Nitroimidazole conjugates of bis(thiosemicarbazonato)⁶⁴Cu(II) – Potential combination agents for the PET imaging of hypoxia

Abstract

Combination agents comprising two different pharmacophores with the same biological target have the potential to show additive or synergistic activity. Bis(thiosemicarbazonato)copper(II) complexes (e.g. ⁶⁴Cu-ATSM) and nitroimidazoles (e.g. ¹⁸F-MISO) are classes of tracer used for the delineation of tumor hypoxia by positron emission tomography (PET). Three nitroimidazole-bis(thiosemicarbazonato)copper(II) conjugates were produced in order to investigate their potential as combination hypoxia imaging agents. Two were derived from the known bifunctional bis(thiosemicarbazone) H₂ATSM/A and the third from the new precursor diacetyl-2-(4-N-methyl-3-thiosemicarbazone)-3-(4-N-ethylamino-3-thiosemicarbazone) – H₂ATSM/en. Oxygen-dependent uptake studies were performed using the ⁶⁴Cu radiolabelled complexes in EMT6 carcinoma cells. All the complexes displayed appreciable hypoxia selectivity, with the nitroimidazole conjugates displaying greater selectivity than a simple propyl derivative used as a control. Participation of the nitroimidazole group in the trapping mechanism is indicated by the increased hypoxic uptake of the 2- vs. the 4-substituted ⁶⁴Cu-ATSM/A derivatives. The 2-nitroimidazole derivative of ⁶⁴Cu-ATSM/en demonstrated superior hypoxia selectivity to ⁶⁴Cu-ATSM over the range of oxygen concentrations tested. Biodistribution of the radiolabelled 2-nitroimidazole conjugates was carried out in EMT6 tumor-bearing mice. The complexes showed significantly different uptake trends in comparison to each other and previously studied Cu-ATSM derivatives. Uptake of the Cu-ATSM/en conjugate in non-target organs was considerably lower than for derivatives based on Cu-ATSM/A.

1. Introduction

Hypoxia occurs to a variable extent in many solid tumors [1]. Its prominence in cancer is attributable to the impaired vasculature of tumors that occurs as a result of uncontrolled cell proliferation [2,3]. Clinical diagnosis of hypoxia is important because it is correlated with resistance to conventional therapeutic treatments and has a profound impact on patient survival [4]. Hypoxic tumors are often less susceptible to traditional chemotherapy [5] as a result of both limited drug penetration and induced resistance mechanisms [6,7]. Similarly, radiotherapy is usually far less effective in regions of hypoxia [8,9], mainly because oxygen is required to convert the initial DNA free radicals formed by ionisation into strand breaks [10]. Consequently there has been significant interest in developing non-invasive techniques for delineating hypoxia and a number of PET and SPECT agents have been developed for this purpose [11–16].

Nitroimidazoles and bis(thiosemicarbazonato)copper(II) complexes are two classes of hypoxia tracer that have been extensively investigated. It has been shown that 2-nitroimidazoles can be trapped in cells with low pO₂ values [17], and the 2-nitroimidazole compound [¹⁸F]fluoromisonidazole (¹⁸FMISO) has been used for PET imaging of stroke, ischemia myocardium and tumor hypoxia [12,18–20]. 2-Nitroimidazole cyclam derivatives radiolabelled with ^99mTc, ⁶⁴Cu and ⁶⁷Cu have also been investigated as potential PET/SPECT agents for tumor hypoxia [21]. Bis(thiosemicarbazone) ligands derived from 1,2-diones form low molecular weight, neutral copper(II) complexes which are capable of rapid cellular uptake. Diacetyl-bis(N⁴-methylthiosemicarbazonato)copper(II) (Cu-ATSM, where the positron emitting isotopes ⁶⁰Cu, ⁶¹Cu or ⁶⁴Cu are used) has been shown to be selective for hypoxic tissue [22–24] and has been used clinically for the detection of hypoxia in cancer [25–27]. Both classes of tracer are thought to function via a redox trapping mechanism. Having crossed the cell membrane, enzyme mediated reduction of the tracer molecule occurs. In cells with normal oxygen concentrations reoxidation by intracellular oxygen is rapid and the molecule can then leave the cell. In oxygen deprived environments, reoxidation is slow and decomposition of the molecule and/or additional reduction reactions occur leaving the radiolabelled fragments trapped in the cell. In the case of nitroimidazoles initial reduction to the radical anion is reversible, but subsequent reductions result in irreversible fragmentation [19,28]. With bis(thiosemicarbazonato)copper(II) complexes the initial reduction from Cu(II) to Cu(I) occurs reversibly, but the Cu(I) species formed is thought to become protonated and undergo decomplexation and trapping of the copper ion [24]. Derivatives such as Cu-ATSM that contain the diacetyl backbone show the best hypoxia selectivity and it has been suggested that this is due to their reduction being more reversible and occurring at a more negative reduction potential in comparison to other variants [29–31].

