Imaging DNA damage in vivo using γH2AX-targeted immunoconjugates

Abstract

DNA damage responses (DDRs) occur during oncogenesis and therapeutic responses to DNA damaging cytotoxic drugs. Thus, a real-time method to image DNA damage in vivo would be useful to diagnose cancer and monitor its treatment. Toward this end, we have developed fluorophore- and radioisotope-labeled immunoconjugates to target a DDR signaling protein, phosphorylated histone H2A variant H2AX (γH2AX), which forms foci at sites of DNA double-strand breaks. Anti-γH2AX antibodies were modified by the addition of diethylenetriaminepentaacetic acid (DTPA) to allow ¹¹¹In labeling or the fluorophore Cy3. The cell-penetrating peptide Tat (GRKKRRQRRRPPQGYG) was also added to the immunoconjugate to aid nuclear translocation. In irradiated breast cancer cells, confocal microscopy confirmed the expected colocalization of anti-γH2AX-Tat with γH2AX foci. By comparison to non-specific antibody conjugates, ¹¹¹In-anti-γH2AX-Tat was retained longer in cells. Anti-γH2AX-Tat probes were also used to track in vivo DNA damage using a mouse xenograft model of human breast cancer. After treatment with local X-irradiation or bleomycin, the anti-γH2AX-Tat probes produced fluorescent and single photon emission computed tomography (SPECT) signals in the tumors that were proportionate to the delivered radiation dose and the amount of γH2AX present. Taken together, our findings establish the use of radioimmunoconjugates that target γH2AX as a non-invasive imaging method to monitor DNA damage with many potential applications in preclinical and clinical settings.

Introduction

Human cancers are frequently associated with DNA-damage response (DDR) defects (1). This helps explain the extensive DNA damage observed in tumors at all stages, from dysplasia to advanced disease (2, 3). Commonly used anti-cancer treatments, including ionizing radiation (IR) and many chemotherapy drugs, exploit defects in DNA repair by inflicting DNA damage to which the tumor cell is unable to mount a normal response. Small molecule inhibitors of components of the DDR can arrest or reverse tumor growth and many are in pre-clinical or clinical development (4). It follows that an imaging probe capable of revealing DNA damage in vivo could provide useful prognostic information and be used to predict sensitivity to treatment or monitor response to therapy. The ability to image DNA damage would facilitate evaluation of drugs designed to cause tumoral DNA damage or to inhibit its repair.

DNA damage can be measured in human tumors ex vivo by immunohistochemistry using fluorophore-labeled antibodies that bind specific DNA repair proteins (5) or using flow cytometry (6). However, direct quantification of DNA damage in vivo is not currently possible. In contrast, investigators have reported efforts to image cell death in vivo using, for example, reporter constructs that are activated by caspase-3 cleavage or radiolabeled ligands such as annexin-V with affinity for apoptotic cells (7, 8). In some cases only weak correlation between in vivo detection of apoptosis and outcome for a particular tumor was noted (9). This may be because cancer cell death can result from processes other than apoptosis such as mitotic catastrophe, senescence and autophagy.

DNA double-strand breaks (DNA dsb) are caused directly by IR and some radiomimetic drugs, or indirectly through replication fork stalling (4). The ability to image DNA dsb in vivo would be particularly informative, as they are extremely deleterious and their number and persistence reflects the likelihood of eventual cell death (10). Also, because it is an early event following genotoxic stress, DNA dsb formation would likely predict cell fate, whatever the nature of the initiating insult. There are, however, two potential limitations to DNA dsb as a target for imaging. DNA dsb are generally low in number, limiting the sensitivity achievable with most imaging modalities. Also, because DNA dsb exist within the nucleus, they are separated from circulating imaging probes by the cell and nuclear membranes, rendering them inaccessible, particularly to high molecular weight, antibody-based agents. Both hurdles are surmountable. While DNA dsb themselves may not be abundant, they do lead to the accumulation of DNA repair proteins which may provide tractable targets. One is the histone H2A variant, H2AX, which is phosphorylated on Ser139, starting immediately after DNA dsb formation. Hundreds of copies of phosphorylated H2AX (γH2AX) accumulate in foci at DNA dsb, measuring up to 40 Mbp (11). Detection of these foci using an anti-γH2AX antibody forms the basis of a sensitive in vitro assay for DNA dsb (12).

Generally, basal γH2AX expression is high in cancers and exposure to genotoxic stress results in induction of γH2AX which is more prominent and protracted in cancer compared to normal cells (3). Recent reports suggest that enumeration of γH2AX foci in clinical samples may correlate with outcome (13). These observations support investigation of γH2AX as a biomarker of response to DNA damaging agents.

