Optoacoustic Detection of Early Therapy-Induced Tumor Cell Death Using a Targeted Imaging Agent

DNA sequencing of tumor biopsies or of circulating tumor DNA (ctDNA) and detection of early treatment response through increases in the levels of ctDNA are expected to play an increasingly important role in selecting treatment for individual patients (1, 2). However, tumor heterogeneity remains an obstacle for both approaches. Biopsies may not detect all clones present within a tumor and the presence of residual treatment-resistant clones may not be reflected in the levels of ctDNA released following treatment. Imaging methods that can detect treatment response throughout the entire tumor volume are likely, therefore, to play an important role in treatment selection (3).

Tumor cell death represents a generic downstream marker of treatment response and consequently there has been considerable interest in developing imaging methods and agents to detect tumor cell death in vivo (4). We have developed a targeted imaging agent based on the C2A domain of Synaptotagmin-I, which binds with nanomolar affinity to the phosphatidylserine (PS) exposed on the plasma membrane of dying cells (5). The protein was used initially as a glutathione-S-transferase (GST) fusion protein, where imaging labels were attached to lysine residues, and used to detect tumor cell death using magnetic resonance and radionuclide imaging (6–9). Subsequently, we developed a much smaller derivative of C2A (16 kDa) based on the isolated C2A domain (C2Am), in which we introduced a unique cysteine residue distant from the PS binding site. This allowed site-specific attachment of imaging labels, without affecting binding affinity, and the production of a chemically homogeneous preparation of the agent. The smaller size allows for relatively rapid renal clearance, the production of a homogeneous agent rules out the possibility of labeled material that does not bind PS, or shows only weak binding, and the removal of the GST tag, by lowering potential immunogenicity, increases clinical translatability. C2Am, when labeled with a fluorophore, showed better specificity for binding to dead cells than a similarly labeled Annexin V derivative (5). Fluorescently labeled Annexin V is widely used to detect cell death in vitro and, in radionuclide-labeled form, has also translated to the clinic. However, clinical use of this agent was limited by poor specificity and relatively high levels of background binding (10).

The relatively high sensitivity, minimal tissue absorption, and low cost of near infrared (NIR) fluorescence imaging have made this a widely used technique for non-invasive imaging in small animals (11) and increasingly in the clinic (12). However, image depth and resolution are limited by scattering, with depths of 1–2 cm and resolutions of 1–2 mm (13). Optoacoustic imaging can give three-dimensional (3D) tomographic images and can generate relatively high-resolution images at depths of a few centimeters. In this technique, NIR radiation is administered in the form of nanosecond laser pulses. Absorption by chromophores in endogenous materials, such as oxy- and deoxyhemoglobin and melanin, or in contrast agents introduced exogenously, results in thermoelastic expansion and the production of a pressure wave that can be detected non-invasively using ultrasound transducers. As pressure waves are relatively low frequency, they are scattered much less than light or NIR radiation and so can be used to generate much higher resolution images. Detectors operating around 5 MHz enable penetration depths of several centimeters with resolutions between 100 and 400 μm, while detectors operating in the 10 to 100 MHz frequency range penetrate 1–10 mm with resolutions in the tens of micrometers (14). In multispectral optoacoustic tomography (MSOT) measurements are made at several NIR wavelengths in less than a second, which enables real-time imaging of individual chromophores within entire tumor volumes. With the introduction of a hand held device, bed-fitted devices and the development of real-time three-dimensional (3D) optoacoustic tomography (14, 15), clinical applications of MSOT have started to emerge, including screening of dense breasts for the presence of tumors (16, 17), monitoring the metastatic status of sentinel lymph nodes in melanoma patients (18) and assessment of inflammatory disease (19).

Tumor necrosis factor (TNF)-related apoptosis-inducing ligand receptor 2 (TRAILR2) is a death receptor upregulated in a wide range of human tumors (20–22), which upon ligation triggers apoptosis through activation of a caspase cascade. MEDI3039 is a multivalent TRAILR2 agonist that can induce cell death in tumor cells at picomolar concentrations (23, 24).

