Targeted Radionuclide Therapy Using Auger Electron Emitters: The Quest for the Right Vector and the Right Radionuclide

Auger electron emitters (AEEs) are attractive tools in targeted radionuclide therapy to specifically irradiate tumour cells while sparing healthy tissues. However, because of their short range, AEEs need to be brought close to sensitive targets, particularly nuclear DNA, and to a lower extent, cell membrane. Therefore, radioimmunoconjugates (RIC) have been developed for specific tumour cell targeting and transportation to the nucleus. Herein, we assessed, in A-431CEA-luc and SK-OV-31B9 cancer cells that express low and high levels of HER2 receptors, two 111In-RIC consisting of the anti-HER2 antibody trastuzumab conjugated to NLS or TAT peptides for nuclear delivery. We found that NLS and TAT peptides improved the nuclear uptake of 111In-trastuzumab conjugates, but this effect was limited and non-specific. Moreover, it did not result in a drastic decrease of clonogenic survival. Indium-111 also contributed to non-specific cytotoxicity in vitro due to conversion electrons (30% of the cell killing). Comparison with [125I]I-UdR showed that the energy released in the cell nucleus by increasing the RIC’s nuclear uptake or by choosing an AEE that releases more energy per decay should be 5 to 10 times higher to observe a significant therapeutic effect. Therefore, new Auger-based radiopharmaceuticals need to be developed.


Introduction
Targeted radionuclide therapy (TRT) is an attractive approach to treat cancer because it allows the specific irradiation of tumour cells. As, theoretically, TRT spares healthy tissues that do not express high levels of the targeted receptor, it represents a method of choice for treating diffuse and metastatic disease [1,2]. Historically, radiopharmaceuticals administered to treat patients have been based on beta-particle emitters, such as iodine-131 ( 131 I) in thyroid carcinoma and phosphorus-32 ( 32 P) in ovarian cancer. Later, [ 90 Y]Y-ibritumomab tiuxetan (Zevalin) and [ 131 I]I-tositumomab (Bexxar) were approved for radioimmunotherapy of non-Hodgkin B lymphoma. More recently, [ 177 Lu]Lu-DOTATATE was also approved for therapy of neuroendocrine tumours [3], [ 177 Lu]Lu-PSMA-617 for metastatic castration-resistant prostate cancer [4], and many clinical trials are currently assessing this radiopharmaceutical. Beta-particle emitters are largely available, at reasonable costs, and they are easy to chelate. However, although their relatively long range (≥1 mm) can counterbalance heterogeneity in radiopharmaceutical tumour distribution, it can also cause non-specific irradiation of healthy cells/tissues, thus limiting their interest for TRT. Moreover, beta particles are low linear energy transfer (LET) particles (like X and gamma rays). This means that they have a low ionising power and produce simple lesions in cells that can be repaired. Therefore, they can show limited efficacy against radioresistant solid tumours. This can be overcome by using high LET alpha particles, such as those (1.85 energetic photons of 171.28 and 245.35 keV per decay) that can cause non-specific irradiation and raise radiation protection concerns, respectively.
Another difficulty with AEEs is the need to build radioimmunoconjugates (RIC) that can specifically target cancer cells and also drive activity at least into the nucleus, if not into DNA. Indeed, as Auger electrons with high LET have a range of few nm, nuclear localisation without incorporation into DNA would lead to a significant decrease in efficacy, although the results might still be acceptable.
Herein, we assessed two new 111 In-RICs based on the anti-HER2 antibody trastuzumab functionalised with two cationic peptides (NLS and TAT) harbouring a nuclear localising sequence. Pioneering groups have shown that these peptides allow the nuclear transportation of the functionalised antibodies in AEE-based TRT settings [11,26,27]. Although, other authors highlighted the need for alternative approaches to improve their delivery [28][29][30], none of these approaches has been evaluated beyond the preclinical stage. Moreover, only one study investigated [ 111 In]In-DTPA-human epidermal growth factor in patients [31] but did not confirm the promising results obtained in animal models [28]. In this study, we evaluated whether NLS-/TAT-immunoconjugates are suitable vectors to specifically deliver AEE-based TRT to the cancer cell nucleus. We hypothesised that the 111 In-trastuzumab RIC specifically targets HER2-positive cancer cells before internalization in the cell cytoplasm. When conjugated to NLS or TAT peptides, the nuclear localisation signal should drive this RIC through the nuclear pore complex to enhance Auger electron cytotoxicity. To test this hypothesis, we thoroughly investigated the subcellular localisation, activity uptake, and cytotoxic effects of the developed 111 In-RICs.

