Background
Cancer prodrugs are typically small molecules that are essentially nontoxic but can be converted to a cytotoxic compound (referred to from hereon as the “drug”) by enzyme-catalyzed reactions [
1]. A class of these (the “N-prodrugs”) relies on enzymes native to humans, which are expressed at a higher level in malignant compared to normal cells. An example of this class is Mitomycin C (MMC), which is reductively activated by nitroreductases, particularly the mammalian NQO1, whose concentration is up-regulated in cancer cells [
2,
3], making them more vulnerable to its action. However, as the normal cells also produce such enzymes, they too activate MMC, resulting in serious off-target toxicity with this and other N-prodrugs.
Another class of prodrugs (the “F-prodrugs”) requires targeting to tumors of a foreign gene that encodes the enzyme needed to generate the drug. This approach is referred to as gene-delivered enzyme prodrug therapy (GDEPT). It holds the promise of largely avoiding off-target toxicity if the delivery of the gene and activation of the prodrug are confined specifically to the tumor, and considerable effort has been underway to develop this therapeutic approach [
4‐
7]. The prodrug Genciclovir (GC), which is activated by the herpes simplex virus 1 thymidine kinase (TK1), was examined in a 4-year Phase III clinical trial involving 248 glioblastoma multiforme patients [
5]; and another prodrug, 5-aziridinyl-2,4-dinitrobenzamide (CB1954), which requires the
Escherichia coli nitroreductase enzyme (NTR), is in clinical trial for prostate cancer [
8]. These studies have indicated that the success of GDEPT depends, apart from the obvious importance of specificity of gene delivery to cancer, also on: a) high level gene transfer; b) extended duration of gene expression; c) increasing the potency of the activating enzyme; and d) an efficient bystander effect (BE). (BE refers to the spread of the activated drug from the transformed cells capable of producing it to the neighboring cells lacking this capacity, and is critical to the efficacy of any GDEPT therapy because no method of gene delivery can transform all the cancer cells in a tumor.)
Attaining these objectives would be facilitated by a prodrug regimen whose drug product could be visualized non-invasively in living mice, as the resulting ‘observational approach’ would minimize the need for mouse sacrifice and the use of more involved tests, such as LC/MS/MS; native fluorescence in a drug is also superior to attaching a fluorophore to visualize it, as the fluorophore may affect the drug in unpredictable ways [
9].
We have previously reported the discovery of such a regimen [
10,
11], consisting of the prodrug 6-chloro-9-nitro-5-oxo-5H-benzo-(a)-phenoxazine (CNOB), and the newly discovered bacterial nitroreductase (also referred to as chromate reductase), ChrR. We have improved the latter several-fold, generating ChrR6 and its humanized version HChrR6 [
12,
13]. The activated cytotoxic product of CNOB, 9-amino-6-chloro-5H-benzo[a]phenoxazine-5-one (MCHB), is fluorescent and has been successfully visualized non-invasively in living mice; this is illustrated in Additional file
1: Figure S1 (reproduced from reference [
11] for convenience). The figure shows that in tumors producing ChrR6, MCHB is visible following iv CNOB injection, but not in tumors lacking ChrR6 [
11]. The CNOB/ChrR6 regimen (referred to from hereon as ‘CNOB-GDEPT’) is effective not only in killing several different cancer cell lines in vitro, but also in treating implanted 4T1 murine mammary tumors in mice with 40 % complete survival on day 140 (10 mg/kg CNOB administered in three daily doses of 3.3 mg/kg); all the untreated mice in this study were dead by day 25 [
11]. These 4T1 tumors represent human stage IV breast cancer model, reflective of both disease progression and metastatic characteristics [
14]. CNOB alone, even at high concentrations (up to 20 mg/kg), showed no significant toxicity, as determined by blood chemistry panel values. MCHB has an impressive BE and kills cells by intercalating with mitochondrial DNA, causing apoptosis involving the mitochondrial pathway, and likely kills both growing and non-growing cells [
11].