The problem of inaccessibility of γH2AX to imaging probes, due to its intranuclear location, can be solved through incorporation of cell-penetrating peptides (CPP) and nuclear localizing signals (NLS) into probe design. Tat peptide is a NLS-containing domain of the HIV-1 transactivator of transcription that has cell-penetrating properties (14). Tat (GRKKRRQRRRPPQGYG) mediates trans-membrane movement of various cargoes and the presence of NLS (underlined) enables Tat to bind to importins for nucleocytoplasmic trafficking (15). We have shown previously that Tat-containing radioimmunoconjugates (RIC-Tats) can penetrate cell and nuclear membranes and are retained to an extent that correlates with abundance of the molecular target (16, 17).

As a first step to targeting γH2AX for imaging, we attached a Tat-peptide covalently to a fluorophore- or ¹¹¹In-labeled anti-γH2AX antibody. We report here that two agents, Cy3-anti-γH2AX-Tat and ¹¹¹In-DTPA-anti-γH2AX-Tat, specifically target DNA dsb. Cy3-anti-γH2AX-Tat co-localizes with γH2AX in the nuclei of irradiated cells. Both Cy3-anti-γH2AX-Tat and ¹¹¹In-DTPA-anti-γH2AX-Tat accumulate in irradiated cancer cells in vitro and in tumors in vivo following DNA damage. While further development is required to optimize probe design, the proof-of-principle studies presented here show that non-invasive imaging for spatiotemporal tracking of DNA damage may be feasible using γH2AX-targeted radioimmunoconjugates that contain CPPs.

Materials and Methods

Synthesis of RICs

A diagram of the synthesis pathway is shown (Fig. 1A). ¹¹¹In-DTPA-anti-γH2AX-Tat and fluorophore-anti-γH2AX-Tat, probes designed to target DNA dsb, were synthesized. γH2AX antibody (100 μg; Calbiochem, Nottingham, UK) or IgG from mouse serum (mIgG) (Sigma-Aldrich, Dorset, UK) was dissolved in 2-(N-morpholino)ethanesulfonic acid (0.1 M). Tat (GRKKRRQRRRPPQGYG) incorporation was achieved using N-(3-dimethylaminopropyl)-N’-ethylcarbodiimide/ N-hydroxysuccinimide (EDC/NHS; Pierce, Rockford, IL, USA) activation. Tat-peptide was added in 5-fold molar excess for 2 h at room temperature. Unconjugated Tat was removed using sephadex G50 gel filtration (SEC) columns, giving Tat:IgG ratio of 5:1. Anti-γH2AX-Tat was incubated with isocyanatobenzyl-DTPA (SCN-Bn-DTPA) (Macrocyclics, Dallas, TX) or was fluorophore-labeled using Cy3- or AF488/AF555-NHS ester. DTPA-conjugation using cyclic DTPA anhydride, ¹¹¹In labeling, instant thin layer chromatography (ITLC) and determination of DTPA conjugation efficiency was performed as previously described (18). Tat:IgG-conjugation ratio was determined by radioiodination of the Tat-peptide as previously described (19). DTPA-anti-γH2AX-Tat conjugate was radiolabeled using ¹¹¹In-chloride and radiolabeling yield, determined by ITLC, was ≥95%. The specific activity range was 0.5-6 MBq/μg. Chemical and radiochemical purity determined by size exclusion radio-HPLC of the final products was ≥95%. Control probes, ¹¹¹In-DTPA-anti-mouseIgG-Tat (¹¹¹In-DTPA-anti-mIgG-Tat) and Cy3- or AF488/AF555-anti-mIgG-Tat, were also synthesized. Tat-conjugated anti-γH2AX or mIgG were labeled with Cy3, AF555 or AF488 using Cy3-NHS (Amersham), AF555-NHS or AF488-NHS (Invitrogen, Paisley, UK). Fluorophore-labeling was efficient, with Cy3:IgG of 6:1.

Figure 1.

A, Schematic overview of the synthesis of ¹¹¹In-DTPA-anti-γH2AX-Tat. B, Radioimmunoassay showing binding of ¹¹¹In-DTPA-anti-γH2AX-Tat to γH2AX, present in whole cell lysates derived from irradiated 231-H2N cells, in competition with increasing concentrations of anti-γH2AX or anti-γH2AX-Tat. C, Internalization of ¹¹¹In-DTPA-anti-γH2AX-Tat and ¹¹¹In-DTPA-mIgG-Tat in irradiated and control MDA-MB-468 whole cells and in, D, nuclei. E, Retention of ¹¹¹In-DTPA-anti-γH2AX-Tat and ¹¹¹In-DTPA-mIgG-Tat in MDA-MB-468 cells, exposed to IR (4 Gy) or in control (unirradiated) cells. Experiments were repeated 3 times with 3 replicates. Error bars show standard deviation (SD).