Bioconjugation is an attractive approach for improving the in vivo characteristics of hypoxia selective agents, since it has the potential to provide additional transport or targeting properties. For this reason we have been investigating strategies to introduce a functional group for conjugation into diacetyl-bis(N⁴-methylthiosemicarbazones) without disrupting the diacetyl backbone or significantly altering the Cu(I/II) redox potential of the complex. The first derivative we developed which fulfils these requirements – H₂ATSM/A – utilizes a hydrazine group at one N-terminus [32]. A variety of ⁶⁴Cu-ATSM/A conjugates have been synthesized and we recently reported preliminary in vitro and in vivo results for a number of these. A series of lipophilic imine conjugates demonstrated significant hypoxia selectivity in vitro in EMT6 cells; however their uptake under normoxic conditions was considerably higher than that of Cu-ATSM [33]. Ex vivo biodistribution results from mice bearing EMT6 tumors showed high accumulation in the liver, suspected to be a consequence of the high lipophilicity or hydrolytic instability of the complexes. We have also studied a glucose conjugate of ⁶⁴Cu-ATSM/A [34]. This hydrophilic complex showed improved in vitro hypoxia selectivity in HeLa cells in comparison to ⁶⁴Cu-ATSM, due to a reduced level of uptake under normoxic conditions. In vivo results from PET imaging of rats bearing P22 carcino-sarcoma showed a similar pattern of biodistribution for both the ⁶⁴Cu-ATSM/A glucose conjugate and ⁶⁴Cu-ATSM. The significant effects of glucose conjugation were to exclude the tracer from the brain and shift its excretion pathway from primarily hepatointestinal to partially renal.

In principle nitroimidazole conjugates of bis(thiosemicarbazonato)copper(II) could show additive or synergistic selectivity for tumor hypoxia compared to their individual components. Interplay between the two differing redox trapping mechanisms might lead to properties such as a faster rate of uptake in hypoxic tissue, sensitivity to a broader range of oxygen concentrations or recognition of hypoxia in more diverse tumor types. In this paper we describe the synthesis of three bis(thiosemicarbazone)–nitroimidazole conjugates and their copper(II) complexes. Two of these are based on the previously described H₂ATSM/A platform. The third employs a novel bifunctional bis(thiosemicarbazone) – H₂ATSM/en – which possesses an ethylamino linker group. Radiolabelling with ⁶⁴Cu was carried out and the results of in vitro hypoxia selectivity and ex vivo biodistribution studies on the radiolabelled complexes are presented.