We describe here in vivo imaging measurements of early tumor responses to MEDI3039, and also 5-FU, using NIR fluorophore-labeled C2Am, using both optical imaging and MSOT. Non-specific probe binding was assessed using a site-directed mutant that is inactive in PS binding (iC2Am; ref. 25). The MSOT signal from C2Am was increased significantly in MEDI3039 and 5-FU-treated Colo205 tumors as early as 3 hours after probe injection, whereas there was negligible retention of iC2Am in MEDI3039-treated tumors. Subsequent histologic analyses showed that the C2Am signal observed in vivo was strongly correlated with immunostaining of dead cells in tumor sections.

Human colorectal adenocarcinoma Colo205 cells were cultured in RPMI-1640 medium (Life Technologies Ltd.) and HT-29 in McCoy's 5A (Modified), GlutaMAX medium (Life Technologies Ltd.). Both media were supplemented with 10% fetal bovine serum (FBS; Lonza). The cell lines were purchased from the ATCC in August 2014 and were used within 4 to 8 passages from the original stocks. Both lines tested negative for mycoplasma by a RNA capture ELISA-based method. Cell lines were cultured in a humidified incubator at 37°C and 5% CO₂. The sensitivities of Colo205 and HT-29 cells to MEDI3039 treatment were assessed by measuring cell viability at 22 hours after drug treatment. Briefly, the cells were collected after drug treatment and resuspended to a density of 1 × 10⁶ cells/mL in medium. Ten microliter cell samples were mixed with 10 μL Trypan blue dye solution and cell viability assessed using a cell viability analyzer (Luna Automated Cell Counter, Logos Biosystems). Cell viability was also assessed in Colo-Dual cells, which expressed luciferase (see below), by the addition of D-luciferin solution (25 μg/μL; SynChem Inc.) to a 96-well plate with 5,000 cells per well. Bioluminescence imaging (BLI) measurements were performed 10 minutes later using an IVIS 200 series camera (PerkinElmer) with an F-stop of 1 and open filter. All the samples were prepared in triplicate.

Cells were transduced with a lentiviral vector in which the EF1 promoter drives transcription of the red fluorescent protein, mStrawberry, and firefly luciferase (26). The mStrawberry coding sequence is separated from the luciferase coding sequences by an E2A sequence (EF1-L-S), which results in similar levels of expression of mStrawberry and luciferase. Lentiviruses were produced by co-transfecting HEK 293T cells with the EF1-L-S plasmid and packaging plasmids. Supernatants containing lentiviruses were collected 72 hours after transfection, mixed with polybrene (8 μg/mL), and used to infect Colo205 cells. After 72 hours, cells displaying similar levels of red fluorescence were sorted, using a BD FACSAria cell sorter (BD Biosciences).

Conjugates of C2Am and iC2Am with VivoTag-S 750-MAL (excitation wavelength: 750 ± 5 nm; emission wavelength: 775 ± 5 nm, PerkinElmer) and iC2Am with IRDye 680RD (excitation wavelength: 672 nm; emission wavelength: 694 nm, LI-COR Biosciences) were prepared as described previously (5). This conjugation protocol has been shown previously to produce conjugates of C2Am and iC2Am that were fully modified, yielding a single molecular species on electrospray ionization mass spectrometry (5, 25). The optical absorption spectrum of 1 μmol/L C2Am-750 in phosphate-buffered saline (PBS) was measured using a PHERAStar FS spectrometer (BMG LABTECH).

