Activity-Based NIR Bioluminescence Probe Enables Discovery of Diet-Induced Modulation of the Tumor Microenvironment via Nitric Oxide

Abstract

Nitric oxide (NO) plays a critical role in acute and chronic inflammation. NO’s contributions to cancer are of particular interest due to its context-dependent bioactivities. For example, immune cells initially produce cytotoxic quantities of NO in response to the nascent tumor. However, it is believed that this fades over time and reaches a concentration that supports the tumor microenvironment (TME). These complex dynamics are further complicated by other factors, such as diet and oxygenation, making it challenging to establish a complete picture of NO’s impact on tumor progression. Although many activity-based sensing (ABS) probes for NO have been developed, only a small fraction have been employed in vivo, and fewer yet are practical in cancer models where the NO concentration is <200 nM. To overcome this outstanding challenge, we have developed BL₆₆₀-NO, the first ABS probe for NIR bioluminescence imaging of NO in cancer. Owing to the low intrinsic background, high sensitivity, and deep tissue imaging capabilities of our design, BL₆₆₀-NO was successfully employed to visualize endogenous NO in cellular systems, a human liver metastasis model, and a murine breast cancer model. Importantly, its exceptional performance facilitated two dietary studies which examine the impact of fat intake on NO and the TME. BL₆₆₀-NO provides the first direct molecular evidence that intratumoral NO becomes elevated in mice fed a high-fat diet, which became obese with larger tumors, compared to control animals on a low-fat diet. These results indicate that an inflammatory diet can increase NO production via recruitment of macrophages and overexpression of inducible nitric oxide synthase which in turn can drive tumor progression.

Short abstract

BL₆₆₀-NO is an activity-based NIR bioluminescent probe for NO. Its application in vivo enabled detection of diet-induced modulation of the tumor microenvironment through aberrant NO production.

Introduction

Nitric oxide (NO) is a radical species that has been implicated in vasodilation,¹ neuronal signaling,² immunology,^3,4 and cancer.⁵ These diverse physiological and pathological roles warrant the development of tools for in vivo detection. This is especially true in the context of cancer, in which NO functions in various, sometimes contradictory, signaling pathways within the tumor microenvironment (TME).⁶ For example, high expressions of nitric oxide synthase in activated macrophage exert cytostatic or cytotoxic effects. On the other hand, resident macrophage within the TME (i.e., tumor-associated macrophage) tend to produce low steady-state concentrations of NO (<200 nM) and are associated with a more aggressive phenotype as well as a poorer clinical prognosis.^7,8 These concentration-dependent effects suggest that other factors that modulate inflammation, such as diet, may be important in the context of cancer prevention and/or treatment.⁹⁻¹³

In vivo molecular imaging is an attractive approach to examine the contributions of inflammation in driving cancer. This extends beyond traditional means, such as measuring inducible nitric oxide synthase (iNOS) expression levels or the extent of macrophage infiltration via immunohistostaining, to provide a direct, real-time link between NO production and cancer within live animals. NO’s fleeting nature remains a limiting challenge in terms of live animal imaging.⁶ Current techniques for directly monitoring NO in vivo have proven useful but have drawbacks, such as low resolution (EPR),^14,15 poor sensitivity (MRI),^16,17 or invasiveness (amperometry).¹⁸ A complementary approach is the use of activity-based sensing (ABS) probes^19,20 which exploit the chemical reactivity of the target analyte to report on activity with high selectivity.²¹⁻²⁴ Until recently, most ABS probes for NO have been developed for fluorescence imaging in cellular systems.²⁵ In contrast, their use in vivo is restricted by the significant scattering of light. The transition to longer excitation and emission wavelengths (visible to NIR) reduces these effects and has facilitated several successful studies in live animals; however, most of these examples were in lipopolysaccharide (LPS) models where the concentration of NO is 10³-fold higher than those found in cancer.²⁶ Our group, as well as others, has been interested in addressing these limitations in terms of both depth penetration and sensitivity. Along these lines, we developed the first ABS probe for photoacoustic imaging of NO and applied it to an LPS-induced murine inflammation model.²⁷ Since our work, different strategies have been reported and validated in similar models.^28,29 By rationally tuning our first-generation probe, we have designed light-activatable NO donors,^30,31 as well as new photoacoustic probes that exhibit an improved limit of detection which has allowed us to visualize endogenous NO in a murine model of breast cancer.^32,75

Unlike fluorescence imaging (light-in, light-out) and photoacoustic imaging (light-in, sound-out), bioluminescence (BL) imaging does not rely on external light excitation to generate a readout.^33,34 As such, BL imaging offers improved sensitivity by circumventing incident light, as well as decreased background that results from autofluorescence or endogenous chromophores that are photoacoustic-active.³⁵ Although numerous ABS probes for BL imaging have been developed,^{36,37,46,47,38−45} deep tissue imaging of NO in the TME has remained elusive. BioLeT is a notable BL probe that was employed for the detection of exogenous NO (delivered in the form of a NO donor) in vivo.⁴⁸ BioLeT is based on an amino-luciferin scaffold which emits visible light and therefore is best-suited for imaging at shallow depths. Recently, the development of near-infrared (NIR) BL technologies such as red-shifted substrates⁴⁹⁻⁵² and optimized luciferase enzymes⁵³⁻⁵⁵ has presented exciting opportunities for deeper tissue imaging. Despite these advances, no ABS probes, for any analyte, exist for NIR BL imaging.

