Purpose: Apoptosis, or programmed cell death, can be leveraged as a surrogate measure of response to therapeutic interventions in medicine. Cysteine aspartic acid–specific proteases, or caspases, are essential determinants of apoptosis signaling cascades and represent promising targets for molecular imaging. Here, we report development and in vivo validation of [18F]4-fluorobenzylcarbonyl–Val–Ala–Asp(OMe)–fluoromethylketone ([18F]FB-VAD-FMK), a novel peptide-based molecular probe suitable for quantification of caspase activity in vivo using positron emission tomography (PET).
Experimental Design: Supported by molecular modeling studies and subsequent in vitro assays suggesting probe feasibility, the labeled pan-caspase inhibitory peptide, [18F]FB-VAD-FMK, was produced in high radiochemical yield and purity using a simple two-step, radiofluorination. The biodistribution of [18F]FB-VAD-FMK in normal tissue and its efficacy to predict response to molecularly targeted therapy in tumors was evaluated using microPET imaging of mouse models of human colorectal cancer.
Results: Accumulation of [18F]FB-VAD-FMK was found to agree with elevated caspase-3 activity in response to Aurora B kinase inhibition as well as a multidrug regimen that combined an inhibitor of mutant BRAF and a dual PI3K/mTOR inhibitor in V600EBRAF colon cancer. In the latter setting, [18F]FB-VAD-FMK PET was also elevated in the tumors of cohorts that exhibited reduction in size.
Conclusions: These studies illuminate [18F]FB-VAD-FMK as a promising PET imaging probe to detect apoptosis in tumors and as a novel, potentially translatable biomarker for predicting response to personalized medicine. Clin Cancer Res; 20(8); 2126–35. ©2014 AACR.
Deviations from normal cell death programs tend to promote cell survival and are frequently associated with cancer. Many anticancer medicines aim to selectively induce cell death in tumor cells, but highly validated noninvasive biomarkers to assess such molecular events are lacking. This study reports a novel positron emission tomography (PET) molecular imaging agent that allows noninvasive, targeted detection of caspase activation and tumor cell death following effective drug treatment. The probe evaluated here was studied within the context of molecularly targeted therapy, but could be equivalently effective to evaluate response to conventional therapeutics and thus used to predict individualized responses in patients and accelerate the development of improved cancer therapies.
Cell death proceeds through multiple, mechanistically distinct processes that include necrosis, autophagy, mitotic catastrophe, and apoptosis (1, 2). Apoptosis, or programmed cell death, is an orchestrated process that facilitates elimination of unnecessary, damaged, or compromised cells to confer an overall advantage to the host organism. As such, apoptosis is an essential component of embryonic development, tissue homeostasis, and immunologic competence. Deviations from normal apoptotic programs are frequently associated with human diseases such as cancer (3). Because many anticancer therapies aim to selectively induce apoptosis in tumor cells (4, 5), quantitative, noninvasive imaging biomarkers that reflect apoptosis represent promising tools to improve drug discovery and predict early responses in patients (6–10).
Clinically robust molecular imaging biomarkers of apoptosis have been sought after for many years, but none have yet proven optimal. Classically, molecular imaging measures of apoptosis have relied upon labeled forms of the 36-kDa protein Annexin V, which binds to externalized phosphatidylserine on the plasma membrane of cells undergoing apoptosis (7, 11). Although functionalization of Annexin V for optical (9, 10), single-photon emission computed tomography (SPECT; refs. 12, 13), and positron emission tomography (PET; refs. 14, 15) imaging have been reported, imaging probes based on Annexin V generally suffer from limitations that include suboptimal biodistribution and pharmacokinetics, calcium ion dependency (16), and a lack of specificity (17–19). Another promising approach that capitalizes upon cell membrane alterations associated with apoptosis uses a small molecule known as 18F-ML-10, which has been evaluated in a limited number of patients (20, 21). The strengths and weaknesses of this approach are under investigation at a number of institutions.
