicals along the tracks of Auger electrons, local ionization events caused directly by the residual ion after Auger electron emission (short range effects), free radicals diffusing significant distances from the Auger electron track and residual ion, and minor ionization and free radicals caused by gamma photons (long range effects).
DNA damage produced by 67 Ga was significantly reduced by chelation with EDTA, DTPA or citrate and incubation with hydroxyl radical scavenger DMSO. The protective effect of chelating the radionuclides with EDTA, DTPA or citrate, on DNA may be a distance effect; assuming unchelated positively-charged In 3+ and Ga 3+ bind directly to negatively charged DNA, as has previously been shown for 111 In [33]. Complexing 67 Ga with EDTA, forming a negatively charged 1:1 6coordinate complex, would completely envelope the 67 Ga atom, preventing metal coordination by plasmid DNA [34]. However, 111 In Table 1 Cell-bound activity per cell (Bq/cell) required for a 50% (A 50 ) and 90% (A 10 ) reduction in viability and clonogenicity in DU145 cells compared to untreated cells (determined by interpolation of data in Figs No A 10 value exists for 111 In-oxine (viability assay), as loss of membrane integrity was not achieved in more than 75% of cells.  forming a 1:1 complex with EDTA would leave some possibility for DNA to coordinate directly to the radiometal which would remain coordinatively unsaturated because of its larger ionic radius [35]. DTPA offers more protection than EDTA against DNA damage by 111 In (Fig. 2); this may be because being octadentate it more completely fills the coordination sphere of indium than EDTA does [36]. The degree of protection (EDTA b DTPA b citrate 2 ) is also in line with the negative charge of the resulting complex: (1-, 2-, 5-). Free radical scavengers such as DMSO are unlikely to protect against the short-range effects. Previous studies with free radical scavengers have focused on 125 I, where DMSO reduced strand breakage by 40% if 125 I was not bound to DNA. When bound to DNA, damage induced by 125 I was not diminished by DMSO [37]. Also, non-plasmid-bound 99m TcO 4 − caused several fold lower induction of single strand breakage in the presence of DMSO [38,39].
Overall, direct incubation of radionuclides with plasmid DNA in a cell-free system is useful to understand DNA damage by radionuclide emissions and decay only. However, in these experiments radionuclides can directly bind DNA, thus overestimating the potential damage compared to the cellular environment, where direct binding to DNA is less likely.
Results obtained in cell studies showed that 111 In and 67 Ga induced high clonogenic toxicity only if incorporated into the cell; external radionuclides and other variables had little effect. Nonetheless decayed radionuclides, producing amongst other compounds zinc and cadmium, did influence both viability and clonogenicity. External irradiation via gamma emissions produced more DNA damage for 111 In-chloride than 67 Ga-chloride due perhaps to higher gamma ray exposure rate constants. Surprisingly, cell viability was decreased for noninternalized 111 In (60.9 ± 8.4%) and 67 Ga (47.2 ± 8.4%) compared to untreated cells, while clonogenic toxicity was not. This highlights that they measure different aspects of cytotoxicity and are complementary rather than alternative methods.
The clonogenic toxicity of incorporated radioactivity is similar for the two radionuclides and for all three cell lines, with an A 10 of approximately 1 Bq/cell. This similarity should be interpreted cautiously, since the cellular toxicity of Auger electron emitters is likely to be highly dependent on the intracellular distribution of the radionuclides, which we have not determined and which may not be the same for the two radionuclides. Nevertheless this figure may be a useful guide to how much radioactivity must be incorporated into cancer cells in vivo for effective targeted radionuclide therapy (tRT) and could be used to assess feasibility of clinical tRT.
It should be noted that 67 Ga-oxine is not a very effective method of incorporating 67 Ga into cells. Results were, however, consistent with previous trends, including radiolabeling yields for oxine with 67 Ga [30] and cell labeling numbers of 111 In-oxine [31] and 67 Ga-oxine [40]. Efficient cell labeling with 111 In is probably due to 111 In-oxine diffusing into the cell cytoplasm and dissociating whereupon 111 In binds intracellular macromolecules and is trapped within the cell [31]. In leukocytes, 111 In-oxine also partly localizes to the nucleus [41]. If 67 Ga-oxine forms a more stable complex [40], the radionuclide may diffuse in but become trapped less readily due to slower dissociation. In order to achieve comparable cellular uptake (Bq/cell), the radioactive concentration of 67 Gaoxine was increased compared to 111 In-oxine.
Future studies should focus on targeted approaches as well as in vivo therapeutic studies comparing 67 Ga with 111 In as well as beta emitters such as 177 Lu. Interestingly, the higher energy Auger electrons emitted by 67 Ga compared to other Auger electron-emitting radionuclides may provide an advantage by relaxing the requirement for 67 Ga to be localized in specific sub-cellular compartments (in particular the nucleus) in order to be effective as a therapeutic. The critical observation that 67 Ga (and similarly 111 In) has to be cell bound to be effective suggests that future targeting studies can focus on the feasibility in vivo of achieving target uptake of around 1 Bq per cancer cell required for effective cell killing.

67
Ga damages plasmid DNA in a manner that may be dependent on distance to the DNA, which in turn may be affected by the chemical form of the radionuclide. Neither 67 Ga nor 111 In showed substantial toxicity unless incorporated into the cells. The threshold cellular uptake of 67 Ga to achieve substantial cell kill is of the order of 1 Bq per cell. 67 Ga deserves further evaluation for radionuclide therapy, especially in the context of a theranostic pairing with 68 Ga.