Cell culture and Western blot analysis

Human breast cancer cell lines MDA-MB-468 (LGC Standards, Teddington, Middlesex, UK), MDA-MB-231 (LGC Standards), MDA-MB-231 cells stably transfected with HER2 (231-H2N, from Dr Robert Kerbel, Sunnybrook Research Centre, Toronto, Canada), and mouse embryonic fibroblasts (MEF) (H2AX −/− and wild-type [WT]) (from Dr Andre Nussenzweig, NCI, Bethesda, MD) were cultured in 5% CO₂ in DMEM (Sigma-Aldrich, Dorset, UK) supplemented with 10% fetal calf serum (Invitrogen, Paisley, UK) and penicillin/streptomycin, 100 units/mL (Invitrogen, Paisley, UK). Cells were tested and authenticated by the providers. The cumulative length of culture was <6 months following retrieval. MDA-MB-468 cells (10⁶) were irradiated (0-10 Gy) using an IBL-637 ¹³⁷Cs irradiator (Cisbio International, Gif Sur Yvette, France; dose-rate 1.0 Gy/min). After incubation for 1 h, cells were lysed and Western blot performed, probing for γH2AX and β-Actin.

Radioimmunoassay

231-H2N cells were irradiated (4 Gy) and after 1 h lysed (10 min on ice; 50 mM HEPES, 150 mM NaCl, 2.5 mM EGTA, 1 mM EDTA, 1 mM DTT, 0.1% Tween-20, 10% glycerol, with added protease and phosphatase inhibitor). Lysates were used to coat wells of a RIA plate, incubated overnight at 4°C, washed and blocked (PBS, 2% BSA, 0.5% Tween-20). ¹¹¹In-DTPA-anti-γH2AX (1 nM) plus increasing concentrations of anti-γH2AX antibody or DTPA-anti-γH2AX-Tat in PBS (100 μL) were added for 2 h at 4°C. Plates were washed and radioactivity in each well determined. IC₅₀ values for competition binding were determined.

Internalization and retention

Internalization, nuclear localization and retention of ¹¹¹In-DTPA-anti-γH2AX-Tat and ¹¹¹In-DTPA-mIgG-Tat in MDA-MB-468 cells were determined as previously described (19).

Microscopy

MDA-MB-468 cells or MEFs were grown on coverslips and exposed to fluorophore-anti-γH2AX-Tat or fluorophore-mIgG-Tat (0.125 μg/mL). After incubation at 37°C for 1 h cells were irradiated (4 Gy) and 1 h or 23 h later washed, fixed using 4% formaldehyde, mounted with DAPI or permeabilized using 1% TritonX-100, blocked (1 h, 37°C; 0.1% Triton X-100, 2% BSA in PBS) and stained for γH2AX using an AF633-labeled secondary antibody. To investigate the spatial relationship between fluorphore-anti-γH2AX-Tat and γH2AX foci, MDA-MB-231 cells were grown on 0.9 μm mylar film (DuPont Teijin Films UK Ltd, Middlesbrough, UK) and incubated with AF555-anti-γH2AX-Tat or AF555-mIgG-Tat (0.125 μg/mL). After 1 h, cells were exposed to 1.5 keV X-Rays through a gold mask with 1 μm slits every 10 μm resulting in DNA damage in “bands” (20). The mean dose to cells was ~5 Gy (local dose is approximately 50 Gy). After incubation for 1 h, cells were stained for γH2AX using an AF488-conjugated secondary antibody.

Mouse models

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, UK). RICs (10 μg; 5 MBq) were administered intravenously (i.v.). γH2AX was induced by intraperitoneal (i.p.) administration of bleomycin (10 μg) 2 h prior to injection of RIC or by X-irradiation of the tumor (10 Gy) 1 h after injection of RIC. Radiation was delivered using a Gulmay 320 kV X-irradiator; 2.0 Gy/min. For optical imaging mice were anesthetized 24, 48 or 72 h after RIC injection using isoflurane. Fluorescence imaging was performed using an IVIS200 system (Caliper LifeSciences, Runcorn, UK). Autofluorescence correction was performed using the built-in correction filter set. For single photon emission computed tomography (SPECT) imaging mice were anesthetized using isoflurane at 24, 48 and 72 h after RIC injection, and SPECT-CT performed using a nanoSPECT-CT scanner (Bioscan, Washington DC, USA). At 72 h, mice were euthanized and organs removed, weighed and counted for radioactivity.