Cell Lines
A-431 vulvar squamous carcinoma and SK-OV-3 ovarian carcinoma cells were obtained from the American Type Culture Collection (ATCC, Rockville, MD, USA). HER1and HER2-expressing A-431 cells were transfected with the genes encoding CEA and luciferase to obtain the A-431 CEA-luc cell line, as previously described [14]. HER1-and HER2-expressing SK-OV-3 cells were transfected with the gene encoding CEA to obtain the SK-OV-3 1B9 cell line. All cell lines were grown in DMEM supplemented with 10% heatinactivated foetal bovine serum, 1% penicillin/streptomycin, and 200 µg/mL geneticin. Hygromycin (100 µg/mL) was added to the medium for A-431 CEA-luc cells. Cell lines were routinely tested for mycoplasma contamination using the MycoTect assay from Life technologies (Thermo Fisher Scientific, Waltham, MA, USA).
The 16-mer TAT peptide [GRKKRRQRRRPPQGYG] harbouring the nuclear localisation sequence (underlined) of the HIV-1 tat protein was synthesised and provided by Genscript Biotech (Leiden, The Netherlands) with a purity ≥ 90%. The 13-mer Simian Virus 40 T-Ag-derived NLS peptide [CGYGPKKKRKVGG] (the nuclear localisation sequence is underlined) was synthesised and provided by Proteogenix (Schiltigheim, France) with a purity of 90.40%. Both peptides were redissolved in chelexed PBS (pH 7.2) at a concentration of 7 mg/mL and kept at −80 • C until use.
Matrix-assisted laser desorption ionisation-time of flight (MALDI-TOF) was used to determine the number of DTPA, SMCC, NLS, or TAT moieties per antibody. The m/z difference between the native mAbs and their conjugates was divided by the molecular weight of the added functional group (DTPA, sulpho-SMCC, or peptide) to find the chelate/peptide-to-antibody ratio.

Immunoconjugates Cell Binding and HER2 Cell Expression
Immunoconjugate cell binding was assessed by flow cytometry. Cells were fixed with 4% (v/v) paraformaldehyde in PBS at RT for 15 min, permeabilised with Triton X-100 (0.5% in PBS) at RT for 15 min, and blocked with PBS/bovine serum albumin (BSA) blocking solution (10 mg/mL) at 37 • C for 1 h. Then, cells were incubated with each immunoconjugate (10 µg/mL) at RT for 1 h, followed by incubation with an AlexaFluor-488-conjugated anti-human secondary antibody (1:200; (F(ab ) 2 -goat anti-human IgG (H+L) Cross-Adsorbed, Thermo Fisher Scientific) at RT in the dark for 1 h. Cells were then washed three times and suspended in PBS for analysis using a Gallios cytometer (Beckman Coulter, Villepinte, France).

RIC Stability
A 100 µL aliquot of each radiopharmaceutical was added to 900 µL of culture medium or human serum. Then, samples were incubated on a shaker at 37 • C. At the desired time points, the radiopharmaceutical stability was evaluated by iTLC using a 50 mM EDTA (pH 5.5) solution as eluant. The retention factor (Rf) of the RIC was 0 and that of free indium was 1. Each experiment was repeated twice and the percentage of radionuclides bound to proteins for each RIC was calculated after measurement with a γ-ray counter.

Determination of the RIC Immunoreactive Fraction
RIC immunoreactive fractions were assessed using a bead-based assay method described by Sharma et al. [33]. Briefly, 20 µL of the magnetic HisPur Ni-NTA bead slurry (Thermo Scientific) was incubated with 1 µg His-Tag human recombinant HER2 protein (Sino Biological). Coated beads were then incubated with 1 ng of RIC in the presence (or not) of unlabelled trastuzumab (blocked condition; ×1000-fold excess) on a shaker at RT for 30 min. Beads were isolated, and the supernatant containing unbound RICs was collected. Next, beads were washed twice with PBS-Tween (0.05%) to remove non-specifically bound RICs. Finally, the activity in each fraction (beads alone, supernatant, and wash) was measured using a γ-ray counter, and the immunoreactive fractions were calculated.

