Relationship of In Vitro Toxicity of Technetium-99m to Subcellular Localisation and Absorbed Dose

Auger electron-emitters increasingly attract attention as potential radionuclides for molecular radionuclide therapy in oncology. The radionuclide technetium-99m is widely used for imaging; however, its potential as a therapeutic radionuclide has not yet been fully assessed. We used MDA-MB-231 breast cancer cells engineered to express the human sodium iodide symporter-green fluorescent protein fusion reporter (hNIS-GFP; MDA-MB-231.hNIS-GFP) as a model for controlled cellular radionuclide uptake. Uptake, efflux, and subcellular location of the NIS radiotracer [99mTc]TcO4− were characterised to calculate the nuclear-absorbed dose using Medical Internal Radiation Dose formalism. Radiotoxicity was determined using clonogenic and γ-H2AX assays. The daughter radionuclide technetium-99 or external beam irradiation therapy (EBRT) served as controls. [99mTc]TcO4− in vivo biodistribution in MDA-MB-231.hNIS-GFP tumour-bearing mice was determined by imaging and complemented by ex vivo tissue radioactivity analysis. [99mTc]TcO4− resulted in substantial DNA damage and reduction in the survival fraction (SF) following 24 h incubation in hNIS-expressing cells only. We found that 24,430 decays/cell (30 mBq/cell) were required to achieve SF0.37 (95%-confidence interval = [SF0.31; SF0.43]). Different approaches for determining the subcellular localisation of [99mTc]TcO4− led to SF0.37 nuclear-absorbed doses ranging from 0.33 to 11.7 Gy. In comparison, EBRT of MDA-MB-231.hNIS-GFP cells resulted in an SF0.37 of 2.59 Gy. In vivo retention of [99mTc]TcO4− after 24 h remained high at 28.0% ± 4.5% of the administered activity/gram tissue in MDA-MB-231.hNIS-GFP tumours. [99mTc]TcO4− caused DNA damage and reduced clonogenicity in this model, but only when the radioisotope was taken up into the cells. This data guides the safe use of technetium-99m during imaging and potential future therapeutic applications.


Introduction
Auger electron (AE) therapy is an attractive option to treat disseminated cancers [1]. AEs have a very short range (<1 µm) with linear energy transfer (LET) reaching 26 keV/µm, setting them apart from αand β-particles [1,2]. Consequently, AE-emitters induce cell toxicity when in close proximity to radiosensitive structures (i.e., nuclear DNA). Targeting AE-emitters into cancer cells can therefore specifically kill cancer cells while theoretically minimising radiation toxicity to surrounding healthy tissues [1]. Currently, [ 123 I]I-MAPI 2 of 13 is the most promising AE-radiopharmaceutical undergoing clinical translation for the treatment of glioblastoma multiforme [3].
Technetium-99m has a half-life of 6.02 h, wide availability, and emits γ-rays (140 keV) ideal for single-photon emission computed tomography (SPECT) [4]. Importantly, it also decays by internal conversion, resulting in the average emission of 1.1 internal conversion electrons and 4.4 AEs per decay [1], thus potentially being a theragnostic tool in AE-therapy. Previous studies showed technetium-99m to damage DNA, reduce cell survival, and slow tumour growth [5][6][7].
Technetium-99m in the chemical form of [ 99m Tc]TcO 4 − is specifically taken up by cells expressing the human sodium/iodide symporter (hNIS), which is a transmembrane glycoprotein that normally mediates iodide uptake into thyroid follicular cells [8]. hNIS presents with endogenous expression restricted to the thyroid, gastric mucosa, salivary, and lacrimal glands [8]. Its anion transport selectivity is low, and [ 99m Tc]TcO 4 − and other imaging radiotracers are also transported. hNIS has been exploited for imaging, [ 131 I]I − -therapy and as a reporter gene [9]. hNIS has also been used for controlled uptake of [ 99m Tc]TcO 4 − to determine its radiotoxicity [10,11]. However, a radiobiology-informed characterisation of technetium-99m effects to predict the efficacy of potential technetium-99m-based AE-therapy and to assess its radiobiological risk when used as an imaging agent remains elusive [12,13].
Here, we used breast cancer cells with defined hNIS expression as a model of controlled [ 99m Tc]TcO 4 − uptake to determine cellular technetium-99m radiotoxicity and nuclear-absorbed dose and to provide reference data in comparison to external beam radiation therapy (EBRT).

Technetium-99m and EBRT-Induced DNA Damage
Average numbers of γ-H2AX foci per nucleus significantly increased 0.5 h after EBRT with 2 Gy irrespective of the cell type compared to unirradiated cells (Figure 2A). These decreased 24 h post-irradiation with no significant difference compared to control nonirradiated cells at 24 h ( Figure 2A). No significant difference in radiosensitivity between parental and hNIS-GFP-expressing cells was observed (p = 0.9191).