Statistical analysis

Statistical analysis and curve fitting was performed using the GraphPad Prism software package (GraphPad Software, San Diego, CA, USA). Results were compared using 1-way ANOVA with post-hoc Tukey test. Fitted parameters from non-linear regression curve fits were compared using the F-test. A p-value of <0.05 was considered significant.

Results

Synthesis, internalization and retention of γH2AX-targeted RIC-Tats

The cell line, MDA-MB-468, was used for evaluation of RIC-Tats. We previously showed induction of γH2AX in MDA-MB-468 cells in response to IR, using an immunofluorescence assay for γH2AX foci (21), and confirmed this by Western blot analysis (Supplementary Fig. S1). To confirm affinity of binding of ¹¹¹In-DTPA-anti-γH2AX-Tat to γH2AX, a competition radioimmunoassay (RIA) was performed (Fig. 1B). 231-H2N cells were irradiated (4 Gy) and incubated for 1 h to allow formation of γH2AX. Whole cell lysates, containing γH2AX, were used to coat wells of an RIA plate. ¹¹¹In-DTPA-anti-γH2AX (1 nM) plus either unlabeled DTPA-anti-γH2AX-Tat or anti-γH2AX (range 0-1000 nM) was added. The radioactivity in each well was determined by quantitative autoradiography. The ability of DTPA-anti-γH2AX-Tat and anti-γH2AX to compete with ¹¹¹In-DTPA-anti-γH2AX for binding to γH2AX was similar with Log(IC₅₀) of 0.33±0.12 and 0.42±0.14 (mean±SD), respectively (p=0.70), indicating that the affinity of the modified antibody, DTPA-anti-γH2AX-Tat, for the target epitope is similar to that of the unaltered antibody, anti-γH2AX.

To investigate the ability of the RIC-Tats to penetrate the cell membrane and accumulate in the nuclei of cells, cell radioactivity internalization and retention experiments were performed. To evaluate internalization, MDA-MB-468 cells were irradiated (4 Gy) or sham-irradiated and exposed to ¹¹¹In-DTPA-anti-γH2AX-Tat or ¹¹¹In-DTPA-mIgG-Tat for 0-4 h. At selected times cells were harvested, fractionated and the radioactivity in whole cells and in membrane, cytoplasmic and nuclear fractions measured (Fig. 1C and D; Supplementary Fig. S2A and B). The amount of ¹¹¹In that internalized and translocated to nuclei peaked at 15-30 min in irradiated and non-irradiated samples, reaching approximately 0.13% of the added radioactivity. The nuclear ¹¹¹In content slowly diminished, reaching approximately 0.1% of the total added radioactivity by 4 h. To investigate retention of ¹¹¹In, MDA-MB-468 cells were loaded with ¹¹¹In-DTPA-anti-γH2AX-Tat or ¹¹¹In-DTPA-mIgG-Tat by exposing them to radiolabeled probe for 1 h. Cells were then irradiated (4 Gy) or left unirradiated. Radiopharmaceutical-containing medium was replaced with fresh medium and the amount of ¹¹¹In in the cells was measured for up to 50 h. In non-irradiated cells, exposed to either ¹¹¹In-DTPA-mIgG-Tat or ¹¹¹In-DTPA-anti-γH2AX-Tat, the amount of retained ¹¹¹In fell rapidly reaching background levels by 10 h (Fig. 1E). This was also true of irradiated cells, in which γH2AX expression was induced, when exposed to ¹¹¹In-DTPA-mIgG-Tat (half-life 0.23±0.09 h). However, retention of ¹¹¹In did occur in irradiated compared to non-irradiated cells exposed to ¹¹¹In-DTPA-anti-γH2AX-Tat (half-life 27.71±11.01 h vs. 1.81±0.87 h, respectively; p = 0.0076, F-test). Thus the anti-γH2AX-Tat probe, was retained only in cells in which γH2AX, was induced.

