Molecular Radiotherapy Using Cleavable Radioimmunoconjugates That Target EGFR and γH2AX

DNA double-strand breaks (DSB) are highly deleterious and their formation is a major determinant of the cytotoxicity of radiation therapy and several widely used anticancer chemotherapeutic agents (1). An early event following the formation of DSBs is the phosphorylation of the X isoform of histone H2A at serine-139 by the phosphoinositide 3-kinases (PI3K; DNA-PKcs, ATM, and ATR), resulting in γH2AX (1, 2). Many hundreds of copies of γH2AX form foci at DSB sites. We previously identified γH2AX as an attractive target for molecular imaging and quantification of DSBs in vivo (3). This exclusively nuclear protein can be targeted using anti-γH2AX antibodies, conjugated to the cell-penetrating peptide (CPP), TAT, which also harbors a nuclear localization sequence (NLS), to allow cellular internalization and nuclear translocation. Anti-γH2AX-TAT was then conjugated to fluorophores or the radionuclide ¹¹¹In to allow fluorescence microscopy or single-photon emission computed tomographic (SPECT) imaging, respectively.

Apart from being a target for imaging DSBs, we have shown that γH2AX is also a good target for Auger electron radiotherapy (4). Auger electrons are low-energy electrons (1–50 eV) with very limited track length (in the nm to μm range) that, when emitted in close proximity of DNA, are highly damaging (5). When anti-γH2AX-TAT was radiolabeled with the Auger electron emitter ¹¹¹In at high specific activity (SA), it caused amplification of DNA damage induced by chemotherapy and ionizing radiation (IR), reduction of clonogenic survival in vitro, and inhibition of tumor growth in vivo. However, none of the constituent parts of anti-γH2AX-TAT ensure tumor-specific uptake of the conjugate, and any nonspecific internalization in normal cells would be undesirable as it may cause toxicity.

Here, we report on two ¹¹¹In-labeled radioimmunoconjugates that target not only γH2AX but also through the incorporation of the EGF peptide, the EGF receptor (EGFR), which is frequently overexpressed in cancer. For example, MDA-MB-468 human breast cancer cells express 1.3 × 10⁶ EGFR/cell (6), but it is expressed to a much lower extent in normal tissues, for example, normal hepatocytes express 8 × 10⁴ to 3 × 10⁵ EGFR/cell (7). These dual-specific immunoconjugates incorporate the NLS of SV-40 large T-antigen to promote nuclear accumulation of the γH2AX seeking moiety. Two different synthetic strategies were pursued. First, EGF conjugated to NLS was covalently bound to anti-γH2AX antibody via a noncleavable polyethylene glycol (PEG) linker. In addition, EGF was linked to NLS-conjugated anti-γH2AX antibody via a cleavable disulfide bond (8). We report on the in vitro and in vivo characteristics of both compounds and show the potential of the ¹¹¹In-labeled cleavable compound as an Auger electron therapeutic radiopharmaceutical.

MDA-MB-468 (breast cancer) and SQ20b (head and neck cancer) cells were obtained from Cell Services, CR-UK London Research Institute. MDA-MB-231 human breast cancer cells, stably transfected with the HER2 gene, yielding MDA-MB-231/H2N cells (hereafter referred to as 231-H2N) were a gift from Robert Kerbel (Sunnybrook Health Sciences Centre, Toronto, ON; ref. 9). These cells were tested and authenticated by the provider, using short tandem repeat profiling. The length of time in culture of these cells was less than 6 months after retrieval from liquid nitrogen storage. The number of EGFR/cell for MDA-MB-468 and 231-H2N cells is 1.3 × 10⁶ and 2 × 10⁵, respectively (10). Cells were cultured in 5% CO₂ in Dulbecco's Modified Eagle Medium (DMEM) cell culture medium (Sigma-Aldrich) supplemented with 10% fetal calf serum (Invitrogen), and 100 units/mL penicillin/streptomycin (Invitrogen).

