Development of a Bioluminescent Nitroreductase Probe for Preclinical Imaging

Bacterial nitroreductases (NTRs) have been widely utilized in the development of novel antibiotics, degradation of pollutants, and gene-directed enzyme prodrug therapy (GDEPT) of cancer that reached clinical trials. In case of GDEPT, since NTR is not naturally present in mammalian cells, the prodrug is activated selectively in NTR-transformed cancer cells, allowing high efficiency treatment of tumors. Currently, no bioluminescent probes exist for sensitive, non-invasive imaging of NTR expression. We therefore developed a "NTR caged luciferin" (NCL) probe that is selectively reduced by NTR, producing light proportional to the NTR activity. Here we report successful application of this probe for imaging of NTR in vitro, in bacteria and cancer cells, as well as in vivo in mouse models of bacterial infection and NTR-expressing tumor xenografts. This novel tool should significantly accelerate the development of cancer therapy approaches based on GDEPT and other fields where NTR expression is important.


Introduction
The nitroreductase (NTR) family of enzymes are widespread amongst bacteria and are known to metabolize nitrosubstituted compounds and quinones using NADH or NADPH as reducing agents [1][2][3][4]. They are important for the development of novel antibiotics being the main target for the treatment of infections caused by bacteria, e.g. Mycobacterium tuberculosis [5], Helicobacter pylori [6] and by parasites, e.g. Trypanosoma [7], Giardia and Entamoeba [8]. Their enzymatic activity in gut microbiota is linked to carcinogen production and etiology of colorectal cancer [9,10]. In addition, they are used in biotechnology for degradation of environmental contaminants [1]. Due to their absence in mammalian cells they are also utilized as activating enzymes in gene-directed enzyme prodrug therapy (GDEPT) approaches for cancer chemotherapy [11] where the NTR gene is used to selectively transform cancer cells, providing unique targeted therapy of tumors over normal tissues [12]. Nitroaromatic prodrug CB1954 in the complex with bacterial NTR is promising for GDEPT and has reached clinical trials for prostate cancer [13].
Following the recent first approval in Europe of a gene therapy medicine, the potential for clinical application of GDEPT is increasing [14]. However, both preclinical and clinical development of NTR-based GDEPT systems has been severely hampered by the lack of imaging tools that allow sensitive in vivo evaluation of transgene expression in living subjects. Quantification of the level of transgene expression is extremely important because it is directly linked to the effectiveness of the therapy.
Bioluminescence (BL) is currently the most sensitive optical in vivo imaging modality available, and has been applied to visualize multiple biological processes in small animals [15,16]. It obviates most of the limitations of in vivo fluorescent imaging [17], such as high tissue-derived autofluorescence, photobleaching, limited tissue penetration and lack of quantification. Several activatable fluorescent probes for NTR imaging in vitro have been previously described [18][19][20][21][22]. However, the only reported probe for imaging of NTR in vivo relies on fluorescence (CytoCy5S) [23][24][25] and therefore possesses the limitations mentioned above.
Similar to bioluminescent imaging, in vivo chemiluminescent imaging offers the advantages of high sensitivity due to low background and high signal-to-noise ratios. Prior studies have elegantly demonstrated the application of chemiluminescence for imaging of myeloperoxidase activity [26] and beta-galactosidase activity [27] in vivo. However, although chemiluminescence has the additional advantage of not requiring luciferase transfected cells for the generation of light, expression of this enzyme allows more elaborate disease models to be developed due to the researcher's ability to define its spatial localization and regulation. Moreover, most chemiluminescent agents suffer from low quantum yield, short maximal photon wavelength emission and high instability. For example, the quantum yield of aqueous luminol chemiluminescence is 1.23 ± 0.20% [28] with a maximal emission of 424 nm [29], while the reported quantum yield of firefly BL is 41.0 ± 7.4% [30], that is about 40 times higher, with Dhydroxyluciferin and D-aminoluciferin having wavelength at 560 nm and 603 nm respectively [31]. Recently, Zhang et al. [32] showed the advantage of using both near-infrared fluorescent and chemiluminescent imaging in combination, while addressing the wavelength issue associated with luminol chemiluminescence in vivo by shifting it into the near-infrared region utilizing quantum dots.
BL is based on the interaction of a small molecule D-luciferin with firefly luciferase that results in the generation of photons of light. The system can be "tuned" through "caging" of the luciferin scaffold to image and quantify activities of biological molecules. Target-mediated selective removal of the caging moiety leads to production of free D-luciferin and subsequent generation of photons by luciferase, that can be quantified [33]. While this strategy was previously used by us and others to study biological processes (delivery and biodistribution of cellpenetrating peptide conjugates [34], cell surface glycosylation [35], hydrogen peroxide fluxes [36], fatty acids uptake [37]) and image enzyme activity (beta-galactosidase [38], caspases [39][40][41], furin [42] and beta-lactamase [43], no bioluminescent probes have been previously reported for imaging of NTR. Here, we describe the development of novel NTR-specific bioluminescent probe, termed Nitroreductase Caged Luciferin (NCL). Our results demonstrate that this probe can be used for non-invasive real-time imaging of NTR activity in vitro, in live bacteria and mammalian cells, as well as in preclinical models of cancer and bacterial infection.