In vitro imaging of DNA dsb using γH2AX-targeted RIC-Tats

MDA-MB-468 cells were exposed to AF488-anti-γH2AX-Tat or AF488-mIgG-Tat for 1 h and irradiated (4 Gy) or sham-irradiated. After 1 h or 23 h, cells were fixed and mounted with DAPI and stained for γH2AX (Fig. 2A and B). AF488-anti-γH2AX-Tat was synthesized using an anti-γH2AX antibody raised in rabbit whereas γH2AX foci were detected using mouse-anti-γH2AX antibody and secondary AF555-labeled anti-mouse IgG. MDA-MB-468 cells exposed to AF488-mIgG-Tat, showed membrane and cytoplasmic staining at 1 h (attributed to Tat-mediated internalization of probe) that disappeared by 23 h post IR (Fig. 2A and B). At no time was nuclear fluorescence observed in cells exposed to AF488-mIgG-Tat, whether irradiated or not. Fluorescence was not observed in the nuclei of sham-irradiated cells exposed to AF488-anti-γH2AX-Tat. However, cells exposed to AF488-anti-γH2AX-Tat followed by IR showed focal nuclear uptake of AF488 at 1 and 23 h. Immunostaining for γH2AX showed that many γH2AX foci formed by 1 h following IR and were still detectable at 23 h. γH2AX foci colocalized with the fluorescence resulting from nuclear accumulation of AF488-anti-γH2AX-Tat (Fig. 2A and B). This was confirmed by generating a cytofluorogram (CFG) obtained by plotting, for each pixel, AF488 fluorescence (green) on the x-axis and AF555 fluorescence (red), denoting presence of γH2AX foci, on the y-axis. Coincidence of the two signals was observed.

Figure 2.

Co-localization of fluorphore-anti-γH2AX-Tat with γH2AX foci in vitro. MDA-MB-468 cells were exposed to AF488-anti-γH2AX-Tat or AF488-mIgG-Tat and after 1 h were sham-irradiated or irradiated (4 Gy). At (A) 2 h or (B) 24 h after addition of RICs, cells were fixed, permeabilized, stained for γH2AX foci (red) and mounted with Vectashield containing DAPI (blue). Images were acquired using confocal microscopy. Nuclear fluorescence due to AF488 (green) was seen only in irradiated cells exposed to AF488-anti-γH2AX-Tat. Cytofluorograms (CFG) shows co-localization of AF488-anti-γH2AX-Tat with γH2AX foci.

It was shown that the presence of fluorophore-anti-γH2AX-Tat at the time of irradiation does not itself interfere with DNA damage repair, as the number of γH2AX foci at 23 h after IR is the same whether fluorophore-anti-γH2AX-Tat is present or not (Supplementary Fig. S3A).

To evaluate dependence of fluorophore-anti-γH2AX-Tat signal on the presence of γH2AX foci, a slit-irradiation technique was used (20). Cells were grown on mylar film, exposed to AF555-anti-γH2AX-Tat or AF555-mIgG-Tat and irradiated through a gold film mask, resulting in cells being irradiated in 1 μm stripes. MDA- γH2AX foci formed in a stripes and AF555-anti-γH2AX-Tat, but not AF555-mIgG-Tat, co-localized with γH2AX foci (Fig. 3A).

Figure 3.

A, MDA-MB-231 cells, exposed to AF555-anti-γH2AX-Tat or AF555-mIgG-Tat for 1 h, were irradiated using 1.5 keV X-Rays (5 Gy), through a gold mask and then incubated for 1 h, fixed, and stained for γH2AX foci and mounted in Vectashield with DAPI. In merged images, co-localization of fluorophore-labeled RIC with γH2AX foci is shown as white. B, WT and H2AX −/− MEFs were exposed to Cy3-anti-γH2AX-Tat or Cy3-mIgG-Tat and after 1 h were sham-irradiated or irradiated (4 Gy). At 2 h after addition of RICs, cells were fixed, permeabilized and stained for γH2AX foci. Co-localization of Cy3-anti-γH2AX-Tat with γH2AX foci is shown in the merged image in WT MEFs and in the cytofluorogram (CFG).

To confirm that retention of anti-γH2AX RIC depends on the presence of γH2AX, WT and H2AX-null MEFs were exposed to Cy3-anti-γH2AX-Tat for 1 h, irradiated or sham-irradiated and after 1 or 23 h, fixed, mounted with DAPI and stained for γH2AX foci (Fig. 3B; Supplementary Fig. S3B). Co-localization of γH2AX foci with Cy3 was observed in irradiated WT MEFs, but not H2AX −/− MEFs, exposed to Cy3-anti-γH2AX-Tat. Efficient DNA damage repair in WT MEFs results in rapid resolution of γH2AX foci, accounting for the absence of fluorescence due to Cy3-anti-γH2AX-Tat and AF488 (γH2AX foci) in nuclei at 23 h (Supplementary Fig. S3B).

