Monitoring Caspase-3 Activation with a Multimodality Imaging Sensor in Living Subjects

A large majority of the chemotherapeutic drugs exert their effects by inducing a final common pathway that leads to apoptosis (1, 2). Apoptosis may occur via either a death receptor–independent (intrinsic or mitochondrial) or death receptor–dependent (extrinsic) pathway. Agents that are commonly used against melanoma, such as temozolomide (methylating guanine agent), cisplatin (cross-linking DNA), and staurosporine (protein kinase C inhibitors; refs. 3–6) induce apoptosis by damaging mitochondria and releasing cytochrome c, which binds and oligomerizes apoptotic protease activating factor-1, forming apoptosome to activate the effector caspase-3, caspase-6, and caspase-7 (7). Another set of cytotoxic agents that act through the interaction with the death receptors present on the melanoma cells results in aggregation of the receptors and binding to the adaptor proteins (7). This in turn activates the death receptor–mediated apoptotic pathways involving activation of Bid, caspase-8, and, finally, activation of the effector caspase-3 and execution of cell death (7). Examples of this class of agents are tumor necrosis factor-α, FAS ligand, and tumor necrosis factor-α–related apoptosis-inducing ligand (TRAIL). Caspases are the cysteine proteases that specifically play a role in executing different steps of apoptosis resulting in cell death (8). To date, 14 caspases have been identified in humans and can broadly be divided into initiator and effector caspases. Each caspase remains in an inactive procaspase form in the normal cell and upon apoptotic stimuli becomes activated and cleaves its cellular substrates by binding to a small peptide sequence. The classic example is the DEVD (aspartic acid-glutamic acid-valine-aspartic acid) peptide sequence that is present in many cellular proteins [poly(ADP-ribose) polymerase, lamins, etc.] and is recognized and cleaved by caspase-3 (the central effector caspase; refs. 9, 10).

Reporter genes have emerged as powerful tools in molecular imaging that can report the occurrence of a specific molecular event at the cellular or subcellular level in an intact organism. Molecular imaging techniques using reporter genes can be classified into five primary categories (11). These are optical imaging (fluorescence and bioluminescence imaging), radionuclide imaging [positron emission tomography (PET) and single-photon emission computed tomography], and magnetic resonance imaging. Each of these modalities has their own advantages and disadvantages. Integration of two or more reporter genes in a single biological vector (plasmid or viral) can use relative advantages of all the reporter genes and could be used successfully in multimodality molecular imaging. Multimodality imaging approaches have been successfully used in the detection of cancer metastasis, cell trafficking, cardiac cell therapy, and many other areas (12–14). The combination of optical and microPET modalities possibly generate the most sensitive and depth-independent information specifically from small animals. However, to date, only bioluminescence imaging has been applied to image caspase-3 activation indirectly and noninvasively from living mice (15–17). In this study, we attempted to noninvasively image the drug-induced caspase-3 activation in melanoma cells using a new multimodality fusion reporter vector composed of a monomeric red fluorescent protein (mrfp1; fluorescence reporter), HSV1-sr39 thymidine kinase (ttk; PET reporter), and firefly luciferase (fl; bioluminescence reporter) joined by a small peptide linker (DEVD). In the fused form, all the three reporter proteins had markedly attenuated activity. However, upon cleavage by activated caspase-3, a significant gain in mRFP1, FL, and TK activity are observed by fluorescence-activated cell sorting (FACS), enzyme-based assays, and in vivo imaging techniques. To the best of our knowledge, this is the first report of imaging caspase-3 activation from single live cells to small living animals using one single vector and various imaging technologies (fluorescence microscopy, FACS, optical imaging, and microPET imaging).

Chemicals. [8-³H]Penciclovir and d-luciferin were purchased from Moravek Biochemicals and Biosynth International, respectively. Tumor necrosis factor-α, cycloheximide, and TRAIL were purchased from Sigma. Fluorine-18 labeled 9-(4-[¹⁸F]fluoro-3-hydroxymethylbutyl)guanine (FHBG) was synthesized at Stanford, as previously detailed (18). Staurosporine, caspase-3 inhibitor (Z-DEVD-FMK), and tumor necrosis factor-α were obtained from Calbiochem, BD Pharminogen, and Sigma, respectively. UCN-01 was a kind gift of Dr. Sausville (National Cancer Institute).