Herein, we report the development of BL₆₆₀-NO: the first ABS probe for NIR BL detection of NO in cancer. BL₆₆₀-NO was successfully applied for imaging in live cells and murine models of liver and breast cancer. Moreover, we employed BL₆₆₀-NO to investigate the effect of diet on the TME, in which a high-fat diet resulted in increased NO production as compared to a low-fat diet in mice. This work highlights the importance of studying the influence of proinflammatory stimuli in the context of cancer prevention, treatment, and progression.

Results and Discussion

Design and Synthesis of BL₆₆₀-NO

BL₆₆₀-NO was designed to satisfy the following criteria to maximize in vivo performance. First, the emission maxima should be in the NIR range (>650 nm) to enable deep tissue imaging where scattering and attenuation of emitted light is limited. Second, the substrate should turnover rapidly in the presence of wild-type firefly luciferase (herein referred to as luciferase) to leverage established cell lines and animal models. Third, the design should be synthetically accessible, modular, and easy to diversify. With these considerations in mind, we selected AkaLumine⁴⁹ as the starting point. By exchanging the N,N-dimethyl group with other N,N-dialkyl moieties, the lipophilic character of the resulting substrates can be tuned to maximize tumor accumulation.⁵⁶ For instance, in comparison to AkaLumine, the N,N-diethyl analogue (BL₆₆₀) produced significantly higher BL signals when A549-Luc2 and 4T1-Luc cells were incubated with equimolar concentrations of both substrates (Figures S1 and S2). Moreover, we found that BL₆₆₀ was stable in cell culture media for up to 12 h (Figure S3). We hypothesized that capping the carboxylate group with a NO-responsive unit (e.g., o-phenylenediamine trigger)⁵⁷ would prevent enzymatic activity with luciferase. However, the reaction with NO (via the active species N₂O₃) is expected to be rapid en route to generate an acyl triazole intermediate, which can undergo spontaneous rate-limiting hydrolysis to yield BL₆₆₀. The resulting substrate can then react with luciferase (K_m = 5.54 ± 0.95 μM) to generate BL in the NIR range (λ_em = 660 nm) (Figure 1a and Figure S4). This is critical because it means that our probe is compatible with any luciferase expressing cell line. An added benefit of this design strategy is that it masks the negative charge and should facilitate rapid uptake, as opposed to substrates such as luciferin.

Figure 1.

(a) Schematic representation of the BL₆₆₀-NO reaction with NO and wild-type luciferase to produce an NIR bioluminescent signal. (b) Synthesis of BL₆₆₀-NO.

The synthesis^49,58 of BL₆₆₀-NO began with the Horner–Wadsworth–Emmons olefination reaction between 4-diethylaminobenzaldehyde and triethyl 4-phosphonocrotonate to afford the conjugated trans-diene ester 1 in 53% yield. Hydrolysis of the ester to the corresponding carboxylic acid proceeded quantitatively in acetonitrile–water to yield acid 2. Condensation of 2 with S-trityl protected d-cysteine methyl ester, followed by cyclization in the presence of trifluoromethanesulfonic anhydride and triphenylphosphine oxide, furnished the thiazoline methyl ester intermediate 3 in 82% yield. Hydrolysis of the methyl ester group with pig liver esterase afforded the free acid 4 (BL₆₆₀) in 43% yield. Finally, 4 was coupled to o-phenylenediamine using O-(benzotriazole-1-yl)-N,N,N′,N′-tetramethyluronium hexafluorophosphate (HBTU) to afford the final probe, BL₆₆₀-NO, in 33% yield (Figure 1b). We also prepared a nonresponsive control compound (Ctrl-BL₆₆₀-NO) to confirm the proposed NO- and luciferase-dependent activation. Ctrl-BL₆₆₀-NO was synthesized using the same reaction sequence above; however, we employed m-phenylenediamine, instead of the o-phenylenediamine trigger, because it cannot form the labile acyl triazole intermediate necessary to generate BL₆₆₀ (Figure S5).

In Vitro Characterization

With BL₆₆₀ and BL₆₆₀-NO in hand, we first obtained a bioluminescent spectrum of BL₆₆₀ which revealed that the emission maximum is separated from the corresponding maximum of luciferin (λ_em = 560 nm) by 100 nm (Figure 2a). Next, we examined the response of BL₆₆₀-NO to NO and recombinant luciferase in vitro by incubating the probe with a NO donor (DEA NONOate, 250 μM) for 30 min (pH 8 and 37 °C). The reaction mixture was then treated with ATP-MgSO₄ and luciferase (0.05 mg/mL). We confirmed via LCMS analysis that BL₆₆₀ and the benzotriazole byproduct were being generated, consistent with the proposed activation mechanism (Figure S6). We further demonstrated that both NO and luciferase were essential to generate a BL signal (Figure 2b). Linear relationships were observed between the BL signal and either probe or NO concentration while holding the other parameter constant (Figure 2c,d).

Figure 2.

(a) Spectra of luciferin and BL₆₆₀ in the presence of recombinant luciferase. (b) In vitro assay demonstrating that the probe, NO, and luciferase must all be present to generate a signal: BL₆₆₀-NO (5 μM), DEA NONOate (250 μM), and luciferase (0.05 mg/mL). (c) Bioluminescent signal as a function of BL₆₆₀-NO concentration (0, 1.25, 2.5, 5 μM). (d) Bioluminescent signal as a function of DEA NONOate concentration (0, 62.5, 125, 250, 500 μM). (e) Selectivity assay against a panel of biologically relevant analytes. All analytes were present in 75-fold excess relative to BL₆₆₀-NO (5 μM). All data is reported as the mean ± standard deviation (n = 3).