These in vitro studies indicate that anti-γH2AX RICs accumulate specifically at γH2AX foci in the nuclei of irradiated cells. Low magnification microscopy images of irradiated cells exposed to AF55-anti-γH2AX-Tat, indicate that co-localization of the probe with γH2AX foci is seen in all nuclei that contain γH2AX foci (Supplementary Fig. S3C).

In vivo imaging of DNA dsb following radiation or chemotherapy

To test whether anti-γH2AX-Tat RIC can visualize DNA damage in vivo, MDA-MB-468 xenograft-bearing mice were scanned using an IVIS-300 optical camera at 24, 48 and 72 h following i.v. administration of Cy3-mIgG-Tat or Cy3-anti-γH2AX-Tat and exposure to IR (10 Gy), delivered to the xenograft-bearing hind-limb, to induce intratumoral γH2AX. Accumulation of Cy3-anti-γH2AX-Tat was detectable in irradiated but not control tumors. There was no uptake of Cy3-mIgG-Tat in irradiated or control tumors (Fig. 4). Accumulation of Cy3-anti-γH2AX-Tat in irradiated tumors peaked at 48 h and returned to baseline by 72 h.

Figure 4.

Mice, bearing MDA-MB-468 xenografts, received Cy3-mIgG-Tat or Cy3-anti-γH2AX-Tat i.v. and tumors were sham-irradiated or irradiated (10 Gy). Images were acquired at 24, 48 and 72 h p.i. Cy3-mIgG-Tat did not accumulate in either control or irradiated tumors but Cy3-anti-γH2AX-Tat accumulated in irradiated tumors.

SPECT was performed 24, 48 and 72 h following administration of ¹¹¹In-DTPA-anti-γH2AX-Tat or ¹¹¹In-DTPA-mIgG-Tat to mice treated with bleomycin (10 μg) by i.p. injection or IR (10 Gy) delivered to the xenograft. Transverse and coronal images through tumors are shown (Fig. 5A; Supplementary Fig. S4A). Volume of interest analyses (VOI) of tumors were performed (Fig. 5B). In untreated controls tumor uptake of ¹¹¹In-DTPA-anti-γH2AX-Tat was greater than ¹¹¹In-DTPA-mIgG-Tat at each time point. The uptake of ¹¹¹In-DTPA-anti-γH2AX-Tat expressed as percentage of the injected dose per gram of tumor (%ID/g) at 24 h p.i. was 3.6±0.1 compared to 1.2±0.1 for ¹¹¹In-DTPA-mIgG-Tat (p<0.0001). Tumor uptake of ¹¹¹In-DTPA-mIgG-Tat did not increase significantly following bleomycin or IR compared to controls, with uptake at 24 h p.i. of 1.7±0.7 % and 1.7±0.2 versus 1.2±0.1 %ID/g, respectively; p>0.05. However, both bleomycin and IR led to significantly increased intratumoral accumulation of ¹¹¹In-DTPA-anti-γH2AX-Tat compared to controls, with uptake at 24 h p.i. of 5.4±1.2 and 5.2±0.6 versus 3.6±0.1 %ID/g, respectively; p<0.0001. At 72 h uptake of ¹¹¹In-DTPA-anti-γH2AX-Tat in irradiated tumor had returned to basal levels whereas uptake in bleomycin-treated tumors remained elevated. There was no evidence of accumulation of ¹¹¹In-DTPA-anti-γH2AX-Tat in the normal tissues of the irradiated hind-limb compared to the unirradiated hind-limb over the time course investigated, including at an early time point (2 h) (Supplementary Fig. S4B). ¹¹¹In-anti-γH2AX-Tat, at the specific activities used for SPECT imaging (<1 MBq/μg), did not reduce cell survival (evaluated in clonogenic assays) (Supplementary Fig. S5).

Figure 5.

A, SPECT images following ¹¹¹In-DTPA-anti-γH2AX-Tat or ¹¹¹In-DTPA-mIgG-Tat in mice bearing MDA-MB-468 xenografts (white circles). Mice received PBS (control), i.p. bleomycin or the xenograft was irradiated (10 Gy). ¹¹¹In-DTPA-anti-γH2AX-Tat or ¹¹¹In-DTPA-mIgG-Tat (10 μg, 1 MBq/μg) was administered i.v. and SPECT scans performed at 24, 48 and 72 h. Transverse images through the tumor are shown. Coronal images are shown in Supplementary Fig. S4A and S4B. B, Volume of Interest (VOI) analysis of xenografts from (A). Results shown are mean SUV values ± SD (n = 3) (* p < 0.0001).