Bioluminescent imaging of NTR with NCL in enzyme assay
In vitro imaging studies were performed in clear bottom black 96 well plates from Becton Dickinson and Company. An IVIS Spectrum (PerkinElmer) was used to measure the amount of bioluminescent imaging (BLI) signal production. The data are presented as pseudocolor images indicating light intensity (red being the most intense and blue the least intense), which are superimposed over the grayscale reference photographs. Bioluminescence was quantified using region of interest (ROI) analysis of individual wells and the average signal expressed as the total number of photons emitted per second per cm 2 per steradian (p/sec/cm 2 /sr) from each of the three wells was calculated by using the Living Image software. Total luminescence was calculated by integrating the area under corresponding kinetic curves. Luciferase buffer was prepared as following: 2 mM ATP, 5 mM MgSO 4 in PBS (pH 7.4). Stock solutions of luciferase in luciferase buffer, NADH, NCL, luciferin and NTR in PBS (pH 7.4), were freshly prepared and aliquoted in a 96-well plate to give the following final concentrations in the total volume of 100 μL/well: luciferase (60 μg mL -1 ), NADH (100 μM), NTR (10 μg mL -1 ), luciferin or NCL (0.25-5 μM); luciferin and NCL were added at the last step with a multichannel pipette from the additional 96-well plate. Bioluminescence signal from the plate was acquired immediately every 1 min with 0.5 s integration time for 30 min.
Bacterial strains, plasmids and culture conditions E. coli K-12 MG1655 (-luc gene) and E. coli K-12 AB1157, containing the luciferase expressing pUC57 Click beetle red (CBR) plasmid (+ luc gene) was a kind gift from Daniel Ansaldi (Perkin Elmer). The NTR triple mutant, E. coli K-12 AB502NemA, was a kind gift from Dr. Antonio Valle (University of Cádiz, Cádiz, Spain) and was transformed with the pUC57 CBR plasmid for production of luciferase. E. coli MG1655 lux, expressing lux luciferase, was generated as previously described [44]. All strains were grown aerobically at 37°C in Luria Bertani (LB) medium supplemented with 100 μg mL -1 ampicillin (Amp).

Bioluminescent imaging of NTR activity by NCL in E. coli
An IVIS-100 (PerkinElmer) was used to measure the amount of BLI signal production. Stock solutions of luciferin and NCL in PBS (pH 7.4) were freshly prepared and aliquoted in a 96-well plate to give the final concentrations (1-250 μM) in the total volume of 200 μL/well. The volume of bacterial suspension was 150 μL/well. Bioluminescence signal from the plate was acquired immediately every 2 min with 10 s integration time for 1 h.