Other intracellular molecular targets within the apoptosis signaling cascade represent opportunities for molecular probe development. For example, as regulators of extrinsic and intrinsic apoptosis, caspases have been suggested as promising targets for molecular imaging and for drug development (6, 22). The goal of this study was to explore the utility of a peptide-based pan-caspase inhibitor, the modified tripeptide sequence Val–Ala–Asp(OMe)–fluoromethylketone (VAD-FMK; refs. 23, 24), to serve as the basis for developing a PET imaging probe for detecting apoptosis. To this end, we report that [18F]4-fluorobenzylcarbonyl–Val–Ala–Asp(OMe)–fluoromethylketone ([18F]FB-VAD-FMK), a novel and potentially translatable molecular probe, enables quantification of caspase activity in vivo using PET imaging. Using small animal microPET imaging of multiple models of human colorectal cancer, we demonstrate that in vivo tumor accumulation of [18F]FB-VAD-FMK accurately reflects elevated caspase-3 activity in response to Aurora B kinase inhibition as well as a multidrug regimen that combined an inhibitor of mutant BRAF and a dual PI3K/mTOR inhibitor. Furthermore, though these studies illuminate [18F]FB-VAD-FMK as a promising tool compound, more generally, these studies suggest that labeled VAD-FMK peptides warrant further development as potential scaffolds for development of translational molecular imaging probes.
Materials and Methods
Co-crystal structures for the caspase-3 protein with an aza-peptide epoxide inhibitor (PDB ID 2CNN; ref. 25) and a covalently bound β-strand peptidomimetic inhibitor (PDB ID 3KJF; ref. 26) were obtained from the Brookhaven Protein Data Bank and were used to evaluate the potential of prosthetically labeled VAD-FMK, and deviations thereof, to be accommodated within the active caspase-3–binding domain. Structural alignment of both caspase-3 inhibitor co-crystal structures was performed using PyMOL (Molecular Graphics System, Version 22.214.171.124; Schrödinger, LLC) to reveal a minor 0.22-Å α-carbon protein backbone coordinate root mean squared deviation (RMSD). Both caspase-3 inhibitor–binding sites were treated as energetically equivalent and the smaller β-strand peptidomimetic inhibitor structure (PDB ID 3KJF) was chosen as the starting point for molecular docking calculations. Following removal of the co-crystallized covalent inhibitor ligand, molecular docking was performed using the high resolution, flexible SurflexDock GeomX protocol as implemented in SYBYL-X 2.0 (Tripos International), with a protomol target based on the β-strand peptidomimetic inhibitor structure with the addition of a 2.0-Å bloat parameter. Best scoring energy-minimized docked poses were ranked using the Tripos SYBYL Surflex-Dock Total Score and PF (protein flexibility) Score using a sort by decreasing Crash scoring component. FB-VAD-FMK, benzyloxycarbonyl-Val-Ala-Asp(OMe)-fluoromethylketone (Z-VAD-FMK), 4-iodobenzyloxycarbonyl-Val-Ala-Asp(OMe)- fluoromethylketone (IZ-VAD-FMK), and VAD-FMK caspase-3 docked complex scores were compared as physically-plausible, low-energy-minimized conformations in the bound state.
Peptide and derivatives
[19F]FB-VAD-FMK and [127I]IZ-VAD-FMK were synthesized as described in the Supplementary Data. VAD-FMK (American Peptide) and Z-VAD-FMK (TOCRIS Bioscience; #2163) were obtained commercially and used without further purification. Relative enzyme selectivity of [19F]FB-VAD-FMK compared to parent peptide was evaluated as described in the Supplementary Data. Lipophilicity measurements (Log P7.5) of peptide derivatives were determined analogously to reported methods (27) and described in the Supplementary Data.
Biochemical caspase-3 inhibition assay
Caspase-Glo 3/7 (Promega) was used as a readout of inhibition whereby a caspase-cleavable protected luciferase substrate is directly sensitive to caspase-3/7 activity and quantifiable by bioluminescence (28). Caspase inhibitor dissolved in dimethyl sulfoxide (DMSO; 0, 0.1, 1, 10, 100, 1,000, or 10,000 nmol/L) was combined with recombinant human caspase-3 enzyme (C1224–10UG; Sigma; 100 nmol/L) in 1× PBS in microcentrifuge tubes, vortexed, and incubated at 37°C for 30 minutes. After incubation, Caspase-Glo 3/7 reagent (Promega) was added in accordance with the manufacturer's instructions. Solutions were vortexed, incubated for an additional 30 minutes, and dispensed into opaque wall/bottom 384-well plates (BD Biosciences) for measurement of caspase activity. Sample luminescence was measured using a Synergy 4 plate reader (BioTek).