2. Materials and methods

2.1. General

All chemicals, unless otherwise stated, were purchased from Sigma–Aldrich Chemical Co. (St. Louis, MO, USA). Water was distilled and then deionized (18 MΩ/cm²) by passing through a Milli-Q water filtration system (Millipore Corp., Milford, MA, USA). HPLC characterisation of the ligands and cold copper complexes was performed on a Hamilton-PRP1 reverse phase column, (4.1 mm × 150 mm, particle size 5 μm, SN 10843) with UV detection at λ = 254 nm using one of two gradient elution methods. Method A: 0.9 mL/min (MeCN/H₂O): start 5% MeCN, hold until 1 min, gradient to 95% MeCN at 15 min, hold until 25 min, reverse gradient to 5% MeCN at 30 min, hold to 35 min. Method B: 1.0 mL/min (MeCN/H₂O): start 5% MeCN, gradient to 95% MeCN at 15 min, hold until 20 min, gradient to 5% MeCN at 25 min, then hold until 30 min. NMR spectroscopy was performed on a Varian Mercury VX300 spectrometer (300 MHz), and spectra were referenced to the residual solvent peak. The splitting of proton resonances in the reported spectra is defined as s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, t = doublet of triplets, td = triplet of doublets, br = broad. Mass spectrometry (MS) was performed using a Bruker Micromass LCT time-of-flight mass spectrometer using electrospray ionisation with positive ion (ES⁺) or negative ion (ES⁻) detection. When possible high resolution mass spectrometry (HRMS) data was obtained and accurate masses are reported to four decimal places using tetraoctylammonium bromide (466.5352 Da) as an internal reference. Copper-64 was produced on a CS-15 biomedical cyclotron at the Washington University School of Medicine using published methods [35]. Radioactivity was counted with a Beckman Gamma 8000 counter containing a NaI crystal (Beckman Instruments, Inc., Irvine, CA, USA). HPLC characterisation of radiochemical compounds was performed on a Vydac Protein and Peptide column (C18 monomeric, 300 Å silica base, 4.6 mm × 100 mm) with UV detection and NaI scintillation crystal detection. A 25 min gradient elution method was employed (MeCN/H₂O, 0.1% TFA): start 0% MeCN hold until 5 min, gradient to 95% MeCN at 10 min, hold until 20 min, reverse gradient to 0% MeCN at 25 min, hold until 30 min. The retention time of the ⁶⁴Cu analogues were compared to those of the previously prepared copper analogues run under the same HPLC conditions. Radiochemical purity was tested radio-thin layer chromatography (radio-TLC) using 100 mm × 25 mm 250 μM reversed-phase hydrocarbon impregnated silica-gel plates with acetonitrile/water (4:1) as the mobile phase. In each case any unreacted copper remained at the origin (R_f: 0–0.1) while the radiolabelled complexes had R_f values of 0.8–0.9. TLC plates were analyzed on a BIOSCAN System 200 imaging scanner (Washington, DC, USA).

2.2. Synthesis

Boc-ethylenediamine [36], diacetyl-bis(N⁴-methylthiosemicarbazonate) (H₂ATSM) [37], Cu-ATSM [24], 3-(4-nitroimidazol-1-yl) propionic acid [38], 3-(2-nitroimidazol-1-yl) propionic acid [39], diacetyl-2-(4-N-methyl-3-thiosemicarbazone) [32], and diacetyl-2-(4-N-methyl-3-thiosemicarbazone)-3-(4-N-amino-3-thiosemicarba-zone) (H₂ATSM/A) [32] were prepared by previously reported procedures.

2.2.1. Methyl-N-(2-^tbutoxycarbonylaminoethyl)dithiocarbamate

Carbon disulphide (1.90 mL, 31.6 mmol) was added dropwise to a stirring solution of boc-ethylenediamine (5.06 g, 31.6 mmol) and triethylamine (4.41 mL, 31.6 mmol) in ethanol (100 mL) whilst maintaining the reaction at 25 °C with a water bath. After stirring for 90 min, iodomethane (1.97 mL, 31.6 mmol) was added and the resulting mixture stirred for a further 1¾ h. The solvent was removed and the residue re-suspended in ethyl acetate. This was washed with 1 M HCl (100 mL), sat. sodium bicarbonate solution (100 mL) and water (100 mL). The organic phase was dried (MgSO₄) and solvent removed under reduced pressure to give an off-white solid (7.10 g, 90%, 28.4 mmol). ¹H NMR (300 MHz, CDCl₃): δ ppm 8.33 (1H, br s, NHC=S), 4.94 (1H, br s, NHBoc), 3.73 (2H, dt, ³J_HH = 5.5, 5.2 Hz, CH₂NHC=S), 3.36 (2H, dt, ³J_HH = 6.1, 5.5 Hz, CH₂NHBoc), 2.53 (3H, s, CH₃S), 1.39 (9H, s, Boc); MS (ES⁺): m/z 273.1 {M + Na}⁺ (100%).

2.2.2. 4-N-(2-^tbutoxycarbonylaminoethyl)-3-thiosemicarbazide

A solution of methyl-N-(2-^tbutoxycarbonylaminoethyl)dithiocarbamate (7.00 g, 27.9 mmol) and hydrazine hydrate (2.20 mL, ca. 38.5 mmol) was dissolved in ethanol (100 mL) and heated under reflux for 150 min. The solvent was removed under reduced pressure and the residue redissolved in chloroform (20 mL). This was passed through a plug of silica, washed with chloroform (20 mL) and the product eluted with methanol (30 mL). The methanol fraction was evaporated to provide an oil which solidified over a few days (6.24 g, 96%, 26.8 mmol). ¹H NMR (300 MHz, CDCl₃): δ ppm 7.76 (1H, br s, NHNH²), 7.58 (1H, br s, CH₂NHC=S), 4.98 (1H, br s, CH₂NHBoc), 3.76 (2H, br s, NHNH₂), 3.69 (2H, dt, ³J_HH = 6.3, 5.4 Hz, CH₂NHC=S), 3.32 (2H, dt, ³J_HH = 6.0, 5.4 Hz, CH₂NHBoc), 1.38 (9H, s, Boc); ¹³C{¹H} NMR (75.5 MHz, CDCl₃): δ ppm 182.6, 156.6, 79.6, 44.4, 40.1, 28.3; HRMS (ES⁺): m/z (calcd) 257.1040 (257.1048) {M + Na}⁺.