Cell lines and cell culture
Cell line MDA-MB-231-NTR-Fluc-EGFP (NTR+luc+) used for in vitro and in vivo experiments in this study was kindly provided by Dr. Ramasamy Paulmurugan (Stanford University School of Medicine, Stanford, USA) [25]. The cells were generated as described below with a reference to Sekar et al. [25]. The cloning vectors, expressing bacterial nitroreductase gene (NTR2) and Fluc-EGFP fusion constructs, were from the plasmid bank (Cellular Pathway Imaging Laboratory, Stanford). To make MDA-MB-231 stable cell line, modified pcDNA3.1 (PURO) vector expressing NTR was transfected using Lipofectamine 2000 Transfection Reagent (Invitrogen). 24 hours later medium was changed and the cells were treated with 100 ng mL -1 of puromycin. The process was continued until no further cell death was observed. The cells were plated in low dilution (1 cell/100 μL) in a 96 well plate. Single colonies of cells expressing NTR were expanded for further transduction with lentivirus expressing Fluc-EGFP fusion protein. To control the level of Fluc-EGFP at near equal expression, cells were sorted by FACS in a similar window after transduction. MDA-MB-231 stable cells were maintained in puromycin stress throughout the study. Single colonies of stable cells were evaluated for the functionality of NTR enzyme by incubating with a CytoCy5S (red-shifted NTR substrate; GE Healthcare) for the detection of fluorescent signal (λ ex/em = 628 nm/ 638 nm).
Cell line MDA-MB-231-luc-D3H2LN Bioware (NTR-luc+), used as NTR negative control for in vitro and in vivo experiments in this study, was purchased from PerkinElmer, and maintained in Dulbecco's Modified Eagle's Medium supplemented with 10% (v/v) heat-inactivated FBS, 1% (v/v) penicillin/streptomycin (all reagents purchased from Life Technologies).

Bioluminescent imaging of NTR by NCL in stable cell lines
MDA-MB-231-NTR-Fluc-EGFP cells and control MDA-MB-231-luc cells were plated at a density 3 × 10 4 cells/well in two black 96-well plates with clear bottom, after 48 h the growth medium was removed, and 100 μL of NCL probe or luciferin solutions (1-100 μM) in cell culture medium was added to the wells. The plates were immediately placed in IVIS Spectrum and were imaged every 1 min for 1 h. Observed BLI signal was quantified using ROI analysis with Living Image software.

Ethics statement
All animal procedures on imaging of bacterial NTR in a mouse model of thigh infection were performed in accordance with the national ethical guidelines prescribed by the Health Products Regulatory Authority (HPRA). Protocols were approved by the animal ethics committee of University College Cork (AERR #2010/003 and #2012/015). Experiments on imaging of NTR in a mouse model of subcutaneous cancer were carried out in strict accordance to the Swiss regulation on animal experimentation and the protocol (#2363) was approved by the authority of the Canton Vaud, Switzerland (EXPANIM (Expérience sur animaux)-SCAV, Département de la sécurité et de l'environnement, Service de la consommation et des affaires vétérinaires). All efforts were made to minimize suffering.

Bacterial administration and imaging of bacterial nitroreductase in mice
Bacteria were grown at 37°C in a shaking incubator until reaching OD 600 of 0.6 in LB medium, containing 100 μg mL -1 Amp. Cultures were harvested by centrifugation (4000 × g for 10 min) and washed three times in PBS. After washing, bacteria were resuspended in one tenth volume of PBS. Mice were kept at a constant room temperature (22°C) with a natural day/night light cycle in a conventional animal colony. Standard laboratory food and water were provided ad libitum. Mice were afforded an adaptation period of at least 7 days before the beginning of experiments. Female BALB/c mice (Harlan, Oxfordshire, UK) in good condition, without infections, weighing 18-22 g and 6-8 weeks old, were kept as previously described [45] and were included in experiments. BALB/c mice were anaesthetized and the fur on the rear legs was removed. Mice were injected directly into the right quadriceps muscle at a depth of approximately 5 mm with 50 μL of bacteria suspended in PBS. The concentration of bacterial suspensions used for injection ranged from 10 6 to 10 9 bacteria/mL. Mice also received an intramuscular injection of 50 μL sterile PBS in the left rear quadriceps (control) and lux MG1655 E. coli . Mice were imaged for bioluminescence at regular intervals beginning immediately after probe injection using IVIS 100. Mice that were infected with bacteria for experimental purposes were monitored for signs of illness for the duration of the experiment. No adverse symptoms were reported. Following bioluminescence imaging, or at experiment end, animals were euthanized by cervical dislocation.