Our previous work established that CNOB fluorescence indicated its presence, but to what degree the fluorescence represented MCHB quantity was not addressed. As quantitative representation of MCHB by its fluorescence would enhance its utility in preclinical studies, we have investigated this here. We report that MCHB fluorescence does quantitatively correspond to its concentration; we also provide information on aspects of the pharmacokinetics (PK) of the CNOB-GDEPT and predictions on more effective CNOB dosages.
Methods
Construction of 4T1/HChrR6 cells
4T1 cells (ATCC) were transfected with humanized
chrR6 (
HchrR6) gene using
Sleeping Beauty transposon method as described before [
11]. Briefly,
HchrR6 gene was cloned into pKT2/UXbG using
HindIII/
ApaI restriction sites, creating pKT2/hU-
HchrR6-SN. Cells were grown to 90–95 % confluence in DMEM without antibiotics in a six-well plate. Transposase vector (pUb-SB11; 0.8 μg) and transposon DNA (pKT2/hU-
HchrR-SN and pKT2/BGL; 7.2 μg) were added to 0.5 mL Opti-MEM (Invitrogen). In another vial, 20 μL of Lipofectamine 2000 (Invitrogen) were added to 0.5 mL of Opti-MEM, and incubated at room temperature (5 min). The medium was aspirated and cells were washed once with PBS. The above solutions were combined, added to each well (total of 1 mL/well), and incubated for 18 to 24 h. The transfection solution was then aspirated and replaced by complete DMEM. Cells were incubated for an additional 48 h and selected with geneticin (Invitrogen; 2 mg/mL; this concentration was predetermined as the minimal killing dose for 4T1 cells). To ensure homogeneity of
HchrR6 expression, cells expressing luciferase were diluted to ~30 cells per 10 mL DMEM, supplemented with geneticin, and 100 μL aliquots were dispensed into a 96-well plate. This dilution generates a ~30 % probability of a well receiving a single cell, so that colonies in a well would develop from a single cell.
In vitro cell viability and fluorescence assays
4T1 cells transfected to constitutively express HChrR6 (‘4T1/HChrR6’ cells) were incubated (37 °C) with 15 μM CNOB for the specified time periods. MCHB fluorescence was measured as described below. Viability was determined at corresponding time periods by the MTS assay.
In vivo studies
Female nude (nu/nu) mice were inoculated subcutaneously in mammary fat pad number 9 with 4T1/HChrR6 cells, i.e., cells endogenously generating HChrR6 (1 × 106 cells in 50 % PBS/50 % matrigel). Tumors were allowed to grow for 10-14 days before injecting CNOB (3.3 mg/Kg) and subsequent imaging and fluorescence- or LC/MS/MS-based quantification of MCHB. To minimize background fluorescence, mice were fed purified rodent diet (AIG093, Dyets Inc.). Tumor burden was measured by caliper.
For detecting off-target activation of CNOB, firefly luciferase (F-Luc)-expressing untransfected 4T1 cells (not generating HChrR6) were used for tumor implantation and the tumors were visualized and imaged 5 min after intraperitoneal (ip) injection of luciferin (150 μL of 30 mg/mL solution); the (
luc promoter
-controled)
chrR6 gene was delivered using SL7838-
chrR6 bacteria [
11,
15]; the bacteria contained the Lux operon, permitting their visualization, as before [
11]. The bacteria were treated with IPTG to induce the enzyme before tail vain injection.
Imaging
Ninety-six-well black color plates with transparent bottom (Costar) and a plate reader (SpectraMax, Molecular Devices) were used for in vitro fluorescence imaging. For in vivo experiments, mice carrying 4T1 tumors were injected with CNOB (3.3 mg/kg iv) prior to imaging. The images were acquired by IVIS Spectrum (Perkin Elmer Inc.) and quantified by Living Image 3.2 software (Perkin Elmer Inc.). The exposure time for photography was 1 s. A standard curve was constructed in vitro based on the quantified photon intensity at various MCHB concentrations (0, 1.5, 15, 150, 1500 μM). Bioluminescence was measured using the same instrument. We note that the IVIS instrument is widely used for quantitative fluorescence/bioluminance imaging, as it possesses a built-in calibration system. As already stated, this permitted generation of a standard curve that linearly related MCHB photon yield with its concentration. The imaging experiments were performed at least four times.