Construction of a caspase-3 sensor fusion (mrfp1-_DEVD-ttk-_DEVD-fl or mtf) reporter vector. PCR amplification and standard cloning techniques were used to build the mrfp1-_DEVD-ttk-_DEVD-fl fusion reporter vector. Briefly, the mrfp1 and the ttk genes were amplified with the following primers (mrfp1 primers 5′ CCC AAG CTT GCC ACC ATG GCC TCC TCC GAG GAC, 3′ CG GGA TCC ATC CAC CTC ATC GGC GCC GGT GGA GTG GCG and ttk primers 5′ CG GGA TCC ATG CCC ACG CTA CTG CGG, 3′ CG GAA TTC ATC CAC CTC ATC GTT AGC CTC CCC CAT CTC) with 3′ primers carrying the DEVD (bolded) linker. Using suitable restriction sites, these two amplified products were cloned into a pcDNA3.1 vector that already contained the fl reporter gene. The fusion vector was completely sequenced to verify that all the three reporter genes were in frame and linked by the two DEVD linkers. Note that all genes are italicized and the corresponding proteins are capitalized.

Cell lines, transfection, and apoptotic induction procedures. 293T human embryonic kidney cells, MCF-7 human breast cancer cells, and B16F10 murine melanoma cells were purchased from American Type Culture Collection. The MCF-7 cells were cultured in RPMI 1640 supplemented with 10% fetal bovine serum and 1% penicillin (100 μg/mL)/streptomycin (292 μg/mL). The B16F10 and MCF-7 cells were cultured as described earlier (19, 20). All transient transfections were carried out using the superfect transfection reagent (Qiagen). B16F10 cells stably expressing the mtf fusion vector were selected using G418 sulfate (500 μg/mL). These stable cells were then transduced with a lentivirus carrying a bicistronic vector containing humanized Renilla luciferase and green fluorescence protein reporters (CMV-hrl-IRES-gfp) and FACS sorted to create B16F10-mtf-hrl cells. All the microscopic observations were done using a Zeiss Axiovert 25 microscope. To induce apoptosis, cells either transiently or stably expressing mtf vector were treated with different concentrations staurosporine (2, 4, 8, 16, and 32 μmol/L) and UCN-01 (2, 5, 10, 20, and 50 μg/mL) for different time periods (6, 24, 30, 48, and 72 h) and then assayed for reporter protein activities. To inhibit caspase-3 action, 20 μmol/L of caspase-3 inhibitor (Z-fmk-DEVD) was added to the cells 3 h before induction. Control cells were incubated with appropriate concentrations of DMSO.

TK, FL, and RL activity. Thymidine kinase, Renilla, and firefly luciferase enzyme activity assays were done, as previously described (12). Ratiometric analyses of caspase-3 activation and cell viability (B16F10-mtf-hrl cells) were done by normalizing the FL [relative light output (RLU)/μg protein] or TK (percentage of conversion of ³H PCV/μg protein/min) activity with RL activity (RLU/μg protein) measured from the same samples.

FACS analysis of mRFP1 activity. 293T cells were cotransfected with mtf fusion and CMV-egfp plasmids and treated with staurosporine for 24 h. Cells were trypsinized, washed in PBS, and analyzed for RFP expression on the FL3 channel after gating on high-GFP expression using a FACS Caliber (BD Biosciences). Analysis was done using FlowJo Software (Tree Star).

Lentiviral production. The lentivirus carrying the CMV-hrl-IRES-gfp reporter was produced as detailed earlier (21).

Western blot analysis. Western blots were done using anti-TK polyclonal (gift of Dr. M. Black, Washington University) and monoclonal caspase-3 antibodies (IMGENEX) following procedure, as detailed earlier (22).

Bioluminescence imaging of cells and living mice.B16F10-mtf and B16F10-mtf-hrl cells were plated in six-well dishes and, after stimulating apoptosis by various inducers, were first imaged for RL expression by adding 1 μg/mL coelenterazine (dissolved in methanol at 1 μg/μL concentration) in PBS using a Xenogen IVIS-50 optical imaging system (Caliper). Cells were thoroughly washed and, after 30 min, imaged for FL activity by adding d-luciferin solution (15 μg/mL in PBS). Regions of interests (ROI) were drawn over the well(s) and quantified by using the Living Image Software version 2.5. Bioluminescence signal was recorded as average radiance with the unit photons/s/cm²/sr.