Mice and tumor induction
Swiss nu/nu mice were obtained from Charles River Labs. Mice were maintained at the EPFL UDP animal facility under pathogen free conditions and group housed in individually ventilated cages at 22°C with 12/12 light cycle. Before experiments the mice were afforded an adaptation period of at least 7 days. Female mice in good condition, weighing 18-25 g and 6 weeks of age were randomly divided in two experimental groups (n = 5 per group). For tumor induction, 1 × 10 6 cells in 100 μL of FBS-free medium/Matrigel (BD Biosciences) (50:50) was injected subcutaneously into the flank of the mice. One group was injected with MDA-MB-231-NTR-Fluc-EGFP cells and the other group-with MDA-MB-231-luc cell (control). The viability of cells used for inoculation was more than 95% as determined by Trypan Blue Dye Exclusion (Gibco). Following tumor establishment, the health parameters that may lead to the endpoints were carefully monitored in xenograft mice three times a week: cachexia (acute weight loss), lack of activity and loss of appetite. Weight was measured and general behavior as well as body conditions was assessed. Tumor size was carefully monitored to ensure that it doesn't exceed maximal allowed size of 1 cm 3 . Tumor volume was measured by caliper and calculated according to the formula 1/2(length × width 2 ). When tumors reached approximately 0.1 cm 3 in volume, the mice were imaged with luciferin (1.5 mg/mouse in 50 μL of PBS IP) for 1 h once a week to estimate the light emission and the optimal imaging window. Bioluminescence was acquired using IVIS Spectrum, every 1 min for 1 h with the auto-exposure mode.
Following bioluminescence imaging at experiment end animals were euthanized by CO 2 inhalation or cervical dislocation.

Imaging of nitroreductase in a mouse model of subcutaneous cancer
Potassium salt of D-luciferin (Intrace Medical) was dissolved in PBS (pH 7.2), solution was filter sterilized through 0.22 μm filter, aliquoted and kept at -20°C. NCL probe was dissolved in PEG400 (Sigma-Aldrich) and diluted with sterile PBS (pH 7.2) 1:5 (20% (v/v) of PEG400 in the total volume of 200 μL), fresh solution was prepared before every imaging. The dose of NCL probe (1.9 mg/mouse) was equivalent (4.7 μmol) to the dose of luciferin (1.5 mg/mouse). Mice were anesthetized prior to injection and during imaging via inhalation of isoflurane (Piramal Critical Care, Inc). When the total photon flux over 1 h from the mice imaged with luciferin reached 1 × 10 8 , the mice entered the experiment. On day 1 of the experiment all mice were injected IP with luciferin (1.5 mg/mouse in 50 μL of PBS) and bioluminescence was acquired immediately every 1 min for 1 h with the auto-exposure mode. On day 2 of the experiment (24 h after the luciferin imaging) all mice were injected IP with NCL probe (1.9 mg/mouse in 200 μL of PBS containing 20% (v/v) PEG400) and bioluminescence was acquired immediately every 1 min for 1 h with the auto-exposure mode.

Data analysis from a cancer model experiment
Bioluminescence was quantified using ROI analysis of the tumor area individually for each mouse. Total luminescence was calculated by integrating the area under corresponding kinetic curves. Percent of the NCL probe uncaging was calculated individually for each mouse using the formula:

Statistical analysis
Two-tailed Student's t-test was used to determine statistical significance (GraphPad Prism 6.03, GraphPad Software).