Following treatment with 1 pmol/L MEDI3039 for 22 hours, Colo205 and HT-29 cells were resuspended in 100 μL of pre-cooled HEPES-buffered saline (10 mmol/L HEPES, 150 mmol/L NaCl, 2 mmol/L CaCl₂, pH 7.4) with 1% FBS. The cell suspensions were incubated with a 1:1 mixture of C2Am-750 and iC2Am-680 at a concentration of 300 nmol/L, and SYTOX Green (Invitrogen; 50 nmol/L) for 20 minutes at 37°C, and then washed twice and kept briefly on ice before being analyzed using an LSR II cytometer (BD Biosciences). For blocking experiments, the cell suspensions were first incubated with an excess of unlabeled C2Am (30 μmol/L) for 30 minutes before the addition of labeled C2Am and iC2Am. Data were analyzed using FlowJo software. MEDI3039 was supplied as a stock solution of 10 mg/mL in PBS.

Colo-Dual cells were cultured in a glass bottomed culture dish (MatTek Crop.) until 80% confluent. Cell death was induced by incubation of the cells with 10 pmol/L MEDI3039 for 5 hours. The cells were then incubated in 200 μL of RPMI-1640 medium with a 1:1 mixture of C2Am-750 and iC2Am-680 at a concentration of 500 nmol/L for 20 minutes at 37°C. The cells were then washed twice gently with phenol red-free RPMI-1640 medium (Life Technologies Ltd.) and imaged using a Leica TCS SP5 confocal microscope (Leica Microsystems Ltd.). Intrinsic NADH autofluorescence was measured using an Argon-UV laser, with an excitation wavelength range of 340 ± 30 nm, and an emission bandpass filter of 411 to 491 nm. mStrawberry fluorescence was measured using a HeNe 543 laser, with an excitation peak at 574 nm, and an emission bandpass filter of 560 to 620 nm.

Female nude mice at approximately 20 g (BALB/c nu/nu, ∼8 weeks old) were acquired from Charles River Laboratories. All surgical and imaging procedures were performed under isoflurane gas anesthesia (3% for induction, 2% for maintenance). Animals were sacrificed by cervical dislocation at the endpoint. All animal experiments were carried out under the authority of project and personal licenses issued by the United Kingdom Home Office under the United Kingdom Animals (Scientific Procedures) Act, 1986, and had been reviewed by the Cancer Research UK, Cambridge Institute Animal Welfare and Ethical Review Body.

For fluorescence imaging two groups of mice (n = 5 each) were implanted with 5 × 10⁶ Colo205 cells. Two weeks later, one group received a single intravenous (i.v.) injection of 0.4 mg/kg MEDI3039. Another group, which served as an untreated control, was injected with solvent vehicle (PBS). After 16 hours, the mice were injected i.v. with a mixture of 0.1 μmol/kg C2Am-750 and 0.1 μmol/kg iC2Am-680 (0.2 mL of a solution containing 10 μmol/L of each probe) and fluorescence images were acquired 0.5 and 3 hours later using an IVIS 200 series camera with an F-stop of 2. The wavelengths used for fluorescence imaging were dictated by the available filter sets on the IVIS camera. (i)C2Am-750 was imaged with an excitation bandpass filter of 705 to 780 nm and an emission bandpass filter of 810 to 885 nm. iC2Am-680 was imaged with an excitation bandpass filter of 615 to 665 nm and an emission bandpass filter of 695 to 770 nm. A correction factor of 3.98 was applied to signals from C2Am-750, when comparing them with iC2Am-680, due to the differences in fluorescence emission from the two fluorophores. The correction factor was calculated by measuring the fluorescence intensities, using the IVIS camera, of iC2Am-680 and C2Am-750 diluted to a concentration of 2 μmol/L in PBS containing 1% FBS. The average radiant efficiency of iC2Am-680 was then divided by the average radiant efficiency of C2Am-750 to obtain a factor that could be used to correct for the more intense fluorescence of the 680 dye. Similar fluorescence measurements were made in groups of mice (n = 3) implanted with 5 × 10⁶ HT-29 cells. Regions of interest were analyzed using Living Image software (PerkinElmer).