Animal care and euthanasia were done with the approval of the Administrative Panels on Laboratory Animal Care of Stanford University. For in vivo bioluminescence imaging, 4-wk-old to 6-wk-old nude mice were subcutaneously implanted with 2 × 10⁶ of B16F10-mtf-hrl cells, and tumors were allowed to grow to 4 to 6 mm in diameter. Mice (n = 8) were anesthetized and injected first with 100 μL of coelenterazine (50 μg/mouse) i.v., and whole body images were acquired for 1 min using the Xenogen IVIS-200 imaging system. After 30 min, mice were imaged to check for the residual hrl signal and then injected with 100 μL (15 mg/mL) of d-luciferin and imaged for 1 min. They were then separated into two groups (n = 4), and each group received intratumoral injection of either 50 μmol/L of staurosporine (dissolved in 50 μL DMSO; treated) or 50 μL DMSO (control). The same sequence of bioluminescence imaging was repeated at different time points (0, 6, 24, and 48 h) of staurosporine or UCN-01 (125 μg/mouse) treatments. ROIs were drawn over the tumors, and bioluminescence signal was recorded as average radiance. To determine the in vivo ratio of caspase-3 activation and cell viability, each of the ROI values of FL signal was normalized with ROI values of RL signal from the corresponding tumor/well imaged at same time point.

MicroPET imaging of mice. After bioluminescence imaging, the same group of mice (n = 8) were injected with [¹⁸F]FHBG (∼200 μCi/mouse) and scanned using a microPET (R4, Siemens) at different time points before and after (0, 6, 24 h) intratumoral staurosporine/DMSO injection. The microPET images were reconstructed with the ordered subset expectation maximization algorithm and were analyzed by using a Medical Imaging Data Examiner (AMIDE) software (23). ROIs were drawn over the tumors, and the mean activities were recorded from the entire ROI. The percentage-injected dose per gram (%ID/g) was calculated by dividing the ROI counts by the injected dose (decay corrected). Ratiometric analyses of μPET imaging were done as mentioned for bioluminescence imaging.

Statistical analysis. Data are expressed as mean ± SD of measurements of n experiments conducted on different assays. Statistical significance was assessed using a Student's unpaired t test, with P < 0.05 considered as significant. Correlations between two data sets were determined by regression analysis using Microsoft Excel.

A multimodality fusion reporter protein shows increased activities for all three reporters when cleaved by activated caspase-3 in apoptosing 293T cells. We have developed a multimodality fusion reporter (MTF) composed of a bioluminescence (FL), a fluorescence (mRFP1), and a PET reporter protein (tTK), wherein each protein is fused through a small peptide linker (DEVD; Fig. 1). Apoptosis was induced at different concentrations of staurosporine in 293T cells transiently expressing the mtf reporter to obtain the highest level of induction in FL and TK activity from the cleaved MTF protein. As shown in Fig. 2B and C, 8 μmol/L of staurosporine were able to exert the highest level of induction for TK [1.5-fold; P = not significant (NS)] and FL (2.8-fold; P < 0.05) from cells expressing MTF protein. The level of induction between FL and TK activity shows a moderate correlation (R² = 0.64) across different concentrations of the inducer.

A 2-fold shift/increase in mRFP1 fluorescence was observed in 293T cells transiently expressing mtf fusion and CMV-egfp reporter genes at 4 μmol/L staurosporine treatment than the untreated cells (P < 0.05) followed by gradual decrease in fluorescence at higher doses (Fig. 2A). At 16 and 32 μmol/L of staurosporine, a second peak corresponding to the dead cells were evident. Fluorescence microscopy of these cells (before FACS) revealed changes in cellular morphology and release of apoptotic bodies from the dying cells (Fig. 2F).

Western blot analysis of the same samples using the anti-TK antibody showed gradual cleavage of the triple fusion vector (MTF, 130-kDa protein) into its three components (mRFP1-tTK 65 kDa, tTK-FL 100 kDa, and tTK 36 kDa; Fig. 2D). Another Western blot from the cell lysates expressing mtf and probed with anti–caspase-3 antibody revealed increase in active form of caspase-3 (14 kDa) with incremented doses, but not in the control or inhibitor-treated wells (Fig. 2E).