LC/MS/MS analysis
Tumor tissue was weighed and homogenized (using 3x volume of PBS/unit weight) using Pro250 homogenizer (Pro Scientific Inc.). For the preparation of standards, various freshly prepared CNOB/MCHB combinations (0, 0.25, 0.5, 1, 2, 2.5, 5, 10, and 20 ng of each) were mixed either with 0.2 mL of blank tumor homogenate, or spiked to the blank plasma. All samples were further spiked with 25 μL of fresh internal standard (Chembridge ID 6066331) plus 10 μL of NH4OH (100 mM), and extracted in 2 mL of ethyl acetate (vortexing and centrifugation at 1400 × g for 10 min). Samples in ethyl acetate phase were evaporated and re-constituted in acetonitrile for LC/MS/MS analysis. Compounds were separated and quantified by the Micromass Quattro Premier triple quadrupole HPLC-MS by experts at Stanford University Mass Spectrometry Lab. The extraction efficiency was 95-99 %. Samples were stored at -80 °C and the LC/MS/MS analysis was performed within 2 weeks; CNOB and MCHB remained stable during this time.
Prediction of alternate modes of administration and statistical analysis
The Phoenix WinNonlin software (version 6.3, Certara, Princeton, NJ) was used to make projections using the PK data obtained here for predicting a potentially more effective dosage regimen of CNOB-GDEPT. All data were calculated and analyzed by the GraphPad Prism software. Statistics were determined using Student’s t-test and correlation analysis; p values of less than 0.05 were considered significant.
The statistical analysis of the AUC data (Table
1) was done by log-transformed raw AUC data, followed by two-tail paired
t-test between groups of two tumor types. The log transformation gives more normally distributed data that better fit the assumptions of the
t-test. This method is recommended by the US FDA for analyzing AUC of bioequivalent drugs. The reason the
t-test instead of the z-test (as done for most bioequivalent studies) was used is because of the sample size (<30).
Discussion
The fluorescence of CNOB is negligible at the emission wavelength of MCHB, meaning that the fluorescence thus measured is that of MCHB alone. Further, measurements of MCHB concentration by its fluorescence and by LC/MS/MS gave similar results. This was shown in vitro by cell killing kinetics, and in vivo for both transfected and untransfected implanted 4T1 tumors (with and without endogenous HChrR6 expression) down to MCHB levels of 2 ng/g tumor, as well as for the plasma. We conclude that the non-invasive MCHB fluorescence imaging is a reliable indicator of its concentration.
As mentioned in the Background, certain conditions must be met for successful development of an F-prodrug regimen, and the fluorescence characteristics identified here provide a powerful tool for attaining these conditions for the CNOB-GDEPT. Two examples will suffice to highlight the importance of this tool in this context. First, to attain specific targeting that confines the CNOB-activating capability to the tumor would require testing a variety of methods involving the use, for instance, of different delivery vehicles, and targeting ligands for a given cancer. As Additional file
1: Figures S1 and S2 illustrate, noninvasive imaging of MCHB fluorescence in living mice can provide a rapid screen of the relative success of methods differing in these, as well as other aspects, that may need to be tested for specific targeting of the tumor. Of course, final confirmation will necessitate the use of
ex-situ and more robust methods, such as LC/MS/MS, immunohistochemical, Western and others, but a quick initial ‘observational’ screen will greatly narrow the outcomes that would require the use of these involved and labor intensive techniques.