For BLI, two groups of mice (n = 5 each) were implanted with 5 × 10⁶ Colo-Dual cells. After 11 days, images were acquired from both groups of animals 10 minutes after intraperitoneal (i.p.) injection of D-luciferin (150 mg/kg) using the IVIS camera with an F-stop of 1 and an open filter. The treated group was then injected with a single dose of 0.4 mg/kg MEDI3039. Untreated animals (n = 5) were injected with drug vehicle (PBS) and images were acquired two to three times per week from both groups.

For optoacoustic imaging, a Multispectral optoacoustic tomography (MSOT) inVision 256-TF small animal imaging system (iThera Medical) was used. Briefly, a tunable optical parametric oscillator (OPO) pumped by an Nd:YAG laser provides excitation pulses with a duration of 9 ns at wavelengths from 660 to 1,300 nm at a repetition rate of 10 Hz with a wavelength tuning speed of 10 ms and a peak pulse energy of 90 mJ at 720 nm. Ten arms of a fiber bundle provide uniform illumination of a ring-shaped light strip of approximately 8 mm width. For ultrasound detection, 256 toroidally focused ultrasound transducers, with a center frequency of 5 MHz (60% bandwidth) and organized in a concave array of 270 degree angular coverage and a radius of curvature of 4 cm, are used (27). Two groups of mice (n = 5 each) were implanted with 5 × 10⁶ Colo205 cells and 2 weeks later received a single dose of 0.4 mg/kg MEDI3039. After 4 hours, the mice were injected with either 0.2 μmole/kg of C2Am-750 or iC2Am-750 and imaged. Another group (n = 3), that were not treated with MEDI3039, were injected with 0.2 μmole/kg of C2Am-750. For imaging, animals were anesthetized and placed on a heated pad before transferring into a custom-made cling film holder (iThera Medical). A small amount of distilled water was placed between the skin of the mouse and the layer of cling film for ultrasound coupling. The holder was placed in the MSOT scanner in a water bath maintained at a temperature of 36°C. MSOT measurements were performed before and at 3, 5, 7, and 24 hours after probe injection, according to a protocol described previously (28). Cross-sectional multispectral image datasets were acquired at different wavelengths in the NIR window (660, 665, 670, 680, 690, 700, 710, 720, 730, 740, 750, 755, 760, 770, 780, 800, 825, 850, and 900 nm) through the entire tumor region in 0.5 mm steps in the z-direction and at a single position where the kidneys were located. Scans at each illumination wavelength took 100 ms to acquire, and 10 averages were used to produce a final image, amounting to 20 seconds acquisition time per slice. Images were reconstructed using a model-based reconstruction algorithm, after which linear spectral unmixing was applied to each set of multi-wavelength images to resolve the biodistribution of the different tissue chromophores, that is, (i)C2Am-750, oxygenated and deoxygenated hemoglobin. Data were processed and quantified using the ViewMSOT software (iThera Medical). Volumetric quantification was done by comparing the signal-to-background ratio after drawing 3D reconstructed volumes of interest (VOI) in the tumor area. Briefly, a signal-to-background ratio was calculated from the mean pixel intensity (MPI) of a VOI, drawn in an area where there was maximum signal, divided by the MPI of an adjacent size-matched VOI with low signal intensity. The volume percentage of tumor bound by C2Am-750 was taken as the volume with C2Am-750 signal above a threshold value (VT), where this was defined as the mean intensity of the background signal in the tumor volume before probe injection, divided by the total tumor volume (V). This was compared with cleaved caspase-3 (CC3) staining, where this was determined in tumor sections taken every 2 to 3 mm through the tumor volume. Corresponding fluorescence imaging measurements were made using an IVIS camera immediately after the MSOT measurements.

Similar fluorescence and MSOT measurements were also made in Colo205 tumor-bearing animals treated with 5-fluorouracil (5-FU). Mice (n = 5) were implanted with 5 × 10⁶ Colo205 cells and then 2 weeks later, fluorescence and MSOT measurements were performed before and 3 hours after i.v. injection of 0.2 μmole/kg of C2Am-750. Twenty-four hours later, the animals received a single i.p. injection (250 mg/kg) of 5-FU and then 24 hours after drug treatment the fluorescence and MSOT measurements were repeated. The mean MSOT voxel intensity in the tumor area was compared before and after 5-FU treatment. For the biodistribution study, the tumors and major organs, including kidney, spleen, heart, lung, liver, muscle, and blood were excised immediately after completion of the in vivo imaging, weighed and then their fluorescence imaged using the IVIS-200 camera.