Next, we tested the selectivity of BL₆₆₀-NO by incubating it with various biologically relevant analytes in the presence of luciferase. For instance, we evaluated its reactivity with several reactive carbonyl species (i.e., formaldehyde, glyoxal, acetaldehyde, dehydroascorbic acid) which can generate various cyclized products.⁵⁹⁻⁶¹ Moreover, we tested a panel of reactive oxygen species (i.e., hydrogen peroxide, tert-butyl hydrogen peroxide, hypochlorite, superoxide anion, hydroxyl radical) and reactive nitrogen species (i.e., nitrite, nitrate, nitroxyl, peroxynitrite) that could react with the substrate component or the trigger to give undesirable activation. Negligible changes (<6%) in the bioluminescent signal were observed with up to 75-fold molar excess of each species, consistent with exceptional selectivity (Figure 2e). In contrast, treatment of BL₆₆₀-NO with NO resulted in a robust signal enhancement. Lastly, prior to incubation with luciferase, we treated BL₆₆₀-NO with fatty acid amide hydrolase (FAAH) and cyclooxygenase-2 (Cox-2) since both of these enzymes can potentially hydrolyze the amide bond of BL₆₆₀-NO. Relative to activation by DEA NONOate (50-fold), there was only an 18.6% signal enhancement with FAAH and no change with Cox-2 (Figure S7).

Detection of NO in Cancer Cells with BL₆₆₀-NO

Next, we evaluated the responsiveness of BL₆₆₀-NO toward endogenous NO in human A549-Luc2 lung cancer cells and murine 4T1-Luc breast cancer cells. Of note, the detection of basal levels of NO in cell culture is challenging owing to the absence of an external stimulus (e.g., TME) that can induce the overexpression of iNOS and subsequent overproduction of NO. Cells were treated with 10 μM BL₆₆₀-NO and imaged immediately. Within a few minutes, the cells became bioluminescent compared to vehicle controls (Figure 3a–d). To determine if this turn-on response was due to NO, we pretreated cells with L-NMMA, a reversible competitive inhibitor of human and murine NOS, before application of BL₆₆₀-NO.⁶² We observed a statistically significant decrease in the bioluminescent signal for both cell lines under these conditions (Figure 3a–d). Moreover, we found that L-NMMA did not exhibit any inhibitory effect on the BL of BL₆₆₀ (Figure S8). Together, these results are important because they indicate that BL₆₆₀-NO exhibits the exceptional sensitivity necessary to detect NO at basal levels, even within cell cultures. Before applying BL₆₆₀-NO in vivo, we performed standard MTT assays to assess its potential cytotoxicity. Cells were incubated with BL₆₆₀-NO at various concentrations (0–20 μM) for up to 3 h. Even at the highest concentration of probe, we did not observe significant loss of cell viability (Figure S9), indicating that BL₆₆₀-NO would be suitable for in vivo applications.

Figure 3.

(a) Representative images of BL signals from A549-Luc2 cells upon treatment with vehicle (DMSO), BL₆₆₀-NO (10 μM), or pretreatment with inhibitor L-NMMA (1 mM) for 30 min followed by BL₆₆₀-NO (10 μM). (b) Quantified data from panel a. (c) Representative images of BL signals from 4T1-Luc upon treatment with vehicle (DMSO), BL₆₆₀-NO (10 μM), or pretreatment with inhibitor L-NMMA (1 mM) for 30 min followed by BL₆₆₀-NO (10 μM). (d) Quantified data from panel c. All data is reported as the mean ± standard deviation (n = 3). Color scales represent luminescence counts. The exposure time, 60 s; emission, open; binning factor, 8; and f number, 1. Dotted white circles represent ROIs used for quantification. A statistical analysis was performed using Student’s t-test (α = 0.05), **: p < 0.01.

Evaluation of BL₆₆₀-NO in Liver and Breast Cancer Models

First, we inoculated Nu/J mice with A549-Luc2 cells using an intrahepatic procedure to generate heterotopic lung tumors.⁷⁶ This experiment served to answer whether BL₆₆₀-NO exhibits sufficient sensitivity to extend into deep tissue (i.e., beyond the subcutaneous space). After 11 weeks, mice were treated with BL₆₆₀-NO and imaged using the IVIS imaging system. The BL signals from the liver were clearly evident while there was no signal from nontumor control animals (Figure S10). To confirm that BL₆₆₀-NO can detect endogenous NO in A549-Luc2 tumors, we inoculated a second set of mice via subcutaneous injection. After 4 weeks, tumors were intratumorally treated with either saline (control) or L-NMMA 1 h prior to imaging with BL₆₆₀-NO (Figure 4a). Although not statistically significant, the resulting trend demonstrates a reduction in the BL signal for the animals receiving the iNOS inhibitor and thus suggests that BL₆₆₀-NO is being activated by NO (Figure 4b). While these two studies allowed us to evaluate our probe in a human cancer cell line, it is important to note that nude mice lack a functional immune system which is critical for establishing the TME. For this reason, subsequent experiments were performed in syngeneic murine models.

Figure 4.

(a) Representative BL images of mice pretreated with a vehicle control (saline) or L-NMMA (35 mM, 50 μL). (b) Normalized data from panel a. Vehicle treatment data is reported as the mean ± standard deviation (n = 3), and L-NMMA treatment data is reported as the mean ± standard deviation (n = 5). Exposure time, 25 s; emission, 660 nm; binning factor, 8; and f number, 1. Dotted white ovals represent ROIs used for quantification. A statistical analysis was performed using Student’s t-test (α = 0.05). No statistical significance was observed.