Clonogenic Assays
Cell survival after incubation with 111 In-RICs or unlabelled immunoconjugates was assessed using a standard clonogenic assay. Briefly, a defined number of cells, from 100 to 800, were plated in 6-well plates in 2 mL of fresh culture medium. The next day, cells were incubated with an increasing amount of activities of 111 In-RICs (0 to 4 MBq/mL) or 10.8 µg/mL unlabelled immunoconjugates at 37 • C/5% CO 2 for 48 h, followed by three washes with 2 mL PBS and the addition of fresh medium (2 mL). Then, cells were cultured for 12 days, and colonies were fixed (acetic acid:methanol [1:3]) and stained with Giemsa diluted in H 2 O (1:5; Sigma-Aldrich, Saint-Quentin Fallavier, France). Colonies were manually counted, and the surviving fraction was calculated. All experiments were repeated at least three times in triplicate for each condition.

In Vitro Measurement of DNA Damage
Cells were seeded on coverslips in 6-well plates. The next day, cells were incubated with 4 MBq/mL of HER2-specific or irrelevant 111 In-RICs for 48 h. After RIC removal, cells were washed with PBS three times and fixed with 4% (v/v) paraformaldehyde in PBS at RT for 15 min, permeabilised with Triton X-100 (0.5% in PBS) at RT for 15 min, and blocked with PBS/BSA blocking solution (10 mg/mL) at 37 • C for 1 h. Next, cells were incubated with an anti-phosphorylated histone H2AX (γH2AX) antibody diluted in PBS (1:200; 05-636, Sigma Aldrich, Saint-Quentin Fallavier, France) at 37 • C for 1 h, followed by washing and incubation with a goat anti-mouse IgG (H+L) FITC-conjugated antibody diluted in PBS/BSA (1:400; 12-506, Sigma-Aldrich, Saint-Quentin Fallavier, France) at 37 • C for 1 h. Finally, cells were washed with PBS/Tween-20 (0.1%) and then with PBS (three times/each). Coverslips were mounted on slides with Vectashield antifade mounting medium and DAPI (Vector laboratories, Les Ulis, France) and imaged using a ×63 or ×40 objective on a Leica fluorescence inverted microscope (Leica Microsystems, Wetzlar, Germany), as described previously [15].

Radioactivity Uptake and Cell Fractionation
To isolate nuclei, the chaotropic method described by Pouget et al. [34] was used. Briefly, 1 × 10 6 cells grown in T25 flasks overnight were incubated with 0.1 MBq/mL (3 mL) 111 In-RICs for 48 h. At various time-points (1, 2, 3, and 6 days), cells were trypsinised and rinsed with PBS three times to remove unbound RICs. The activity of isolated whole cells was determined with a γ-ray counter (Hidex Automatic gamma counter, Villebonsur-Yvette, France). Next, cells were resuspended in 1 mL lysis buffer (320 mM sucrose, 5 mM MgCl 2 , 10 mM Tris-HCl, 0.1 mM deferoxamine, and 1% Triton X-100 in H 2 O) and centrifuged at 1500× g for 10 min. The supernatant (cytoplasm + membrane) was collected. The pellet was again resuspended in 1 mL lysis buffer, centrifuged at 1500× g for 10 min, and the supernatant was collected. The pellet (cell nuclei) was rinsed with 1 mL PBS three times to remove all traces of cytoplasm or membrane. Finally, the activity present in both fractions (nuclei and cytoplasm + membrane) was measured using a γ-ray counter. The cumulative number of decays (Bq.s) per cell over 6 days was then calculated as well as the cumulative number of decays (Bq.s) per fraction (nucleus and cytoplasm + membrane).

