Hypoxia-Activated Prodrug Derivatives of Carbonic Anhydrase Inhibitors in Benzenesulfonamide Series: Synthesis and Biological Evaluation

Hypoxia, a common feature of solid tumours’ microenvironment, is associated with an aggressive phenotype and is known to cause resistance to anticancer chemo- and radiotherapies. Tumour-associated carbonic anhydrases isoform IX (hCA IX), which is upregulated under hypoxia in many malignancies participating to the microenvironment acidosis, represents a valuable target for drug strategy against advanced solid tumours. To overcome cancer cell resistance and improve the efficacy of therapeutics, the use of bio-reducible prodrugs also known as Hypoxia-activated prodrugs (HAPs), represents an interesting strategy to be applied to target hCA IX isozyme through the design of selective carbonic anhydrase IX inhibitors (CAIs). Here, we report the design, synthesis and biological evaluations including CA inhibition assays, toxicity assays on zebrafish and viability assays on human cell lines (HT29 and HCT116) of new HAP-CAIs, harboring different bio-reducible moieties in nitroaromatic series and a benzenesulfonamide warhead to target hCA IX. The CA inhibition assays of this compound series showed a slight selectivity against hCA IX versus the cytosolic off-target hCA II and hCA I isozymes. Toxicity and viability assays have highlighted that the compound bearing the 2-nitroimidazole moiety possesses the lowest toxicity (LC50 of 1400 µM) and shows interesting results on viability assays.


Introduction
Intratumoral heterogeneity, a main feature of solid tumours, is one of the causes of the intractability of cancers [1]. There are multiple mechanisms driving tumor heterogeneity, including genetic, epigenetic and microenvironmental factors such as hypoxia [2]. The presence of oxygen deprivation areas (typically less than 1%), defined as hypoxic domains, and resulting from inadequate tumor vascularization, have been identified in a wide variety of human tumours [3]. The adaptive cellular response to low

Carbonic Anhydrase Inhibition Assay
The hypoxia-activated prodrug carbonic anhydrase IX inhibitors (HAP-CAIs) reported here were assayed using the CO2 hydrase assay against three physiologically relevant human CA isoforms, the cytosolic hCA I and II and the transmembrane, tumor-associated hCA IX (Table 1) [32]. Table 1. Inhibitory activity of compounds 1b-5b, 2c-3c, and the clinically used sulfonamide inhibitor acetazolamide (AAZ), against hCA I, hCA II, and hCA IX using a stopped flow CO2 hydrase assay. The clinically used sulfonamide acetazolamide (AAZ, 5-acetamido-1, 3, 4-thiadiazole-2sulfonamide) has been taken along as standard in these measurements. All compounds acted as inhibitors against the three isoforms hCA I, II and IX, although with variable potency (Table 1). Against the abundant, cytosolic isoform hCA I compound 3c showed weak inhibition potency (2180 nM). Compound 1b showed moderate inhibition activity towards hCA I (166.7 nM) and hCA II (30.6 nM) isoforms and higher inhibition (7.6 nM) towards hCA IX, while compounds 2b and 2c strongly inhibited all CA isoforms (KI ranging from 2.3 to 14.1 nM) with no differences observed in the selectivity ratios hCA II over hCA IX (2b = 0.35, 2c = 0.28). Compounds 3b and 3c showed a moderate inhibition of hCA IX, whilst compound 3b binding to hCA II was very tight (3.1 nM). The large difference in the binding capacity or selectivity ratios (3b = 0.09, 3c = 0.94) of compounds from the same family (nitrogen mustard) supports the substitution effect on the carbamate linker. Compounds 4b and 5b showed moderate inhibition activity towards hCA I, respectively (84 nM) and (64.7 nM), while they strongly inhibited hCA II and hCA IX (KI ranging from 5.7 to 19.8 nM). Nevertheless,