Time kinetics of the induction level of the fl and ttk gene expression using 293T cells expressing mtf (8 μmol/L of staurosporine) show a maximum of 4.5-fold induction and ttk gene expression showed a 2.2-fold induction at 72 hours (data not shown). The induction level of FL and TK activity correlates well over time (R² = 0.89).

We also compared the effect of caspase-3 inhibitor on FL or TK activity in 293T cells expressing mtf, CMV-fl, and CMV-ttk vectors and treated with 8 μmol/L staurosporine for 24 hours. The staurosporine-treated cell lysates expressing mtf showed a 2.8-fold (P = 0.03) higher FL and a 1.5-fold (P = NS) higher TK activity than the control cell lysates. However, FL and TK activities from the caspase-3 inhibitor-treated cell lysates exhibited 1.8-fold (P = NS) and 1.09-fold (P = NS) higher activity than the controls (Supplementary Fig. S1A and B). The fusion vector has 32-fold lower fl expression than CMV-fl and 22-fold lower ttk expressions than CMV-ttk vector. To ensure that the observed induction of TK and FL activity occurred due to cleavage of MTF fusion protein by activated caspase-3, MCF-7 cells (deficient of functional caspase-3) were transiently transfected with mtf and fl vectors and treated with staurosporine for 24 hours. Cells expressing either mtf or fl vectors showed a decrease in FL activity upon staurosporine treatment compared with control cells (Supplementary Fig. S1C).

Ratiometric analysis of caspase-3–mediated activation of FL and TK reporter activity in B16F10-mtf-hrl cells 3 by in vitro assays and in vivo cell imaging. To validate the use of this unique multimodality caspase sensor vector in living organisms, we generated a dual sensor B16F10-mtf-hrl cell stably expressing the mtf and a CMV-hrl-IRES-egfp vector. This B16F10-mtf-hrl stable cell is able to measure two independent molecular events upon treatment: (a) caspase-3 activation and (b) decrease in number of viable cells. Whereas increase in fl or ttk gene expression from mtf vector (as a result of cleavage at the DEVD site) indicates the level of caspase-3 activation, the decrease in hrl gene expression reports the degree of cell death. All our further in vivo experiments were done using this B16F10-mtf-hrl cell.

To explore the optimal dose for induction of FL and TK reporter, we treated B16F10-mtf-hrl cells with different concentrations of staurosporine. The treated cell lysates showed a higher FL signal compared with the control cell lysates with highest signal (3.4-fold) reached at 16 μmol/L (P < 0.05) followed by a drop (Fig. 3B). The hrl expression, on the other hand, decreased with increasing concentration of staurosporine (P < 0.05; Fig. 3C). After normalizing the FL activity with RL activity, we observed a much higher induction level (8-fold) in the treated (16 μmol/L) than the control cells (P < 0.05; Fig. 3D). Similar effect has also been observed with the induction in TK activity level. Without normalization with RL activity, the highest induction in TK activity was 1.2-fold compared with control, but after normalization, the induction level increased to 2.6-fold (P < 0.05; data not shown). Good correlation (R² = 0.78) was found between the induction level of TK and FL activity.

Bioluminescence imaging of live B16F10-mtf-hrl cells treated with 16 μmol/L of staurosporine showed RL expression decreased significantly in the treated wells than the control wells at 24 hours (Fig. 3A, bottom), 30 hours, and 48 hours (P < 0.05; Table 1). However, the FL showed 3-fold higher activity in the same treated wells than the control wells for all the time points (P < 0.05; Fig. 3A, top; Table 1). A ratiometric analysis (FL/RL) revealed a much higher induction in FL activity (8.2-fold to 8.3-fold at 24-fold and 7.7-fold at 48 hours) level, which is comparable with in vitro data (Fig. 3E).

Normalization of FL and TK activity with RL activity from UCN-01–treated cells (10 μg/mL) showed a 6.3-fold induction in FL and 3-fold induction in TK activity compared with the control ones. When analyzed at different time points, a treatment for 6 hours with 10 μg/mL of UCN-01 exhibited highest induction level of FL and TK activities (data not shown).