Evaluation of Nuclear Fraction Contamination by Western Blotting
Nuclei were rinsed twice with PBS and lysed in RIPA buffer (Santa Cruz; Santa Cruz, CA, USA) at 4 • C for 30 min. The total protein concentration in the nuclear fraction was determined using the BCA Protein Assay Reagent (BCA Assay Kit, Thermo Fisher scientific, Illkirch, France). Then, 6 µg of nuclear proteins were separated by SDS-PAGE (12% polyacrylamide gel) and electrotransferred onto nitrocellulose membranes (Bio-Rad, Grabels, France). Membranes were incubated with anti-α-tubulin (11H10) (1/1000; Cell Signaling Technology, Leiden, The Netherlands) and anti-phosphorylated histone 3 (Ser10) antibodies (1/1000; Cell Signaling Technology, Leiden, The Netherlands) diluted in Tris-buffered saline/BSA (5%). A horseradish peroxidase-conjugated anti-rabbit IgG (Cell Signaling Technology, Danvers, MA, USA) was used as a secondary antibody. Proteins were revealed using an enhanced chemiluminescence system according to the manufacturer's instructions (Clarity, Bio-Rad, Marnes-La-Coquette, France). Protein expression was imaged with a PXi analyser (Ozyme, St Quentin en Yvelines, France).

Immunofluorescence Analysis of the Immunoconjugate Subcellular Localisation
As described in Section 2.8, cells were plated on coverslips in 6-well plates, and, the day after, they were incubated with different unlabelled immunoconjugates for 48 h. Then, cells were fixed and permeabilised (see Section 2.8). After blocking with PBS/BSA, an AlexaFluor-488-conjugated anti-human secondary antibody (1:200; (F(ab ) 2 -goat antihuman IgG (H+L) Cross-Adsorbed, Thermo Fisher Scientific) was added at 37 • C in the dark for 1 h. Coverslips were then mounted with Vectashield with DAPI (see Section 2.8) and imaged using a ×63 or ×40 objective and a Leica fluorescence inverted microscope.

Energy Deposition Calculations
The energy deposition of Auger electrons, CEs, gamma-and X-rays in spheres was evaluated using the PENELOPE code system [35] that simulates the interactions of radiations as they pass through the medium, in this case water. The detailed energy spectra of the radiations were generated with the BrIccEmis [36,37] code and typically using 5 million decay events. All simulations were done using the Monte Carlo approach.

Statistical Analysis
Data were analysed using GraphPad Prism version 8.4.3 (686). Data were described using means and standard deviation (SD). For gamma-H2AX, data were described using means and the standard error of the mean (SEM). Data were compared using the Student's t test. A p value < 0.05 was considered significant. The Bliss independence model was used to determine the percentage of specific irradiation. In this method, the global efficacy of [ 111 In]In-trastuzumab was dissociated in specific and non-specific efficacy using the following formula, wherein the non-specific efficacy was determined as [ 111 In]In-IgG efficacy:
The in vitro stability of RICs was assessed by incubation in cell-culture medium and human serum at 37 • C. Results showed excellent stability with percentages of 111 In bound to protein > 82% over 72 h (Figure 2a). Flow cytometry analysis confirmed that HER-2 expression was higher in SK-OV-3 1B9 cells (G-mean: ×116 compared to control) than in A-431 CEA-luc cells (G-mean: ×4.93 compared to control) (Figure 2b). Immunoreactivity assay confirmed that the binding of NLS-or TAT-immunoconjugates was predominantly non-specific because [ 111 In]In-IgG-NLS 5-10 and [ 111 In]In-IgG-TAT 1-3 bound to HER2-coated beads. Moreover, the binding of [ 111 In]In-trastuzumab-NLS 5-10 and [ 111 In]In-trastuzumab-TAT 1-3 was not strongly blocked in the presence of a large excess of unlabelled trastuzumab compared with [ 111 In]Intrastuzumab (Figure 2c, left panel). Flow cytometry analysis of the immunoconjugate cell binding also showed that trastuzumab functionalisation with NLS 5-10 or TAT 1-3 peptides resulted in an increase of the fluorescence signal (G-mean: 104,042, 506,000, and 265,795 for trastuzumab, trastuzumab-NLS 5-10 , and trastuzumab-TAT 1-3 , respectively), indicating higher cell binding. However, this effect was predominantly non-specific because it was observed also after functionalisation of the non-specific IgG mAb (Figure 2c, right panel).
Energy deposit calculations indicated that 111 In emitted 7.19 Auger electrons per decay with a mean energy of 0.94 keV released over a maximal range of about 12 µm. It also emitted 0.16 CEs per decay, which corresponded to a mean energy of 176 keV (maximum range of 573 µm). The energy deposition in a sphere of 20 µm, which corresponds roughly to the diameter of a SK-OV-3 1B9 cell, was about 5.97 keV. This increased up to 37 keV for a sphere of 100 µm in diameter. 111 In also emitted 1.85 energetic photons of 150.8 and 245.3 keV per decay, depositing only about 0.114 keV in a 100 µm diameter sphere that did not contribute to cell killing in vitro. Therefore non-specific cytotoxicity could be due to CEs co-emitted by 111 In.