A schematic overview of the synthesis of noncleavable immunoconjugates is shown in Supplementary Fig. S1A. A full description of the synthesis is available in the Supplementary Methods. Anti-γH2AX antibody or nonspecific IgGs from rabbit serum was conjugated to benzyl-DTPA (BnDTPA) to allow ¹¹¹In-labeling or to Cy3 or AlexaFluor555 (AF555) to allow fluorescence microscopy. This was reacted with (SM(PEO)₆), yielding maleimide-activated conjugates, BnDTPA-anti-γH2AX-(PEO)₆-mal or BnDTPA-rIgG-(PEO)₆-mal. EGF was conjugated to the NLS peptide, GGPKKKRKVGYGCG, using EDC/NHS chemistry, yielding EGF-NLS. EGF-NLS was conjugated to BnDTPA-anti-γH2AX-(PEO)₆-mal or BnDTPA-rIgG-(PEO)₆-mal. The size of BnDTPA-anti-γH2AX-PEO₆-NLS-EGF was determined using polyacrylamide gel electropheresis (PAGE) followed by Coomassie staining. ¹¹¹In labeling was achieved by addition of ¹¹¹In-chloride (¹¹¹InCl₃; Perkin Elmer). For brevity and because BnDTPA was used as the radiometal chelator for all radioimmunoconjugates reported here, BnDTPA-anti-γH2AX-PEO₆-NLS-EGF and BnDTPA-rIgG-PEO₆-NLS-EGF are referred to hereafter as anti-γH2AX-PNE and rIgG-PNE, respectively.

A schematic overview of the synthesis of cleavable immunoconjugates is shown in Supplementary Fig. S1B. A full description of the synthesis is available in the Supplementary Methods. Anti-γH2AX antibody or nonspecific IgGs from rabbit serum was conjugated to BnDTPA to allow ¹¹¹In labeling or to Cy3 or AF588 to allow fluorescence microscopy. This was reacted with SANH for addition of a hydrazine moiety, and subsequently with an N-terminal serine-containing NLS-peptide (SGGPKKKRKVGYGCG) and sodium periodate. EGF was modified with a disulfide function using SMPT. EGF-SMPT was added to BnDTPA- or fluorophore-tagged IgG-NLS, yielding BnDTPA-anti-γH2AX-NLS-SS-EGF and BnDTPA-rIgG-NLS-SS-EGF (referred to hereafter as anti-γH2AX-N-SS-E and rIgG-N-SS-E). ¹¹¹In labeling and determination of radiolabeling yield was achieved as described (Supplementary Methods). The size of BnDTPA-anti-γH2AX-NLS-SS-EGF was determined using PAGE followed by Coomassie staining. To study the influence of the NLS sequence, the same conjugate was prepared using a disrupted NLS sequence (dNLS, SGGPGGKRKVGYGCG), denoted as ¹¹¹In-anti-γH2AX-dN-SS-EGF.

EGFR and γH2AX binding by anti-γH2AX-PNE and anti-γH2AX-N-SS-E were compared with that of unmodified EGF or anti-γH2AX antibody in competition assays against ¹²³I-EGF or ¹²³I-labeled anti-γH2AX antibody as previously described (10). ¹²³I-labeling of EGF (Peprotech) and anti-γH2AX antibody (Merck) was achieved using the Iodogen method (11). See Supplementary Methods.

To determine whether the disulfide linker was cleavable under reducing conditions, a cell-free assay was conducted using glutathione as the reducing agent. rIgG-N-SS-E was synthesized, as described above, using ¹²³I-EGF in place of EGF, to yield rIgG-N-SS-*E. Size exclusion chromatography (SEC) was conducted on IgG-N-SS-*E using a G50 sephadex minicolumn, and the amount of ¹²³I in serial fractions was measured in an automated γ-counter. IgG-N-SS-*E was exposed to 1 mmol/L glutathione for 2 hours at room temperature, and SEC conducted using a G50 sephadex minicolumn. See Supplementary Methods.

231-H2N, SQ20b, or MDA-MB-468 cells, expressing 2 × 10⁵, 1.04 × 10⁶, or 1.3 × 10⁶ EGFR/cell, respectively, were seeded on coverslips and allowed to adhere overnight. To determine the intracellular distribution of conjugates, cells were exposed to 16 nmol/L AF555-labeled immunoconjugate with or without 16 nmol/L AF488-EGF (Invitrogen), for up to 4 hours at 37°C. In some cases, cells were irradiated (4 Gy) 1 hour after addition of fluorophore-labeled immunoconjugate, after which the growth medium was replaced with fresh medium. Cells were washed twice with PBS, fixed for 10 minutes at room temperature with 4% paraformaldehyde (Sigma), washed again with PBS, permeabilized at room temperature using 1% Triton X-100 in PBS (Sigma), and blocked [1 hour at 37°C; 2% bovine serum albumin (BSA) in PBS]. For γH2AX immunostaining, fixed cells were incubated with anti-γH2AX primary antibody raised in mouse (JBW301; Millipore; 1:800 dilution) for 1 hour at 37°C. Following 3 washes, cells were exposed to goat anti-mouse antibody (Invitrogen; 1:250 dilution) labeled with AF488 or AF555 for 1 hour at 37°C. After 3 washes, coverslips were mounted on slides using Vectashield plus DAPI (Vector Laboratories). Confocal microscopy was conducted on a Zeiss 530 confocal microscope (Zeiss). To determine the distribution of EGFR, cells were fixed and permeabilized as above and stained for EGFR using anti-EGFR antibody (Invitrogen; 1:250 dilution) and AF555-labeled goat anti-mouse antibody (Invitrogen; 1:250 dilution).