Cellular caspase inhibition assay
DiFi cells were propagated in Dulbecco's Modified Eagle Medium (DMEM; Mediatech) and supplemented with 10% FBS (Atlanta Biologicals) and 1 mg/mL gentamicin sulfate (Gibco) in a 95% humidity, 5% CO2, at 37°C atmosphere. Cells were seeded as subconfluent monolayer cultures at a density of 1 × 104 per well into 96-well, black wall/bottom plates (BD Biosciences) and allowed to adhere for 24 hours at 37°C. For evaluation of caspase-3/7 inhibition concomitant with drug exposure, DiFi cells were incubated with cetuximab (0.5 μg/mL) for 24 hours, based on our experience from previous work (9). Cell media was then replaced with 1× PBS containing inhibitor (0, 10, 100, 1,000, and 10,000 nmol/L) and incubated at 37°C for 30 minutes. Inhibition was assessed using Caspase-Glo 3/7 reagent in accordance with the manufacturer's instructions. Luminescence was quantified using a Synergy 4 plate reader.
Detailed chemical and radiochemical methods and characterization can be found in the Supplementary Data. The radiochemical intermediate [18F]N-succinimidyl-4-fluorobenzoate ([18F]SFB), was prepared using commercial GE TRACERlab-Synthesizer MX production kits (ABX) with a BIOSCAN Coincidence radiochemistry module. [18F]SFB was then conjugated to VAD-FMK forming [18F]FB-VAD-FMK.
Studies involving animals were conducted in accordance with federal and institutional guidelines. Biodistribution studies were performed using male C57BL/6 mice. A detailed description of in vitro drug response studies that support the subsequently described animal models and drug therapies is provided in the Supplementary Data.
Dual xenograft-bearing mice were sequentially injected with 1 × 107 SW620 and 5 × 106 DLD-1 human colorectal cancer cells subcutaneously onto the left or right hind limbs (respectively) of 5- to 6-week-old female, athymic nude mice (Harlan Sprague-Dawley). Palpable tumors were observed for 2 to 3 weeks following inoculation. Animals bearing SW620 and DLD-1 xenograft tumors were treated with vehicle or AZD-1152 (25 mg/kg, DMSO) via intraperitoneal injection once daily for 5 days, analogous to previous work (29). In vivo PET imaging studies were performed on day 5, approximately 4 to 6 hours after treatment.
COLO-205 and LIM-2405 xenografts were generated by subcutaneously injecting 1 × 107 cells onto the right flank of 5- to 6-week-old female, athymic nude mice (Harlan Sprague-Dawley). Palpable tumors were observed within 3 weeks following inoculation. For BEZ-235 (Selleckchem) and PLX-4720 (synthesized using reported methods; ref. 30) single-agent treatment, animals were administered vehicle, BEZ-235 (35 mg/kg, 0.1% Tween 80, and 0.5% methyl cellulose), or PLX-4720 (60 mg/kg, DMSO) via oral gavage once daily for 4 days. For BEZ-235 and PLX-4720 combination treatment, BEZ-235 (35 mg/kg, 0.1% Tween 80, and 0.5% methyl cellulose) and PLX-4720 (60 mg/kg, DMSO) were administered via oral gavage as separate doses (respectively) approximately 7 to 8 hours apart. To monitor tumor growth, tumor volumes were measured on days 1 and 4 of treatment using a previously established ultrasound imaging–based methodology (31). In vivo imaging studies were performed on the fourth day (PET), based on our previous experience with BRAF inhibition in vivo (32).
In vivo PET imaging and analysis
Imaging acquisition and processing was performed analogously to our previously reported methods (32). Further details are reported in the Supplementary Data.
Tumor tissues were harvested immediately following conclusion of imaging, fixed for 24 hours in 5% buffered formalin, and blocked in paraffin. Immunohistochemistry for cleaved caspase-3 (Cell Signaling Technology, #9664) was carried out as previously described (9). Tissues were stained using standard hematoxylin and eosin (H&E) methods and reviewed by an expert gastrointestinal pathologist (M.K. Washington). Images displayed are representative of three randomly selected high-power fields (×40). Semiquantitative immunohistochemical (IHC) analysis was performed using the image-processing software ImageJ and is described in the Supplementary Data.
Unless otherwise stated, experimental replicates are reported as the arithmetic mean ± SD. Statistical significance of in vitro and in vivo datasets was evaluated using an unpaired, two-tailed t test. Differences were assessed within the GraphPad Prism 6.01 software package and considered statistically significant if P < 0.05.