Carbonic Anhydrase Inhibition Assay
The hypoxia-activated prodrug carbonic anhydrase IX inhibitors (HAP-CAIs) reported here were assayed using the CO 2 hydrase assay against three physiologically relevant human CA isoforms, the cytosolic hCA I and II and the transmembrane, tumor-associated hCA IX (Table 1) [32]. Table 1. Inhibitory activity of compounds 1b-5b, 2c-3c, and the clinically used sulfonamide inhibitor acetazolamide (AAZ), against hCA I, hCA II, and hCA IX using a stopped flow CO 2 hydrase assay. The clinically used sulfonamide acetazolamide (AAZ, 5-acetamido-1, 3, 4thiadiazole-2-sulfonamide) has been taken along as standard in these measurements. All compounds acted as inhibitors against the three isoforms hCA I, II and IX, although with variable potency (Table 1). Against the abundant, cytosolic isoform hCA I compound 3c showed weak inhibition potency (2180 nM). Compound 1b showed moderate inhibition activity towards hCA I (166.7 nM) and hCA II (30.6 nM) isoforms and higher inhibition (7.6 nM) towards hCA IX, while compounds 2b and 2c strongly inhibited all CA isoforms (K I ranging from 2.3 to 14.1 nM) with no differences observed in the selectivity ratios hCA II over hCA IX (2b = 0.35, 2c = 0.28). Compounds 3b and 3c showed a moderate inhibition of hCA IX, whilst compound 3b binding to hCA II was very tight (3.1 nM). The large difference in the binding capacity or selectivity ratios (3b = 0.09, 3c = 0.94) of compounds from the same family (nitrogen mustard) supports the substitution effect on the carbamate linker. Compounds 4b and 5b showed moderate inhibition activity towards hCA I, respectively (84 nM) and (64.7 nM), while they strongly inhibited hCA II and hCA IX (K I ranging from 5.7 to 19.8 nM). Nevertheless, considering the difficulty of obtaining small compounds with a better affinity for the tumor-associated isozyme (hCA IX) over hCA II, the selectivity obtained for these series is comparable or better than that of the clinically used CA inhibitor acetazolamide AAZ.

Stability of Carbamate Linker under Acidic Conditions
The stability to chemical hydrolysis of compounds 1b-3c was evaluated by measuring the peak area or retention time of the compound after incubation at varying pH conditions up to 8 h. No degradation was observed at any pH condition for all tested compounds, indicating significant stability. Additionally, all tested compounds were found to be stable for at least 8 h to harsh acid-catalyzed hydrolysis (pH = 2.0, 37 • C, data not shown). Literature studies have shown that the electron-withdrawing nitro group conjugated with the carbamate linker resulted in a remarkable decrease in stability, which may explain the apparent low inhibitory potency of these compounds toward fatty acid amide hydrolase (FAAH), due to decomposition under the assay conditions [33]. Of all compounds (1b-3c), the nitro group was not in direct conjugation with the carbamate linker, thereby showing no stability loss when incubated under various pH conditions.

Cell Viability and Clonogenic Assays
The cytotoxicity of all compounds (1b-3c) was determined in a panel of human tumor cell lines. Compound 1b showed an IC 50 of 204.5 µM (p < 0.0001) in HT29 cells under anoxia (IC 50 A) and no detectable cytotoxicity (IC 50 > highest tested concentration) was observed under normoxia ( Figure 1). Furthermore, in HCT116 cells, compound 1b resulted in an IC 50 of 148.6 and 59.3 µM under normoxic and anoxic conditions, respectively, resulting in a hypoxia selectivity cytotoxicity ratio (HCR) of 2.5 ( Table 2). All other compounds, except compound 3b in HCT116 cells, did not show any cytotoxicity at the tested concentrations. Compound 3b resulted in a cell-dependent cytotoxicity with an HCR of 2.7 in HCT116 cells, while no cytotoxicity in HT29 cells was observed. considering the difficulty of obtaining small compounds with a better affinity for the tumorassociated isozyme (hCA IX) over hCA II, the selectivity obtained for these series is comparable or better than that of the clinically used CA inhibitor acetazolamide AAZ.

