Nitric oxide is a small messenger molecule utilized by nature in cell signalling and the non-specific immune response. At present, nitric oxide releasing prodrugs cannot be efficiently targeted towards a specific body compartment, which restricts their therapeutic applications. To address this limitation, we have designed two photolabile nitric oxide releasing prodrugs, tert-butyl S-nitrosothiol and tert-dodecane S-nitrosothiol, which are based on the S-nitrosothiol functionality. By modulating the prodrugs’ hydrophobicity, we postulated that we could increase their stability within the cell by preventing their interaction with hydrophilic thiols and metal ions; processes that are known to inactivate this prodrug class. Our data demonstrate that these prodrugs have improved nitric oxide release kinetics compared to currently available S-nitrosothiols, as they are highly stable in vitro in the absence of irradiation (t1/2 > 3 h), while their rate of decomposition can be regulated by controlling the intensity or duration of the photostimulus. Nitric oxide release can readily be achieved using non-laser based light sources, which enabled us to characterize photoactivation as a trigger mechanism for nitric oxide release in A549 lung carcinoma cells. Here we confirmed that irradiation induced highly significant increases in cytotoxicity within a therapeutic drug range (1–100 μm), and the utility of this photoactivation switch opens up avenues for exploring the applications of these prodrugs for chemical biology studies and chemotherapy.
Nitric oxide (NO) is a signalling molecule that regulates multiple physiological and pathological processes. NO and downstream nitrogen oxides, collectively termed Reactive Nitrogen Species (RNS), display limited molecular recognition and engage in reactions with many different cellular components (1,2). Due to this lack of specificity, the cellular effects of RNS are determined by the concentration and duration of the NO signal. At low concentrations, NO signals through activation of guanylate cyclase, while at higher levels RNS can act as cellular toxins through nitrosation and nitration reactions, which cause damage to cell membranes and DNA (3). Additionally NO binding can directly alter the function of metalloproteins (4) and sequester iron from iron-sulphur proteins (5).
The main drawback to the in vivo use of NO donor drugs is that they do not display organ or tissue specificity. Therefore it is exceedingly challenging to selectively deliver NO to a target compartment, with non-specific NO release causing changes to vascular dynamics that result in systemic hypotension (6). An alternative approach is to deliver NO via the site specific activation of a prodrug, which minimizes adverse drug reactions. One of the most promising strategies is photodynamic therapy (PDT), whereby an inert precursor drug is activated by irradiation (7), followed by the decomposition of the excited electronic state to release NO. The major advantage of this technique is that, by selective irradiation, it should be possible to confine NO release to a specific region of the body.
This approach has been tested with thionitrites, also known as S-nitrosothiols (SNTs), of glutathione (GSH) forming S-nitrosoglutathione (GSNO) (8,9) and penicillamine (10), as well as photolabile metal-NO complexes (11,12). However, for cancer chemotherapy, there are several limitations to the use of photoactives. GSNO has been the most widely studied drug, due to its biological relevance (13,14) and available clinical trial data (15,16). During irradiation, the energy supplied causes the S-N bond to undergo homolytic cleavage, resulting in the formation of the corresponding thiyl radical and NO (RS-NO + hv → RS• + •NO) (10), whereas in the absence of irradiation, thermal decomposition pathways predominate, which do not involve generation of the thiyl radical (17).
For cytotoxicity, high concentrations of GSNO (LD50 > 200 μm) and light intensities (300 W) are required (8). While other SNTs have much faster NO release kinetics than GSNO (18) and could potentially be more effective for PDT, a major limitation to their use is that they undergo trans-nitrosation reactions with cellular thiols (R1S-NO + R2SH → R1SH + R2S-NO) (17). Given the high cellular concentrations of GSH in most cell types (1–10 mm), the rate constant for the trans-nitrosation reaction (ranging from 10 to 20/m/s) (17), and that the thermodynamics favour GSNO as the more stable SNT species (17), this pathway has the potential to result in the formation of GSNO as the predominant intracellular SNT, regardless of the original parent prodrug administered. Therefore, while SNT prodrugs could conceivably be designed with optimal kinetic parameters for NO release, the trans-nitrosation pathway would be predicted to switch the nitrosonium (NO+) group from the prodrug to GSH to form GSNO, at which stage kinetic control would be lost. Additionally a major factor influencing the rate of decomposition of SNTs is catalysis by transition metal species such as Cu+ (19,20), which restricts their suitability for PDT.