Following acquisition of images at 72 h mice were euthanized and selected organs removed and radioactivity counted. This confirmed significant retention of radioactivity in tumors of bleomycin-treated animals at 72 h (Supplementary Fig. S6; Supplementary Table 1).

To demonstrate that tumor uptake of anti-γH2AX RIC correlates with the abundance of γH2AX, xenografts were exposed to IR (0-4 Gy) following administration of ¹¹¹In-anti-γH2AX-Tat. SPECT images acquired 23 h post IR show that intratumoral accumulation of the probe increased with increasing IR dose. Transverse images through tumor are shown (Fig. 6A). Tumors and muscle were removed and radioactivity counted for calculation of tumor:muscle (T/M) ratio. A positive correlation between IR dose and intratumoral accumulation of ¹¹¹In was observed (Spearman r=0.9; p=0.042) (Fig. 6B). Tumors were sectioned, stained for γH2AX, showing that tumors exposed to higher doses of IR contained the greatest number of γH2AX foci (Fig. 6C).

Figure 6.

A, Mice received ¹¹¹In-DTPA-anti-γH2AX-Tat as in Fig. 5A, xenografts (white circles) were sham-irradiated or irradiated (1-4 Gy) and SPECT scans performed at 24 h p.i. Transverse images through the tumor are shown. B, Tumor accumulation of ¹¹¹In-DTPA-anti-γH2AX-Tat expressed as %ID/g and as tumor:muscle ratio. A positive correlation between IR dose and intratumoral accumulation of ¹¹¹In was observed (Spearman r = 0.9; p = 0.042). C, Tumors from (A) were harvested and stained ex vivo for γH2AX foci. Representative 0.8 μm thick slices are shown. The number of γH2AX foci increases with dose of IR.

Discussion

Many common anti-cancer treatments cause DNA damage. These agents are efficacious because cancer cells are often DDR-impaired and, therefore, are unable to recover from genotoxic stress. IR and several types of chemotherapy, including alkylators, topoisomerase inhibitors and replication inhibitors, cause DNA dsb, leading to cell death. DNA dsb also result from replication fork stalling which can be caused by alkylating and antimetabolite drugs (4). Phosphorylation of H2AX by the PI3K-like kinases, to form γH2AX, is an early and almost universal feature of the eukaryotic response to DNA dsb. Detection and quantification of γH2AX expression is therefore used as a measure of DNA damage.

A recently proposed model of tumorigenesis implicates activated oncogenes as a cause of replication stress and DNA damage (1). This phenomenon, ‘constitutive DNA damage’, may account for the frequent observation of high γH2AX expression in pre-cancerous lesions and cancers but not in adjacent normal tissue. High γH2AX expression has been described in many types of cancer and in their precursor lesions (2, 22-24). The kinetics of γH2AX induction differs in malignant compared to normal cells (10). The number of γH2AX foci per nucleus increases rapidly following irradiation of normal tissues, returning to baseline levels by approximately 24 h post IR (25). In tumors, however, elevation of γH2AX expression is protracted and evident up to 72 h post IR (26). These differences in normal versus malignant tissue have stimulated interest in γH2AX as a diagnostic and pharmacodynamic biomarker (27). However, the techniques currently available to study γH2AX expression rely on the analysis of clinical samples ex vivo (28). A method for whole-body imaging of γH2AX would, therefore, be a powerful tool with which to study the spatiotemporal dynamics of DDR.

An average cell contains 30 × 10⁶ nucleosomes (29), with two H2A proteins per nucleosome i.e. 60 × 10⁶ H2A per cell. Approximately 10% of H2A is present in the H2AX isoform in mammalian cells (although up to 25% has been reported) (30). This means that there are about 6 × 10⁶ copies of H2AX per cell. It has been estimated that roughly 2% of H2AX molecules become phosphorylated for every Gy of IR to which cells are exposed (30). Therefore, the number of copies of γH2AX formed following doses of 2 to 10 Gy IR would be 2.4 × 10⁵ to 1.2 × 10⁶. In cells with a higher proportion of H2AX (25%), the numbers would be 6 × 10⁵ and 3 × 10⁶. Each copy of γH2AX could be bound by a Tat-conjugated anti-γH2AX RIC. This number of targets (approximately 10⁶ per cell) is similar to that of other tumor antigens that have been explored for molecular imaging such as HER2 and EGFR. Also cancers overexpress γH2AX compared to surrounding normal tissues, even in the absence of exposure to a genotoxic treatment (2, 22-24, 31). Therefore it is possible that the number of copies of γH2AX in some cancers is already high and is then induced further following treatment. These lines of evidence suggest that the abundance of γH2AX expression in cancers would be sufficient for molecular imaging.