2.2.3. H₂ATSM/en-boc

4-N-(2-^tbutoxycarbonylaminoethyl)-3-thiosemicarbazide (0.750 g, 3.21 mmol) was suspended in ethanol (30 mL) and stirred at 50 °C. Diacetyl-2-(4-methyl-3-thiosemicarbazone) (0.555 g, 3.21 mmol) was added in portions over 30 min. After the final addition, 2 drops of conc. HCl was added and the reaction heated under reflux for 3½ h. The cooled solution was poured into water (70 mL) to produce a white precipitate. This was collected by filtration, washed with water (2 × 30 mL) and diethyl ether (1 × 10 mL), and dried in vacuo to leave the product as a white solid (0.851 g, 2.07 mmol, 68%). ¹H NMR (300 MHz, DMSO-d₆) δ ppm 10.31 (1H, s, NHN=C), 10.29 (1H, s, NHN=C), 8.45 (1H, t, ³J_HH = 5.1 Hz, CH₂NHC=S), 8.38 (1H, q, ³J_HH = 4.5 Hz, CH₃NHC=S), 7.03 (1H, t, ³J_HH = 5.1 Hz, NHBoc), 3.59 (2H, td, ³J_HH = 5.4, 5.1 Hz, CH₂NHC=S), 3.17 (2H, td, ³J_HH = 5.4, 5.1 Hz, CH₂NHBoc), 3.02 (3H, d, ³J_HH = 4.55 Hz, CH₃NHC=S), 2.23 (3H, s, CH₃C=N), 2.21 (3H, s, CH₃C=N), 1.36 (9H, s, Boc); ¹³C{¹H} NMR (75.5 MHz, DMSO-d₆): δ ppm 178.5, 178.0, 156.2, 148.3, 147.8, 77.9, 44.5, 39.1, 31.2, 28.2, 11.8, 11.7; HRMS (ES⁺): m/z (calcd) 412.1555 (412.1565) {M + Na}⁺. HPLC (Method B): R_t 12.37 min.

2.2.4. H₂ATSM/en

TFA (2 mL) was added to diacetyl-2-(4-N-methyl-3-thiosemicarbazone)-3-(4-N-^tbutylethylcarbamate-3-thiosemicarbazone) (0.750 g, 1.93 mmol) and the resulting colourless solution stirred for 1½ h at room temperature. The solvent was removed to leave a sticky residue and sat. aqueous NaHCO₃ (10 mL) added slowly to produce a white suspension. After 20 min stirring at room temperature the solid was collected by filtration and washed with water (3 × 20 mL) and diethyl ether (2 × 20 mL). Drying in vacuo gave the product as off-white solid (0.431 g, 1.49 mmol, 77%). ¹H NMR (300 MHz, DMSO-d₆ with added TFA) δ ppm 10.53 (1H, d, ³J_HH = 8.21 Hz, NHN=C), 10.24 (1H, d, ³J_HH = 12.0 Hz, NHN=C), 8.50 (1H, q, ³J_HH = 6.7 Hz, CH₃NHC=S), 8.39 (1H, t, ³J_HH = 4.4 Hz, CH₂NHC=S), 7.87 (3H, br s, NH₃⁺), 3.83 (2H, td, ³J_HH = 6.1, 5.9 Hz, CH₂NHC=S), 3.05 (2H, tq, ³J_HH = 6.3, 6.1 Hz, CH₂NH₃⁺), 3.02 (3H, d, ³J_HH = 6.3 Hz), CH₃NHC=S), 2.22 (6H, br s, 2 × CH₃C=N); ¹³C{¹H} NMR (75.5 MHz, DMSO-d₆): δ ppm 178.4, 177.6, 147.6 (2 carbons), 46.3, 40.3, 31.2, 11.7, 11.5; HRMS (ES⁺): m/z (calcd) 290.1228 (290.1222) {M + H}⁺; HPLC (Method B, run with 0.1% TFA): R_t 6.08 min.