Functionalisation with NLS 5-10 Is Associated with High and Moderate Non-Specific Cytotoxicity
On the basis of the hypothesis that the non-specific cytotoxicity of [ 111 In]In-trastuzumab-NLS 17-22 could be due to the mAb's high degree of functionalisation with NLS peptides,

[ 111 In]In-Trastuzumab-NLS 5-10 Is Associated with the Highest Activity Uptake per Cell or Nucleus
A nucleus-isolation technique was used to determine the activity incorporated into the cell nucleus and the activity remaining in the extranuclear fraction (cell cytoplasm and membrane). As these experiments required high cell numbers and large volumes, the activity of 0.1 MBq/mL was chosen. Western blot analysis showed that α-tubulin (marker of the cytoplasmic fraction) was not or barely detectable, whereas histone H3 (nuclear fraction marker) was strongly expressed in the nuclear fraction of both cell lines (Figure 4a), validating our nucleus-isolation technique (<4% contamination from the cytoplasmic fraction).
Then, the nuclear and extranuclear activity curves showed a similar trend in both cell lines, with a progressive increase and maximal uptake values at 24 or 48 h after RIC addition ( Figure 4b). As HER2 receptor expression is higher in SK-OV-3 1B9 cells (Figure 2b), uptake was approximately one log higher in this cell line with a slower activity decrease at 72 h (i.e., 24 h after radioactivity removal) (Figure 4b).
From these curves, the cumulative number of decays (Ã) occurring in the whole cell (Ã cellular ), in the cytoplasm + membrane (Ã extranuc ), and in the nucleus (Ã nuclear ) over 6 days was determined (Figure 4c). In the absence of a specific peptide sequence targeting the cell nucleus, Ã nuclear and Ã cellular for [ 111 In]In-trastuzumab were 128 and 898 Bq.s in A-431 CEA-luc cells and 10 times higher or more (1503 and 12,704 Bq.s) in SK-OV-3 1B9 cells. The Ã cellular values for [ 111 In]In-IgG were <59 Bq.s in A-431 CEA-luc and <410 Bq.s in SK-OV-3 1B9 cells. The addition of NLS 5-10 or TAT 1-3 sequences globally improved RIC subcellular uptake (in all compartments) and in both cell lines with a more pronounced effect from NLS 5-10 than from TAT 1-3 . HER2 expression level (high versus low) influenced the Ã cellular values, and also the Ã nuclear values, which were multiplied by 6.1 in A-431 CEA-luc (low expression; trastuzumab versus trastuzumab-NLS 5-10 ) and only by 2.0 in SK-OV-3 1B9 (high expression; trastuzumab versus trastuzumab-NLS 5-10 ).

TAT and NLS Peptides Do Not Increase the Percentage of Activity Reaching the Nucleus
Although, the raw Ã extranuc and Ã nuclear values increased with the addition of NLS and TAT peptides, the ratio between these values did not highlight any specific increase in nuclear targeting (Figure 5a). The proportion of radioactivity in the nucleus remained between 6 and 18% for all RICs, indicating that nuclear targeting was not specific. It must be noted that the nuclear proportion of [ 111 In]In-IgG-TAT 1-3 was higher (37% and 40% in A-431 CEA-luc and SK-OV-3 1B9 cells, respectively), but this was mainly due to low cellular uptake such that the percentage was increased. The same conclusion can be drawn for [ 111 In]In-IgG in A-431 CEA-luc (Figure 5a).

