In vitro cytotoxicity of Auger electron-emitting [67Ga]Ga-trastuzumab

Introduction Molecular radiotherapy exploiting short-range Auger electron-emitting radionuclides has potential for targeted cancer treatment and, in particular, is an attractive option for managing micrometastatic disease. Here, an approach using chelator-trastuzumab conjugates to target radioactivity to breast cancer cells was evaluated as a proof-of-concept to assess the suitability of 67Ga as a therapeutic radionuclide. Methods THP-trastuzumab and DOTA-trastuzumab were synthesised and radiolabelled with Auger electron-emitters 67Ga and 111In, respectively. Radiopharmaceuticals were tested for HER2-specific binding and internalisation, and their effects on viability (dye exclusion) and clonogenicity of HER2-positive HCC1954 and HER2–negative MDA-MB-231 cell lines was measured. Labelled cell populations were studied by microautoradiography. Results Labelling efficiencies for [67Ga]Ga-THP-trastuzumab and [111In]In-DOTA-trastuzumab were 90% and 98%, respectively, giving specific activities 0.52 ± 0.16 and 0.61 ± 0.11 MBq/μg (78–92 GBq/μmol). At 4 nM total antibody concentration and 200 × 103 cells/mL, [67Ga]Ga-THP-trastuzumab showed higher percentage of cell association (10.7 ± 1.3%) than [111In]In-DOTA-trastuzumab (6.2 ± 1.6%; p = 0.01). The proportion of bound activity that was internalised did not differ significantly for the two tracers (62.1 ± 1.4% and 60.8 ± 15.5%, respectively). At 100 nM, percentage cell binding of both radiopharmaceuticals was greatly reduced compared to 4 nM and did not differ significantly between the two (1.2 ± 1.0% [67Ga]Ga-THP-trastuzumab and 0.8 ± 0.9% for [111In]In-DOTA-trastuzumab). Viability and clonogenicity of HER2-positive cells decreased when each radionuclide was incorporated into cells by conjugation with trastuzumab, but not when the same level of radioactivity was confined to the medium by omitting the antibody conjugation, suggesting that 67Ga needs to be cell-bound or internalised for a therapeutic effect. Microautoradiography showed that radioactivity bound to individual cells varied considerably within the population. Conclusions [67Ga]Ga-THP-trastuzumab reduced cell viability and clonogenicity only when cell-bound, suggesting 67Ga holds promise as a therapeutic radionuclide as part of a targeted radiopharmaceutical. The causes and consequences of non-homogeneous uptake among the cell population should be explored.