Stability of Carbamate Linker under Acidic Conditions
The stability to chemical hydrolysis of compounds 1b-3c was evaluated by measuring the peak area or retention time of the compound after incubation at varying pH conditions up to 8 h. No degradation was observed at any pH condition for all tested compounds, indicating significant stability. Additionally, all tested compounds were found to be stable for at least 8 h to harsh acidcatalyzed hydrolysis (pH = 2.0, 37 °C, data not shown). Literature studies have shown that the electron-withdrawing nitro group conjugated with the carbamate linker resulted in a remarkable decrease in stability, which may explain the apparent low inhibitory potency of these compounds toward fatty acid amide hydrolase (FAAH), due to decomposition under the assay conditions [33]. Of all compounds (1b-3c), the nitro group was not in direct conjugation with the carbamate linker, thereby showing no stability loss when incubated under various pH conditions.

Cell Viability and Clonogenic Assays.
The cytotoxicity of all compounds (1b-3c) was determined in a panel of human tumor cell lines. Compound 1b showed an IC50 of 204.5 µM (p < 0.0001) in HT29 cells under anoxia (IC50A) and no detectable cytotoxicity (IC50 > highest tested concentration) was observed under normoxia ( Figure 1). Furthermore, in HCT116 cells, compound 1b resulted in an IC50 of 148.6 and 59.3 µM under normoxic and anoxic conditions, respectively, resulting in a hypoxia selectivity cytotoxicity ratio (HCR) of 2.5 ( Table 2). All other compounds, except compound 3b in HCT116 cells, did not show any cytotoxicity at the tested concentrations. Compound 3b resulted in a cell-dependent cytotoxicity with an HCR of 2.7 in HCT116 cells, while no cytotoxicity in HT29 cells was observed.   Based on its selective cytotoxicity under anoxia in both cell lines, compound 1b was selected for further studies, investigating its effects on cell survival. Compound 1b did not reduce clonogenic cell survival under normoxia or anoxia at the tested concentrations ( Figure 2). TH-302, a 2-nitroimidazole based hypoxia-activated prodrug-alkylating agent was used as positive control (data not shown) [34]. Based on its selective cytotoxicity under anoxia in both cell lines, compound 1b was selected for further studies, investigating its effects on cell survival. Compound 1b did not reduce clonogenic cell survival under normoxia or anoxia at the tested concentrations ( Figure 2). TH-302, a 2-nitroimidazole based hypoxia-activated prodrug-alkylating agent was used as positive control (data not shown) [34]. and anoxia (black bars) when exposed to compound 1b. The data represent the average ± SEM of three independent biological repeats.

Toxicity Evaluation on Zebrafish
During development, zebrafish embryos are easily affected by chemical compounds compared to adult zebrafish or other animal models or cell models and are therefore suitable for assessing the subtle toxic effects of chemicals [28,30]. The toxicity of compound 1b, 4b and 5b was determined by using 24-h post-fertilization zebrafish embryos [35]. In these tests, the LC50, zebrafish phenotypic parameters and the swim pattern were analyzed for each compound.

Determination of Half Maximal Lethal Concentration 50 (LC50)
The lethality of 1b, 4b and 5b on the developing zebrafish embryos was concentration dependent ( Figure 3). Among the three compounds tested for the toxicity, 1b was the less toxic and did not cause any mortality of the larvae, even at a 1 mM concentration at the end of 5 day of exposure to the compound. Compound 4b was more toxic compared to the other two compounds and caused significant mortality even at a 500-µM concentration ( Figure 3). The LC50 values of the prodrugs at the end of the 5 days of exposure were in the range of 500 (4b), 1000 (5b) and 1400 µM (1b), as shown in Figure 3. The LC50 concentration of the compounds were higher compared to the inhibitors that we screened in our earlier studies [36,37], suggesting that these compounds can be characterized further for developing as drugs.  anoxia (black bars) when exposed to compound 1b. The data represent the average ± SEM of three independent biological repeats.
Abbreviations: IC 50 : Concentration of an inhibitor where the response is reduced by half; Normoxia (N); Anoxia (A); Hypoxia selectivity cytotoxicity ratio (HCR).

Toxicity Evaluation on Zebrafish
During development, zebrafish embryos are easily affected by chemical compounds compared to adult zebrafish or other animal models or cell models and are therefore suitable for assessing the subtle toxic effects of chemicals [28,30]. The toxicity of compound 1b, 4b and 5b was determined by using 24-h post-fertilization zebrafish embryos [35]. In these tests, the LC 50 , zebrafish phenotypic parameters and the swim pattern were analyzed for each compound.