To investigate the intracellular distribution of immunoconjugates, aliquots of 2 × 10⁵ MDA-MB-468 or 231-H2N cells in 500 μL of growth medium were exposed to 16 nmol/L ¹¹¹In-anti-γH2AX-N-SS-E or ¹¹¹In-anti-γH2AX-dN-SS-E in 200 μL DMEM (0–6 MBq/μg). At selected times, supernatant was removed from the cells, cells were washed with 0.1 mol/L glycine-HCl, pH 2.5, to remove cell surface–bound radioactivity, the cytoplasmic membrane was lysed (25 mmol/L KCl, 5 mmol/L MgCl₂, 10 mmol/L Tris-HCl, and 0.5% NP-40), and cell nuclei were pelleted and lysed using 0.1 mol/L NaOH, as previously described (12). Radioactivity in cytoplasmic and nuclear fractions was counted in a γ-counter. Retention of ¹¹¹In-labeled compounds in MDA-MB-468 and 231-H2N cells was determined using a load-chase assay as previously described (12). Briefly, cells were seeded in a 24-well plate (2 × 10⁵ cells/well) and left to adhere overnight. Cells were exposed to 0.25 μg/mL of ¹¹¹In-labeled compound for 1 hour, irradiated (4 Gy) or sham-irradiated, and then washed twice with PBS and supplied with fresh culture medium. At selected time points, the amount of ¹¹¹In remaining in cells was determined as previously described (12).

Aliquots of MDA-MB-468 (1.3 × 10⁶ EGFR/cell) or 231-H2N cells (2 × 10⁵ EGFR/cell) were exposed to various concentrations of ¹¹¹In-anti-γH2AX-N-SS-E (0–6 MBq/μg) and ¹¹¹In-rIgG-N-SS-E (0–6 MBq/μg) either alone or in combination with IR. The highest concentration used, 0.5 μg/mL, can be easily achieved in vivo by injection of a 20 g mouse with 10 μg of conjugate. After incubation at 37°C for 24 hours, 2 × 10³ cells were seeded per well in triplicate on a 6-well plate and incubated with fresh growth medium (2 mL). After 7 to 14 days, plates were washed with PBS, and cell colonies were stained with methylene blue (2% methylene blue in water:methanol 1:1) and counted using a Gelcount-automated plate reader (Oxford Optronics).

All animal procedures were carried out in accordance with the UK Animals (Scientific Procedures) Act 1986 and with local ethical committee approval. MDA-MB-468 xenografts were established in female BALB/c nu/nu mice (Harlan). ¹¹¹In-anti-γH2AX-N-SS-E or ¹¹¹In-rIgG-N-SS-E (10 μg; 5 MBq) was administered intravenously (i.v.). γH2AX was induced by X-irradiation of the tumor (10 Gy) 1 hour postinjection (p.i.) using a Gulmay 320 kV X-irradiator; 2.0 Gy/min. Apart from the tumor, the body of the mouse was shielded from radiation using 1-cm-thick brass. The extent of γH2AX expression was shown by immunohistochemistry on sections from excised xenografts, 24 hours after irradiation, processed as previously reported (3). Images were acquired using a confocal microscope as above. Enumeration of foci was conducted manually, on a single 1-μm optical section through the tumor, in at least 100 cells. To show targeting of γH2AX by ¹¹¹In-anti-γH2AX-N-SS-E, tumor sections were stained for γH2AX as before (3), using an antibody raised in mouse and an AF488-conjugated goat anti-mouse secondary antibody. An AF555-conjugated goat anti-rabbit antibody was used to stain the IgG moiety of ¹¹¹In-anti-γH2AX-N-SS-E. In some cases, a 100-fold excess of EGF was co-injected to block EGFR. For SPECT imaging, mice were anesthetized using isoflurane at 24 hours p.i. and SPECT-CT images were acquired using a nanoSPECT-CT scanner (Bioscan). Volume of interest (VOI) analysis on SPECT images was conducted using the Inveon Research Workplace Software Package (Siemens).

All statistical analyses and nonlinear regression were conducted using Graphpad Prism (Graphpad Software Inc.). One- or 2-way ANOVA was used for multiple comparisons. The F test was used to compare parameters between curves.