Introduction
With regulatory approval of beta particle-emitting radiopharmaceuticals such as [ 177 Lu]Lu-DOTATATE [1], early successes in trials of alpha particle-emitting therapy 223 Ra [2] and an increasing choice of therapeutic radionuclides, chelators and radiochemistry techniques to choose from, this is an exciting time for molecular radiotherapy. Examples of therapeutic radiopharmaceuticals in routine clinical use include 131 I for thyroid cancer, [ 131 I]mIBG for neuroblastoma, 90 Y-and 177 Lu-radiopeptides for neuroendocrine tumours, and 90 Y-labelled microspheres (albeit defined for regulatory purposes as a medical device rather than a radiopharmaceutical) for hepatic tumours [3,4]. Most prominent among new beta particleemitting radiopharmaceuticals undergoing clinical trials is [ 177 Lu]Lu-PSMA [5]. The potential of high linear energy transfer alpha particle-emitters, such as 213 Bi, 225 Ac, 211 At, and 227 Th [6][7][8] to treat low volume tumours effectively while limiting healthy tissue toxicity is being explored. By comparison with alpha particles (<100 μm particle range), the short range (often <1 μm) of Auger electrons and related secondary electrons would be expected to increase the therapeutic ratio and further mitigate non-target toxicity [9]. Although likely to be less effective than beta emitters in large tumours because of the lack of "cross-fire" contribution to the tumour radiation dose, Auger electron-emitters could be effective against single cells and micrometastases which are ineffectively treated by betaemitters [10] and are responsible for recurrence of several tumour types [11,12].
Among therapeutic Auger electron-emitters, 111 In has so far been closest to clinical translation and has been evaluated in both breast cancer [13] and neuroendocrine tumours [14]. 67 Ga, which has been used clinically for SPECT imaging of inflammation and infection [15,16] also emits Auger electrons but its use as a therapeutic radionuclide is relatively unexplored [17][18][19][20][21][22]. Despite promising in vitro results, including its use in a radiopharmaceutical targeted towards the CD74 receptor on human B-lymphoma cells [21,22], low specific activities and a lack of purpose-designed gallium chelators have hindered progress with 67 Ga therapy. The recent development of several new chelators [23][24][25][26][27][28][29][30], including the tris(hydroxypyridinone) chelators, allow kit-based radiolabelling of biomolecules, including proteins, with gallium radionuclides at high specific activities [31][32][33]. This development provides renewed opportunities for clinical translation to assess the tumouricidal potential of 67 Ga preclinically and clinically as a radionuclide for molecular radionuclide therapy. Our recent studies have shown that 67 Ga caused more DNA damage per Bq than 111 In in a cell-free system. The cellular radioactivity required to kill 50% and 90% of HCC1954 breast cancer cells was 6-and 1.5-fold less, respectively, for 67 Ga than for 111 In [9,34]. Lower plasmid DNA damage was caused by 67 Ga than 111 In when the radionuclides were physically separated from the DNA, suggesting that 67 Ga would cause less unwanted damage to surrounding non-target cells than 111 In [34]. These data suggest that 67 Ga could be selectively cytotoxic if targeted to tumour cells, with minimal damage to nearby non-targeted cells.
A widely explored tumour-specific molecular target is HER2, a receptor expressed on breast cancer cells. The antibody trastuzumab binds to HER2 and so is a potential basis for targeted delivery of radionuclides and evaluation of the therapeutic potential of 67 Ga. Even in patients with primary tumours classified as HER2-negative, their metastases can still prove HER2-positive [35]. HER2-positivity in patients with breast cancer is associated with aggressive disease, poor prognosis and shortened overall survival [36]. A combination of trastuzumab, pertuzumab and docetaxel was able to partially overcome trastuzumab resistance and extend survival of patients with metastatic HER2-positive breast cancer from 15 (trastuzumab alone) to 56.5 months [37]. This drug combination is now used as first-line therapy for metastatic HER2-positive breast cancer. Another successful approach uses antibody-drug conjugates, such as the trastuzumab conjugate T-DM1, which delivers a toxic payload of emtansine to HER2-positive breast cancer cells [38]. The success of T-DM1 suggests that the efficacy of using a radionuclide, such as an Auger electron-emitter, as the toxic payload, should be explored [38]. In this study, trastuzumab was used to target 67 Ga to HER2-positive breast cancer cells to determine the ability of 67 Ga to kill these cells selectively compared to [ 111 In]In-DOTA-trastuzumab.

Radionuclide preparation
Radiolabelling was carried out with 67 Ga used either as supplied ([ 67 Ga]Ga-citrate at a concentration of 80-160 MBq in 2.2 mL; Mallinckrodt, Netherlands) or after converting to [ 67 Ga]GaCl 3 as previously described [39]. For conversion to [ 67 Ga]GaCl 3 , [ 67 Ga]Ga-citrate was diluted to 5 mL with distilled water and concentrated on a SEP-PAK silica light 120 mg column (Waters, USA) at 1 mL/min. Trapped 67 Ga was eluted with 0.1 M HCl in 50 μL fractions. The radioactivity concentration in the highest fraction (usually fraction 6 or 7) was 1010 ± 388 MBq/mL [ 111 In]InCl 3 (110-200 MBq) was used as supplied in 0.3 mL hydrochloric acid (Mallinckrodt, Netherlands).

THP-trastuzumab-Trastuzumab
(435 μL, 22.96 mg/mL) in physiological saline was incubated with 50 mM ethylenediaminetetraacetate sodium (EDTA) solution for 30 min to remove any adventitious metal ions. The mixture was filtered at 4000 g in a Vivaspin 2 filter (Sartorius, Germany). The filter was washed with metal-free HEPES and the antibody thus recovered had a concentration of 12.46 ± 0.05 mg/mL (as measured by NanoDrop spectrophotometer) in metal-free HEPES. This solution of trastuzumab (3.23 mg, 260 μL, pH 8.7) was incubated with THP (21 μL of 10 mM solution in DMSO) at 10:1 chelator:antibody molar ratio at room temperature (2 h) and then 4 °C (2 h). The mixture was applied to a Sephadex PD MidiTrap G-25 size exclusion column (GE Healthcare) and eluted with 0.2 M ammonium acetate. The fraction eluting between 1 mL and 2.5 mL was applied to a Sephadex PD-10 size exclusion column (GE Healthcare), and eluted with 0.2 M ammonium acetate solution in 500 μL fractions. Fraction 8 (protein concentration 3.5 mg/mL, measured by NanoDrop) was used for radiolabelling.