Tumors were fixed in 4% formaldehyde and embedded in paraffin. Five-μm sections were cut and imaged using the 700 and 800 nm channels of an Odyssey Infrared Imaging scanner (LI-COR). Images were generated at a resolution of 21 μm using two lasers, one emitting at 685 nm and the other at 785 nm. Consecutive sections were stained for CC3, where a rabbit monoclonal anti-CC3 antibody (Cell Signaling Technology) and a donkey anti-rabbit secondary biotinylated antibody (Jackson ImmunoResearch Laboratories) were used in a Polymer Refine Kit on an automated Bond platform (Leica Biosystems Ltd); or they were stained using TdT-mediated dUTP Nick-End Labeling (TUNEL) using a DeadEnd Colorimetric system kit (PromegaBenelux BV). Stained sections were scanned on an Aperio AT2 (Leica Biosystems) at ×20 magnification, with a resolution of 0.5 μm per pixel. All annotations were performed with ImageScope (Leica Biosystems) and the stained surface area was quantified using the algorithm “Positive Pixel Count v9” in Imagescope.

Statistical analyses were performed in GraphPad Prism (GraphPad Software). Data are shown as mean ± SD, unless stated otherwise. A two-tailed Student t test was used for pairwise comparisons. Pearson r test was used to assess the significance of the correlation between CC3 staining and tumor binding of C2Am-750. P values of <0.05 were considered significant.

HT-29 cells were largely resistant to MEDI3039, showing minimal loss of viability up to 10 pmol/L drug concentration and only approximately 25% loss at much higher drug concentrations, whereas Colo205 cells showed approximately 50% loss of viability at low drug concentrations (EC₅₀ 1.3 pmol/L) and complete loss of viability at higher drug concentrations (Fig. 1A). This loss of viability, measured by Trypan blue dye exclusion, was mirrored in Colo-Dual cells by loss of bioluminescence (Fig. 1B and C). Colo-Dual cells were transduced with a lentiviral vector to express firefly luciferase and the red fluorescent protein, mStrawberry. The bioluminescence signal from these cells is indicative of intracellular ATP concentration. Binding of C2Am labeled with a fluorescent dye excited at 750 nm (C2Am-750) and a site-directed mutant inactive in PS binding labeled with a fluorescent dye excited at 672 nm (iC2Am-680) to MEDI3039-treated cells was characterized using confocal microscopy (Fig. 2A) and flow cytometry (Fig. 2B). The specificity of C2Am-750 for PS binding was confirmed by a blocking study with unlabeled C2Am (Supplementary Fig. S1). Confocal microscopy measurements with Colo-Dual cells treated with MEDI3039, showed that when compared with untreated cells, there were relatively low levels of both NADH (intrinsic autofluorescence) and mStrawberry fluorescence, indicating cell death. These cells showed binding of C2Am-750 to the plasma membrane but negligible binding of iC2Am-680. There was negligible binding of C2Am-750 to untreated cells (Fig. 2A). Flow cytometry of MEDI3039-treated TRAIL-sensitive Colo205 cells, but not TRAIL-resistant HT-29 cells, showed increased binding of C2Am-750, which was correlated with a decrease in NADH autofluorescence and an increase in binding of a cell necrosis marker, Sytox Green (Fig. 2B). iC2Am-680 showed negligible binding to either cell line, regardless of drug treatment.