These drawbacks have driven us to develop NO donors that combine thermal and chemical stability with improved kinetics of NO release during photoactivation. Here we present our design of SNTs attached to a hydrophobic alkyl framework (Figure 1). We hypothesized that, by attaching the SNT functionality to a hydrophobic scaffold, we would generate a membrane soluble prodrug. This would be a highly desirable property for in vivo PDT, as a prodrug partitioned into the plasma membrane or intracellular membranes would be shielded from interactions with cytosolic thiols such as GSH or cysteine, as well as labile transition metal species.
Materials and Methods
BSA, propidium iodide (PI), 3-(4,5-Dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) and all cell culture reagents were purchased from Invitrogen (Auckland, New Zealand). Annexin-FITC was purchased from BD Biosciences (North Ryde, NSW, Australia). All other chemicals were obtained from Sigma-Aldrich (Auckland, New Zealand) unless stated otherwise.
Quantum mechanical calculations for bond geometry optimization were performed with the MOPAC package using the AM1 semiempirical Hamiltonian force field (Chem3D Pro v11.0, CambridgeSoft Corporation, Cambridge, MA, USA). Calculated octanol:water partition coefficients were derived using the CLogP driver module of ChemBio3D Ultra 12.0 (CambridgeSoft Corporation, Cambridge MA, USA).
SNTs were formed by the reaction of the precursor thiol and 1.1 equivalents of tert-butyl nitrite at 0 °C under an argon atmosphere following an established procedure (21). Following synthesis the prodrugs were dissolved in DMSO and stored at −80 °C, under which conditions they were stable for several months. To allow for decomposition during storage, SNT concentration was re-determined at the start of each experiment using λ = 340 nm (ε340 = 675/m/cm) (22). The prodrug was then diluted into the appropriate buffer (see sections below for final DMSO concentrations).
The kinetics of SNT breakdown were followed spectrophotometrically at λ = 340 nm at room temperature. Due to the low SNT extinction coefficient at this wavelength, these measurements were conducted using 550 μm prodrug. Compared to the NO release and cell culture experiments (0.1% v/v DMSO), this required an increase in DMSO to 20% v/v (in PBS) in order to prevent drug precipitation.
NO electrochemical detection
NO concentrations were quantified using a free radical analyzer with polarographic NO electrode (ISO-NOP, WPI, Sarasota, FL, USA). All NO release measurements were conducted in an incubation chamber that was shielded from light. The electrode cell was sealed to prevent NO diffusion, and all experiments were conducted at 37 °C in NaPi buffer (pH 7.4, 100 mm NaPi, 2 mm DTPA, 0.1% DMSO). The electrode was calibrated by the addition of 1–100 μm of NaNO2 to a reducing solution (0.1 m H2SO4, 0.1 m KI) to generate NO (23).
Commercially available horse heart myoglobin consists of a mix of oxy (MbO2) and met (met-Mb) myoglobin. This mixture was reduced to Mb by the addition of excess sodium dithionite in NaPi buffer (pH 7.4, 100 mm NaPi, 2 mm DTPA). To re-oxygenate and purify MbO2, the solution was passed though a PD10 desalting column (Sephadex G-25M, GE Healthcare, Auckland, NZ) (24). MbO2 concentration was determined at λ = 542 nm (ε542 = 13 900/m/cm) (25).
A549 human lung carcinoma cells were obtained from the ATCC and cultured at 37 °C in an atmosphere of 5% CO2, 95% air, using Advanced DMEM medium supplemented with GlutaMAX-I (1%), antibiotic-antimycotic liquid (1%), and fetal bovine serum (2%).