In particular, we established a heterotopic 4T1-Luc breast cancer allograft model in BALB/c mice via subcutaneous injection in the flank. The tumors were allowed to grow to 300–400 mm³ (∼30 days). BL₆₆₀-NO was then administered systemically, and the BL signal was monitored (Figure 5a). We observed a signal enhancement in the tumor region that persisted for at least 2 h. As before, to verify that this was due to NO, we divided the tumor-bearing mice into two groups which were treated with a vehicle control (saline) or L-NMMA prior to administration of BL₆₆₀-NO. We hypothesized that the NOS inhibitor would attenuate the BL signal if NOS-derived NO was responsible for the probe activation. We observed a turn-on response of 2.2 ± 0.3 (defined as the ratio of the BL signal at 1 h relative to immediately after BL₆₆₀-NO treatment) in the vehicle group. In comparison, we did not note any change in the signal intensity of the L-NMMA treatment group (Figure 5b). To further corroborate our results, we introduced Ctrl-BL₆₆₀-NO, a nonresponsive isomer of our probe, and BL₆₆₀-NO via intratumoral injection. We selected this route of administration as opposed to systemic injection to account for possible differences in biodistribution and uptake into the tumors. As anticipated, we did not observe any signal enhancement relative to BL₆₆₀-NO (Figure S11).

Figure 5.

(a) Representative BL images of mice pretreated with a vehicle control (saline) or L-NMMA (35 mM, 50 μL). (b) Normalized data from panel a. All data is reported as the mean ± standard deviation (n = 3). Exposure time, 60 s; emission, open; binning factor, 8; and f number, 1. Dotted white ovals represent ROIs used for quantification. A statistical analysis was performed using Student’s t-test (α = 0.05), *: p < 0.05.

Effect of Diet on NO Generation in a Heterotopic Breast Cancer Model with BL₆₆₀-NO

Numerous studies have suggested that high-fat diets are a risk factor for breast cancer because they promote chronic inflammatory states.^12,63−68 For instance, a recent study of 337 000 women found that those who ate the most saturated fat were approximately 30% more likely to develop breast cancer compared to their counterparts who ate the lowest levels.⁶⁹ However, relating these findings to a molecular entity, such as NO, has not been possible to date. With the development of BL₆₆₀-NO, we sought to design an experiment to determine the impact of a high-fat diet on inflammation, NO generation in the TME, and tumorigenesis. Specifically, we elected to employ an orthotopic breast cancer model since this would allow us to study the effects of diet in the native TME. First, we randomly divided female BALB/c mice into two groups to be fed low-fat and high-fat diets, in which 10% and 60% of calories are from fat, respectively, for 12 weeks. Of note, we did not control their caloric intake by allowing the mice to feed freely to mimic the variable food consumption behavior of humans. After this period, 4T1-Luc cells were injected into the mammary fat pads, and the same diets were continued until the end of the study (Figure 6a). Consistent with previous reports, the mice in the high-fat group became obese and were 60% heavier (39.7 ± 3.9 g vs 23.1 ± 0.8 g). Moreover, the tumor volumes of the low-fat and high-fat diet groups were measured to be 176.7 ± 41.5 and 223.9 ± 74.3 mm³, respectively. This raises an important consideration; that is, since the tendency for high-fat animals is to have larger tumors on average, would the presence of more luciferase expressing cancer cells give confounding results since it would be difficult to determine if a higher BL signal is due to more NO or more luciferase.

Figure 6.

(a) Schematic representing the workflow for the generation of mouse models to study the effect of diet on tumorigenesis and NO production by BL imaging. (b) Representative BL images of mice on low-fat and high-fat diets for 24 weeks, respectively, upon treatment with BL₆₆₀-NO. (c) Quantified data from panel b. All data is reported as the mean ± standard deviation (n = 4). Exposure time, 60 s; emission, open; binning factor, 8; and f number, 1. Dotted white ovals represent ROIs used for quantification. A statistical analysis was performed using Student’s t-test (α = 0.05), *: p < 0.05. Representative images of tumors excised from mice fed a (d) low-fat diet and (e) high-fat diet for 24 weeks with CD68 staining. Scale bar = 25 μm. Representative images of tumors excised from mice fed a (f) low-fat diet and (g) high-fat diet for 24 weeks with iNOS staining. Scale bar = 25 μm.

To distinguish this, we designed a cell-based study where we performed BL imaging with BL₆₆₀-NO and luciferin. We hypothesized that the ratio between luciferin emission and BL₆₆₀-NO emission (560/660) would be constant at a given NO concentration, and it should not depend on the number of cells present if the sensing of NO is rate-limiting. In contrast, if the number of cancer cells effects the 560/660 ratio, this would indicate that BL₆₆₀-NO was reporting on luciferase levels instead of NO and, thus, may be ineffective when tumor sizes differ. We tested this by treating one, two, or three million 4T1-Luc cells/mL with either 62.5, 125, or 250 μM DEA NONOate. The cells were then incubated with BL₆₆₀-NO and imaged using the 660 nm filter set on the IVIS imaging system. Likewise, a second set of cells were treated with luciferin and imaged using the 560 nm filter set. We found that the 560/660 ratio did not change as a function of cell number (Figure S12a). However, as the DEA NONOate concentration increased from 62.5 to 125 to 250 μM, the mean 560/660 ratio (obtained by averaging the ratio at different cell numbers) decreased in a concentration-dependent manner (Figure S12b). To confirm that DEA NONOate was not exerting an unanticipated effect on luciferase activity, we preincubated 4T1-Luc cells with either a vehicle or the NO donor prior to treatment with luciferin or BL660. In both instances, the BL signal intensity was not impacted by the presence of DEA NONOate (Figure S13). Moreover, we implanted 4T1-Luc cells in a new group of animals and performed imaging at three tumor volumes (104.9 ± 20.0, 156.7 ± 24.4, and 253.5 ± 19.0 mm³). We hypothesized that we would not observe a tumor volume-dependent increase in the BL signal if differential activation of BL₆₆₀-NO was due to changes in NO levels. Indeed, we did not record statistically significant differences as a function of tumor volume (Figure S14). Together, these results are consistent with NO detection being rate-limiting, which is an important feature of our probe design. Moreover, we designed a control experiment to investigate whether factors independent of NO (e.g., uptake and accumulation of the probe in fatty tissue) could yield a difference in BL. Specifically, BL₆₆₀ (the turned over product) was employed to image tumors in the high-fat and low-fat diet groups. As shown in Figure S15, the two groups were indistinguishable.