Multimodality imaging of activated caspase-3 by induction of FL and TK signal from B16F10-mtf-hrl tumors with bioluminescence and microPET. Our key interest in developing this multimodality caspase sensor vector is to indirectly image caspase-3 activation using different imaging techniques in living subjects. Four nude mice carrying B16F10-mtf-hrl tumors were imaged for hrl (Fig. 4A, 1 and 2), fl (Fig. 4A, 5 and 6), and ttk (Fig. 4A, 3 and 4) gene expression sequentially before and after stauorsporine treatment, as described in Materials and Methods. Whereas RL signal showed significant decrease [(5.57 ± 0.2 versus 3.19 ± 0.05) × 10⁴ photons/s/cm²/sr; P < 0.05], the FL signal showed a slight increase [∼1.4-fold; (2.11 ± 0.1 versus 2.86 ± 0.2) × 10⁶ photons/s/cm²/sr; P = NS] from the tumors after staurosporine treatment (Fig. 4B). A ratiometric analysis of the signals (FL/RL) revealed a 2.4-fold increase (P < 0.05) in FL activity from the tumors by staurosporine treatment (Fig. 4C).

After the similar trend of FL activation by staurosporine, we observed a slight increase in FHBG uptake (∼1.2-fold; before and after stauorsporine treatment) in the tumors of treated group [(2.61 ± 0.04 versus 3.29 ± 0.07) %ID/g; P = NS; Fig. 4B]. When normalized for cell death (TK/RL), the treated tumors showed a 2.2-fold increase (P < 0.05) in FHBG uptake as a result of cleavage of MTF fusion protein by activated caspase-3 (Fig. 4C). Injection of DMSO in the control (n = 4) group of mice did not show increase in FL or TK activity (Supplementary Fig. S2). A slight decrease in both FL and TK signals were observed after normalizing them with RL signals, which did not reach statistical significance.

Serial bioluminescence imaging of caspase-3 activation by staurosporine and UCN-01 in living mice. UCN-01 (7-hydroxy staurosporine) is a potent PKC inhibitor drug that is now being evaluated for treatment of glioma, melanoma, and several other cancers and is in second phase of clinical trials (21–23). We, therefore, aimed to image the effects of staurosporine and UCN-01 on caspase-3 activation using our dual sensor B16F10-mtf-hrl cells in a tumor xenograft mouse model. For the staurosporine regimen, treated mice showed a slight decline in RL signals at 24 and 48 hours compared with 4 hours, whereas the control mice maintained similar level at all time points except at 48 hours. In contrast, the FL signals from the tumors of treated group increased gradually over time but the signals from control group were maintained at before, 4 hours, and 24 hours with a slight increase at 48 hours (Table 2). When FL signals were normalized with the RL signals for the treated mice, highest induction of fl gene expression (3.8-fold compared with the control group) were observed at 48 hours (P = 0.03; Fig. 5A). The tumors then dissected out and were homogenized and analyzed for rl and fl gene expression. Expression of fl [(41595 ± 1934 versus 21305 ± 2398) RLU/μg protein] and hrl [(21844 ± 1131 versus 35430 ± 5071) RLU/μg protein] genes from treated and control tumor lysates also followed the similar trend (3.1-fold induction in FL activity when analyzed ratiometrically), as found in imaging experiments.

Treatment by UCN-01 showed a rapid loss of RL signal at 24 hours in the treated group compared with the control group. The FL signal, however, showed a slight increase at 4 hours followed by a decrease in signal at 24 hours for both the treated and control groups (Table 3). When analyzed ratiometrically (FL/RL), a 4.2-fold induction of FL activity was found at 4 hours (P = NS; Fig. 5B). We did not pursue the 48-hour imaging time point due to very low bioluminescence signals in the treated tumors.