2.2.5. Zn-ATSM/en

Zn(OAc)₂·2H₂O (0.23 g, 1.0 mmol) was added to a solution of H₂ATSM/en (0.25 g, 0.86 mmol) in methanol (10 mL) and the mixture stirred at room temperature. for 1 h. Water (20 mL) and triethylamine (1 mL) were added causing the formation of a yellow precipitate. The solution was filtered, the solid washed with water and diethyl ether and dried in vacuo to give the product as a yellow solid (0.29 g, 0.69 mmol, 80%). The ¹H NMR (300 MHz, DMSO-d₆) gave broad peaks possibly due to protic equilibria or self-association in solution; δ ppm 7.32 (1H, br s), 7.16 (1H, br s), 3.27 (2H, br m) 2.82 (3H, d, ³J_HH = 3.5 Hz) 2.42 (2H, br m), 2.22 (3H, s), 2.14 (3H, s). HRMS (ES⁺): m/z (calcd) 352.0362 (352.0356) {M + H}⁺; Crystals suitable for X-ray diffraction were grown by the diffusion of H₂O into a DMSO solution of the product. The complete crystallographic data can be obtained at http://www.ccdc.cam.ac.uk/(deposition number: CCDC 710238).

2.2.6. N-hydroxysuccinimide propanoate

This procedure was adapted from the procedure of Hirata et al. [40] DCC (8.00 g, 38.9 mmol) was added to a solution of N-hydroxysuccinimide (4.60 g, 40.0 mmol) and propionic acid (3.00 mL, 39.5 mmol) in THF (100 mL) and the mixture stirred at −78 °C for 1 h. After stirring for a further 2 h at room temperature, the solution was filtered, the solvent removed and the residue partitioned between dichloromethane and water. The organic layer was washed with a sat. NaHCO₃ solution (2 × 25 mL), dried with MgSO₄ and the solvent removed to leave N-hydroxysuccinimide propanoate as a viscous oil which solidified to a white solid over a few hours (5.13 g, 30.0 mmol, 77%). ¹H NMR (300 MHz, DMSO-d₆): δ ppm 2.76 (4H, s, CH₂–N(–O)–CH₂), 2.66 (2H, q, ³J_HH = 7.5 Hz, O=C–CH₂CH₃), 1.11 (3H, t, ³J_HH = 7.5 Hz, O=C–CH₂CH₃); ¹³C{¹H} NMR (75.5 MHz, DMSO-d₆): δ ppm 171.1, 170.8, 25.4, 24.3, 9.8; MS (EI+): 171.1 (5%, {M}⁺), 87.1 (60%, {M–(O=C–CH₂–CH₂–C=O)}⁺), 57.0 (100%, {M–(N-hydroxysuccinimide)}⁺).

2.2.7. N-hydroxysuccinimide 3-(4-nitroimidazol-1-yl)propanoate

EDC (520 mg, 2.80 mmol) was added to a mixture of N-hydroxysuccinimide (320 mg, 2.78 mmol) and 3-(4-nitroimidazol-1-yl) propionic acid (500 mg, 2.75 mmol) in THF (25 mL) and the mixture stirred for 1 h at −78 °C. The mixture was stirred for a further by 15 h at room temperature before the solvent was removed and water/methanol (1:1, 10 mL) added to the residue. The resulting solid collected, cold water (2 mL) and diethylether (2 × 20 mL) to leave pure N-hydroxysuccinimide 3-(4-nitro-1H-imidazol-1-yl) propanoate as a white solid (611 mg, 2.16 mmol, 80%). ¹H NMR (300 MHz, DMSO-d₆): δ ppm 8.44 (1H, d, ³J_HH = 1.4, imidazole), 7.87 (1H, d, ³J_HH = 1.3, imidazole), 4.40 (2H, t, ³J_HH = 6.8), 3.40 (2H, t, ³J_HH = 6.8, O=C–CH₂CH₂-imidazole), 2.76 (4H, s, CH₂–N(–O)–CH₂); ¹³C{¹H} NMR (75.5 MHz, DMSO-d₆): δ ppm 170.8, 167.5, 147.7, 138.2, 122.3, 42.8, 32.0, 25.9; HRMS (ES⁺): m/z (calc) 283.0684 (283.0679) {M + H}⁺. Elem. Anal.(%): Found: C 42.5, H 3.6, N 19.8, Calc. (for C₁₀H₁₀N₄O₆): C 42.6, H 3.6, N 19.9. HPLC (Method A): R_t 8.00 min.