With these results, we proceeded to image NO with BL₆₆₀-NO, 60 min after probe administration. We found that the BL signal was on average 2.63-fold higher in mice fed the high-fat diet (9883.7 ± 4014.4) than in those fed the low-fat diet (3756.4 ± 3063.4) (Figure S16). Animals were sacrificed after imaging to harvest their tumors for further analysis. Tumors were subjected to immunohistochemical analyses of CD68, a protein overexpressed in circulating macrophages and tissue macrophages. We observed positive CD68 staining in the tumors of the high-fat mice that was higher than that in the low-fat tumors (Figures S17 and S18). These results indicate that a high-fat diet is correlated with increased macrophage infiltration. Likewise, we noted significantly greater iNOS staining in the tumors of mice from the high-fat diet condition (Figures S19 and S20). Together, these results suggest that the consumption of a high-fat diet over a 12 week period promotes an inflammatory response within the TME that is linked to an increase in the number of tumor-associated macrophage and the overexpression of iNOS, which in turn are responsible for the elevated levels of NO that was detected by BL₆₆₀-NO. Despite these promising results, we designed and conducted a second study where the diet-priming period was increased from 12 to 24 weeks since numerous reports have demonstrated that this would result in a more profound inflammatory response.^70,71 Additionally, the time frame between probe administration and image acquisition was decreased from 60 min (initial study) to 30 min. We identified this optimized incubation time using Ctrl-BL₆₆₀-NO, which was administered systemically to mice fed high-fat and low-fat diets. Specifically, at 30 min, there was no activation or a statistically significant difference between the two groups (Figures S23 and S24). Results from this second study revealed a greater BL signal increase in the mice fed a high-fat diet (14.8 ± 4.7-fold; 3.4 ± 2.5 × 10⁵ cf. 2.3 ± 0.5 × 10⁴) compared to the 12 week study owing to higher levels of NO. Likewise, when we performed immunohistostaining for CD68 (26.2 ± 3.0% cf. 8.9 ± 3.4%) and iNOS (46.2 ± 10.1% cf. 16.1 ± 5.3%), we again observed significantly elevated levels in the mice fed a high-fat diet which corroborated our initial findings (Figure 6d–g, Figures S27 and S28). When we compared the BL intensity and iNOS expression levels in each animal, we did not observe any correlation. This highlights the importance of activity-based sensing probes such as BL₆₆₀-NO since enzyme activity can be regulated by factors which are independent of protein expression (e.g., cofactor and substrate availability).

Conclusion

The detection of fleeting biological species, such as NO, using ABS probes in vivo is an immense challenge. For instance, in the context of cancer, NO levels have been reported to be less than 200 nM. We hypothesized that we could detect subtle differences in endogenous NO levels by leveraging the low intrinsic background of BL imaging, as well as reduced scattering and attenuation of emitted light in the NIR range. Although it is noteworthy that one example exists where the emission tail of the probe (λ_em = 599 nm) could be employed to detect fatty acid amide hydrolase activity via NIR BL imaging,^45,51 BL₆₆₀-NO is the first ABS probe where the λ_em is NIR. Upon evaluation, we found BL₆₆₀-NO to be selective, biocompatible, and highly sensitive as we could visualize basal levels of NO in systems ranging from cell cultures to numerous animal models.

Unlike fluorescence, photoacoustic, or chemiluminescence imaging, the generation of a BL readout requires the presence of luciferase.^34,72 In a typical BL imaging experiment, a higher signal indicates that there are more cells, and this is commonly used to track tumor progression. However, in the case of analyte sensing, the dependence on luciferase activity can confound results. Therefore, it could be difficult to discern whether a higher BL signal is due to more NO or more luciferase. To resolve this, we established a protocol where we would record the signal of BL₆₆₀-NO and luciferin to calculate a ratio. Because NO sensing is rate-limiting, the ratio remains the same at a given NO concentration regardless of the number of cells. These results have important implications for the development of other BL probes.

Lastly, using BL₆₆₀-NO in an orthotopic model of breast cancer, we were able to study the impact of a high-fat diet on the TME. Although it has long been thought that a high-fat diet can promote tumor progression by creating an abnormal inflammatory TME,^73,74 the link to NO generation has been elusive owing to a dearth of real-time detection strategies of NO in vivo. Our data indicates that a high-fat diet can lead to greater macrophage infiltration of the TME, which in turn generate more NO. Understanding this relationship at the molecular level can help us better combat cancer by devising strategies to reduce an inflammatory response in the TME (e.g., by developing diets with low fat for cancer prevention). Moreover, with BL₆₆₀-NO in hand, it is now possible to evaluate the efficacy of anti-inflammatory drugs or NO scavengers using any luciferase expressing cell line or animal model.