For polarographic electrode measurements, light intensity was regulated by a 100 W halogen lamp cold light source with variable power output (LG-PS2; Olympus, Auckland, New Zealand). The condenser lens of the light guide was positioned approximately 10 cm from the centre of the electrode chamber. For all other studies, irradiation was initiated using a desktop lamp (100 W) positioned approximately 15–50 cm from the cell or plate containing the SNT. Light intensity was detected using a P-9710 optometer fitted with an RW-3705 detector head (Gigahertz-Optik, Munich, Germany).
Effect of SNT photoactivation upon A549 cell viability
A549 cells were plated at 10 000 cells/well into 96 well plates and left to adhere for 48 h. Cells were then treated with tert-dodecane S-nitrosothiol (tDodSNO) dissolved in medium containing DMSO (0.1% v/v). All wells contained the same volume of DMSO. Following prodrug treatment, the cells were washed in PBS and incubated for a further 24 h in medium. MTT (0.4 mg/mL) was then added for 3 h, the medium removed, and the residual crystals dissolved in DMSO. Cell viability was calculated at λ = 550 nm (26).
Cells were plated at 150 000 cells/well into six well plates and left to adhere for 48 h. Cells were then treated with the tDodSNO for 60 min prior to photoactivation. Following photoactivation, the cells were washed with PBS and harvested using trypsin. The cells were then re-suspended in Annexin binding buffer (pH 7.4, 10 mm Hepes, 140 mm NaCl, 2.5 mm CaCl2) at a concentration of 1 × 106 cells/mL. Annexin V–FITC and PI were then added according to the manufacturer’s instructions and cell labelling quantified using a FACScalibur flow cytometer (BD Biosciences, San Jose, CA, USA) measuring 10 000 cells per treatment condition.
Statistical analyses were performed using OriginPro 8 (OriginLab Corporation, Northampton, MA, USA). Statistical significance was accepted when p < 0.05. All data were expressed as mean values ± SEM unless otherwise stated.
SNT prodrug design criteria
As tertiary SNTs are generally more stable than their primary or secondary analogues (27), we studied the photorelease of NO from two prodrug candidates that contained a SNT group linked to a tertiary carbon centre (Figure 1). Our main objective was to prevent the transfer of the SNT group to other thiol species within the cell; to accomplish this we applied two design criteria:
• The SNTs should possess a substantial lipophilic nature (assessed by octanol:water partition coefficients), to predominantly partition the prodrug away from the cytosol and into cellular membranes.
• The SNT functionality should have steric shielding, to protect it from side reactions with other membrane components.
We hypothesized that this combination of properties would enhance prodrug stability and so, by preventing loss of the SNT functionality via trans-nitrosation or transition metal catalyzed decomposition, allow us to use PDT to regulate the kinetics of NO release.
Quantum mechanical derivation of physicochemical parameters
We initially compared the calculated properties of the SNT prodrug candidates tert-butyl S-nitrosothiol (tBuSNO) and tDodSNO (Table 1). Both SNTs were calculated to possess identical S-N bond lengths (1.69 Å) and similar C-S-N and S-N-O bond angles (approximately 104° and 121° respectively). In line with our prodrug design criteria, the CLogP values for these SNTs displayed major differences, with tBuSNO having substantial hydrophobic character (CLogP +1.73) which increased as the hydrophobic tail was extended by eight carbon units to tDodSNO (CLogP +5.31). Both of these SNTs were several orders of magnitude more hydrophobic than GSNO (CLogP −3.17), indicating that they were unlikely to partition into the same intracellular compartments.