Technetium-99m and EBRT-Induced DNA Damage
Average numbers of γ-H2AX foci per nucleus significantly increased with 2 Gy irrespective of the cell type compared to unirradiated cells (Fi    Figure 3A and Figure S3). There was also no SF reduction in MDA-MB-231.hNIS-GFP treated with decay product [ 99 Tc]TcO 4 − ( Figure 3C).

Dosimetry
When exposed to EBRT, the SF of MDA-MB-231.hNIS-GFP and MDA-MB-231 cells significantly decreased with increasing absorbed dose, with no significant difference between the two cell lines (p = 0.2045) ( Figure 4A). Calculation of dosimetry for molecular

Dosimetry
When exposed to EBRT, the SF of MDA-MB-231.hNIS-GFP and MDA-MB-231 cells significantly decreased with increasing absorbed dose, with no significant difference between the two cell lines (p = 0.2045) ( Figure 4A). Calculation of dosimetry for molecular radiotracers fundamentally requires knowledge of the subcellular distribution of the radioisotope. Classical cell fractionation methodology showed that only 4 Figure 4B,C). Using the subcellular fractionation-derived value as input, an estimated absorbed dose to the nucleus of 0.79 Gy was delivered to reach SF 0.37 (95%CI = [SF 0.29 ;SF 0.45 ]; Figure 4B). In contrast, relying on the volumetric approach, we estimated the delivered nuclear-absorbed doses to obtain SF 0.37 depending on the assumed location of [ 99m Tc]TcO 4 − within the cells, as shown in Figure 4C.
radiotracers fundamentally requires knowledge of the subcellular distribution of the radioisotope.  Figure 4B). In contrast, relying on the volumetric approach, we estimated the delivered nuclear-absorbed doses to obtain SF0.37 depending on the assumed location of [ 99m Tc]TcO4 − within the cells, as shown in Figure 4C.

Discussion
Only few in vitro and in vivo radiotoxicity studies have been performed w netium-99m so far [6,10,11,[15][16][17][18][19], and the correlation of technetium-99m-induce toxicity with intracellular activity remained elusive. Here, using an hNIS expre forded model for controlled [ 99m Tc]TcO4 − uptake into cells, we found that techneti induced DNA damage and reduced cell survival manifested only when the radi was taken up into cells (Figures 2B-E and 3A). Notably, the technetium-99m dec uct, the β-emitter technetium-99, did not exert any DNA damage under these con further demonstrating the specificity of the observed effects to technetium-9