Determination of Half Maximal Lethal Concentration 50 (LC 50 )
The lethality of 1b, 4b and 5b on the developing zebrafish embryos was concentration dependent (Figure 3). Among the three compounds tested for the toxicity, 1b was the less toxic and did not cause any mortality of the larvae, even at a 1 mM concentration at the end of 5 day of exposure to the compound. Compound 4b was more toxic compared to the other two compounds and caused significant mortality even at a 500-µM concentration ( Figure 3). The LC 50 values of the prodrugs at the end of the 5 days of exposure were in the range of 500 (4b), 1000 (5b) and 1400 µM (1b), as shown in Figure 3. The LC 50 concentration of the compounds were higher compared to the inhibitors that we screened in our earlier studies [36,37], suggesting that these compounds can be characterized further for developing as drugs.
Different phenotypic parameters were also analyzed (hatching, heartbeat, edema, swim bladder development, yolk sac utilization and body shape) from the developing larvae of 1-5 days, after exposure to the compounds. Figure 4 shows the representative images of larvae treated with the concentrations considered safe for further characterization. Among the compounds screened, the prodrug 1b was found to be less toxic with no or minimal phenotypic abnormalities, even at 1 mM. In our earlier study, compounds 1b-3c also showed minimal or no phenotypic abnormalities at 1 mM concentrations. In the present study, compounds 4b and 5b showed phenotypic defects, such as edema and the absence of a swim bladder (arrows) at lower concentrations compared to 1b, as shown in compound. Compound 4b was more toxic compared to the other two compounds and caused significant mortality even at a 500-µM concentration (Figure 3). The LC50 values of the prodrugs at the end of the 5 days of exposure were in the range of 500 (4b), 1000 (5b) and 1400 µM (1b), as shown in Figure 3. The LC50 concentration of the compounds were higher compared to the inhibitors that we screened in our earlier studies [36,37], suggesting that these compounds can be characterized further for developing as drugs.  Different phenotypic parameters were also analyzed (hatching, heartbeat, edema, swim bladder development, yolk sac utilization and body shape) from the developing larvae of 1-5 days, after exposure to the compounds. Figure 4 shows the representative images of larvae treated with the concentrations considered safe for further characterization. Among the compounds screened, the prodrug 1b was found to be less toxic with no or minimal phenotypic abnormalities, even at 1 mM. In our earlier study, compounds 1b-3c also showed minimal or no phenotypic abnormalities at 1 mM concentrations. In the present study, compounds 4b and 5b showed phenotypic defects, such as edema and the absence of a swim bladder (arrows) at lower concentrations compared to 1b, as shown in Figure 4. . Images of zebrafish larvae in the control and prodrug-treated groups. Representative images of 2-5 dpf zebrafish larvae exposed to different concentrations (80 µM-1000µM) of 1b, 4b and 5b. The upper panel shows images for normal development of zebrafish larvae in the control group (not treated with inhibitors) and the 1% dimethyl sulfoxide (DMSO)-treated group. The lower panel shows the larvae treated with concentrations of the compounds at which they induced minimal or no phenotypic defects. Prodrug 4b showed (arrow) an absence of a swim bladder at a 250-µM concentration at 5 days. Prodrug 5b induced edema (arrow) as early as 2 days post exposure to the compound. Compound 1b showed an absence of a swim bladder (arrow) at a 1000-µM concentration at 5 days.
Furthermore, the toxic effects of compounds across the concentrations (20 µM-2 mM) were assessed on individual parameters of the 5 days post exposed zebrafish larvae. Figure 5 presents plot graphs of the dose-dependent effects of prodrugs on larvae. The results indicate that the 1b Figure 4. Images of zebrafish larvae in the control and prodrug-treated groups. Representative images of 2-5 dpf zebrafish larvae exposed to different concentrations (80 µM-1000 µM) of 1b, 4b and 5b. The upper panel shows images for normal development of zebrafish larvae in the control group (not treated with inhibitors) and the 1% dimethyl sulfoxide (DMSO)-treated group. The lower panel shows the larvae treated with concentrations of the compounds at which they induced minimal or no phenotypic defects. Prodrug 4b showed (arrow) an absence of a swim bladder at a 250-µM concentration at 5 days. Prodrug 5b induced edema (arrow) as early as 2 days post exposure to the compound. Compound 1b showed an absence of a swim bladder (arrow) at a 1000-µM concentration at 5 days. Furthermore, the toxic effects of compounds across the concentrations (20 µM-2 mM) were assessed on individual parameters of the 5 days post exposed zebrafish larvae. Figure 5 presents plot graphs of the dose-dependent effects of prodrugs on larvae. The results indicate that the 1b compound showed minimal or no adverse effects on the observable parameters of the larvae ( Figure 5A-E). However, prodrugs 4b and 5b exhibited adverse effect on hatching, swim bladder development, the utilization of the yolk sac, and the shape of the body and induced pericardial edema, as shown in Figure 5A-E.