Quality control-Labelling
efficiencies were determined by paper chromatography on Whatman 1 paper (1 cm × 6.5 cm) with a mobile phase of 1 mM DTPA and were almost always >95%. Preparations in which the radiochemical purity was ≤95% were purified using a PD-10 column in 1 mL fractions of PBS containing 0.1% BSA.

Cell culture
MDA-MB-231 (HER2-negative) and HCC1954 (HER2-positive) breast cancer cell lines were grown as monolayers at 37 °C in a humidified atmosphere with 5% CO 2 in high glucose (4.5 g/L) Dulbecco's Modified Eagle Medium (DMEM; PAA Laboratories) and

Europe PMC Funders Author Manuscripts
RPMI-1640, respectively. Media were supplemented with 1.5 mM L-glutamine (PAA Laboratories, Austria), 10% foetal bovine serum (Invitrogen) and penicillin/streptomycin (Invitrogen) to create full medium. Cells were harvested using trypsin-EDTA for experimental use. In-DOTA-trastuzumab (0.08 and 0.09 MBq, respectively) in 1 mL medium for 1 h at 37 °C. Unbound activity in the medium was collected and combined with washings from two PBS wash steps after centrifuging at 1500 rpm for 5 min. Cell pellets were subjected to an acid wash by resuspending in 1 mL 500 mM NaCl and 200 mM NaOAc (pH 2.5) for 5 min at 4 °C [41]. Cells were then centrifuged to separate membraneassociated activity (supernatant) from internalised activity (pellet). Radioactivity was measured as above. . Slides were then dipped in 1% acetic acid for 30 s and then in 0.5% gelatin and dried at room temperature. Finally, autoradiographs were fixed in 30% sodium thiosulfate for 5 min, washed for 10 min and left to dry. Prior to imaging, 50 μL DAPI-containing mounting solution was applied to the slide and a coverslip was placed on top. Images were acquired with an EVOS FL Cell imaging microscope (Life Technologies, USA) and analysed with Image J (NIH, USA). The total number of silver grains around each cell was measured (within a circle 25 μm in diameter around the cell centre) in cells identified as DAPI-positive. The amount of signal in cells was corrected for background in an equivalent cell-free area of the image.

Radiolabelling
The

Viability (trypan blue staining)
Treatment of HER2-positive HCC1954 cells with 100 nM non-radiolabelled trastuzumab did not affect viability (relative viability 101 ± 3.7% compared to untreated control cells whose relative viability was defined as 100%  Fig. 2A, B). Treatment of HCC1954 cells with [ 67 Ga]Ga-THP-trastuzumab at 100 nM (2 MBq/mL) gave an average cellular radioactivity of 0.14 Bq/cell and produced significant reduction in cell viability (to 66.5 ± 4.8% of the control value which was defined as 100 ± 8.6% at 0 nM antibody concentration, p = 0.007) ( Fig. 2A-C). Under the same conditions, [ 111 In]In-DOTA-trastuzumab, gave an average cellular uptake of 0.10 Bq/cell, and reduced viability to 66.2 ± 6.7% of the control value. Thus, the effect of the two radiopharmaceuticals (at similar average cell-bound activity per cell) on viability was not significantly different (p > 0.9). Treatment of HER2-positive cells with non-antibodyconjugated [ 67 Ga]Ga-THP (which did not bind to cells) did not measurably reduce cell viability, while non-antibody-conjugated [ 111 In]In-DOTA (which also was not cell-bound) marginally decreased cell viability to 85.2 ± 4.0% of the control (Fig. 2D). Thus, both radionuclides were significantly less toxic to HER2-positive HCC1954 cells when not cellbound (p = 0.0003 for 67 Ga and p = 0.0146 for 111 In). The difference in toxicity of non-cellbound 67 Ga and 111 In was not significant (p = 0.09). The results suggest that 67 Ga and 111 In need to be bound to cells to achieve significant reduction in cell viability.