TAT and NLS Increase the Non-Specific Cellular Uptake
The addition of NLS 5-10 or TAT 1-3 resulted in an overall increase of cellular, and subsequently nuclear, uptake of the non-specific IgG RIC (Figures 4c and 5 -trastuzumab). The HER2-specific/non-specific Ã cellular ratios for NLS 5-10 were 1.4 in A-431 CEA-luc and 1.5 in SK-OV-3 1B9 cells. The HER2-specific/non-specific Ã cellular ratios for TAT 1-3 were 1.9 in A-431 CEA-luc and 3.7 in SK-OV-3 1B9 cells.

Immunoconjugates Do Not Localise in the Nucleus
Although less sensitive than radioactive detection, the localisation of the immunoconjugates was next assessed by immunofluorescence analysis (Figure 6). DTPA-trastuzumab was detected at the cell surface and in the cytoplasm. The amount of foci in the cytoplasm increased for DTPA-trastuzumab-NLS 5-10 and DTPA-trastuzumab-TAT 1-3 . Conversely, DTPA-IgG mAb was not detected at the cell surface or in the cytoplasm. However, when functionalised with NLS 5-10 or TAT 1-3 , the number of cytosolic DTPA-IgG mAb foci increased. Few foci of NLS-and TAT-functionalised mAbs were detected in the nucleus, but their number was much lower than in the cytoplasm.