Caspase-3 active site accommodates N-terminal functionalization of VAD-FMK
Imaging labels should impart minimal, if any, effects upon the biologic and chemical properties of the parent molecule. Given this, we initially used a molecular modeling approach to explore multiple N-terminally functionalized analogues of VAD-FMK (Fig. 1A), two of which would result in PET/SPECT imaging probes: FB-VAD-FMK (Fig. 1B), Z-VAD-FMK, IZ-VAD-FMK (33), and VAD-FMK. Molecular docking of each peptide into an irreversible inhibitor-bound protein structure of caspase-3 (PDB ID, 3KJF; ref. 26) enabled predictions to be drawn about the accommodation of modified peptides into the caspase-3 active site. Flexible docking calculations were performed using postdocking energy minimization for each peptide to assess potential energetic perturbations that result from N-terminal modification. Only docked poses that positioned the C-terminal fluoro-methyl ketone moiety within 5.0 Å of the caspase-3 active site cysteine 163 thiol moiety were considered in our analysis. All four peptides exhibited predicted binding energy scores that fell within the micromolar to nanomolar range (Total Score) and ranked by the steric bump parameter (Crash) using SurflexDock in Tripos SYBYL-X 2.0 (Supplementary Table S1). The Surflex-Dock scoring term for Polar contacts suggested that the FB-VAD-FMK peptide had greater potential to form hydrogen bonds with active caspase-3 site residues relative to other candidate structures, including the parent peptide, while displaying less internal ligand Strain scores. Of note, IZ-VAD-FMK produced docking scores that reflected a lower-scoring Polar term and demonstrated that more poses buried the larger, lipophilic iodo-moiety toward the protein active site, suggesting greater lipophilicity than FB-VAD-FMK. Although exploratory in nature, these studies suggested both tolerance to functionalization of VAD-FMK at the N-terminus with the FB prosthesis and that the resulting PET imaging probe might possess favorable physical and chemical properties relative to parent peptide and the IZ-labeled form.
Enzyme selectivity of VAD-FMK peptide analogues
From the binding mechanism of VAD-FMK-type peptides (23, 34), we anticipated that FB-VAD-FMK would exhibit caspase selectivity similar to that of the parent peptide. To explore this, we evaluated the relative affinity of [19F]FB-VAD-FMK against activated caspases-3/6/7/8 (Supplementary Fig. S1). Addition of the FB prosthesis had little impact on caspase selectivity compared with that of the parent peptide, where both peptides inhibited caspases-3/6/7 with single micromolar potency and caspase-8 with slightly greater potency. Caspase-3 was used in subsequent characterization and validation studies as a representative of other relevant caspases.
Lipophilicity of VAD-FMK peptide analogues
To validate the predicted physical properties of the labeled VAD-FMK derivatives, lipophilicity studies were undertaken using [19F]FB-VAD-FMK (Supplementary Fig. S2A), Z-VAD-FMK, and [127I]IZ-VAD-FMK (Supplementary Fig. S2B). We determined that the FB-modified peptide (log P7.5 = 1.41) was between 50 and 100 times less lipophilic than both Z-VAD-FMK (log P7.5 = 2.20) and [127I]IZ-VAD-FMK (log P7.5 = 2.38), in support of the predicted Polar contact scores determined by molecular modeling. These findings also agree with a previous report of IZ-VAD-FMK, which suggested this compound to be too lipophilic for in vivo use (33).
[19F]FB-VAD-FMK potently inhibits active caspase activity
The biologic activity of labeled VAD-FMK derivatives was validated with recombinant caspase-3 enzyme using a commercially available chemiluminescent caspase activity assay. Nonlinear regression analysis of the resultant inhibitory profiles yielded a mean (n ≥ 3) IC50 of approximately 225 ± 70 nmol/L for [19F]FB-VAD-FMK (Fig. 1C). Although all peptides exhibited reasonable potencies that were in line with predicted affinities, [19F]FB-VAD-FMK demonstrated the greatest potency toward caspase-3 inhibition. Similar to biochemical analysis, [19F]FB-VAD-FMK inhibited caspase-3/7 activity at nanomolar concentrations analogously to the parent in colorectal cancer cells (DiFi) in log-phase growth (Fig. 1D and E). When apoptosis was induced in DiFi cells by exposure to the EGF receptor (EGFR) monoclonal antibody cetuximab (9), [19F]FB-VAD-FMK inhibited caspase-3/7 activity with comparable efficacy to the parent peptide. Combined with studies demonstrating acceptable physical properties, these investigations illustrated that [19F]FB-VAD-FMK exhibited caspase affinity similar to the parent VAD-FMK peptide and was subsequently prioritized for radiochemical development.