The second example that illustrates the advantage of MCHB fluorescence concerns the fact that gene expression has proved a limiting factor in the success of F-prodrug therapy (Background). Several different approaches would need to be tested to address this problem. Although DNA has been primarily used in gene delivery, there may be compelling advantages in using mRNA instead. For DNA-mediated gene delivery, transport into the nucleus is required for expression, and It is well established that DNA transport to the nucleus is highly inefficient [
23,
24]; in contrast, mRNA expression can occur directly in the cytosol. Studies have indeed shown the superiority of mRNA over DNA in gene transfer in both proliferating and non-proliferating cells [
25,
26]. Direct protein transfer may also need to be considered along with measures to enhance the stability and duration of expression. As is seen in Fig.
4, the relative effectiveness of these approaches in improving the level and duration of expression of the gene (or mRNA) and its peak levels can also be quickly gauged with this prodrug regimen by imaging, minimizing the need for the
ex situ involved techniques. Visualization approaches can also be applied to rapidly assess the extent of transfection of cells in tumors to generate the bystander effect required for effective therapy.
The PK parameters measured here enabled us to predict ways of administering CNOB to make the therapy more effective. This information will aid in the clinical transfer of this regimen. Indeed, we have already succeeded in specific delivery of ChrR6 mRNA to the HER2 +ve BT474 cells conferring on them the capacity to convert CNOB into MCHB (ms in preparation). Further studies for transfer to the clinic of the exosome-based regimen will utilize more sophisticated PK/PD models, for example that developed by Simeoni et al (
http://www.ncbi.nlm.nih.gov/pubmed/14871843) to link plasma concentration over time data to tumor growth.
Conclusion
The fluorescence intensity of the cytotoxic product of CNOB-GDEPT, MCHB, quantitatively reflects its exposure level in the tumor and plasma. This feature provides a powerful tool to rapidly screen a variety of approaches to make this regimen a successful anticancer therapy; it permits noninvasive imaging of MCHB generation in live mice, thereby greatly narrowing the outcomes that would need to be followed up by the use of more rigorous but also more labor intensive approaches. The prediction of a more optimal dose regimen of CNOB reported here will also facilitate attaining this end.
Abbreviations
AUC, area under the curve; AUC0-24, AUC between 0 and 24 h; BE, bystander effect; CB1954, 5-aziridinyl-2,4-dinitrobenzamide; ChrR, chromate reductase; ChrR6, improved form of ChrR; CL, clearance; Cmax, maximum observed concentration; CNOB, 6-chloro-9-nitro-5-oxo-5H-benzo-(a)-phenoxazine; DMEM, Dulbecco’s modified eagle medium; Emax, maximum effect attributable to the drug; F-Luc, firefly luciferase; F-prodrugs, prodrugs requiring a foreign enzyme for activation; GC, Gencicloir; GDEPT, gene-delivered enzyme prodrug therapy; HER2-positive breast cancer, breast cancer overexpressing human epidermal growth factor recptor 2; ip, intraperitoneal; IPTG, isopropyl β-D-1-thiogalactopyranoside; iv, intravenous; LC/MS/MS, liquid chromatography/tripe quadruple mass spectroscopy; Luc, luciferase; MCHB, 9-amino-6-chloro-5H-benzo[a]phenoxazine-5-one; MEM, minimum essential medium; MMC, Mitomycin C; N-prodrugs, prodrugs activated by native tumor enzymes; NQO1, another name for mammalian NAD(P)H quinone dehydrogenase 1; NTR, bacterial nitroreductase; PK/PD, pharmacokinetics/pharmacodynamics; RFU, relative fluorescent units; SD, standard deviation; SE, standard error; t1/2, half life; TK1, herpes simplex virus 1 thymidine kinase; USFDA, United States Food and Drug Administration; Vss, volume of distribution at steady state
Acknowledgments
We thank Dr. Ludmila Alexandrova of Stanford University Mass Spectrometry Lab for help in conducting LC/MS/MS analysis of CNOB and MCHB.