Swim Pattern Analysis to Assess the Subtle Toxic Effects of the Prodrugs
To assess the subtle toxic effects of the prodrugs on the swim patterns of the larvae, they were assessed at the end of 5 days of exposure to the compounds. The swim pattern analysis showed no abnormal or ataxic movement pattern in the larvae exposed to the concentration (160 µM) that did not induce any of the phenotypic defects and is considered to be safe. In our earlier studies, the nitroimidazole-based compounds DTP338 and DTP348 showed an ataxic movement pattern even at a 100-µM concentration due to neurotoxicity [28,30]. Therefore, the prodrug screening in the current study can be considered safe for further preclinical characterization and development as potential drugs.

Discussion
We developed here new HAPs incorporating a benzenesulfonamide through a carbamate linker, designed with different bio-reducible moieties for the selective delivery of anticancer drugs on hypoxic tumors. Through this study, we demonstrated that these HAPs can selectively inhibit the hCA IX at the nanomolar level. In contrast to other nitroaromatic drugs, HAPs 1b-3c did not show stability loss when incubated under various pH conditions as the nitro group was not in direct conjugation with the carbamate linker. Compound 1b, harboring a 2-nitroimidazole as a reducible moiety, showed the best cytotoxic action against HT29 and HCT116 cells (IC50 of 204.5 and 59.36 µM

Swim Pattern Analysis to Assess the Subtle Toxic Effects of the Prodrugs
To assess the subtle toxic effects of the prodrugs on the swim patterns of the larvae, they were assessed at the end of 5 days of exposure to the compounds. The swim pattern analysis showed no abnormal or ataxic movement pattern in the larvae exposed to the concentration (160 µM) that did not induce any of the phenotypic defects and is considered to be safe. In our earlier studies, the nitroimidazole-based compounds DTP338 and DTP348 showed an ataxic movement pattern even at a 100-µM concentration due to neurotoxicity [28,30]. Therefore, the prodrug screening in the current study can be considered safe for further preclinical characterization and development as potential drugs.

Discussion
We developed here new HAPs incorporating a benzenesulfonamide through a carbamate linker, designed with different bio-reducible moieties for the selective delivery of anticancer drugs on hypoxic tumors. Through this study, we demonstrated that these HAPs can selectively inhibit the hCA IX at the nanomolar level. In contrast to other nitroaromatic drugs, HAPs 1b-3c did not show stability loss when incubated under various pH conditions as the nitro group was not in direct conjugation with the carbamate linker. Compound 1b, harboring a 2-nitroimidazole as a reducible moiety, showed the best cytotoxic action against HT29 and HCT116 cells (IC 50 of 204.5 and 59.36 µM under hypoxia) and an HCR of approximately 2.5. Our data also show that 1b is the less toxic, with no mortality observed on zebrafish larvae even at a 1-mM concentration at the end of 5 days of exposure. Of note, prodrug 1b also did not induce phonotypic abnormalities in the larvae as well as abnormal or ataxic swim movement at the concentration that did not induce any of the phenotypic defects and is considered as safe. As other nitroimidazole derivatives, such as DTP338 and DTP348 [30], induced a neurotoxic effect even at a 100-µM concentration, the data gathered here suggest a potential for derivative 1b to be optimized for the development of new safer HAPs.