In vivo normal tissue uptake of [18F]FB-VAD-FMK
Radiochemical preparation of [18F]FB-VAD-FMK (Supplementary Fig. S3) was performed as described in the Supplementary Data. The in vivo biodistribution of [18F]FB-VAD-FMK was evaluated by ex vivo tissue counting at 60 minutes after administration and correlative PET imaging up to 75 minutes after administration. Upon intravenous administration, [18F]FB-VAD-FMK was widely distributed throughout a range of normal tissues, with the greatest activity in kidneys and liver after 60 minutes of uptake (Fig. 2A). Very little radioactivity was found in brain, lung, or bone. Summed dynamic PET acquisitions following [18F]FB-VAD-FMK administration, 0 to 75 minutes (Fig. 2B and C), agreed with tissue-counting measurements and provided evidence of renal and hepatobiliary excretion. This was confirmed by high-performance liquid chromatography (HPLC) radiometabolite analysis, using a protocol analogous to that described in the Supplementary Data for radiochemical purity analysis, which revealed largely parent compound, approximately 50% or greater, in bile and urine samples (n = 3) at 60 minutes after injection.
[18F]FB-VAD-FMK PET reflects AZD-1152–dependent caspase-3 activity in tumors
In vivo uptake of [18F]FB-VAD-FMK in tumor was evaluated in SW620 and DLD-1 human colorectal cancer cell line xenografts given the in vitro data which demonstrated, in concert with polyploidy (Fig. 3A), AZD-1152-HQPA concentration-dependent increases in cleaved PARP and cleaved caspase-3 levels (Fig. 3B). These results, in addition to previously reported in vivo findings (35), suggest that quantification of caspase activity may reflect response to Aurora B kinase inhibition in these models.
In vivo [18F]FB-VAD-FMK PET was explored as a means to reflect response to AZD-1152 compared with vehicle. Animals were treated for 5 days and subjected to imaging after treatment on the fifth day. Although vehicle treatment did not result in significant accumulation of [18F]FB-VAD-FMK in either xenograft model, AZD-1152 treatment increased [18F]FB-VAD-FMK uptake relative to vehicle in SW620 xenografts (0.79 ± 0.15%ID/g and 0.42 ± 0.25%ID/g respectively; P = 0.030; Fig. 4A–C), which was in agreement with in vitro studies. Interestingly, [18F]FB-VAD-FMK uptake was absent in areas of central necrosis, as evident from three-dimensional PET images (Fig. 4A). Interestingly, unlike in vitro studies that illustrated the sensitivity of DLD-1 cells to AZD-1152, probe accumulation was similar between AZD-1152- and vehicle-treated DLD-1 xenografts (0.57 ± 0.14 %ID/g and 0.49 ± 0.22 %ID/g respectively; P = 0.510).
To validate imaging, xenograft tissues were harvested for histology immediately following imaging. In agreement with [18F]FB-VAD-FMK PET, caspase-3 immunoreactivity seemed modest in both vehicle-treated SW620 and DLD-1 xenografts (Fig. 4D). Furthermore, AZD-1152 treatment led to elevated caspase-3 immunoreactivity compared with vehicle treatment in SW620 but not DLD-1 xenografts as verified with semiquantitative IHC analysis (Supplementary Fig. S4A). AZD-1152–treated SW620 xenografts demonstrated evidence of drug exposure given the presence of enlarged nuclei evident by H&E staining (35). Conversely, the lack of in vivo effects of AZD-1152 in DLD-1 xenografts, predicted by [18F]FB-VAD-FMK PET, could possibly be attributed to poor drug exposure as little evidence of polyploidy was observed by H&E staining (Fig. 4D).