Two radioimmunoconjugates were synthesized, based on anti-γH2AX antibodies and EGF. The linkage between the antibody and EGF moieties was made either using a noncleavable PEG linker or using a cleavable disulfide bond. All compounds were synthesized so that they contained, on average, a 1:1 ratio of the various moieties. Radiolabeling yield for ¹¹¹In labeling was more than 95%. The affinity for EGFR of EGF-containing conjugates was compared with that of ¹²³I-EGF. Similarly, the affinity of anti-γH2AX–containing conjugates was compared with ¹²³I-anti-γH2AX. IC₅₀ values were not significantly different between the conjugates and native EGF or anti-γH2AX (Supplementary Fig. S2). To show reductive cleavage of the disulfide, rIgG-N-SS-*E was exposed to glutathione (Supplementary Fig. S3). After addition of glutathione, ¹²³I became associated with low-molecular-weight fractions, compared with high-molecular-weight species before addition. This was ascribed to cleavage of the disulfide bond. See Supplementary Methods.

MDA-MB-468 cells were exposed to the fluorophore-labeled, noncleavable anti-γH2AX and EGF-containing immunoconjugate, AF555-anti-γH2AX-PNE and AF488-EGF. A focal pattern of uptake of both fluorophores was observed in the cytoplasm, mainly in the perinuclear region, but not in the nucleus at 1- and 4-hour time points (Fig. 1A). AF555-anti-γH2AX-PNE colocalized closely with AF488-EGF, suggesting that the EGF moiety of AF555-anti-γH2AX-PNE is a major determinant of its intracellular distribution. γH2AX foci were observed in irradiated, but not sham-irradiated MDA-MB-468 cells, at 4 and 24 hours after irradiation (Fig. 1B, green). The number of foci, counted in a single focal plane, in MDA-MB-468 cells was 14.80 ± 0.76 versus 9.14 ± 1.03 γH2AX foci/cell at 4 versus 24 hours after irradiation. In 231-H2N cells, 14.45 ± 0.70 versus 6.12 ± 0.45 γH2AX foci/cell were counted at 4 and 24 hours post IR, respectively. In SQ20b cells, 22.13 ± 0.36 versus 8.54 ± 0.60 γH2AX foci per cells were counted, respectively. Neither AF555-anti-γH2AX-PNE nor AF555-rIgG-PNE were detected in the nuclei of irradiated or unirradiated cells (Fig. 1B, red) and therefore failed to colocalize with γH2AX foci present in irradiated cells. AF555-anti-γH2AX-PNE was not taken up into 231-H2N cells, consistent with their low EGFR density, and did therefore not colocalize with γH2AX foci (Supplementary Fig. S4A).

In contrast to the noncleavable immunoconjugate, AF555-anti-γH2AX-N-SS-E did colocalize with γH2AX foci in the nuclei of irradiated EGFR-positive MDA-MB-468 and SQ20b cells, 4 and 24 hours after IR (Fig. 2A, Supplementary Fig. S5). At the 4-hour post-IR time point, AF555-anti-γH2AX-N-SS-E was visible predominantly in the cytoplasm of MDA-MB-468 cells but also in the nucleus where it colocalized with γH2AX foci. By 24 hours, cytoplasmic staining was no longer present, and all AF555 signal colocalized closely with γH2AX foci in the nuclei. AF555-anti-γH2AX-N-SS-E colocalized with every observed γH2AX focus. This is in contrast to AF555-anti-γH2AX-PNE, which showed no colocalization with γH2AX foci. This suggests that in the case of AF555-anti-γH2AX-N-SS-E, intracellular cleavage at the S-S bond does occur as intended, releasing AF555-anti-γH2AX-NLS, which is able to bind γH2AX in the nucleus where it is retained. There was no accumulation of AF555-anti-γH2AX-N-SS-E in the nuclei of sham-irradiated cells, which lacked γH2AX foci, and the nonspecific conjugate AF555-rIgG-N-SS-E did not localize in nuclei, whether cells were irradiated or sham-irradiated. To investigate the specificity of EGFR binding by immunoconjugates, MDA-MB-468 cells were exposed to an excess of EGF. This effectively blocked binding of AF555-rIgG-N-SS-E and AF488-EGF to the membrane (Fig. 2B) and prevented internalization, confirming that cellular uptake of AF555-rIgG-N-SS-E is EGF-dependent. In addition, AF555-anti-rIgG−N-SS-E was not taken up into 231-H2N cells, consistent with their low EGFR density (Supplementary Figs. S4 and S6).