Chemistry
General. All reagents and solvents were of commercial quality and used without further purification unless otherwise specified. All reactions were carried out under an inert atmosphere of nitrogen. TLC analyses were performed on silica gel 60 F254 plates (Merck Art. no. 1.05554). Spots were visualized under 254 nm UV illumination or by ninhydrin solution spraying. 1 H-and 13 C-NMR spectra were recorded on a Bruker DRX-400 spectrometer (Bruker, Hanau, Germany) using dimethyl sulfoxide-d6 (DMSO-d 6 ) as solvent and tetramethylsilane as an internal standard. For 1 Hand 13 C-NMR spectra, chemical shifts are expressed in δ (ppm) downfield from tetramethylsilane, and coupling constants (J) are expressed in hertz. All compounds that were tested in the biological assays were analyzed by high-resolution mass spectra (HRMS) using a Bruker micrOTOF-Q II mass spectrometer (Bruker, Hanau, Germany) fitted with an electrospray ion source General procedure for preparation of chloroformate: The corresponding hydroxy starting compound (1a, 2a and 3a) (2 mmol) in tetrahydrofuran (10 mL) was added to phosgene (4 mL, 8 mmol) and tetrahydrofuran (THF) (15 mL) at 0 • C. The reaction mixture was stirred for 16 h, then the solvent was removed in vacuo. The crude chloroformate was used without further purification.
General procedure for synthesis of 1b, 2b, 2c, 3b, and 3c: A suspension of chloroformate (23.4 mmol) in dry THF (100 mL) was treated with trimethylamine (50 mmol). The corresponding amine (23.4 mmol) was added afterwards and the reaction was stirred overnight at room temperature. The solvent was removed in vacuo. The residue was dissolved in chloroform and washed twice with 1.0 N sodium hydroxide and water. The organic phase was dried over anhydrous sodium sulfate, filtered and concentrated under vacuum. The residue was purified by chromatography on silica gel using methylene chloride−methanol 98:2 as eluent.
General procedure for synthesis of 4b and 5b: To a suspension of carbonyldiimidazole (CDI) (0.31 mmol) in dry THF (5 mL) the corresponding hydroxy starting compound (4a and 5a) (0.28 mmol) was added. The reaction was stirred for 3h at room temperature and the 4-(2-aminoethyl) benzene sulfonamide was added. The reaction mixture was stirred overnight at room temperature and concentrated under reduced pressure. The residue was dissolved in EtOAc (12.5 mL) and washed with water (2 × 7.5 mL) and brine (7.5 mL). The organic phase was dried over anhydrous sodium sulfate, filtered and concentrated under vacuum. The residue was purified by chromatography on silica gel using cyclohexane-ethyl acetate 3:7 as eluent.   155.87, 151.72, 143.53, 142.68, 138.45, 133.07, 127.26, 125.65, 62.11

Carbonic Anhydrase Inhibition Assays
An Sx.18Mv-R Applied Photophysics (Oxford, UK) stopped-flow instrument was used for assaying the CA catalyzed CO 2 hydration activity. Phenol red (at a concentration of 0.2 mM) was used as an indicator, working at the absorbance maximum of 557 nm, with 20 mM HEPES (pH 7.5) as buffer, and 20 mM Na 2 SO 4 (for maintaining the constant ionic strength), following the initial rates of the CA-catalyzed CO 2 hydration reaction for a period of 10-100 s. The CO 2 concentrations ranged from 1.7-17 mM for the determination of the kinetic parameters and inhibition constants. In particular, CO 2 was bubbled in distilled deionized water for 30 min so that the water was saturated (the concentration at a specific temperature is known from literature). In addition, a CO 2 assay kit (from Merck, Darmstadt, Germany) was used to measure the concentration in variously diluted solutions obtained from the saturated one (which was kept at the same temperature and a constant bubbling during the experiments). For each inhibitor, at least six traces of the initial 5%-10% of the reaction were used for determining the initial velocity [32]. The uncatalyzed rates were determined in the same manner and subtracted from the total observed rates. Stock solutions of inhibitor (0.1 mM) were prepared in distilled deionized water and dilutions up to 0.01 nM were done thereafter with distilled deionized water. Inhibitor and enzyme solutions were pre-incubated together for 15 min-2 h (or longer, i.e., 4-6 h) at room temperature (at 4 • C for the incubation periods longer than 15 min) prior to assay, in order to allow for the formation of the EI complex. The inhibition constants were obtained by non-linear least-squares methods using PRISM 3 and the Cheng-Prusoff equation, as reported earlier [38,39], and represent the mean from at least three different determinations. hCA I was purchased by Merck (Darmstadt, Germany) and used without further purification, whereas all the other hCA isoforms were recombinant ones obtained in-house as reported earlier [40].