Next, we investigated the capability of C2Am-750 to detect tumor cell death in vivo, in Colo205 and HT-29 xenografts in BALB/c nu/nu mice. Three hours after probe injection, there was a significant increase in C2Am-750 retention in treated as compared with untreated Colo205 tumors, which was observed both in fluorescence measurements made in vivo (5.97-fold, P < 0.001) and on excised tumors ex vivo (4.1-fold, P value < 0.001; Fig. 3). Compared with C2Am-750, the retention of iC2Am-680 was significantly lower in both treated (P < 0.001) and untreated (P < 0.001) tumors. These results were similar to those obtained previously with 5-FU-treated Colo205 tumors (250 mg/kg of body weight; injected i.p.), in which C2Am and iC2Am, both labeled with AlexaFluor-750 C5-maleimide, were injected separately (25). There was no significant difference in C2Am-750 retention in treated and untreated HT-29 xenografts, which was comparable with the retention observed in untreated Colo205 tumors (Supplementary Fig. S2). iC2Am-680 retention was much lower in both groups. Fluorescence images of Colo205 tumor sections showed strong fluorescence signals from C2Am-750 (green) that co-localized with regions stained with TUNEL (a marker of necrotic cell death) and cleaved caspase-3 (CC3, a marker of apoptotic cell death). There was negligible signal from iC2Am-680 (blue), confirming the specificity of C2Am-750 for binding to dead and dying cells (Fig. 3B). Measurements of bioluminescence from Colo-Dual tumors showed that at this single dose of MEDI3039 (0.4 mg/kg), there was a complete loss of signal within 24 hours of drug treatment, which was maintained for up to 45 days after treatment (Fig. 4A). The rapid loss of bioluminescence in treated tumors, presumably due to loss of ATP (Fig. 4B), preceded decreases in tumor volume, which started 1 week after treatment.

Next, we used MSOT to image retention of C2Am-750 and iC2Am-750 in MEDI3039-treated Colo205 tumors. The system used here had a spatial resolution of approximately 150 μm at a penetration depth of 3 cm. Spectral unmixing was used to generate images of C2Am-750, iC2Am-750, and intrinsic tissue chromophores (deoxygenated and oxygenated hemoglobin) throughout the entire tumor volume. Representative MSOT images show that there was some retention of C2Am-750 in untreated tumors (Fig. 5A). However, signal from C2Am-750 was increased markedly at 7 hours after drug treatment and 3 hours after probe injection. Maximal signal from C2Am-750 appeared in the center of the treated tumors (Fig. 5B), whereas there was negligible retention of iC2Am-750 (Fig. 5C). Signal from kidney cortex confirmed that similar concentrations of C2Am-750 and iC2Am-750 had been injected in these animals. Axial, coronal, and sagittal views showed that C2Am-750 was distributed throughout the entire tumor volume (Fig. 5D). Images reconstructed in three dimensions are shown in Figure 5E (video representations of these images are in supplementary data). An MSOT absorption spectrum from a region of high-signal intensity at the center of a MEDI3039-treated tumor obtained 3 hours after injection of C2Am-750 was similar to the optical absorption spectrum of C2Am-750 measured in vitro (Supplementary Fig. S3), demonstrating that the MSOT signal observed in vivo arose predominantly from C2Am-750.

Dynamic MSOT measurements with C2Am-750 and iC2Am-750 in MEDI3039-treated Colo205 tumors and parallel fluorescence images acquired from the same animals are shown in Fig. 6. Both the MSOT (Fig. 6A) and fluorescence images (Fig. 6B) showed tumor retention of C2Am-750 for up to 24 hours following probe injection, whereas there were only very low levels of iC2Am-750 in the tumor at 3 hours and this had cleared by 24 hours. At 24 hours, the tumor signal-to-background ratio in animals injected with C2Am-750, in regions where there were high levels of C2Am-750 binding, increased from 1.3 ± 0.5 prior to probe injection to 15.0 ± 4.0 following probe injection (ratio ± SEM, n = 5), whereas this ratio did not change in animals injected with iC2Am-750 (Fig. 6C). The average volume occupied by bound C2Am in drug-treated tumors at 24 hours after probe injection was correlated with CC3 staining of tumor sections obtained post mortem, where the tumors were excised immediately following imaging (R² = 0.97, P value < 0.01; Fig. 6D). Similar fluorescence signal intensities for C2Am-750 and iC2Am-750 were observed in the kidneys of these animals, confirming again that similar concentrations of the two probes had been injected.