Discussion
Only few in vitro and in vivo radiotoxicity studies have been performed with technetium-99m so far [6,10,11,[15][16][17][18][19], and the correlation of technetium-99m-induced radiotoxicity with intracellular activity remained elusive. Here, using an hNIS expression-afforded model for controlled [ 99m Tc]TcO 4 − uptake into cells, we found that technetium-99m induced DNA damage and reduced cell survival manifested only when the radioisotope was taken up into cells ( Figure 2B-E and Figure 3A). Notably, the technetium-99m decay product, the β-emitter technetium-99, did not exert any DNA damage under these conditions, further demonstrating the specificity of the observed effects to technetium-99m. Importantly, observed differences in DNA damage induction between hNIS-GFP-expressing and parental MDA-MB-231 cells were not due to hNIS-mediated changes in radiosensitivity as demonstrated by both cell lines suffering equal DNA damage when exposed to EBRT (Figure 2A).
While we reached up to 1900 ± 200 mBq/cell after 0.5 h incubation with [ 99m Tc]TcO 4 − , a longer incubation time of 24 h was needed to observe radiotoxicity ( Figures 1B and 3A). Therefore, a cellular concentration of 30 mBq/cell technetium-99m was required to reduce survival to 37%. This was about two-fold less than what was previously reported for rat thyroid follicular cells [11], whereby possible expression and uptake rate differences between hNIS-GFP in our cells and rat NIS in rat thyroid follicular cells, as well as differences in radiosensitivity were not accounted for. To better compare studies not performed under identical conditions, we estimated decays/cell, and in our work, 24,430 decays/cell were required to achieve SF 0.37 (95% CI = [SF 0.31 ; SF 0.43 ]). This is substantially more than was required to achieve a 90% reduction of cell survival when using indium-111 and gallium-67 (3240 and 3600 decays/cell, respectively, over 60 min incubation) or thallium-201 (1000-1600 decays/cell over 90 min incubation) [20,21]. While this data stem from different cells, the discrepancies in toxicity effectiveness are likely to be caused by the lower emission of average AE energy released per decay for technetium-99m compared to indium-111, gallium-67, and thallium-201 (0.9 vs. 6.6, 6.9, and 20.9 keV, respectively [1]). Technetium-99m also presents with a larger ratio between total emitted photon energy and electrons per decay compared to other AE-emitters such as platinum-195m (7.78 vs. 0.42 [22]), which increases its likelihood of inducing undesirable radiation effects on normal tissues. Moreover, the short half-life of technetium-99m (6.02 h) is sub-optimal for AE-therapy [22,23]. Despite these disadvantages compared with traditional AE-emitters, technetium-99m remains a contender for theranostic applications as it is widely available, has a long-lived (half-life = 2.1 × 10 5 years) daughter radionuclide, hence less likelihood of inducing undesired harm if detached from a targeting radiopharmaceutical, and is ideally suited for SPECT imaging [4].
The correlation of radiobiological effects with the absorbed dose in EBRT is wellunderstood and crucial to assess radiobiological risks and plan radiotherapy. The doseresponse curves acquired here help to better understand the differences between EBRT and potential AE-therapy with technetium-99m in terms of biological response at the same absorbed radiation dose. Although the LQ-model is adequate for low LET-radiation such as in EBRT [12,24], we showed here that the dose-response curve of internal irradiation with [ 99m Tc]TcO 4 − followed an LQ-model with a short sub-lethal damage repair half-life (high α/β), i.e., negligible quadratic term in the LQ-model (constrain β = 0) ( Figure 4B,C). This difference in the survival curves (β = 0 vs. β > 0) may have been due to heterogeneous dose delivery, protracted exposure and lower dose rate of [ 99m Tc]TcO 4 − over the 24 h incubation period (<1 Gy/h) when compared to a homogeneous high acute dose rate of 5 Gy/min in EBRT, thereby enabling cells to repair sub-lethal damage [24]. Moreover, the non-significant changes in DNA damage with increasing incubation time indicated a protracted exposure to [ 99m Tc]TcO 4 − . In addition, while DNA damage induced by EBRT was significantly repaired to background levels 24 h post-irradiation, DNA damage induced by [ 99m Tc]TcO 4 − was not totally repaired compared to untreated cells for the largest concentration used ( Figure S2A). Hence, our data suggest that lethal damage (α) is the key contribution for technetium-99m radiotoxicity.
For the determination of nuclear-absorbed doses, it was important to estimate the relative amounts of technetium-99m in the nucleus compared to the whole cell volume. Traditionally this was approached by subcellular fractionation. The soluble small molecule nature of [ 99m Tc]TcO 4 − might, however, impact the accuracy of this approach [13]; hence, we also performed microscopy-aided cell volume determinations. First, using subcellular fractionation, a lower nuclear absorbed dose of 0.79 Gy compared to 2.59 Gy with EBRT was estimated to achieve SF 0.37 (95%CI = [SF 0.29 ; SF 0.45 ] and [SF 0.35 ; SF 0.39 ], respectively; Figure 4A,B). Second, using microscopy-aided whole-cell and nucleus volume determinations, we calculated the nuclear-absorbed doses under the following different assumptions: that [ 99m Tc]TcO 4 − was either (i) excluded from the nucleus, (ii) uniformly distributed across the whole cell, or (iii) located only in the nucleus ( Figure 4C). Obtained values for the delivered nuclear-absorbed dose were (i) 0.33 Gy, (ii) 2.51 Gy, and (iii) 11.7 Gy, in line with the expectation of a dependency on relative technetium-99m proximity to the nucleus, i.e., the further away from the nucleus, the lower the estimated delivered nuclear-absorbed dose due to the short-range (<1 µm) energy deposition of AEs. Except for the unrealistic borderline case that all technetium-99m would be present in the nucleus, the obtained absorbed dose values were lower than for EBRT. These results were similar to a previous study in rat follicular cells that reported 1.2 ± 0.1 Gy and 2.6 ± 0.3 Gy for incubation with [ 99m Tc]TcO 4 − (24 h) and irradiation with X-rays, respectively [11]. Notably, there were limitations stemming from a lack of accurate methods to determine precisely the relative distribution of [ 99m Tc]TcO 4 − within cells and limitations in the Medical Internal Radiation Dose formalism (MIRD) formalism (i.e., assumption of uniform radioactivity distribution within cells, spherical shape of cells). Because of these limitations, we did not attempt to compare the relative biological effectiveness between AE-therapy with technetium-99m as modelled here and EBRT. However, the values obtained for nuclear-absorbed doses, the primarily lethal damage α component, and the long-term induced DSBs, suggest enhanced radiobiological effectiveness of technetium-99m compared to EBRT if technetium-99m was internalised into cells.
We also performed a non-exhaustive in vivo study to quantify the relative uptake of [ 99m Tc]TcO 4 − ( Figure 5). Our observations were in line with previously reported data on [ 99m Tc]TcO 4 − uptake in organs [9,25]. Notably, our data showed that the retention of [ 99m Tc]TcO 4 − in tumours was longer than expected from in vitro radiotracer efflux experiments and that [ 99m Tc]TcO 4 − was still retained in the tumours 24 h after administration, i.e., the time required for [ 99m Tc]TcO 4 − to induce radiotoxicity in our in vitro experiments. In our estimation, 0.5 h after administration of 20 MBq and 200 MBq [ 99m Tc]TcO 4 − cellular activities of 21 ± 10 mBq/cell and 390 ± 60 mBq/cell were reached in the primary tumours. Therefore, at high administered [ 99m Tc]TcO 4 − amounts, it may be possible to reach similar cellular radioactivity values in vivo compared to our in vitro experiments ( Figure 1B) that showed substantial radiotoxicity ( Figure 3A).