Clonogenicity
Treatment of HER2-positive HCC1954 cells with 100 nM non-radiolabelled trastuzumab did not affect clonogenicity (relative clonogenicity 0.97 ± 0.13 compared to untreated control cells whose relative clonogenicity was defined as 1.

Discussion
This study extends our previous cell-free system studies using pBR322 plasmids and cells non-specifically labelled using [ 67 Ga]Gaoxine, which showed that 67 Ga causes more DNA Europe PMC Funders Author Manuscripts damage, and a greater decrease in cancer cell clonogenicity when bound to cells, than the same activity of 111 In [34]. Here, we have exploited the targeted delivery of radionuclides to HER2-positive cells using the antibody trastuzumab to investigate the selectivity and cytoxicity of targeted radionuclide therapy with 67 Ga compared to 111 In [10,[42][43][44].
Initial binding studies were carried out to determine the selectivity of uptake and the concentrations of radiolabelled antibodies required to achieve sufficient uptake of radionuclide in cells to affect measurable impact on cell viability and clonogenicity.
Due to difficulty experienced in radiolabelling trastuzumab with 67 Ga to a sufficiently high specific activity (i.e. comparable to that achievable with 111 In) under mild conditions using the DOTA chelators, we used the gallium-specific chelator THP, which can be radiolabelled with gallium radionuclides in under 5 min, at pH 6.5 and room temperature [23] allowing mild and efficient radiolabelling of proteins [33]. Viability studies using trypan blue staining showed that 67 Ga induced significant toxicity, but only if incorporated into the cell; at similar concentrations in the media, non-antibodyconjugated [ 67 Ga]Ga-THP, which did not bind to cells, did not affect cell viability. [ 111 In]In-DOTA, which did not bind to cells, diminished cell viability only slightly, possibly due to a higher fraction of radiation dose coming from gamma emissions in the case of 111 In. This is consistent with our previous observations [34].
Both [ 67 Ga]Ga-THP-trastuzumab and [ 111 In]In-DOTA-trastuzumab at 100 nM (2.5 MBq/mL) decreased clonogenic survival significantly compared to untreated HCC1954 cells and compared to HCC1954 cells incubated with 100 nM non-radiolabelled trastuzumab (Fig. 3), suggesting that the mechanism of cell death is mediated by radiation damage. Interestingly, the trends seen in Fig. 3A-C match what one might expect for a survival curve of tumour cells affected through the bystander response [45]. However, further studies will need to be carried out to determine whether the bystander response is indeed activated by these Auger-emitting radiopharmaceuticals. Our data using 100 nM non-radiolabelled trastuzumab (14 μg/mL, 1 h incubation) are in agreement with previous findings that trastuzumab induces no change in viability of HCC1954 cells even after 7 days of incubation at 15 μg/mL [46].
At similar radioactivity concentrations in the medium, [ 67 Ga]Ga-THP-trastuzumab is 1.4fold more toxic (measured as effect on clonogenicity) than [ 111 In]In-DOTA-trastuzumab (Fig. 3, A and B). It is likely that only part of this apparent difference in toxicity is associated with the 1. Europe PMC Funders Author Manuscripts cell, because when clonogenicity is plotted against the measured average activity per cell (Fig. 3, C), 67 Ga still shows significantly greater suppression of clonogenicity than 111 In.
The differences in toxic effects as measured by viability (trypan blue) and clonogenicity indicate that viability and clonogenic assays measure different aspects of cytotoxicity and are complementary rather than alternative methods.
The modest difference in toxicological profile between the two radioimmunoconjugates could originate from differences in the cellular trafficking, or from differences in the electron and photon emission spectra of the two radionuclides. The 67 Ga conjugate shows greater apparent affinity and uptake than the 111 In conjugate, and although this does not cause any significant difference in the measured internalised fraction, it implies that differences in subsequent intracellular trafficking, including to the cell nucleus, cannot be excluded. Electron emissions (Auger and Coster-Kronig electrons) of 67 Ga and 111 In are similar in total energy (6.24 MeV and 6.75 MeV, respectively) and both also emit internal conversion electrons (0.32 and 0.16 per decay respectively, with a total energy of 28.078 KeV and 25.957 KeV, respectively). However, the energy distribution, and hence range in tissue, of the Auger and Coster-Kronig electrons differ markedly between the two radionuclides. The Supplementary information provides relevant data in the form of plots of yield of electrons per decay against the logarithm of the range [47]. The total energy of the Auger and Coster-Kronig emissions of 67 Ga is distributed among fewer electrons than is the case for 111 In (4.7 and 14.7, respectively [47]) and consequently, electrons from 67 Ga have higher energy per electron, and hence longer range. Fewer Auger electrons emissions per decay than 14.7 [47] have been estimated for 111 In, namely between 5.84 and 7.43, however these values are still higher than for 67 Ga (between 4.56 and 4.96) [48]. 