Probe design
The reduction of nitroaromatic compounds can occur through one-or two-electron mechanism [1]. Two types of bacterial NTRs have been described and they are classified according to the oxygen dependence. The NTRs used in our study (NfsA, NfsB) are type I oxygen-insensitive NTRs, they catalyze the reduction of the nitro group by addition of a pair of electrons, and their activity does not depend on the level of oxygen. However, the oxygen-sensitive NTRs (type II) catalyze the reduction of the nitro group by the addition of one electron, forming the nitro anion radical, which is oxidized back to the nitro group by oxygen. Bacteria contain both types of nitroreductases with type I being the most characterized among other NTRs. The independence of reduction from the level of oxygen in E. coli had been previously demonstrated for an NTR-sensitive coumarin probe (7-nitrocoumarin-3-carboxylic acid) suggesting the prevalent involvement of type I NfsA and NfsB possibly along with other uncharacterized NTRs [46].
We have previously demonstrated the suitability of exploiting NTR activity as being sufficiently selective to distinguish bacterial cells from host background [47]. Several NTR-related enzymes have been identified in mammalian cells and they functionally relate to type I NTRs (NAD(P)H-quinone oxidoreductase (DT-diaphorase EC 1.6.99.2) and xanthine dehydrogenase EC 1.17.1.4). They can potentially contribute to reduction of nitroaromatics, although they are not phylogenetically related and do not exhibit the typical domain characteristic of NTR family. These properties were investigated in the study on FMISO imaging reagent, a derivative of nitroimidazole used as a hypoxia PET tracer [48]. It was reported that under hypoxic conditions xanthine dehydrogenase is converted to xanthine oxidase that reduces FMISO and other nitroimidazole-containing compounds. Similarly, eukaryotic NTRs that are functionally related to type II (aldehyde oxidase EC 1.2.3.1, cytochrome c oxidase EC 1.9.3.1, and NADPH cytochrome P450 reductase EC 1.6.2.4) can potentially reduce nitroaromatics anaerobically and are generally used as targets for hypoxia-activated prodrugs and imaging agents [49].
Therefore, several important factors need to be taken into account when designing the compounds activated selectively by bacterial or mammalian enzymes. For bacterial NTRs the activation is largely dependent on the redox potential of the nitroaromatics. For example, nitrofurans display relatively high redox potentials (reported from -250 to -270 mV) and are reductively activated by NAD(P)H nitroreductases of enteric bacteria. At the same time metronidazole (nitroimidazole) is only activated by anaerobic enzymes showing low redox potentials (-480 mV) in some bacteria and protozoa, making it well tolerated in humans when used as an antibiotic [50].
Substrate specificity of the designed compounds is also important, for example CB 1954 prodrug is efficiently reduced by bacterial NTRs and a human DT-diaphorase (NQO2) while being a poor substrate for a human paralogue NQO1 enzyme [51]. These and other important aspects of selectivity of bioreductive prodrugs are discussed in more details in a recent review by Wilson and Hay [52].
The overall probe design is based on caging of D-luciferin with nitrofuryl moiety resulting in "Nitroreductase Caged Luciferin" (NCL) probe (Fig 1). 5-Nitrofuryl was selected as a cage as its derivatives (nitrofurazone, nitrofurantoin) were shown to be efficiently activated by NTR in bacteria [50]. Upon the reduction of the nitro group by NTR the resulting electron-donating amino group promotes the cleavage of the C-O bond (uncaging), leading to the subsequent release of luciferin which is oxidized by luciferase and a photon of light is emitted. Therefore, release of free luciferin followed by light production is only possible in the presence of NTR.

Bioreductive activation of NCL in cell-free assays
We first investigated the specificity of probe uncaging by incubating NCL with a recombinant NTR enzyme from E. coli (NfsA) in the presence of NADH as a cofactor. The release of luciferin was followed by HPLC-MS analysis as a function of time (S1 Fig). The resulting data demonstrated rapid conversion of NCL into free luciferin under these conditions, the calculated values of rate constant and NCL half-life, were 5.8 × 10 -3 s -1 and 119.5 s respectively (S1 Fig). Next, we determined the Michaelis-Menten kinetic parameters for the NTR-specific cleavage of NCL probe (S2 Fig) using fluorescence to monitor release of luciferin (λ ex/em = 330/530 nm) as the caged probe is not fluorescent. Both V max and K m values were found to be comparable to those previously reported for a NTR fluorescent substrate [20] and were determined to be 0.057 μM s -1 and 24.7 μM respectively. Catalytic efficiency of probe reduction by NTR (k cat / K m ) was determined to be 2.25 × 10 7 M -1 s -1 , which is two orders higher than that of luciferinluciferase reaction (1.07 × 10 5 M -1 s -1 ) [53]. To evaluate if the probe can be used as a reporter of NTR activity, we investigated the effect of the NTR inhibitor dicoumarol (competitive with NADH) on the efficiency of NCL uncaging. A gradual concentration-dependent decrease in signal was observed (Fig 2A) indicating that the uncaging of NCL depends on the activity of NTR. We also assayed quantitative capability of the probe against the amount of NTR by fluorescence and found the detection limit to be 0.15 μg/mL (S2 Fig).
To determine the utility of NCL as a bioluminescent reporter, we measured the light emission from increasing concentrations of NCL (0.25-5 μM) in the presence or absence of NTR and firefly luciferase (Fig 2B). The resulting signal was concentration-dependent and no significant light was produced in the absence of NTR, resulting in high signal to background noise ratios even at relatively low concentrations of the probe in comparison to that previously reported [36,42]. In addition, NCL demonstrated an average 70% conversion into luciferin, the highest uncaging efficiency among existing caged luciferin substrates reported to date [36,42]. We also verified that NTR did not have any effect on the luciferin-luciferase reaction (S4 Fig).