2.2.8. H₂L¹

Triethylamine (150 μL, 1.07 mmol) was added to a solution of H₂ATSM/A (260 mg, 1.00 mmol) in DMSO (3 mL), followed by N-hydroxysuccinimide 3-(4-nitroimidazol-1-yl)propanoate (296 mg, 1.05 mmol). The solution was stirred for 6 h, water (20 mL) added, the solution centrifuged and the crude solid collected. The solid was then washed with ethanol (10 mL) and diethylether (3 × 10 mL) to leave H₂L¹ as an off-white solid (358 mg, 0.84 mmol, 84%). ¹H NMR (300 MHz, DMSO-d₆): δ ppm 10.41 (H, s, S=C–NH–NH–C=O), 10.11 (2H, br s, 2 × S=C–NH–N) 9.99 (H, s, S=C–NH–NH–C=O), 8.37 (2H, m, NH–CH₃ and imidazole), 7.83 (H, s, imidazole), 4.35 (2H, t, ³J_HH = 6.7 Hz, O=C–CH₂–CH₂-imidazole), 3.02 (3H, s, CH₃-NH) 2.80 (2H, t, ³J_HH = 6.7 Hz, O=C–CH₂–CH₂– imidazole), 2.24 (3H, s, N=C–CH₃), 2.17 (3H, s, N=C–CH₃); ¹³C{¹H} NMR (75.5 MHz, DMSO-d₆): δ ppm 180.2, 179.0, 168.7, 150.7, 148.1, 147.5, 138.3, 122.5, 44.3, 34.0, 32.2, 12.4, 12.2; MS (ES⁻) m/z: 427.1 {M – H}^– (100%); HPLC (Method A): R_t 8.69 min. Elem. Anal.(%): Found: C 34.7, H 5.0, N 31.2. Calc. (for H₂L¹·H₂O, C₁₃H₂₂N₁₀O₃S₂): C 35.0, H 5.0, N 31.4.

2.2.9. H₂L²

3-(2-Nitroimidazol-1-yl)propionic acid (88 mg, 0.48 mmol) and then BOP (200 mg, 0.47 mmol) were added to a solution of H₂ATSM/A (100 mg, 0.38 mmol) and DIPEA (100 μL, 0.59 mmol) in DMF (3 mL). After stirring for 12 h, water (20 mL) was added, the precipitate collected and washed with ethanol (5 mL) and ether (3 × 20 mL) to leave H₂L² as an off-white solid (131 mg, 0.31 mmol, 81%). ¹H NMR (300 MHz, DMSO-d₆): δ ppm 10.30 (H, s, S=C–NH–NH–C=O), 10.18 (2H, br s, 2 × S=C–NH–N), 9.91 (H, s, S=C–NH–NH–C=O), 8.40 (H, s, NH–CH₃), 7.49 (H, s, imidazole), 7.15 (H, s, imidazole), 4.60 (2H, t, ³J_HH = 6.6 Hz, O=C–CH₂–CH₂– imidazole), 3.02 (3H, s, CH₃-NH), 2.75 (2H, t, ³J_HH = 6.6 Hz, O=C–CH₂–CH₂-imidazole), 2.22 (3H, s, N=C–CH₃), 2.16 (3H, s, N=CH₃); ¹³C{¹H} NMR (75.5 MHz, DMSO-d₆): δ ppm: 180.5, 178.1, 168.6, 150.4, 148.1, 145.5, 128.2, 128.0, 44.3, 35.0, 31.2, 12.2, 11.9; MS (ES⁻) m/z: 427.1 {M – H}⁻ (100%); HPLC (Method A): R_t 8.57 min. Elem. Anal.(%): Found: C 34.6, H 5.0, N 31.3. Calc. (for H₂L²·H₂O, C₁₃H₂₂N₁₀O₃S₂): C 35.0, H 5.0, N 31.4.