The intracellular distribution of ¹¹¹In-anti-γH2AX-N-SS-E, ¹¹¹In-anti-γH2AX-dN-SS-E, and ¹¹¹In-rIgG-N-SS-E is shown in Fig. 3A. The results are displayed to show the proportion of the total internalized radioactivity that accumulated in the cytoplasm and nucleus for the 2 cell lines, MDA-MB-468 and 231-H2N. EGFR expression on MDA-MB-468 cells or 231-H2N cells was not altered after exposure to IR (Supplementary Fig. S6A). In each cell line, the uptake of ¹¹¹In-anti-γH2AX-N-SS-E, ¹¹¹In-anti-γH2AX-dN-SS-E, or ¹¹¹In-rIgG-N-SS-E after incubation for 4 hours was not significantly different (P > 0.05). The absolute amount of total internalized radioactivity was 7-fold less in 231-H2N (0.15% ± 0.008% of added radioactivity was found to be cell associated after a 1-hour exposure) compared with MDA-MB-468 cells (1.11% ± 0.04%), in line with the >6-fold lower expression of EGFR in 231-H2N cells (Supplementary Fig. S6A). The initial distribution of EGFR in MDA-MB-468 cells is shown in Supplementary Fig. S6B. In irradiated MDA-MB-468 cells, the proportion of the total internalized radioactivity that accumulated in the nucleus was greater for ¹¹¹In-anti-γH2AX-N-SS-E than for ¹¹¹In-rIgG-N-SS-E (P = 0.0003). The amount of intranuclear radioactivity following exposure of cells to ¹¹¹In-anti-γH2AX-N-SS-E was significantly higher in irradiated cells than in sham-irradiated cells (P < 0.05). ¹¹¹In-anti-γH2AX-dN-SS-E, containing a disrupted NLS sequence, showed limited nuclear uptake (10% of the internalized amount of ¹¹¹In was found in the nucleus; P < 0.05). In 231-H2N cells, nuclear localization of ¹¹¹In was 50% higher in irradiated cells exposed to ¹¹¹In-anti-γH2AX-N-SS-E than ¹¹¹In-rIgG-N-SS-E (P < 0.05). In both cell lines, the amount of nonspecific immunoconjugate, anti-rIgG-N-SS-E, that accumulated in the nucleus was lower in irradiated than in nonirradiated cells but this difference did not reach statistical significance (P > 0.05).

Retention of ¹¹¹In-anti-γH2AX-N-SS-E was prolonged in irradiated MDA-MB-468 cells, compared with sham-irradiated cells [retention half-life: 2.9 hours; 95% confidence interval (CI), = 1.787–7.729 vs. 0.4 hour, 95% CI, 0.3670–0.4687, respectively; P < 0.0001; Fig. 3B]. It was confirmed that this observation was not due to an increase in the number of EGFR per cell following radiation (Supplementary Fig. S6A). Retention of ¹¹¹In-rIgG-N-SS-E was modest in both sham-irradiated and irradiated cells [retention half-life: 0.4 hour (0.3718–0.4004) vs. 0.5 hour (0.4314–0.6541); P > 0.05]. In 231-H2N cells, uptake of both immunoconjugates was low and this was reflected in modest retention of ¹¹¹In-anti-γH2AX-N-SS-E and ¹¹¹In-rIgG-N-SS-E, which was not significantly influenced by IR (P > 0.05; Fig. 3C).

Increasing the concentration of ¹¹¹In-rIgG-N-SS-E or ¹¹¹In-anti-γH2AX-N-SS-E had no effect on the SF of sham-irradiated 231-H2N cells (Fig. 4A). Similarly, ¹¹¹In-rIgG-N-SS-E had no effect on SF in irradiated cells. However, the addition of ¹¹¹In-anti-γH2AX-N-SS-E did enhance the cytotoxicity of IR, although only at the highest concentration tested (50% at 0 μg/mL vs. 31% at 0.5 μg/mL; P < 0.01). As for 231-H2N cells, ¹¹¹In-rIgG-N-SS-E did not affect survival in MDA-MB-468 cells (Fig. 4B). In contrast, in irradiated MDA-MB-468 cells, ¹¹¹In-anti-γH2AX-N-SS-E reduced the clonogenic survival in a dose-dependent manner across the range of concentrations tested (75% at 0 μg/mL vs. 19% at 0.5 μg/mL; Spearman P = 0.0028; P < 0.001), whereas ¹¹¹In-anti-γH2AX-N-SS-E caused a less marked but statistically significant reduction in SF in sham-irradiated cells (100% at 0 μg/mL vs. 67% at 0.5 μg/mL; P < 0.001; Fig. 4B).