Imaging of NTR activity in E. coli
We approached the validation of NCL in a live biological system initially utilizing bacteria as a source of both NTR and luciferase. We investigated the potential of NCL probe for imaging NTR in E. coli naturally expressing NTR [2][3][4] and engineered to express luciferase. The resulting signal from NCL or luciferin control was compared between several E. coli strains: 1) E. coli wt, 2) E. coli engineered to stably express luciferase (E. coli luc+), 3) a strain of E. coli (NTR KO luc+) with three well-described NTR genes knocked out (NfsA, NfsB and NemA). First, as shown on Fig 3A, significant signal above background was detected from E. coli luc+ in comparison with E. coli wt, indicating the need of luciferase presence for light production. Both wt and NTR KO luc+ strains showed similar levels of luciferase expression when treated with luciferin. However with NCL the signal from parent strain was significantly higher than from the NTR mutant strain (NTR KO luc+), demonstrating probe selectivity for detection of NTR activity in bacteria. The presence of a signal in wells with NTR KO luc+ strain indicates that reduction of the nitrofuryl cage is not exclusive or specific to any of the three major E. coli NTRs (NfsA, NfsB and NemA) and that cage reduction can be achieved at detectable levels in the presence of the remaining NTRs in E. coli. In E.coli, several nitroreductases (NfsA [2], NfsB [3], YdjA [4]) and reductases (NemA [54]) are characterized, while the exact functions of other reductase proteins remain unclear. Recent studies [4] indicate that E. coli reductases could also have additional nitroreductase activity. We also investigated the signal dependence on different numbers of bacteria and significant signals were evident at concentrations of bacteria as low as 10 4 cells/mL (1.5 × 10 3 cells/well) (Fig 3C). The average efficiency of probe uncaging in bacteria was calculated to be 35% (Fig 3D).

Stability profiling
Since an ideal imaging reagent should be non-toxic and stable in biological environments, we next investigated these parameters. The probe did not induce any toxicity in bacteria or mammalian cells (S10 Fig). The half-life of NCL in mouse plasma in vitro was determined to be 53.

Imaging of bacterial NTR in vivo in a mouse model of intramuscular infection
The utility, and importantly, the specificity, of the probe in vivo was examined in a mouse model of thigh muscle infection, again utilizing bacteria as the source of both NTR and luciferase. Balb/c mice were injected in quadriceps with various numbers of E.coli luc+ (5 × 10 4 -5 × 10 7 ), followed by IP injection of NCL 30 min later (Fig 4A). Animals were imaged at various time points over 24 h (Fig 4C). Signal from the probe was detected 20 min post injection lasting for as long as 24 h and correlated with the amount of probe injected. The intensity increased over the first 4 h reaching plateau afterwards. The total photon flux produced during this time was approximately 1/3 of the total flux detected from the mice injected with luciferin, demonstrating high efficiency of uncaging by bacterial NTR in vivo (Fig 4B). We also compared the probe kinetics after different administration routes in mouse model of E. coli intramuscular infection (S8 Fig). Statistical analysis of the data showed no significant difference in signal from IP and IV administration of the probe in this model.