[18F]FB-VAD-FMK PET reflects response to combination therapy
Rational combination therapy for V600EBRAF melanoma (36, 37) and colon cancer (32) include an inhibitor of mutant BRAF and a phosphoinositide 3-kinase (PI3K) family inhibitor. Leveraging this concept, [18F]FB-VAD-FMK uptake was evaluated in V600EBRAF-expressing colorectal cancer xenograft-bearing mice (COLO-205 and LIM-2405) treated with vehicle, single-agent PI3K/mTOR inhibitor (BEZ-235), single-agent BRAF inhibitor (PLX-4720), or combination therapy (BEZ-235/PLX-4720). Animals were imaged by PET after treatment on day 4. Elevated probe uptake in COLO-205 xenografts, relative to vehicle treatment (0.84 ± 0.16%ID/g), was observed in the dual-agent–treated cohort (1.54 ± 0.55%ID/g; P = 0.007) but not for either the BEZ-235 or PLX-4720 single-agent cohorts (0.94 ± 0.09%ID/g, P = 0.14 and 0.87 ± 0.13%ID/g, P = 0.69, respectively; Fig. 5A and B). Strikingly, probe uptake in LIM-2405 xenografts was nondifferential, relative to vehicle treatment (0.95 ± 0.13%ID/g), across treatment cohorts: 1.08 ± 0.25%ID/g, P = 0.22 (BEZ-235), 0.92 ± 0.23%ID/g, P = 0.81 (PLX-4720), and 0.78 ± 0.20%ID/g, P = 0.10 (BEZ-235/PLX-4720). Representative xenograft tissue was harvested immediately following imaging. In agreement with [18F]FB-VAD-FMK PET, caspase-3 immunoreactivity seemed modest in vehicle- and single agent–treated COLO-205 and LIM-2405 xenografts, whereas combination treatment increased caspase-3 immunoreactivity in COLO-205 but not LIM-2405 xenografts (Fig. 5C). These observations were confirmed with semiquantitative IHC analysis (Supplementary Fig. S4B). In concert with elevated probe uptake, after 4 days of treatment, a significant reduction in tumor size (P = 0.023) was observed in COLO-205 xenografts treated with combination therapy (Fig. 5D). In contrast, statistically significant changes in tumor growth compared with vehicle were not found for single-agent therapies in COLO-205 xenografts or any drug therapies in LIM-2405 cohorts.
Irreversible caspase-binding ligands are frequently used as tool compound inhibitors in vitro and in vivo (23, 24). Among the known inhibitors, we sought to extend the utility of VAD-FMK–type peptides to PET imaging probe development. Mechanistically, VAD-FMK peptides irreversibly bind active caspase through condensation of the fluoromethyl ketone moiety with a cysteine thiol, Cys163 for caspase-3, within the active site, permanently inactivating the enzyme (26, 38). As a molecular imaging probe, irreversible binding may confer certain advantages, such as enhanced retention in apoptotic versus healthy cells, due to slow dissociation kinetics. We believe this study is the first to report development and in vivo validation of a novel PET imaging probe derived from the VAD-FMK peptide sequence. In general, the VAD-FMK sequence is known to be quite versatile and has been previously functionalized with fluorophores (39–42) and a radioisotope (33) for in vitro and certain preclinical applications. The probe developed here possesses all of the inherent advantages of a PET imaging agent, which include sensitivity, depth, and quantification (43, 44). Furthermore, the FB derivative exhibits certain physical properties, such as improved water solubility, which was a previously reported limitation of another reported labeled peptide IZ-VAD-FMK.
Inhibitor–caspase interactions have been reported previously (26) and provide reasonable structural templates for the design of inhibitor-based probes. The molecular modeling approaches used in this study were confirmed by in vitro biochemical studies and suggested that FB-VAD-FMK would be accommodated in the caspase-3 active site without significant perturbation of physically plausible low-energy complex structures. We interpret our results, which identified a best-scoring docked pose for FB-VAD-FMK with the FB prosthesis pointed away from the active site and toward solvent, to suggest that further chemical modification of the N-terminus may be feasible in an effort to tune the physical properties of the peptide and optimize biodistribution in future studies. These studies suggest that the imaging prosthesis should likely be tuned to match the solvent/protein milieu. In support of the modeling hypothesis, in vitro biochemical and cell assays validated that [19F]FB-VAD-FMK maintained similar potency against caspase-3/7 compared with the parent peptide and exhibited favorable physical properties.