Table 1. Physicochemical parameters characterizing S-nitrosothiol (SNT) prodrugs. Θ: bond angle; B: bond length; CLogP: calculated octanol:water partition coefficient; ; half life with photoactivation (16 W/m2); t1/2: half life without photoactivation. SNT half lives were calculated by following prodrug decomposition in PBS (containing 20% v/v DMSO) at λ = 340 nm, data are expressed as mean values ± SEM (N = 3)
10 ± 1
205 ± 35
260 ± 45
3320 ± 160
To further evaluate prodrug properties, we next considered the solvent accessible surface of both molecules (Figure 2). As the trans-nitrosation pathway requires nucleophilic attack of the competing thiol group at the SNT S-N bond, steric hindrance at this position should improve the stability of the initial SNT, even in the presence of competing thiol groups. Examination of the solvent accessible surface revealed that, for both tBuSNO and tDodSNO, the alkyl groups of the carbon chain were able to overlap, and partially shield, the SNT group. This steric crowding was substantially increased for tDodSNO, indicating that this would be the more resistant of the two analogues towards NO+ transfer to competing thiol centres.
Photoactivation allows therapeutic control of the rate of NO release from SNTs
Investigation of the stability of the S-NO bond in our SNT prodrug candidates revealed that, in line with other known tertiary SNTs (27), the compounds were stable under physiological conditions (Table 1), with half-lives (t1/2) substantially >3 h. Irradiation resulted in a major change in half-life (), to approximately 10 min for tBuSNO and 260 min for tDodSNO (Table 1). To our knowledge, such a pronounced shift in t1/2 is unique to these compounds, which suggests that they are well suited to act as molecular switches for photoactivation induced NO release.
As SNT t1/2s reported in the literature are highly dependent upon experimental conditions, we decided to further explore the utility of our SNTs by comparing tBuSNO, our SNT with the shortest t1/2, to GSNO, which has previously been examined as a NO donor for PDT in several other studies (8,9). Under identical conditions, both tBuSNO and GSNO were stable without photoactivation but, in line with our measured t1/2, tBuSNO rapidly decomposed upon photoactivation, while GSNO levels remained constant (Figure 3). This rapid switch in t1/2 confirms that our SNTs are superior in vitro NO donors to GSNO for PDT applications, as a major change in t1/2 can readily be achieved using conventional light sources.
To confirm that the breakdown of our SNTs was actually releasing NO, we then measured the kinetics of NO release from tDodSNO, the SNT with the longest t1/2, using a polarographic NO electrode. All experiments were performed in a sealed reaction chamber that was shielded from ambient light, with photoactivation being initiated by the delivery of light from a cold light source via a flexible light guide fitted with a condenser lens. To accurately quantify NO release, we ensured that there was no head space between the buffer solution and chamber seal, which prevented NO diffusion from solution. In the absence of photoactivation, we were not able to detect any NO release (Figure 4, Panel A), confirming our previous results (Table 1) that our SNTs are stable in buffer systems. When photoactivation was initiated, we then observed a rapid release of NO, which reached a steady state level of around 8 μm after 4 min (Figure 4, Panel A), with the actual value varying slightly (±2 μm) between repeats depending upon the orientation of the light guide.
Given the substantial shift in t1/2 upon photoactivation (Table 1), we then examined the potential for light flux to directly regulate NO flux within the chamber. Control of NO levels between 0 and 10 μm could be readily achieved, either by step increases in the intensity of the applied light (Figure 4, Panel B) or by cycling the light between the on and off states (Figure 4, Panel C). By combining these two approaches, it was possible to poise NO concentration anywhere within the 0–10 μm range, initially by increasing light flux until the desired concentration was achieved, and then cycling the light to maintain constant steady-state NO levels.
To quantitatively confirm that the signal detected using the polarographic electrode was solely due to NO release, we used the scavenger oxy-myoglobin (MbO2), which rapidly reacts with NO (k = 4 × 107/m/s) with 1:1 stoichiometry to form met-myoglobin and nitrate (24). The kinetics of NO release from 100 μm tDodSNO were monitored and, when NO levels in the chamber reached 6 μm, we introduced 5 μm MbO2 (Figure 4, Panel D). In line with the 1:1 stoichiometry of the reaction, MbO2 rapidly decreased NO levels in the chamber by 5 μm. Once the scavenger was fully oxidized, NO levels then began to increase, and subsequent MbO2 additions indicated that the SNT was constantly releasing NO over the experimental time course.