Experimental Details

Synthetic Methods

Ethyl (2E,4E)-5-(4-(Diethylamino)phenyl)penta-2,4-dienoate (1)

Triethyl 4-phosphonocrotonate (1.8 mL, 7.99 mmol, 1.5 equiv) was added to a suspension of NaH (0.43 g, 10.65 mmol, 2 equiv) in anhydrous THF (10 mL) at 0 °C. The reaction was stirred at this temperature for 20 min to give an orange mixture. A solution of 4-(diethylamino)benzaldehyde (0.94 g, 5.32 mmol, 1 equiv) in anhydrous THF (4.7 mL) was added dropwise over 5 min, and it was brought to room temperature and stirred for an additional 3 h. The reaction was quenched with water and extracted with EtOAc (3×). The combined organic layers were washed with brine, dried over Na₂SO₄, filtered, concentrated, and purified via silica gel column chromatography (eluent: 4% EtOAc in hexanes) to afford 1 as a light yellow solid (0.77 g, 2.82 mmol, 53% yield). ¹H NMR (500 MHz, CDCl₃): δ 7.44 (dd, J = 15.1, 11.1 Hz, 1H), 7.33 (d, J = 8.9 Hz, 2H), 6.81 (d, J = 15.4 Hz, 1H), 6.71–6.58 (m, 3H), 5.86 (d, J = 15.2 Hz, 1H), 4.21 (q, J = 7.1 Hz, 2H), 3.38 (q, J = 7.1 Hz, 4H), 1.31 (d, J = 14.3 Hz, 3H), 1.18 (t, J = 7.1 Hz, 6H). ¹³C NMR (125 MHz, CDCl₃): δ 167.74, 148.52, 146.08, 141.42, 129.07, 123.32, 121.28, 117.88, 111.51, 60.14, 44.55.

(2E,4E)-5-(4-(Diethylamino)phenyl)penta-2,4-dienoic acid (2)

A 1 M solution of NaOH (2.75 mL) was added to a suspension of 1 (0.38 g, 1.39 mmol, 1 equiv) in isopropyl alcohol (11 mL) and refluxed for 8 h. The reaction was then cooled to room temperature, and the isopropyl alcohol was removed under a vacuum. The mixture was acidified with a 1 M solution of HCl and extracted with chloroform (3×), washed with brine, dried over Na₂SO₄, filtered, and concentrated to give 2 (0.34 g, 1.38 mmol, 99% yield). ¹H NMR (500 MHz, CD₃OD): δ 7.42 (dd, J = 15.1, 10.9 Hz, 1H), 7.36 (d, J = 8.9 Hz, 2H), 6.85 (d, J = 15.4 Hz, 1H), 6.75 (dd, J = 15.4, 10.9 Hz, 1H), 6.67 (d, J = 9.0 Hz, 2H), 5.83 (d, J = 15.1 Hz, 1H), 3.42 (q, J = 7.1 Hz, 4H), 1.17 (t, J = 7.1 Hz, 6H). ¹³C NMR (125 MHz, CD₃OD): δ 171.30, 149.94, 148.19, 143.12, 130.11, 124.68, 122.03, 118.44, 112.65, 45.37, 12.92.

Methyl (S)-2-((1E,3E)-4-(4-(Diethylamino)phenyl)buta-1,3-dien-1-yl)-4,5-dihydrothiazole-4-carboxylate (3)

A solution of 2 (0.5 g, 2.04 mmol, 1 equiv) and d-cysteine-(S-Trityl)-OMe (0.85 g, 2.25 mmol, 1.1 equiv) in anhydrous DMF (20.4 mL) was treated with EDC (1.32 g, 6.88 mmol, 3.38 equiv) and DMAP (0.64 g, 5.24 mmol, 2.56 equiv). The reaction was stirred at room temperature under N₂ for 24 h, quenched with water, and extracted with EtOAc (3×). The combined organic layers were dried over Na₂SO₄, filtered, concentrated, passed through a silica plug, and used in the next reaction without further purification. In part 2, a solution of triphenylphosphine oxide (1.12 g, 4.02 mmol, 3 equiv) in anhydrous CH₂Cl₂ (14 mL) was cooled to 0 °C under N₂. Tf₂O (0.34 mL, 2.02 mmol, 1.5 equiv) was then slowly added dropwise and stirred for 30 min. The intermediate prepared in part 1 (0.81 g, 1.34 mmol, 1 equiv) was dissolved in anhydrous CH₂Cl₂ (13 mL) and added dropwise at 0 °C. The resulting reaction mixture was stirred for 10 min, quenched with saturated NaHCO₃, and extracted with CH₂Cl₂ (3×). The combined organic layers were washed with brine, dried over Na₂SO₄, filtered, concentrated, and purified by silica gel column chromatography (eluent: 20% EtOAc in hexanes) to give 3 (0.58 g, 1.68 mmol, 82% yield). ¹H NMR (500 MHz, CDCl₃): δ 7.33 (d, J = 8.5 Hz, 2H), 6.94 (dd, J = 15.3, 10.2 Hz, 1H), 6.78–6.65 (m, 2H), 6.62 (d, J = 8.4 Hz, 2H), 6.52 (d, J = 15.3 Hz, 1H), 5.16 (t, J = 9.0 Hz, 1H), 3.82 (s, 3H), 3.62–3.49 (m, 2H), 3.38 (q, J = 7.0 Hz, 4H), 1.18 (t, J = 7.0 Hz, 6H). ¹³C NMR (125 MHz, CDCl₃): δ 171.62, 170.42, 148.42, 144.13, 139.82, 128.96, 123.54, 122.29, 122.12, 111.57, 52.93, 44.58, 34.68, 12.77. HRMS: [M + H]⁺ calcd for C₁₉H₂₅N₂O₂S, 345.1637; found, 345.1632.