Increasing the SA of ¹¹¹In-anti-γH2AX-N-SS-E resulted in an SA-dependent reduction in clonogenic survival in irradiated cells (Fig. 4C and D). This effect was more profound in MDA-MB-468 cells that have high EGFR overexpression but was also observed in 231-H2N cells. Also, at high SA, ¹¹¹In-anti-γH2AX-N-SS-E caused a modest decrease in SF in sham-irradiated MDA-MB-468 but not in 231-H2N cells (Fig. 4C and D). One possible explanation is that MDA-MB-468 cells have a higher background number of γH2AX foci than 231-H2N cells (13), which would be expected to lead to some retention of the anti-γH2AX antibody containing immunoconjugate. ¹¹¹In-rIgG-N-SS-E had no significant effect on SF in irradiated or sham-irradiated EGFR-positive or -negative cells. As a control, ¹¹¹InCl₃ was added to cells in equivalent amounts to the immunoconjugates, but it did not influence clonogenic survival in irradiated or sham-irradiated cells (P > 0.05).

SPECT was conducted 24 hours following intravenous administration of ¹¹¹In-anti-γH2AX-N-SS-E or ¹¹¹In-rIgG-N-SS-E to MDA-MB-468 xenograft–bearing mice treated with IR (10 Gy) delivered to the xenograft. The tumor uptake of ¹¹¹In-anti-γH2AX-N-SS-E appears modest on SPECT images because of the proximity of tumors to the kidneys; however, VOI analyses were indicative of a detectable signal-to-background signal (Fig. 5A). At 24 hours p.i, the tumor uptake of ¹¹¹In-anti-γH2AX-N-SS-E was greater in irradiated xenografts than in unirradiated control animals (6.30 ± 1.47%ID/g vs. 2.57 ± 0.47%ID/g, respectively; P < 0.01). Tumor uptake of the nonspecific control agent ¹¹¹In-rIgG-N-SS-E was significantly lower in irradiated animals than for ¹¹¹In-anti-γH2AX-N-SS-E (2.01 ± 0.01%ID/g vs. 6.30 ± 1.47%ID/g, respectively; P < 0.01). When an excess of EGF was coinjected with ¹¹¹In-anti-γH2AX-N-SS-E to block EGFR, uptake in irradiated xenografts decreased to 1.07 ± 0.14%ID/g (P < 0.001). γH2AX staining of sections of tumor, taken from mice after 10 Gy irradiation, showed marked induction of γH2AX foci (9.8 ± 0.4 foci/cell), even at 24 hours after irradiation, compared with sham-irradiated tumors (2.2 ± 0.3 foci/cell; Fig. 5B). To show γH2AX targeting, tumor sections were stained for γH2AX and anti-γH2AX-N-SS-E (Fig. 5C). Some degree of nonspecific background staining of anti-γH2AX-N-SS-E was observed, but anti-γH2AX-N-SS-E colocalized with every γH2AX focus.

Taken together, these results show that ¹¹¹In-anti-γH2AX-N-SS-E uptake in tumors is dependent on γH2AX expression, anti-γH2AX/γH2AX interaction, and EGF/EGFR binding. Liver and kidney uptake of ¹¹¹In-anti-γH2AX-N-SS-E was relatively low (4.67 ± 1.29%ID/g and 9.00 ± 1.02%ID/g at 24 hours p.i., respectively). There was no significant difference in liver and kidney uptake whether ¹¹¹In-anti-γH2AX-N-SS-E or ¹¹¹In-rIgG-N-SS-E was administered. Liver and kidney uptake was similar in irradiated and control mice for both tracers (P > 0.05).

Radionuclides that emit Auger electrons are of interest as therapeutic agents, due to the high linear energy transfer (LET) and short range in tissue of these particles. As ionizations are clustered within several cubic nanometers around the point of decay of Auger electron-emitting radionuclides, leading to local absorbed radiation doses in excess of 100 Gy, their targeted delivery to the DNA of cancer cells holds promise as a therapeutic strategy in cancer (14, 15). We have previously shown that the combination of IR- and γH2AX-targeted Auger electron exposure is a very effective tumor control measure in vitro as well as in vivo (3, 4, 15). Here, our aim was to increase tumor selectivity of a γH2AX-targeting Auger electron-emitting compound by adding an EGF moiety.