Photoactivation as a molecular switch – induction of cytotoxicity
We then examined the utility of our SNTs as prodrugs for PDT, with the aim of using photoactivation as a molecular switch to release cytotoxic concentrations of NO. As prior literature indicated that cells must be exposed to NO over a time period of hours in order to induce substantial cytotoxicity (28,29), we selected tDodSNO ( = 260 min) as our lead PDT agent. Using A549 lung carcinoma cells as a tumour model, we established a cytotoxic concentration-response profile between 0 and 100 μm tDodSNO (Figure 5). Our data demonstrated that, with photoactivation, a concentration dependent decrease in cell survival was observed, with decreases in cell survival apparent at 1 μm and the trend becoming highly significant at 50 μm (p < 0.01). In the absence of photoactivation, a shift in the tDodSNO concentration-response profile was observed, with the onset of toxicity occurring at much higher prodrug concentrations, and highly significant changes were not observed until 100 μm tDodSNO (p < 0.01). This differential response was observed at all prodrug concentrations, indicating that photoactivation greatly increased the toxicity of the prodrug.
To further probe the mechanism and extent of NO mediated cell toxicity, we performed a flow cytometry analysis using Annexin V and PI double labelling to determine the extent of apoptotic and necrotic cell death following photoactivation (Figure 6). At an intermediate tDodSNO concentration (40 μm), photoactivation resulted in a significant decrease in viable cells by approximately 15% (p < 0.05), with the majority of cell death being associated with apoptosis (Figure 6). NO release as a result of tDodSNO photoactivation is therefore sufficient to cause extensive cytotoxicity in our cancer model, indicating that our SNTs are promising agents for PDT.
Mechanism of photoactivation-induced cell death
In addition to NO release, the photolysis of SNTs also generates equivalent concentrations of the corresponding thiyl radical (RS•). In in vitro studies, this leads to the formation of the disulfide (RS-SR) although radical-radical recombination. However in a cellular context there are alternative molecular targets for sulphur centred radicals. Therefore photoactivation induced prodrug cytotoxicity could potentially be mediated by the thiyl radical and downstream reactive sulphur species (30–34) rather than by NO.
To resolve the extent of NO’s contribution towards cell death, we therefore investigated the protective effects of the NO trap MbO2 on prodrug toxicity (Figure 7). While MbO2 is extracellular, and tDodSNO likely to partition into cellular membranes, the rapid diffusion coefficient of NO means that extracellular scavengers are capable of efficiently quenching intracellular NO levels (35). We therefore considered that, if cytotoxicity was due to NO release, an extracellular MbO2 concentration of 50 μm should be sufficient to inhibit the cytotoxic action of 80 μm tDodSNO over 1 h (as = 260 min). This prediction was supported by our experimental findings (Figure 7), which demonstrated that MbO2 was able to totally protect cells from tDodSNO toxicity. While these results do not rule out potential interactions between the tDodSNO breakdown products and MbO2, they support the mechanistic hypothesis that NO release is implicated in photoactivation induced cell death.
Our data demonstrates that the photoactivation of either of tBuSNO or tDodSNO lead to the release of NO. Both prodrug candidates displayed promising properties for PDT applications; in the absence of irradiation they were relatively stable in physiological buffer systems (t1/2 = 205 and 3320 min for tBuSNO and tDodSNO respectively, Table 1), while photoactivation via a conventional light source resulted in a major decrease in their half-lives ( = 10 and 260 min for tBuSNO and tDodSNO respectively, Table 1). This behaviour is distinct from other SNTs that have been previously examined for PDT applications; for comparison the well characterized GSNO (8,9) was stable under these conditions and did not release detectable amounts of NO during irradiation.