BL₆₆₀ (4)

Pig liver esterase (76 mg) was added to a suspension of 3 (0.2 g, 0.58 mmol, 1 equiv) in ethanol (17 mL) and 10 mM NH₄HCO₃ (50 mL) at pH 7.8. The reaction was stirred at 37 °C for 19 h under argon. The reaction mixture was concentrated, and the residue was suspended in 1:1 v/v MeOH/CHCl₃. The precipitates were removed via filtration and washed. The filtrate was concentrated under a vacuum to give 4 (0.082 g, 0.248 mmol, 43% yield). ¹H NMR (500 MHz, CD₃OD): δ 7.37 (d, J = 8.9 Hz, 2H), 7.13–7.05 (m, 1H), 6.89–6.75 (m, 2H), 6.68 (d, J = 8.9 Hz, 2H), 6.50 (d, J = 15.1 Hz, 1H), 5.02 (t, J = 8.7 Hz, 1H), 3.67–3.55 (m, 2H), 3.42 (q, J = 7.1 Hz, 4H), 1.17 (t, J = 7.1 Hz, 6H). ¹³C NMR (125 MHz, CD₃OD): δ 150.10, 147.22, 142.99, 130.33, 124.79, 122.69, 112.71, 78.38, 45.40, 36.07, 12.93. HRMS: [M + H]⁺ calcd for C₁₈H₂₃N₂O₂S, 331.1480; found, 331.1475.

BL₆₆₀-NO (5)

A solution of 4 (0.21 g, 0.64 mmol, 1 equiv), 1,2-phenylenediame (0.14 g, 1.29 mmol, 2 equiv), HBTU (0.24 g, 0.64 mmol, 1 equiv), and HOBt (0,097 g, 0.63 mmol, 1 equiv) in anhydrous DMF (8.0 mL) was stirred under N₂ for 5 min. DIPEA (0.11 mL, 0.63 mmol, 1 equiv) was added to the reaction mixture and stirred at room temperature overnight. The reaction was diluted with water and extracted with EtOAc (3×). Combined organic layers were dried over Na₂SO₄, filtered, concentrated, and purified using silica gel column chromatography (eluent: 1% MeOH in dichloromethane) to afford BL₆₆₀-NO as a dark red solid (0.088 g, 0.209 mmol, 33% yield). ¹H NMR (500 MHz, CD₃OD): δ 7.90 (s, 1H), 7.36 (d, J = 9.0 Hz, 2H), 7.22 (dd, J = 7.9, 1.5 Hz, 1H), 7.10–7.00 (m, 2H), 6.86 (dd, J = 8.0, 1.2 Hz, 1H), 6.80 (d, J = 8.8 Hz, 1H), 6.76–6.59 (m, 4H), 6.54 (d, J = 15.3 Hz, 1H), 5.25 (t, J = 9.2 Hz, 1H), 3.71–3.60 (m, 2H), 3.42 (q, J = 7.1 Hz, 4H), 1.17 (t, J = 7.0 Hz, 6H). ¹³C NMR (125 MHz, CD₃OD): δ 173.33, 172.27, 149.86, 145.90, 143.30, 141.76, 132.03, 130.00, 128.46, 127.10, 124.85, 124.50, 122.76, 122.42, 119.43, 118.40, 112.67, 79.97, 49.51, 45.36, 35.66, 12.93. HRMS: [M + H]⁺ calcd for C₂₄H₂₉N₄OS, 421.2062; found, 421.2047.

Analyte Selectivity Assay

The response of BL₆₆₀-NO (5 μM) toward a panel of biologically relevant aldehydes, reactive oxygen species, and reactive nitrogen species (75 equiv) in a 96-well culture plate was monitored using the IVIS imaging system. BL₆₆₀-NO was incubated with each analyte at 37 °C for 30 min before the reaction was initiated by adding ATP-MgSO₄ and luciferase. Light was collected immediately after mixing, and the signal enhancement was determined relative to control wells not treated with analyte. The total volume of each well was 100 μL [50 mM Tris buffer (pH 8.0) with 0.5% DMSO]. Formaldehyde solutions were prepared by depolymerizing saturated aqueous solutions at 100 °C before use. Dehydroascorbic acid was prepared by dissolving the solid in water at 65 °C before cooling to room temperature for use. Superoxide anion was added as a solution of potassium superoxide in DMSO. Nitroxyl was generated in situ from a solution of Angeli’s salt in degassed 10 mM potassium hydroxide solution. Peroxynitrite was prepared according to the previously reported literature. NO was generated in situ from a solution of DEA-NONOate in degassed PBS buffer. All other analytes were prepared by dilution or dissolution from commercially available sources.

Cellular Imaging in A549-Luc2 and 4T1-Luc Cells with BL₆₆₀-NO

24-well culture plates were seeded with A549-Luc2 or 4T1-Luc cells (500 μL of 400 000 cells/mL per well) and incubated at 37 °C with 5% CO₂ for 24 h. After this period, cells were ∼85% confluent. Cells were then treated with a 10 μM solution of BL₆₆₀-NO (0.5% DMSO final concentration) and imaged immediately using the IVIS imaging system with a 660 nm filter. To perform inhibition studies, we replaced the media with serum-free RPMI 1640 with or without L-NMMA (final concentration 1 mM). The cells were incubated for 30 min before BL₆₆₀-NO was applied for BL imaging.

Application of BL₆₆₀-NO and Luciferin for BL Imaging

4T1-Luc cells were trypsinized, pelleted, and resuspended in serum-free RPMI 1640 media. The number of cells was determined using a Countess II automated cell counter. 200 μL of cells at a density of one, two, or three million cells/mL was added to 0.6 mL Eppendorf tubes and incubated BL₆₆₀-NO (10 μM) or a vehicle control (to account for the DMSO content used to solubilize BL₆₆₀-NO) for 15 min at 37 °C. Luciferin (0.47 mM) was then added to the tubes treated with the vehicle. DEA NONOate in degassed PBS was added to all tubes at a final concentration of 62.5, 125, or 250 μM, and the tubes were incubated for an additional 30 min at 37 °C. The tubes were then centrifuged for 2 min. The supernatant was discarded, and the cell pellet was resuspended in fresh serum-free RPMI 1640 media. The resuspended cells (100 μL) were transferred to 96-well culture plates and imaged using the IVIS imaging system using a 560 nm filter set for luciferin treated cells and a 660 nm filter set for BL₆₆₀-NO treated cells.