We have shown previously that fluorescently labeled C2Am can detect cell death in EL4 murine lymphoma cells treated with etoposide and MDA-MB-231 human breast cancer cells treated with doxorubicin (5) and that C2Am labeled with fluorescent and radionuclide labels can detect cell death in etoposide-treated EL4 tumors, cyclophosphamide-treated Eμ-myc tumors and Colo205 tumors treated with 5-FU (25). Results of MSOT experiments with C2Am-750 in 5-FU–treated Colo-205 tumors are shown in Supplementary Fig. S4. There was a significant increase in the MSOT signal from C2Am-750 at 24 hours after 5-FU treatment and 3 hours after probe injection. A biodistribution study based on C2Am-750 fluorescence showed a significant increase in C2Am-750 retention in excised tumors compared with that in blood (4.05-fold, P < 0.001). Biodistribution data for fluorophore-labeled (i)C2Am-750 have been reported previously, where we showed that the biodistribution profiles were similar to those of ^99mTc- and ¹¹¹In-labeled C2Am (25).

Cell death, whether by apoptosis or necrosis, has attracted considerable attention as an imaging target as it is a generic marker for the presence of disease and additionally, in the case of cancer, a marker of response to treatment (4, 29). A number of cell death imaging agents have been developed, with some interacting with intracellular targets, such as CC3 (30), cytosolic proteins (31, 32), and the mitochondrial membrane potential (33); or with extracellular targets, including exposed extracellular DNA (34), histones (35) and plasma membrane phospholipids (6, 36). Plasma membrane phospholipids, such as PS and phosphatidylethanolamine (PE), which are normally present on the inner leaflet of the plasma membrane bilayer, are exposed on the surface of apoptotic cells and in necrotic cells by permeabilization of the plasma membrane to the imaging agent (37, 38). These phospholipids are an attractive target as in necrotic cells their exposure is persistent, which makes timing of imaging after cell death less critical, and they are abundant and, therefore, potentially capable of giving high signal-to-noise when bound by the imaging probe (7, 39). In a previous study, titration of isolated apoptotic cells with fluorescently labeled C2Am gave an exposed PS concentration of 100 to 300 pmol/10⁶ cells (5), which assuming a tumor cell density in vivo of approximately 10⁸ cells/mL, corresponds to a PS concentration in apoptotic tumor tissue of 10 to 30 μmol/L. This concentration is considerably in excess of the estimated minimum MSOT-detectable dye concentration in mouse tissue of 0.5 to 2 μmol/L (14).

We have shown previously that C2A, when conjugated to labels detectable by magnetic resonance (6, 9), radionuclide, and fluorescence imaging (25), is capable of detecting tumor cell death in vivo by binding to exposed PS. We have also shown that C2Am is more specific than Annexin V in detecting cell death, which also binds PS, as C2Am shows less binding to viable cells (5). The specificity of C2Am for detecting cell death was confirmed here by fluorescence measurements performed with C2Am and iC2Am in vitro and in vivo using MEDI3039-sensitive (Colo205) and resistant (HT-29) cell lines. Flow cytometry showed negligible binding of iC2Am and increased binding of C2Am to drug treated Colo205 cells but not to HT-29 cells. These findings were confirmed in implanted tumors in vivo and ex vivo by planar fluorescence imaging. Histologic analysis of Colo205 tumor sections confirmed binding of C2Am-750 to apoptotic and necrotic cells.