Using a synthetic strategy similar to the one we have used in the past to target the cyclin-dependent kinase inhibitor protein, p27^kip1 (10), EGF was tagged with NLS and covalently linked to an anti-γH2AX antibody via a noncleavable PEO₆ linker to generate ¹¹¹In-anti-γH2AX-PNE (Supplementary Fig. S1A). ¹¹¹In-anti-γH2AX-PNE bound to EGFR on the surface of breast cancer cells and was internalized and distributed within the cell, mirroring the subcellular distribution of native EGF (Fig. 1A). Although anti-γH2AX-PNE retained affinity for the γH2AX protein (Supplementary Fig. S2B), a fluorophore-labeled conjugate did not colocalize with IR-induced γH2AX foci in cells (Fig. 2B). The most likely explanation for this observation is that the concentration of AF555-anti-γH2AX-PNE in the nucleus was too low to generate a detectable signal.

Therefore, the immunoconjugate was redesigned to allow improved DNA targeting after internalization. A conjugate was synthesized with the EGF and γH2AX-NLS sections separated by a disulfide bond (Supplementary Fig. S1B). Disulfide bonds have been used extensively in antibody–drug conjugate (ADC) design to enable delivery of drugs to cells through linkage to antibody vectors (16). Upon internalization of this conjugate, the disulfide bridge is cleaved through the action of reductive intracellular thiols such as glutathione (8). We have shown that this approach can be used for the release of the DNA-targeting, Auger electron-emitting portion of the radiotherapeutic agent, ¹¹¹In-anti-γH2AX-NLS, from EGF. Reductive cleavage of disulfide bonds is a general property of all cells and takes place after acidification of the early endosome. This explains the recent success of ADCs, where tumor-selective antibodies are linked to cytotoxic drugs that, upon cellular internalization, are cleaved from their antibody carriers (17). Peptide–drug conjugates have also been investigated. For example, camptothecin linked through disulfide conjugation to substance P was cytotoxic in a range of cell lines (U87, U251, MCF-7, and MDA-MB-231) that express NK1R, the receptor for substance P (18).

Several endogenous compounds and enzymes play a role in the cleavage of disulfide bonds in endosomes and lysosomes (19). Of these, glutathione and cysteine constitute the major physiologic thiols responsible for the intracellular reducing environment, with glutathione being the most prevalent cellular thiol accounting for more than 90% of the total non-protein sulfur (20). Another mechanism of endosomal disulfide bond cleavage is the activity of the enzyme, GILT (γ-IFN–inducible lysosomal thiol reductase). One study used folate conjugated to a BODIPY fluorescent probe conjugated to a rhodamine fluorescent probe through a reducible disulfide linkage. It was shown that the disulfide bond was reduced very efficiently throughout the entire intracellular folate trafficking process (21). Here, we chose to show disulfide bond cleavage using glutathione, as it is the most prevalent intracellular reductive agent. Taken together, reductive cleavage of disulfide linkers is a process that has been observed in the endosomes and cytosol of many different cells and, as here, provides an excellent mechanism to deliver nuclear targeting radiopharmaceuticals to cancer cells.

Anti-γH2AX-N-SS-E functions as intended, as it was shown to bind to EGFR (Supplementary Fig. S2A) and to internalize into cells in an EGF/EGFR-dependent manner (Fig. 2). In confocal microscopy experiments, AF555 signal was clearly seen in the nuclei of irradiated cells that expressed abundant γH2AX, suggesting that cleavage at the disulfide bond of AF555-anti-γH2AX-N-SS-E had occurred, with release and translocation to the nucleus of AF555-anti-γH2AX-NLS (Figs. 2B and 3A). This process, which would be expected to be similar for ¹¹¹In-labeled anti-γH2AX-N-SS-E, would result in the localization of the Auger electron-emitting radioisotope, ¹¹¹In, near the DNA, leading to reduced clonogenic survival, as shown in Fig. 4. The proposed mechanism of action of ¹¹¹In-anti-γH2AX-N-SS-E is summarized in Supplementary Fig. S7. First, ¹¹¹In-anti-γH2AX-N-SS-E binds to EGFR, expressed on the cell surface and is internalized by ligand-induced endocytosis into clathrin-coated pits. The disulfide bond is reductively cleaved, with release of ¹¹¹In-anti-γH2AX-NLS, which escapes from the endosome into the cytoplasm. The mechanism of this endosomal escape remains to be elucidated but has been observed before by Costantini and colleagues, who showed increased cytotoxicity of ¹¹¹In-labeled trastuzumab when conjugated to NLS (22). The escape mechanism may be similar to that of the TAT peptide, which permeabilizes membranes by generating topologically active saddle-splay (“negative Gaussian”) membrane curvature through multidentate hydrogen bonding of lipid head groups (23, 24). Then, the NLS tag interacts with importins so that anti-γH2AX-NLS is transported into the nucleus, where it binds to IR-induced γH2AX, expressed in foci at the sites of DSBs.