Formulation of BL₆₆₀-NO for in Vivo Imaging

BL₆₆₀-NO (400 μM) was dissolved in a 1:4 solution of DMSO and sterile saline (v/v) and filtered through a 0.22 μm sterile filter immediately prior to use. The typical injection volume is ∼100 μL at a dose of 0.55 mg/kg in the diet studies and 0.93 mg/kg in all other in vivo experiments. All in vivo experiments involving systemic administration utilized retro-orbital injections which is the standard technique employed by our group unless noted otherwise.

Generation of A549-Luc2 Heterotopic Lung Cancer Model

A549-Luc2 cancer cells were surgically implanted into the liver of 4–5 weeks old Nu/J mice. The mice were anesthetized and aseptic technique was followed throughout the procedure. A laparotomy was performed to expose the liver. A cotton-tipped applicator was used to stabilize the lobe of the liver. The needle was inserted into the liver and slowly injected 5 × 10⁶ cells (50 μL, 1:1 PBS/Matrigel). After injection, 5–0 Vicryl sutures were used to close the abdominal wall with 2 single interrupted sutures. Wound clips were used to close the skin and were then removed after 10 days. Tumor growth was monitored for up to 11 weeks via BL imaging.

BL Imaging of a Heterotopic Lung Cancer Model with BL₆₆₀-NO

Nude mice with and without A549-Luc2 tumors were treated with BL₆₆₀-NO (0.93 mg/kg, 20% DMSO/saline) via retro-orbital injection. After 1 h, the mice were imaged on the IVIS imaging system. Light was collected using the open filter set. An ROI was drawn around the liver, and the signal intensity was quantified using Living Image Analysis software. The counts were averaged and normalized to the non-tumor-bearing mice.

Generation of a 4T1-Luc Breast Cancer Model

Female BALB/c mice (6–8 weeks old) were inoculated with 4T1-Luc cells (100 μL of 1 × 10⁶ cells/mL in 1:1 serum-free RPMI 1640 media and Matrigel) via subcutaneous injection into the flank. Tumor volumes were measured using the caliper method, and the body weight of the mice was monitored over the course of the experiment. After 30 days, the tumors had grown to a final volume of 300–400 mm³.

BL Imaging of a Breast Cancer Model with BL₆₆₀-NO

4T1-Luc tumor-bearing mice were treated with a sterile saline (50 μL) or a 35 mM solution of L-NMMA in sterile saline (50 μL) via intratumoral injection. After 1 h, BL₆₆₀-NO was administered via retro-orbital injection. After an additional hour, the mice were imaged on the IVIS imaging system. Light was collected using the 660 nm filter set. An ROI was drawn around each tumor and the signal intensity was quantified using Living Image Analysis software. Results are reported as a ratio of tumor-bearing flank over nontumor control flank.

Generation of Orthotopic 4T1-Luc Tumors and Diet Study

Female BALB/c mice (6–8 weeks old) were placed on a control diet (Research Diets no. D12450B,) or a high-fat diet (Research Diets no. D12452) where 10% or 60% of calories are from fat, respectively. After 12 (diet study 1) or 24 (diet study 2) weeks, mice were inoculated with 4T1-Luc cells (100 μL of 1 × 10⁶ cells/mL in 1:1 serum-free RPMI 1640 media and Matrigel) via subcutaneous injection into the mammary fat pad. Each group continued receiving their respective diets until the completion of the study. Their body weights were monitored after inoculation. After ∼30 days, mice from both groups were treated with BL₆₆₀-NO, administered via retro-orbital injection and were imaged using the IVIS imaging system.

Glossary

Abbreviations

DEA NONOate

diethylammonium (Z)-1-(N,N-diethylamino)diazen-1-ium-1,2-diolate

EPR

electron paramagnetic resonance

L-NMMA

L-N^G-monomethyl arginine acetate

LCMS

liquid chromatography mass spectrometry

MRI

magnetic resonance imaging

PBS

phosphate buffer saline

ROI

region of interest

TLC

thin layer chromatography

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acscentsci.1c00317.

• Further experimental details, including synthetic procedures, spectral data, and supplemental in vitro and in vivo procedures and data (PDF)

Author Present Address

^‡ C.J.R.: Department of Chemistry, The Skaggs Institute for Chemical Biology, The Scripps Research Institute, La Jolla, California 92037, United States

Author Contributions

^† A.K.Y. and M.C.L. contributed equally. A.K.Y. synthesized BL₆₆₀-NO and Ctrl-BL₆₆₀-NO; performed in vitro characterization, selectivity studies, and cellular imaging; and established a protocol for ratiometric calibration. M.C.L. generated the allograft breast cancer model and performed BL imaging with assistance from C.J.R. M.Y.L. generated the liver model and performed BL imaging with S.S. M.C.L. generated and performed BL imaging in the dietary-induced inflammation breast cancer model with M.Y.L. and S.S. M.Y.L. harvested and prepared tissue for immunohistostaining experiments with M.C.L. All authors analyzed and interpreted the data. A.K.Y. and J.C. wrote the manuscript. All authors have given approval to the final version of the manuscript.

This work was supported by the National Institutes of Health (R35GM133581).

The authors declare no competing financial interest.