Discussion
This study assessed the ability of two highly cationic peptides, the synthetic 13-mer NLS peptide [CGYGPKKKRKVGG] from the simian virus 40 large-T antigen and the synthetic 16-mer TAT peptide [GRKKRRQRRRPPQGYG] from the HIV-1 TAT protein, to drive the 111 In-anti-HER2 (trastuzumab) RIC into the cell nucleus of two HER2-expressing cancer cell lines, A-431 CEA-luc (HER2 + ) and SK-OV-3 1B9 (HER2 ++ ). According to the literature, after HER2-mediated internalisation in the cytoplasm, NLS and TAT peptide sequences are recognised by importin-α and importin-β1 to form a nuclear-pore-targeting complex [38,39]. This complex facilitates RIC passage through the nuclear pore and their translocation into the nucleus where they will be dissociated to release NLS-/TAT-conjugates [11]. In total, eight RICs were obtained and characterised: First, functionalisation of trastuzumab with the TAT or NLS peptides slightly increased its cell binding, but this binding was at least predominantly mediated by non-specific mechanisms. Moreover, non-radioactive immunoconjugates showed very low cytotoxicity, even when used at the highest concentration (10.8 µg/mL; equivalent to tested volumic activity of 4 MBq/mL for 111 In-RIC) (Supplementary Figure S1).
Then, comparison of the different RICs showed that [ 111 In]In-IgG was significantly less cytotoxic than [ 111 In]In-trastuzumab, but more than what was reported in previous studies. The origin of 111 In non-specific cytotoxic effect in vitro could be explained by CE emission. It must be noted that, in several studies showing the absence of non-specific cytotoxicity, clonogenic assays were performed using a protocol different from the one followed in the present study. This previous protocol included two consecutive steps. First, cells were exposed to the radionuclide for several hours before centrifugation and washes. Then, cells were seeded at low concentration for clonogenic assay. It is likely that the first step might have led to the loss of the most damaged cells, and therefore, cell survival was only measured for the most viable cells that had not been eliminated by washing and that could adhere to the flask bottom. Moreover, the non-specific irradiation contribution could be different in vivo, particularly considering that, in the present in vitro experiments, cells were incubated with RICs at high activity for 48 h, a situation that might not reflect the in vivo pharmacokinetics of 111 In-RICs.
Functionalisation with TAT 1-3 peptides did not significantly modify [ 111 In]In-trastuzumab cytotoxicity in A-431 CEA-luc cells and slightly increased it in SK-OV-3 1B9 cells (p < 0.01). The observation that the non-specific [ 111 In]In-IgG-TAT 1-3 was as cytotoxic as [ 111 In]Intrastuzumab-TAT 1-3 indicates that TAT conjugation is associated with the loss of mAb specificity. The lack of or limited additional cytotoxicity provided by functionalisation with TAT 1-3 could be explained by the low increase of activity localisation in the nucleus (Ã nuclear ) after cell exposure to [ 111 In]In-trastuzumab-TAT 1-3 . Ã cellular increase in A-431 CEA-luc and in SK-OV-3 1B9 cells exposed to [ 111 In]In-trastuzumab-TAT 1-3 was associated with nonspecificity because it also increased in cells incubated with [ 111 In]In-IgG-TAT 1-3 . However, the Ã cellular increase in SK-OV-3 1B9 cells was lower than in A-431 CEA-luc cells, suggesting that the higher HER2 expression level of this cell line prevented the passive internalisation of the TAT-RIC.
Functionalisation of radiolabelled mAbs with 17 to 22 NLS/mAb (NLS [17][18][19][20][21][22] ) resulted in a drastic reduction of the surviving fractions for both [ 111 In]In-trastuzumab-NLS [17][18][19][20][21][22] and [ 111 In]In-IgG-NLS [17][18][19][20][21][22] . Again the addition of NLS peptides was associated with a loss of specificity. Reducing the number of NLS to 5-10 (NLS 5-10 ) increased RIC specificity and maintained high cytotoxicity. Nevertheless, even functionalisation with NLS 5-10 was associated with non-specific cytotoxicity in A-431 CEA-luc cells because [ 111 In]In-IgG-NLS 5-10 was more cytotoxic than [ 111 In]In-trastuzumab (p = 0.001), but the difference between the specific and non-specific [ 111 In]In-mAb-NLS 5-10 effects was more pronounced than for [ 111 In]In-mAb-TAT 1-3 . Cell fractionation experiments showed that functionalisation with NLS 5-10 was accompanied by an increase of the Ã nuclear values for both [ 111 In]Intrastuzumab-NLS 5-10 and [ 111 In]In-IgG-NLS 5-10 in A-431 CEA-luc cells and, to a lower extent, in SK-OV-3 1B9 cells. Nevertheless, the most striking observation was that the functionalisation of [ 111 In]In-trastuzumab with NLS 5-10 significantly increased Ã cellular in both A-431 CEA-luc (× 4.7) and in SK-OV-3 1B9 (× 1.8) cells, with a similar trend also for [ 111 In]In-IgG (×52.5 in A-431 CEA-luc cells; ×37.2 in SK-OV-3 1B9 cells), highlighting again the lack of specificity of the NLS-mediated uptake. It has been proven that NLS or TAT peptides pass through plasma membranes in a non-receptor-dependent manner thanks to their highly cationic properties. In the context of our study, NLS or TAT peptides' addition to mAbs will generate a highly cationic macromolecule that will passively penetrate cells or accumulate onto negatively charged cell-surface proteins. Our results showed an increase of cell uptake when the mAb is conjugated with NLS or TAT peptides suggesting a correlation with the addition of NLS or TAT.