We have seen no evidence of accumulation of ¹¹¹In-DTPA-anti-γH2AX-Tat in irradiated normal tissues over the time course investigated. This can be appreciated by comparing the appearance of the right (irradiated) hind-limb with the left (non-irradiated) hind-limb in the images shown in Fig. 4 and Fig. S4. We speculate that this is because the resolution of γH2AX foci in normal tissue is so rapid and complete that the probe does not accumulate sufficiently to produce a detectable signal. The rapid resolution of γH2AX in the normal cell line, WT MEF, compared to the protracted expression of γH2AX in a malignant cell line, MDA-MB-468, after IR suggests that this explanation may be correct (Supplementary Fig. S3A and B). Bleomycin has been shown to accumulate in tumors. Hou et al, for example, showed the tumor:blood ratio for ⁵⁷Co-bleomycin at 48 h was >100 in a murine hepatoma model (32). Intra-tumoral concentration of bleomycin would cause prolonged induction of γH2AX and this may account for the observation that ¹¹¹In-DTPA-anti-γH2AX-Tat signal is more protracted in bleomycin- compared to IR-treated tumors (Supplementary Fig. S3 and S6).

The internalization of anti-γH2AX RIC-Tats into cells is not influenced by γH2AX expression, since Tat peptides are able to penetrate almost any cell (14). However, anti-γH2AX RICs are only retained in the nuclei of cells that express γH2AX, resulting in target/non-target contrast. We have shown that fluorophore-labeled anti-γH2AX-Tat colocalizes with γH2AX foci that were simultaneously visualized using a standard immunofluorescence method. This indicates that anti-γH2AX RIC-Tats bind their molecular target and accounts for their intranuclear retention in γH2AX-expressing cells. In vivo, fluorophore- and radio-labeled anti-γH2AX RIC-Tats specifically accumulate in tumors in which DNA damage has been induced by IR or chemotherapy.

Cy3-anti-γH2AX-Tat is suitable for imaging γH2AX in vitro and superficial tumors in mice but not for imaging deep-seated tumors because of light attenuation by overlying structures. Therefore, we developed an ¹¹¹In-labeled probe. ¹¹¹In, which has gamma emissions of 245 and 171 kV, is commonly used in nuclear medicine. For example, ¹¹¹In-labeled leukocytes, octreotide, anti-CD20 antibody and capromab pendetide are used routinely to image inflammation, neuroendocrine tumors, lymphoma and prostate cancer, respectively. Both the gamma photons and the short-range Auger electrons which ¹¹¹In emits are capable of causing DNA damage, although this is unlikely to be significant when small quantities of isotope are used for imaging. To increase their potential applications it would be possible to label anti-γH2AX RICs with other isotopes such as ^99mTc or ⁸⁹Zr for SPECT and PET imaging.

It has been shown that anti-γH2AX RICs can image DNA damage in vitro and in vivo. Further optimization of the probes is desirable. Currently, cellular internalization of anti-γH2AX RICs is non-specific. Modest tumor uptake of radioactivity was seen following administration of the control agent, ¹¹¹In-DTPA-mIgG-Tat, which may be due to the enhanced permeability and retention (EPR) of tumors (33). The retention of ¹¹¹In-DTPA-anti-γH2AX-Tat but not ¹¹¹In-DTPA-mIgG-Tat in tumor, however, can be ascribed specifically to γH2AX upregulation. It may be possible to improve tumor-specific uptake of anti-γH2AX RICs by incorporation of a tumor-seeking moiety, such as a receptor-binding peptide. We have reported success with this strategy previously (17). The kinetics of γH2AX formation and resolution following genotoxic stress are rapid in normal tissues, with removal of foci starting from about 30 minutes. Injected antibodies on the other hand have long half-lives and penetrate tissue slowly. Therefore, refinement of the anti-γH2AX RIC by replacement of antibody by Fab’ might provide more favorable pharmacokinetics and allow imaging of rapid changes in γH2AX content.

In summary, we report initial evaluation of molecularly-targeted contrast agents designed to image DNA dsb. Labeled anti-γH2AX-Tat immunoconjugates hold promise as clinical tools for non-invasive imaging of DNA damage.