In Vitro Chemical Stability and Analytical Method
The chemical stability of compounds 1b, 2b, 2c, 3b and 3c at various pH (6.2, 6.6, 7.0 and 7.4) was measured on a Waters XEVO QTOF G2 Mass Spectrometer (Waters, Etten-Leur, The Netherlands), with an Acquity H-class solvent manager, Flow-Through Needle (FTN) sample manager and Tunable UV (TUV) detector. The system was equipped with a reversed phase C18 column (Waters, Etten-Leur, The Netherlands, acquity PST 130A, 1.7 µm 2.1 × 50 mm i.d.), with a column temperature of 40 degrees. Mobile phases consisted of 0.1% formic acid in water and 90% acetonitrile. FTN-purge solvent was 5% acetonitrile in water. Mobile phase gradient was maintained starting with 5% acetonitrile to 50% acetonitrile for 15 min at a 220-nm wavelength. Stock solutions of compounds were prepared in DMSO and each sample was incubated at a final concentration of 1-5 µM in pre-thermostated buffered solution. The final DMSO concentration in the samples was kept at 1%. The samples were maintained at 37 • C in a temperature-controlled shaking water bath (60 rpm). At various time points, 100-µL aliquots were removed and injected into the High-Performance Liquid Chromatography (HPLC) system for analysis. Mass was measured in positive sensitivity mode with a mass range between 100 and 1000 Da.

Cells
Human colorectal HCT116 and HT29 carcinoma cells were cultured in Dulbecco's Modified Eagle Medium (DMEM, Lonza) supplemented with 10% fetal bovine serum and 1% PenStrep. Cells were exposed to anoxic conditions for 24 h in a hypoxic chamber (MACS VA500 microaerophilic workstation, Don Whitley Scientific, Bingley, UK) with an atmosphere consisting of ≤0.02% O 2 , 5% CO 2 and residual N 2 . Normoxic cells were grown in regular incubators with 21% O 2 , 5% CO 2 at 37 • C.

Cell Viability Assays
The cytotoxic efficacy of the bio-reducible derivatives was determined based on cell viability assays using alamarBlue ® (Invitrogen). In short, HT29 and HCT116 were seeded in 96-well plates and allowed to attach overnight. On the following day, the plates were exposed to normoxia or anoxia and medium was replaced with pre-incubated normoxic or anoxic DMEM. Compounds were dissolved in DMSO (0.5%, Sigma-Aldrich) and final concentrations were made with pre-incubated anoxic or normoxic DMEM and added to the wells after 24 h of exposure. Cells were exposed to compounds for a total of 2 h, after which the medium was washed off and replaced with fresh medium. Cells were allowed to grow for an additional 72 h under normoxic conditions prior to measurement. Cells were incubated with alamarBlue ® for 2 h during normoxic conditions, which corresponds with their metabolic function, a measure for cell viability. Fluorescence was measured using plate reader (FLUOstar Omega, BMG LABTECH, Ortenberg, Germany) using a fluorescence excitation wavelength of 540-580 nm.

Clonogenic Assays
Clonogenic survival HT29 and HCT116 cells were seeded with a high density for CA IX-dependent extracellular acidification [41]. Compound 1b doses were selected from corresponding IC 50 (anoxic/normoxic) ranging from 10 to 100 µM for the preliminary experiments. These cells were exposed to 23:30 h normoxic or anoxic conditions and 30 min to drugs after which cells were trypsinized and reseeded in triplicate with known cell numbers. Cells were allowed to grow for 9 (HCT116) and 14 (HT29) days to form colonies that were quantified after staining and fixation with 0.4% methylene blue in 70% ethanol. Surviving fractions were calculated and compared to control survival curves produced in the same experiment without compound treatment.