Apoptosis-modulating therapeutics for treatment of different types of human malignancies have shown significant efficacies in preclinical animal models, and many of them are in clinical trials (2, 24). Here, we describe an MTF vector that is able to assess apoptotic events in vivo using different imaging technologies. In our design, three reporter proteins [a bioluminescent (FL), a fluorescent (mRFP1), and a PET (tTK)] are linked by a small peptide linker (DEVD) that can specifically be recognized and cleaved by active caspase-3. In the fused form, all the three reporter proteins have markedly attenuated activity. However, when activated caspase-3 recognizes and cleaves the fusion protein into its individual components, each reporter protein shows an increase in its activity. This vector therefore is able to report the event and levels of caspase-3 activation indirectly by an increase in reporter protein signal. Partial but not complete inhibition in induction of FL and TK activity by the caspase-3 inhibitor in staurosporine-treated 293T cells and presence of the cleaved TK protein band in control samples in Western blots indicates presence of a minimal level of activated caspase-3 under normal cellular conditions. This basal level of activated caspase-3 in unstimulated cells has previously been reported (17). However, specificity of this vector for caspase-3 sensitivity is evident from the results of MCF-7 cells (caspase-3 deficient) where no induction of FL and TK activity is observed upon treatment.

Each of the three reporter genes used in this vector have their own advantages in imaging caspase-3 activation. Apoptosis is routinely detected by FACS through binding of FITC–Annexin V to exposed phosphatidyl serine on membranes of dying cells (25). Shift of the phosphatidyl serine molecules to the outer cell membrane is also a common feature of necrotic cells (25), and thus, a dual staining of Annexin V and propidium iodide is needed to select the apoptotic cells (26, 27). Detection of a 2-fold higher induction in mRFP1 fluorescence in a gated population of live cells by FACS exhibited by the mtf plasmid is, thus, truly identifying the level of active caspase-3. Another sensitive approach of detecting activated caspase-3 using reporter genes involves fluorescence resonance energy transfer between different fluorescence proteins fused through a DEVD linker by FACS (28). Our caspase sensor vector possesses added advantages over this approach by having fl and ttk reporter genes, thereby allowing bioluminescence and microPET imaging of activation of caspase-3. The FL component of the mtf fusion has the dual advantage of imaging apoptosis in living cells, as well as in living animals. Bioluminescence imaging is a very powerful and sensitive technology and is being used for many high throughput screening studies. Laxman et al. (15) had developed a caspase-3 sensor vector using the firefly luciferase reporter gene and validated in mice bearing D54 (human glioma) tumors (stably expressing the sensor vector) by bioluminescence imaging (15). However, their in vivo imaging study was limited to a short treatment (75 minutes) of TRAIL, and they did not consider the longer in vivo half life of FL (1-4 hours) while measuring the caspase-3 mediated augment in FL signal from living mice. Reinjection of d-luciferin within 75 minutes from the initial scan could potentially lead to an apparent increase in signal due to simple cumulative effects of the signals. Thus, their figure showing ∼10-fold induction in FL activity due to caspase-3 cleavage may be misleading, and the actual gain in FL activity in the TRAIL-treated mice was only 3-fold when compared with PBS-treated mice (as also mentioned in their manuscript text). Using our vector, we show a 2-fold gain in both FL and TK signals (treated group) from living mice, which is comparable with their in vivo study. Our control group of mice did not show any increase in signal (Supplementary Fig. S2). In our study, we also tried to indirectly monitor the kinetics of caspase-3 activation with serial bioluminescence imaging from B16F10-mtf-hrl tumors using two protein kinase inhibitors. Although more time points, greater number of mice, and better statistical analysis are still needed, our preliminary results indicate that UCN-01 (4-6 hours) can induce the highest FL activity much earlier than staurosporine (24-48 hours). The caspase sensor vector validated in the current work should be very useful for screening a large number of drugs from live cells at different time points using a cooled CCD camera. The TK component of the mtf vector is more useful for in vivo apoptosis imaging through microPET, especially from greater depths as γ-rays emitted from radiotracers do not suffer attenuation while traveling through tissues as light does and is, thus, applicable for imaging both mice and larger animals. The addition of the TK component to the vector also helped to reduce the overall basal activity of all the proteins in the fusion background. Generation of tomographic information is another added advantage of microPET over bioluminescence imaging. Throughout our study, we have not obtained statistically significant induction in TK activity while using a transient transfection system. This is probably due to partial cleavage of the MTF protein and TK being the middle component, which needs to be cleaved at both DEVD sites by caspase-3 to achieve full functionality. This could also be a possible explanation for achieving moderate correlation between FL and TK activities in transient transfection studies. However, the TK induction level reached significance in the stable cells (B16F10-mtf-hrl) and in tumors (2-fold) especially after normalization with RL activity, indicating the integrity of the three components in the fusion background. To the best of our knowledge, this is the first report of imaging caspase-3 activation by microPET using reporter gene technology. Optimization of this unique caspase sensor vector for better sensitivity is in progress.