Several targeted contrast agents for optoacoustic imaging of tumors have been developed in recent years (14, 40, 41), including dye-conjugated EGF, which was used to image EGF receptor expression in an orthotropic pancreatic cancer model (42), and a dye-conjugated cyclic peptide, which was used to image expression of the integrin, α_vβ_3, in an orthotropic brain tumor model (43). There has been a recent preliminary study in which a PS-binding antibody was used to image cell death in an implanted human breast cancer model in nude mice (44). However, the injected concentration was not specified nor was the statistical significance of the increase in signal intensity, which only appeared to be increased significantly at a time when the tumor was already shrinking. Moreover, there was no histologic validation, or control measurements with a non-PS binding IgG.

We have shown here that MSOT measurements with an NIR fluorophore-labeled PS binding protein, C2Am-750, can be used to create 3D images of tumor cell death in human colorectal xenografts treated with a novel TRAILR2 agonist, MEDI3039. The C2Am-750 signal in treated Colo205 tumors increased as early as 3 hours after probe injection, and at 24 hours, the signal-to-background ratio, for those regions showing the greatest binding, was >10. There was negligible retention of a site-directed mutant of C2Am, which is inactive in PS binding (iC2Am-750; Fig. 6A–C). Positive MSOT signal showed a good correlation with staining of apoptotic cells in tumor sections.

A limitation of the current study was the very high levels of tumor cell death following treatment with MEDI3039. However, we have shown previously that fluorescence imaging with C2Am labeled with AlexaFluor-750 C5-maleimide was capable of detecting 5-FU-induced cell death in Colo205 tumors, where CC3 staining in tumor sections increased from 1% to 2% before treatment to 2% to 5% after treatment (25). Moreover, we have shown here that MSOT imaging with C2Am labeled with another NIR fluorophore was also capable of detecting 5-FU–induced tumor cell death. Tumor cell death in the clinic can range from less than 2% before treatment to 5% to 15% after treatment (45).

Optoacoustic imaging with the C2Am probe described here has the potential to be used in the clinic to assess early treatment response in relatively superficial tumors, such as in the breast (46), and in an endoscopic format (47), for assessing treatment response in tumors in the gastrointestinal tract.

The raw data acquired during this study and on which the results presented in this paper are based can be found at https://doi.org/10.17863/CAM.13025.

M.R. Tomaszewski reports receiving commercial research support from iTheresa Medical. S.R. Mullins has ownership interests (including patents) at AstraZeneca and reports receiving commercial research support from MedImmune. S.E. Bohndiek reports receiving commercial research support from iTheresa Medical. K.M. Brindle reports receiving commercial research support from MedImmune. K.M. Brindle and A.A. Neves are listed as coinventors on a patent on C2 for detecting cell death, which is owned by Cambridge Enterprise. No potential conflicts of interest were disclosed by the other authors.

Conception and design: B. Xie, S.R. Mullins, D. Tice, R.C.A. Sainson, R.W. Wilkinson, K.M. Brindle

Development of methodology: B. Xie, M.R. Tomaszewski, A.A. Neves, S. Ros, S.E. Bohndiek

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): B. Xie, M.R. Tomaszewski, D.-E. Hu, S. McGuire, S.E. Bohndiek

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): B. Xie, A.A. Neves, S.E. Bohndiek, R.W. Wilkinson

Writing, review, and/or revision of the manuscript: B. Xie, A.A. Neves, S.R. Mullins, D. Tice, R.C.A. Sainson, S.E. Bohndiek, R.W. Wilkinson, K.M. Brindle

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B. Xie, S.R. Mullins

Study supervision: B. Xie, A.A. Neves, S.R. Mullins, R.C.A. Sainson, K.M. Brindle

The authors thank the CRUK Cambridge Institute's Histopathology, Flow Cytometry, Microscopy and Research Instrumentation Core Units, for their technical support and James Joseph, Neal Burton, and Luis Santana for their advice on MSOT data analysis.

The work was supported by a Cancer Research UK Programme grant (17242) and a Project grant from MedImmune (K.M. Brindle) and by the CRUK-EPSRC Imaging Centre in Cambridge and Manchester (16465; K.M. Brindle and S.E. Bohndiek).

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.