Using established radiochemical methods, [18F]FB-VAD-FMK was produced with activity and purity suitable for use in small animal PET imaging studies. [18F]FB-VAD-FMK was initially evaluated in vivo in two colorectal cancer cell line models (SW620 and DLD-1) in response to prodrug (AZD-1152) treatment. Upon conversion to the active form of AZD-1152-HQPA in plasma, AZD-1152-HQPA inhibits Aurora B kinase activity (35, 45) and induces 4N DNA accumulation, endoreduplication, and polyploidy (35). As demonstrated here and elsewhere (29), these events lead to apoptosis-induced cell death and elevated caspase activity in SW620 tumor cells. We demonstrated that [18F]FB-VAD-FMK PET could be used to monitor response in this setting and that imaging accurately reflected elevated levels of apoptosis in prodrug- versus vehicle-treated tumors and was validated by immunohistochemical analysis. Caspase-3 activity corresponded accordingly with evidence of drug exposure as determined by tumor polyploidic events in tissue. Interestingly, in DLD-1 xenografts, [18F]FB-VAD-FMK uptake was not differential between drug- and vehicle-treated tumors. Tissues collected from these animals agreed with these findings and revealed little evidence of drug activity, as noted by the absence of polyploidy and caspase-3 activity. Thus, in this model, [18F]FB-VAD-FMK PET quantization reflected a lack of efficacy that seemed to be the result of poor therapeutic delivery.
Next, [18F]FB-VAD-FMK PET was used to evaluate response to a multidrug therapeutic regimen, which included an inhibitor of mutant BRAF and a PI3K/mTOR inhibitor in V600EBRAF-expressing colorectal cancer models. A noninvasive imaging metric that can be used to elucidate the basis of response to complicated therapeutic multidrug regimens would be very attractive within the setting of drug development clinical trials. [18F]FB-VAD-FMK PET imaging demonstrated that combination therapy led to elevated apoptosis in one of two models compared with vehicle- or single agent–treated colorectal cancer xenograft-bearing mice. We found that changes in tumor growth closely correlated with levels of caspase activity detected by both [18F]FB-VAD-FMK PET and IHC analysis.
These studies illuminate the potential for caspase-specific PET imaging probes, such as [18F]FB-VAD-FMK, to be used in the assessment of early clinical response to therapeutics. A possible limitation of this probe is the relatively modest uptake observed in target tissue responding to therapy (0.8–2.5 %ID/g). However, other caspase-targeted PET agents reported in the literature and under clinical development are also known to exhibit modest uptake, including the promising isatin sulfonamide chemical class (46–48) and a PEG-functionalized “DEVD-A” penta-peptide analogue, [18F]-CP18 (49, 50). From the current work, as well as that from previous reports (46–50), it is not clear that improvements in the overall tumor uptake, which might result in greater signal-to-noise ratio, would result in a probe that better reflects the underlying determinants of apoptosis. Future head-to-head comparisons would be beneficial toward sorting this out. Future studies should also evaluate the advantages and limitations inherent in the use of a pan-caspase targeting probe, where extrinsic and intrinsic apoptosis pathways are indistinguishable, compared with more selective compound classes.
These studies illuminate the VAD-FMK peptide as a promising scaffold for molecular imaging of caspase activity and response to molecularly targeted therapy using PET. Among these peptides, [18F]FB-VAD-FMK seems to be an attractive PET imaging probe for noninvasive quantification of apoptosis in tumors and may represent a potentially translatable biomarker of therapeutic efficacy in personalized medicine.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: M.R. Hight, M.L. Nickels, E.S. Dawson, S. Saleh, H.C. Manning
Development of methodology: M.R. Hight, Y.-Y. Cheung, M.L. Nickels, S. Saleh, D. Tang
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.R. Hight, M.L. Nickels, P. Zhao, S. Saleh
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.R. Hight, E.S. Dawson, S. Saleh, R.J. Coffey, H.C. Manning
Writing, review, and/or revision of the manuscript: M.R. Hight, Y.-Y. Cheung, M.L. Nickels, E.S. Dawson, S. Saleh, J.R. Buck, D. Tang, M.K. Washington, R.J. Coffey, H.C. Manning
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P. Zhao
Study supervision: H.C. Manning
The authors gratefully acknowledge Frank Revetta for performing the immunohistochemistry. Donald D. Nolting, Gary L. Johnson, and Ronald M. Baldwin are acknowledged for helpful discussions. Md. Imam Uddin is acknowledged for assistance with chemical synthesis and M. Noor Tantawy and George H. Wilson for assistance with PET acquisitions.
These studies were supported by grants from the NIH (R01-CA140628, RC1-CA145138, K25-CA127349, P50-CA128323, P50-CA95103, U24-CA126588, S10-RR17858, R 41-MH85768, 5P30 DK058404, and R25-CA136440) and the Kleberg Foundation.
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.