67 Ga emits no electrons with a range below 1 nm, whereas 111 In emits 8; and 67 Ga emits around 0.6 electrons per decay with a range encompassing multiple compartments of the cell (1-10 μm), whereas 111 In emits none. Thus, 111 In decay is likely to be biologically effective over a much smaller range (<1 μm) than 67 Ga decay, and the biological effects of decay of 111 In are likely to be much more dependent on its sub-cellular location than those of 67 Ga.
Although DNA damage is known to be an important mechanism by which radionuclide decay induces cell killing, it is a currently open and actively investigated question whether the cell nucleus and DNA is the most important target for radionuclide emission, or whether other organelles such as the mitochondria or cell membrane [49,50] are also important radiosensitive targets. Although not explored here, it is known that [ 111 In]In-trastuzumab can localise to the nucleus [42,44]. The extent to which this happens is likely to be an important factor in the therapeutic efficacy of the radiopharmaceutical, considering the very short range of the majority of its electron emissions. Nuclear localisation peptide sequences have often been used to further increase therapeutic efficacies of Auger-emitting radiopharmaceuticals [42][43][44], including some labelled with 67 Ga [51]. Although other nonnuclear-associated pathways are known that can influence therapeutic efficacy [50], and despite the longer range of the 67 Ga electron emissions, it is possible that an approach using nuclear localisation sequences will also be of benefit for [ 67 Ga]Ga-trastuzumab. The data presented here, however, do not provide insight into the mechanism of cell killing or subcellular target of these radioimmunoconjugates. Microautoradiography demonstrates a wide range of [ 67 Ga]Ga-THP-trastuzumab binding among HCC1954 cells, despite the presumption that the cells are a clonally identical population. In fact, 20% of cells in the highest activity group received around a hundred times higher radiation dose than the 15% in the lowest activity group, and a large subpopulation of the cells would have received negligible radiation dose. This may account for the observation (Fig. 3B, C) that the surviving cell fraction does not greatly diminish below 0.5 as activity of [ 67 Ga]Ga-THP-trastuzumab per cell increases above 0.05 Bq/cell. It implies that the average activity per cell is likely to have very limited utility as a predictor of toxicity. Heterogeneity of target expression is likely to be even greater in cell populations within patients' tumours. This is likely to be the overriding limitation in targeted Auger emitter therapy. The impact of non-homogeneous radiopharmaceutical uptake in limiting anti-tumour effectiveness will need to be considered for any radiopharmaceutical undergoing in vitro and in vivo evaluation and requires further investigation if the antitumour effects of targeted Auger electron-emitting radionuclides are to be better understood.

Conclusion
[ 67 Ga]Ga-THP-trastuzumab specifically reduced cell viability and clonogenicity of HER2expressing cells, confirming the potential of 67 Ga as a therapeutic radionuclide as part of a targeted radiopharmaceutical. Diminished viability and clonogenicity only occurred when radioactivity was directly bound to cells and was not caused by radioactivity in the medium. Highly heterogeneous cellular uptake may compromise the efficacy of Auger electron irradiation and the interpretation of data derived from average cellular radioactivity. This merits further investigation using methods, such as microautoradiography, capable of measuring the radioactivity of individual cells.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material.   Examples of microautoradiography of HCC1954 cells incubated with 4 nM 67 Ga-THPtrastuzumab showing mostly highly variable activity per cell ranging from very low (A) to very high (D). The green arrows show the cell boundaries as determined by light microscopy (green circles) with DAPI staining delineating the nucleus (blue). Silver grains are red in the images. Scale bar is 100 μm.