Radionuclides and EBRT
Technetium-99m was generator-eluted as Na[ 99m Tc]TcO 4 (Guy's and St Thomas' NHS foundation Trust Radiopharmacy, London, UK) and used within one half-life of elution. EBRT was delivered using a gamma-irradiator calibrated using Gafchromic EBT3 films and performed at 5 Gy/min ( Figure S4) and results expressed as intracellular uptake percentage and intracellular radioactivity (Equations (1) and (2)

Assessment of DNA Damage
Cellular assessment of DNA damage was quantified as double-strand breaks (DSB). Therefore, cells were treated as indicated and then fixed and stained with anti-phosphohistone H2AX (Ser139) antibody to detect DSBs in counter-stained nuclei using confocal fluorescence microscopy (for details, see Supplementary Methods).

Dosimetry
Dosimetry calculations were conducted using the MIRD formalism [26]. Briefly, the cell nucleus was chosen as the target, and the absorbed dose to the nucleus was estimated using self-dose. Live-cell confocal microscopy provided cell and nuclear radii of cells and nuclei (11 ± 2 and 6 ± 1 µm, respectively). Using MIRDcell V2.0 [26], S self-dose -values for different source volumes (cytoplasm and nucleus) were obtained to represent the absorbed dose rate in the target volume per unit activity inside each source volume (Table S1). The obtained S self-dose -values accounted for estimated cell and nuclei radii. Cumulative intracellular decays per cell in each source volume were estimated by calculating time-integrated activity per cell via trapezoidal integration using GraphPad Prism (v9.1.0), whereby uptake and efflux assay data served as input as well as data from subcellular fractionation and extra-nuclear/nucleus volumes. Following the MIRD formalism, [ 99m Tc]TcO 4 − was assumed to accumulate uniformly across all cells, and the exponential growth of cells with time was taken into account, as well as the technetium-99m decay. Dose-response curves were fitted to the LQ-model using GraphPad Prism (v9.1.0).

Animal Tumour Model
Animal procedures were performed in accordance with UK legislation (PPL 70/8879) and approved by King's College London AWERB. MDA-MB-231.hNIS-GFP xenografts were established by injecting 1 × 10 6 cells into the left mammary fat pads of young adult female NSG mice. Twenty-eight days later, animals were randomly assigned to experimental groups (n = 2/cohort, 3 cohorts).

Ex Vivo Analysis of Tissue Radioactivity
Relevant tissues were harvested and gamma-counted. Data were expressed as %IA/g tissue. Cohorts that had received 20 or 200 MBq radiotracers were first subjected to imaging, while the 125 MBq cohort was sacrificed 24 h after radiotracer administration.

Statistics
Three independent experiments were performed for all in vitro studies described, and data were represented as mean ± standard deviation. Graphs were plotted using GraphPad Prism v9.1.0. Statistical significance was calculated with either one-way ANOVA with Dunnett's multiple comparisons test or two-way ANOVA with Tukey's multiple comparisons test (5% significance level) unless stated otherwise.

Conclusions
This study in an epithelial breast cancer model informs on technetium-99m-induced in vitro DNA damage and reduction of cell survival. It reveals that tumour retention of [ 99m Tc]TcO 4 − needed to be sufficiently long to cause damage if technetium-99m was internalised into cancer cells. Dosimetry suggested enhanced radiobiological effectiveness of technetium-99m when internalised into cells compared to EBRT. While further radiobiological investigations with different technetium-99m radiotracers are warranted, the values provided here will serve as a reference to develop future AE-therapy with technetium-99m developers and to assess the radiobiological risk from technetium-99m based imaging agents.