2.2.10. H₂L³

As H₂L², except using H₂ATSM/en (110 mg, 0.38 mmol), DIPEA(100 μL, 0.59 mmol), BOP (200 mg, 0.47 mmol), DMF (3 mL) and 3-(2-nitroimidazol-1-yl) propionic acid (90 mg, 0.49 mmol) to give H₂L³ as a white solid (122 mg, 0.27 mmol, 71%). ¹H NMR (300 MHz, DMSO-d₆): δ ppm 10.28 (2H, br s, 2 × S=C–NH–N), 8.40 (2H, br s, CH₃–NH, CH₂–NH–C=S), 8.19 (H, s, NH–C=O), 7.49 (H, s, imidazole), 7.15 (H, s, imidazole), 4.60 (2H, t, ³J_HH = 6.8 Hz, O=C–CH₂–CH₂-imidazole), 3.58 (2H, t, ³J_HH = 6.0 Hz, S=C–NH– CH₂–CH₂–NH–C=O), 3.02 (5H, m, S=C–NH–CH₂–CH₂–NH–C=O, CH₃–NH), 2.75 (2H, t, ³J_HH = 6.8 Hz, O=C–CH₂–CH₂-imidazole), 2.18 (6H, s, 2 × N=C–CH₃); ¹³C{¹H} NMR (75.5 MHz, DMSO-d₆): δ ppm: 179.1, 178.4, 10.3, 148.9, 148.2, 145.5, 128.2, 128.0, 46.5, 44.3, 38.2, 36.2, 31.9 11.8, 11.7; MS (ES⁻) m/z: 455.1 {M – H}⁻ (100%); HPLC (Method A): R_t 9.10 min. Elem. Anal.(%): Found: C 37.9, H 5.4, N; 29.5. Calc. (for H₂L³·H₂O, C₁₅H₂₆N₁₀O₄S₂): C 38.0, H 5.5, N 29.5.

2.2.11. H₂L⁴

As H2L¹, except using triethylamine (150 μL, 1.07 mmol), H₂ATSM/A (260 mg, 1.00 mmol) and N-hydroxysuccinimide propanoate (178 mg, 1.05 mmol) and DMSO (3 mL). H₂L⁴ was isolated as a white solid (250 mg, 0.69 mmol, 69%). ¹H NMR (300 MHz, DMSO-d₆): δ ppm 10.50 (H, s, S=C–NH–NH–C=O, 10.20 (H, s, S=C–NH–NH–C=O), 9.99 (H, br s, S=C–NH–N), 9.93 (H, br s, S=C–NH–N), 8.39 (H, s, NH–CH₃), 3.02 (3H, s, CH₃-NH), 2.20 (8H, m, 2 × N=CH₃ and CH₂–CH₃), 1.07 (3H, t, ³J_HH = 7.5 Hz, CH₂–CH₃). ¹³C{¹H} NMR (75.5 MHz, DMSO-d₆): δ ppm 180.2, 179.0, 168.7, 150.7, 148.1, 32.2, 24.0 12.7, 12.1, 9.8; HRMS (ES⁻) m/z: 361.1011 (361.1014) {M–H}⁻; HPLC (Method A): R_t 8.13 min. Elem. Anal.(%): Found: C 35.8, H 6.4, N 28.8. Calc. (for H₂L⁴·H₂O, C₁₀H₂₁N₇O₂S₂): C 35.8, H 6.3, N 29.2.

2.2.12. Complex 1

CuCl₂·2H₂O (94 mg, 0.55 mmol) in DMF (3 mL) was added dropwise over 5 min to a stirring solution of H₂L¹ (214 mg, 0.5 mmol) in DMF (2 mL). After 20 min, water (30 mL) was added to the solution, the mixture centrifuged if necessary, and the solid residue collected. The solid was washed with water (20 mL), cold ethanol (5 mL), ether (2 × 20 mL) and dried in vacuo to give Compound 1 as a deep red–brown solid (85%). HRMS (ES⁺): m/z (calcd) 490.0387 (490.0379) {M + H}⁺; HPLC (Method A): R_t 9.37 min. Elem. Anal.(%): Found: C 31.6, H 3.7, N 28.4. Calc. (for CuL¹, C₁₃H₁₈CuN₁₀O₃S₂): C 31.9, H 3.7, N 28.6.

2.2.13. Complex 2

As complex 1, except with CuCl₂·2H₂O (47 mg, 0.28 mmol) in DMF (2 mL) and H₂L² (107 mg, 0.25 mmol) in DMF (2 mL). Compound 3 was obtained as a red/brown solid (104 mg, 0.20 mmol, 80%). HRMS (ES⁺): m/z (calcd) 490.0380 (490.0379) {M + H}⁺; HPLC (Method A) R_t: 9.44 min. Elem. Anal.(%): Found: C 31.6, H 3.7, N 28.4. Calc. (for CuL², C₁₃H₁₈CuN₁₀O₃S₂): C 31.9, H 3.7, N; 28.6.