Statistical Analyses
GraphPad Prism (version 5.03, GraphPad Software, San Diego, CA, USA) was used for all statistical analyses. For the cytotoxic compounds, IC 50 values were estimated with the curve of the log (inhibitor) vs. normalized response (Variable slope).

Preparation of Inhibitor Samples
The hypoxia-activated prodrugs were screened for their toxicity using 24-h post-fertilization zebrafish embryos as described earlier [35]. The compounds were either dissolved in Embryonic medium (5.0 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl 2 , 0.33 mM MgSO 4 , and 0.1% w/v Methylene Blue (Sigma-Aldrich, Darmstadt, Germany) or in dimethyl sulfoxide (DMSO) (Sigma-Aldrich, St. Louis, MO, USA) to prepare 100 mM stock solutions. Before the start of each experiment, the series of dilutions were made from the above stock in the embryonic medium.

Maintenance of the Zebrafish
The wild-type adult zebrafish (AB strains) were maintained at 28.5 • C in an incubator [35]. In each breeding tank, about embryos 3-5 pairs of male and female fish were set up overnight for collecting the embryos. The following morning, 1-2-h post-fertilization (hpf) embryos were collected in a sieve and rinsed with embryonic medium and kept the collected embryos in an incubator at 28.5 • C overnight [35]. The toxicity evaluation studies of the compounds were performed using the fish that were 24-hpf. All the zebrafish experiments were performed at the zebrafish core facility, Tampere University, Finland, and according to the protocol used in our laboratory.
Ethical statement. The research unit at Tampere University has an established zebrafish core facility authorized granted by the National Animal Experiment Board (ESAVI/7975/04.10.05/2016). The experiments using developing zebrafish embryos were performed according to the Provincial Government of Eastern Finland Province Social and Health Department Tampere Regional Service Unit protocol # LSLH-2007-7254/Ym-23.

Determination of half Maximal Lethal Concentration 50 (LC 50 )
The LC 50 values for all the three prodrugs were determined using 24-hpf embryos using 810 different concentrations for each compound. We used a minimum of 30 zebrafish larvae for each concentration of the compound [28,30,36,42]. The 24-hpf larvae were exposed to different concentrations of the inhibitors that ranged from 20 µM to mM. A dose response curve (DRC) was calculated using the DRM (model fitting function) of the DRC R package [43]. The control group constituted an equal number of larvae not treated with any compound and the larvae that were treated with 1% DMSO. Toxicological evaluation studies were performed in 24-well plates (Corning V R Costar V R cell culture plates). In each well, we placed 1-2 24-hpf embryos in 1 mL of embryonic medium containing a diluted inhibitor. For control groups, 1% DMSO diluted in either embryonic medium or embryonic medium with no chemical compound. A minimum of three sets of experiments were carried out for each prodrug. The mortality of the larvae was checked every 24 h until 5 days after exposure to the prodrugs.

Phenotypic Analysis of the Larvae
After exposure to the prodrugs, we evaluated the effects of these compound on the zebrafish larvae by analyzing the following eight phenotypic parameters: (1) mortality, (2) hatching (3) edema (4) swimming pattern, (5) yolk sack utilization, (6) heartbeat, (7) body shape, and (8) swim bladder development. The images of the developing larvae were taken using a Lumar V1.12 fluorescence microscope (Carl Zeiss MicroImaging GmbH, Göttingen, Germany) attached to a camera with a 1.5 lens (Carl Zeiss MicroImaging GmbH, Göttingen, Germany). The images were analyzed with AxioVision software versions 4.7 and 4.8 as described in our standard protocol for the assessment of the toxicity and safety of chemical compounds.

Swim Pattern Analysis
The swim pattern of the zebrafish larvae was studied after 5 days of exposure to these compounds. For the analyses of swim pattern, about 15-20 zebrafish larvae after 5 days of exposure to the compounds were placed in a 35 mm × 15 mm petri dish containing embryonic medium, and the larvae were allowed to settle in the petri dish for 1 min. The movement of the zebrafish larvae was observed under the microscope for 1 min. A short video of about 1 min was taken for each group of larvae that were treated with a concentration of the compound that did not show any phenotypic abnormalities. The swim patterns were compared with the control group zebrafish larvae that were not treated with any inhibitor.