A major prodrug design criterion was to generate hydrophobic SNTs, as these molecules would be predicted to partition into different intracellular compartments than either hydrophilic thiols such as GSH or redox active transition metal ions. Therefore highly lipid soluble SNTs should display improved intracellular stability, due to the membrane environment shielding them from interactions with these decomposition pathways. Of the two SNT’s studied, tDodSNO was several orders of magnitude more hydrophobic than tBuSNO, with a CLogP value of +5.31. For comparison, the lipid permeable drug chloroquine, which localizes to the lysozome (36), has a LogP value of +4.63 (37), indicating that differential organelle partitioning is likely to be observed for these compounds. A LogP value of +5 is generally considered to be on the borderline of acceptability for oral administration and this solubility property, along with its low molecular weight (231 Da), and ability to engage in hydrogen bonding, meant that tDodSNO fulfilled the commonly utilized ‘rule of five’ criteria for a successful drug candidate (38). In addition to the favourable drug-like characteristics, the molecule also displayed pronounced steric shielding of the SNT group and had a value that would allow NO release over a time period of several hours, which favours NO acting as a cytotoxic agent. This combination of properties prompted us to consider tDodSNO as the more promising of the two SNTs for cancer PDT, and we progressed this prodrug further into in vitro evaluation.
In principle, the use of a photoactivation trigger should enable the rate of NO release to be continuously modulated, either by varying the intensity or the duration of the photostimulus (9). By employing cycled pulses of light, we were able to modulate NO levels in real-time, allowing precise therapeutic control of NO concentration. Importantly, using tDodSNO as the NO donor, we were able to apply this strategy to rapidly generate micromolar levels of NO within the incubation chamber. These NO release kinetics have important implications for PDT, as at these concentrations NO switches its mechanism-of-action from signalling molecule to cellular toxin. Therefore, by selecting the correct irradiation conditions, it is possible to convert tDodSNO into a cytotoxic species.
When administering tDodSNO as a cytotoxin, it was necessary to select the optimum NO release kinetics in order to maximize cell kill. Prior literature indicated that, in order to induce significant cancer cell death, NO must be released within a limited therapeutic window, with the key factors being the concentration and duration of the NO dose (39). For example, using NO donors with a range of well characterized half-lives, Yamamoto et al. were able to demonstrate that NO displayed cytotoxic properties towards PC12 cells at high drug concentrations (>10 μm) (29). However this effect was only apparent when NO was delivered over a prolonged time period (>1 h), and the same dose of NO assessed at shorter administration intervals did not result in significant toxicity (29). Therefore we decided to examine tDodSNO cytotoxicity over 1 h using a range of therapeutically relevant drug concentrations (1–100 μm).
Our initial screen provided proof-of-concept that, by using irradiation to selectively activate tDodSNO, we could induce extensive cytotoxicity over our concentration range. These data were supported quantitatively by flow cytometry with Annexin-PI double staining, which confirmed that even relatively low concentrations of tDodSNO (40 μm) were able to induce significant cell death (≈15%) following 1 h irradiation. These effects could be blocked by NO specific scavengers, supporting our hypothesis that NO release was essential for prodrug action. However, the ability of an NO scavenger to block tDodSNO toxicity does not exclude the possibility that the thiyl radical, which is also generated during the photolysis of SNTs, is also involved in the cytotoxic mechanism, as some of the downstream products of thiyl radical metabolism could potentially react with NO, which would then generate potent oxidizing and nitrating RNS (1).
Conclusions and Future Directions
The ability to readily control the kinetics of NO release from SNTs opens up a range of PDT applications (40). In particular, regulation of NO flux has the potential to provide therapies for hypoxia-reperfusion disorders and cardiovascular disease (41), while at higher exposure levels these PDT agents could be used to selectively kill malignant cells. As NO effects are dependent upon both the cellular environment and duration of NO exposure (42), it is reasonable to expect differences in prodrug action and toxicity between cell lines, and future studies will explore these effects. The hydrophobic class of NO donor therefore shows promise as a new lead compound for the development of therapeutics to treat a range of diseases. Additionally the utility of the photoactivation trigger for NO release provides a new molecular tool for chemical biology approaches to the study NO signalling.
We are grateful for financial support provided by a Laurenson Award from the Otago Medical Research Foundation and a Dean’s Bequest Grant from the Otago School of Medical Sciences, University of Otago.