Another important aspect of apoptosis/caspase research is the loss of cell viability with increased dose/time of inducers. In our study, although fl gene expression from treated samples exhibited 2-fold to 3-fold higher level than controls when analyzed from cell lysates, bioluminescence imaging of the same treated live B16F10-mtfl cells showed a decrease in signal over time compared with the controls at similar conditions. Higher dose of staurosporine or longer treatments lead to further decrease in FL signal. Whereas our in vitro analyses were done by normalizing the signal level (RLU) per microgram of protein, this was not possible for analyses in living mice. Thus, to normalize the decrease in cell viability effect for in vivo imaging, we created a dual sensor cell line by isolating stable clones of B16F10-mtf cells expressing a hrl-IRES-gfp vector. This unique B16F10-mtf-hrl cell, on the one hand, is able to image caspase-3 activation by FL and TK activity and, on other hand image, the cell viability by RL activity. Normalization of FL activity with RL from treated cells showed an 8-fold increase compared with the control cells by bioluminescence imaging. We also analyzed the TK induction level by normalizing with RL and found a 2.6-fold higher TK activity from cells treated with 16 μmol/L starospourine than control cells. UCN-01, a more specific protein kinase C inhibitor and an analogue of staurosporine, exhibited 4-fold induction of FL at an earlier time point than staurosporine. In vivo imaging of mice bearing tumor xenografts of B16F10-mtf-hrl cells and treated with either DMSO or staurosporine by bioluminescence showed a similar level of FL activity at 24 hours (control versus treated) without normalization with RL activity and 3.4-fold higher signal after normalization. A 2-fold increase in TK activity (as measured by uptake of [¹⁸F]FHBG) was observed in the treated tumors over the control tumors after normalization with RL signal. In a follow-up study with their caspase sensor vector (15), the Michigan group have used bioluminescence imaging to monitor the therapeutic effects of TRAIL and 5-flurouracil over 7 days in tumor-bearing mice (16). However, the authors never considered the extent of cell death as a result of single or combined therapy while measuring the caspase-3–mediated activation of FL signal. Herein, we have taken a normalization approach (caspase-3 activation/cell viability) to account for the loss of signals from dead cells and tried to achieve a more accurate measure of increase in FL and TK signals.

Current clinical imaging for monitoring melanoma treatment involves use of [¹⁸F]FDG (29, 30); however, there is no report of imaging drug efficacy for treatment of melanoma. Melanoma cells are known to have elevated levels of PKCs, specifically the PKC-α in murine melanoma cells (B16a, B16F10). Increase in PKC-α expression leads to cell differentiation, increased doubling time, reduced anchorage-independent growth, and increased melanin production (31). PKC inhibitor drugs are already in clinical trial as an agent of combined melanoma therapy (32). Noninvasive imaging of the caspase-3 activation as a result of PKC inhibitor (staurosporine or UCN-01) in the melanoma cells can predict significant outcome of the therapeutic effect. Our study is the first report to attempt imaging and monitoring apoptotic events and cell viability in response to PKC inhibitors in melanoma cells. This unique caspase sensor vector will also be useful in high-throughput screening of different apoptosis-inducing drugs and monitoring their efficacy noninvasively by molecular imaging. In contrast to the general mechanism, caspase-3 activation has also been found in certain cases of necrosis. This caspase sensor vector would be useful for detecting those necrotic cells specifically for the anticancer drugs that lead to necrosis. Although direct use of this vector in humans will require validation and approvals from federal agencies, we could envision using this sensor for human cell therapy studies, wherein the cells would be marked with the reporter before introduction into patients and then PET imaging could be used to monitor apoptosis of the introduced cells. Further improvement of this strategy to measure/image different steps of cell death simultaneously will also be important and is under exploration.

No potential conflicts of interest were disclosed.

We thank Anitha Junutula for her excellent help in cell culture, Dr. Ricky Tong for his help with the graphics, and Drs. David Dick, Fred Chin, and colleagues in the cyclotron/radiochemistry facility for their help with PET tracer synthesis.