The release of ¹¹¹In-anti-γH2AX-NLS from the delivery moiety, EGF, led to marked nuclear accumulation of ¹¹¹In in irradiated cells with induced γH2AX expression (Fig. 3A) and localization at DSB sites (Fig. 2A). Importantly, it was shown that increased intranuclear accumulation of ¹¹¹In was not the consequence of increased cellular uptake of the compound due to EGFR copy number modulation, which remained unchanged after irradiation, under the experimental conditions used (see Supplementary Fig. S6). The extensive nuclear localization of ¹¹¹In following exposure of cells to ¹¹¹In-anti-γH2AX-N-SS-E translated into a marked reduction in clonogenic survival (Fig. 4B and D). Reduction in survival of EGFR-overexpressing MDA-MB-468 cells that expressed γH2AX through prior exposure to IR, was radiation dose–dependent, as it correlated with the specific activity of the cleavable conjugate, ¹¹¹In-anti-γH2AX-N-SS-E (Fig. 4D).

Given their very short path length, Auger electrons are most radiotoxic when emitted close to DNA. This makes their use distinctly different from previously reported EGFR-targeting radioimmunotherapeutics, which mainly incorporate alpha- or beta-emitting radionuclides. The use of ¹⁷⁷Lu-, ⁹⁰Y-, and ²³¹Bi-labeled anti-EGFR antibodies has shown great promise for radioimmunotherapy (25–28). However, because these agents were investigated using different cell models and given the unique radiobiologic characteristics of Auger electron emitters (29), a direct comparison with the method reported here is not straightforward. Direct comparison between ¹¹¹In-anti-γH2AX-N-SS-E and ¹¹¹In-DTPA-EGF is possible, as they are both labeled with ¹¹¹In (6, 13). In MDA-MB-468 cells, a similar concentration and amount of ¹¹¹In only reduced clonogenic survival to 81% ± 5%, compared to 100% for untreated cells, 74% ± 13% for 4 Gy IR, or 19% ± 3% for ¹¹¹In-anti-γH2AX-N-SS-E (in combination with 4 Gy IR). The main difference between ¹¹¹In-anti-γH2AX-N-SS-E and ¹¹¹In-DTPA-EGF is that the former has a defined intranuclear DNA-associated target, whereas the latter does not, so relies on nuclear accumulation due to EGFR NLS. A similar mechanism accounts for the radiotoxicity of the ¹²⁵I-labeled anti-EGFR mAb 425 antibody (30) or ¹²⁵I-m225 (31). Although several studies show that even ¹²⁵I-labeled noninternalizing antibodies can have potent cytotoxic effects (31), the recently described ¹¹¹In-labeled anti-EGFR antibody, nimotuzomab, which is in itself a cytotoxic agent, conjugated to a nuclear localization sequence, shows very elegantly that conjugation of an NLS to the anti-EGFR antibody results in increased nuclear uptake and therefore leads to increased radiotoxicity (32). It is possible to hypothesize that the cytotoxicity of all these agents could be increased by addition of a DNA-targeting moiety.

In vivo biodistribution studies confirm that ¹¹¹In-anti-γH2AX-N-SS-E uptake in EGFR-expressing MDA-MB-468 xenograft tumors is dependent on γH2AX expression, anti-γH2AX/γH2AX interaction, and EGF/EGFR binding (Fig. 5A and B) and that anti-γH2AX-N-SS-E targets γH2AX in irradiated tumor tissue in vivo (Fig. 5C). Taken together, the results reported in this article show the EGFR-specific, tumor-specific delivery of a γH2AX-targeting radiopharmaceutical for Auger electron therapy and encourage further investigation of this type of bispecific conjugate.

The cleavable radiolabeled immunoconjugate, ¹¹¹In-anti-γH2AX-N-SS-E, binds to EGFR and, following receptor-mediated internalization, associates with γH2AX in the nuclei of cells with DNA DSBs. The resulting exposure of DNA to Auger electron radiation causes DNA damage and reduces cell growth.

No potential conflicts of interest were disclosed.

Conception and design: B. Cornelissen, K.A. Vallis

Development of methodology: B. Cornelissen, K.A. Vallis

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): B. Cornelissen, A. Waller

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): B. Cornelissen, A. Waller, K.A. Vallis

Writing, review, and/or revision of the manuscript: B. Cornelissen, S. Able, K.A. Vallis

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases):

Study supervision: K.A. Vallis

The authors thank Dr. Philip Waghorn for helpful discussions.

This work was supported through a Cancer Research-UK grant to K.A. Vallis (C14521), the CR-UK/EPSRC Oxford Cancer Imaging Centre, and the Oxford Experimental Cancer Medicine Centre.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.