Moreover, the Ã nuclear versus Ã extranuc ratio with [ 111 In]In-trastuzumab-NLS 5-10 was similar to that obtained with [ 111 In]In-trastuzumab, suggesting that the NLS peptide did not improve the nuclear translocation of 111 In-RIC once internalised in the cytoplasm (Figure 4). These results were supported by the fluorescent detection of immunoconjugate foci, mainly in the cytoplasm, and of very few NLS-and TAT-functionalised immunoconjugate foci in the nucleus. This low nuclear localisation might be due to endosomal-lysosomal entrapment, leading to RIC hydrolysis before they can reach the nucleus and might constitutes a key limiting factor for bringing AEEs into the nucleus. Several groups have developed attractive methods to promote endosomal escape. For instance, in hetero-functional constructs, a target-specific antibody is conjugated to cholic acid for endosomal escape and an NLS peptide is added for nuclear targeting [30,40]. Moreover, modular nanotransporters consist of a target-specific module, a diphtheria toxin translocation domain as endosomolytic module, an NLS peptide for transport to the nucleus, and the Escherichia coli haemoglobin-like protein as a carrier module [28,40] [28,41]. These approaches should be further investigated to bypass Auger electron-carrier endosomal entrapment.
At this stage it was not possible to thoroughly investigate the relationship between clonogenic survival and activity uptake in the different cell compartments. Such study would require a comprehensive dosimetric approach (out of the scope of this study) and an assessment of the possible contribution of bystander effects. Moreover, it is important to keep in mind that the cell fractionation experiments were done using activities of 0.1 MBq/mL, and therefore, the only relationship that could be investigated would be with the clonogenic survival data measured at 0.5 MBq/mL.
Interestingly, in a previous study [15], we showed that, in HCT116 cells exposed to [ 125 I]I-UdR, Ã nuclear was about 441 Bq.s and Ã cellular was 455 Bq.s. Using the Bliss model and considering that cytotoxicity was only due to radioactivity in the nucleus, we calculated that 80% of cell killing was due to 125 I located in the DNA [15], while here it was 20% (at 0.5 MBq/mL, a test activity close to the one used for determining the Ã values and for which non-specific irradiation can be neglected) but with about 1.8 (786/441) more decays. This would mean that one decay of 125 I is 7.2 times more efficient than one decay of 111 In. If we consider that the energy deposited per decay and due to Auger electrons is about 8.7 keV ( 125 I) and 5.4 keV ( 111 In) in a 5 µm diameter sphere, then 125 I deposits about 1.6 times more "high LET" energy than 111 In. Moreover, a correction factor also needs to be introduced to counterbalance the localisation differences (nuclear but not bound to DNA versus nuclear and bound to DNA). As a relative biological effectiveness value of seven has been proposed when 125 I is bound to DNA [42,43] and of four [7] when unbound, then the factor that needs to be introduced here is 1.7. Additionally, HCT116 cells are about 1.8 times more radiosensitive than A-431 cells at a dose of 2 Gy [44]. In conclusion, 111 In incorporated in our RICs should be 8.8 times (2.1 × 1.6 × 1.7 × 1.8) less efficient than [ 125 I]I-UdR, a value that can be compared to the value of 7.2 found above.
The preliminary analysis of DNA damage yield (γH2AX foci) in A-431 CEA-luc cells exposed to 4 MBq/mL RIC showed that the highest levels of γH2AX foci were observed with [ 111 In]In-trastuzumab-NLS 5-10 and [ 111 In]In-IgG-NLS 5-10 , which are associated with the highest cytotoxicity and the highest Ã nuclear and Ã cellular values, although the transient DNA DSB yield cannot be strictly correlated with cell survival. As DNA breaks can be considered to be produced mostly by radiation, we investigated the relationship between the cumulative number of γH2AX foci measured over 144 h and the Ã nuclear and Ã cellular values. However, it must be kept in mind that DNA DSBs can also be produced by radioactivity contained in the culture medium or in neighbouring cells, two parameters not considered here. Moreover, the efficacy of DNA repair occurring at the same time as irradiation [16] could be influenced by the nature of DNA DSBs induced by radioactivity localised in or outside the nucleus. Despite these approximations, we found a relationship between the cumulative number of DSBs and the Ã nuclear and Ã cellular values, although none of these parameters seemed to fit the data best. It must be also noted that some DNA DSBs can be produced by a delayed mechanism involving cell membrane irradiation. We showed previously that Auger electron-mediated irradiation of the cell membrane induces lipid raft formation and the subsequent activation of signalling pathways, leading to the formation of ROS via NF-kB and DNA damage [15][16][17]. We could hypothesise that the high Ã extranuc value of [ 111 In]In-trastuzumab-NLS 5-10 (2412 Bq.s) is related to HER2 binding at the cell membrane, which is not possible with [ 111 In]In-IgG. However, this hypothesis needs to be further assessed because we do not know whether enough energy is deposited at the cell membrane.

Conclusions
In this study, we showed that the functionalisation of trastuzumab with NLS 5-10 or TAT 1-3 was accompanied by a rather limited increase in nuclear activity level and therefore, without a drastic decrease in clonogenic survival after exposure to 111 In-RIC. We obtained similar results with the functionalised non-specific IgG, suggesting that the process is associated with a loss of specificity. This loss of specificity was stronger with TAT 1-3 than NLS [5][6][7][8][9][10] . This low nuclear uptake might be due to endosomal-lysosomal entrapment, leading to RIC hydrolysis before they reach the nucleus, and alternative methods to favour endosomal escape are required. Our results also indicate that the non-specific toxicity of 111 In can be a limiting factor for AEE-based TRT, but this needs to be further assessed in vivo in mice for the reasons mentioned above. The comparison with [ 125 I]I-UdR indicates that the amount of energy released in the nucleus by modulating activity uptake or by choosing a better AEE should be multiplied by 10 to obtain significant cell killing (e.g., 99%).