Evaluation of NCL in NTR expressing cancer cells
As the next step we applied NCL for imaging of NTR activity in breast cancer cells, stably transfected with NTR and luciferase (MDA-MB-231-NTR+luc+) [25]. The expression of active NTR was confirmed by using the previously described NTR-specific fluorescent CytoCy5S probe (S9 Fig). Addition of different concentrations of NCL to NTR+ cells resulted in rapid concentration-dependent signal increase, with up to 40 times signal-to-noise ratio at the highest concentration used (100 μM) (Fig 5A). Contrary to this, NTR-cells produced much lower signal of equal intensity at all concentrations used (Fig 5B). The difference in brightness between the two cell lines due to different expression levels of luciferase was taken into account by normalizing the signal from NCL to the signal from equimolar quantities of luciferin control. The selectivity of NCL uncaging was also tested in another NTR-/NTR+ cancer cell line and similar results were observed (S11 Fig). The difference in kinetics of the signal from the probe in bacteria (Fig 4) and mammalian cells (Fig 5) can be explained by the fact that E. coli naturally expresses several types of nitroreductases, while the mammalian cell line used in our experiments was transfected with one type of nitroreductase.

Imaging of NTR activity in subcutaneous xenograft model of cancer
In light of these positive results we decided to investigate the utility of NCL as a reporter of NTR activity in a cancer xenograft model in mice, using the same cell lines previously validated in vitro (Fig 5). Subcutaneous NTR+ or NTR-xenograft tumors were induced in two groups of . As a positive control, mice were injected with equal amounts of E. coli MG1655 expressing lux luciferase that doesn't require exogenous substrate for light production [44]. The signal was collected over 24 h, n = 3 per group.
doi:10.1371/journal.pone.0131037.g004 mice (n = 5 per group) and were grown to an average volume of approximately 0.25 cm 3 . We chose to implant the NTR+ and NTR-cells in different mice to ensure that the signal produced from the probe is specific to the cell type and doesn't result from luciferin diffusion to the neighboring tumor. Another reason to implant the cells separately was the difference in light emission that could preclude accurate measurements for light sources of different intensity on a single mouse.
When caged luciferin probes are used to measure the activity of a biomolecule in various experimental settings, the light output needs to be calibrated to the amount of luciferase [36,55,56]. In our case this calibration takes into account variations between luciferase levels in tumors from one group to another. Therefore, in order to compare the performance of the probe in two different cancer cell lines with a various level of luciferase expression we normalized the signal from of the probe to the signal from equimolar amount of D-luciferin.
Both groups of mice were first injected IP with luciferin (1.5 mg) to determine the overall light emission from each tumor over 1 h. This signal was later used to calculate percentage of NCL uncaging in order to normalize tumor size and level of luciferase expression from NTR-/ + cell lines. 24 h later, all mice were injected IP with equimolar quantities of NCL (1.9 mg), followed by collection of light over 1 h. Since the residual light from luciferin injection could contribute to the signal from the probe, the absence of residual signal was verified by imaging of mice before probe administration. The level of probe uncaging in NTR+ and NTR-groups was determined by quantifying light production from NCL in each mouse relative to luciferin. As shown in Fig 5C the efficiency of probe uncaging in the NTR+ group was approximately an order of magnitude higher than in the NTR-group (11.7±2.5% and 1.5±1.4% respectively; p = 0.0001), demonstrating the utility of this probe for imaging of NTR activity in tumors.
In a previously reported study of NTR imaging with CytoCy5S fluorescence of orthotopic xenograft tumors, probe activation by bacteria in the gastrointestinal tract (GIT) resulted in background fluorescence in that cancer model requiring fluorescence lifetime gating to differentiate tumor signal from GIT signal [24]. Although we have not presented results of orthotopic tumors, it is anticipated that the bioluminescence based imaging tool in our study, permitting detection of the signal only with coexpression of both luciferase and NTR, would eliminate such issues.

Conclusions
In conclusion, these results demonstrate that the NCL probe can be effectively used for noninvasive real-time imaging of NTR activity in vitro, in live bacteria and mammalian cells, as well as in vivo, in preclinical models of cancer and certain bacterial infections (S1 Movie). This novel reagent should significantly simplify screening of prodrugs in vivo and accelerate the preclinical development of enzyme-activatable therapeutics for translation into the clinic.