2.2.14. Complex 3

As complex 1, except with CuCl₂·2H₂O (47 mg, 0.28 mmol) in DMF (2 mL) and H₂L³ (114 mg, 0.25 mmol) in 2 mL DMF. Compound 4 was obtained as a red solid (104 mg, 0.20 mmol, 80%). MS (ES⁺): m/z 517.1 {M + H}⁺ (100%); HPLC (Method A) R_t: 10.21 min. Elem. Anal.(%): Found: C 34.6, H 4.4, N 26.8. Calc. (for CuL³, C₁₅H₂₂CuN₁₀O₃S₂): C 34.8, H 4.3, N 27.0.

2.2.15. Complex 4

As complex 1, except with CuCl₂·2H₂O (94 mg, 0.55 mmol) in DMF (3 mL) and H₂L⁴ (180 mg, 0.50 mmol) in DMF (3 mL). Compound 2 was isolated as a deep brown–red solid (285 mg, 0.38 mmol 75%). HRMS (ES⁺): m/z (calcd) 379.0310 (379.0310) {M + H}⁺; HPLC (Method A) R_t: 9.37 min. Elem. Anal.(%): Found: C 31.4, H 4.7, N 26.7. Calc. (for CuL⁴, C₁₀H₁₇CuN₇OS₂): C 31.7, H 4.5, N 27.0.

2.3. Cyclic voltammetry

Cyclic voltammetry experiments were performed on a CH instruments 600A electrochemical analyser with Chi600 software, a platinum working electrode, a platinum wire supporting electrode and a Ag|AgNO₃ reference electrode. Scans were internally referenced to ferrocene (0.53 V vs. SCE). Solutions of the respective complexes were prepared to approximately 1 mM in degassed DMF containing 0.1 M TBA[BF₄] and were run between 0.3 and −1.6 V (vs. SCE) at a scan rate of 100 mV/s.

2.4. ⁶⁴Cu complexation of ligands

All ligands, with the exception of H₂L³ were labelled with ⁶⁴Cu in an identical fashion. The ligand was dissolved in DMSO to give an overall concentration of 0.1 mg/mL. A total of 200 μL (20 μg) of this solution was then added to 200 μL (10 mCi) of ⁶⁴Cu in chelexed NaOAc buffer (0.1 M), which was typically supplied at a specific activity of around 200 mCi/μg. This solution was then stirred for 30 min at 37 °C, after which it was loaded onto a C18 SepPak-Light^® (Waters Corporation, Milford, MA), which had been conditioned with 5 mL of ethanol and 5 mL of water. After loading the sample, 15 mL of water was passed through to remove the DMSO and any unreacted ⁶⁴Cu. The labelled complexes were then eluted in 350 μL of ethanol (following a 150 μL ethanol elution for the void volume). The radiochemical yield for all radiolabelling reactions was 80–85% and were all purified to >95% purity. Complex 3 was radiolabelled in a similar fashion, except the stock ⁶⁴Cu²⁺ solution was prepared in pH 4.5 acetate buffer (0.1 M).

2.5. Partition coefficients

Octanol/water partition coefficients were calculated as log -P = log(counts in octanol/counts in PBS buffer (pH 7.4)) and were determined in triplicate as described previously [41].

2.6. Media stability

To measure the stability of the labelled complexes, 5 μCi of the required ⁶⁴Cu complex was incubated in 1 mL of media at 37 °C, where at 1, 5, 15, 30 and 60 min the proportion of intact complex was determined using radio-TLC.

2.7. Serum protein binding

To measure the stability of the labelled complexes, 5 μCi of the required ⁶⁴Cu complex was incubated in 1 mL of fresh mouse serum at 37 °C. At 1, 5, 15, 30 and 60 min triplicate 50 μL fractions were removed and 200 μL of ethanol added to each to precipitate the proteins present. The sample was then centrifuged until the precipitated proteins formed a pellet. The supernatant was removed from the pellet and the pellet washed with 500 μL ethanol, and the sample centrifuged again. The combined supernatants and the pellet were counted on a gamma counter and the percentage protein-bound ⁶⁴Cu calculated.

2.8. In vitro hypoxia selectivity

The apparatus and procedures used for these in vitro experiments are based on those previously described [23,42]. The apparatus controls the in vitro incubation conditions of temperature, humidity, pH, and oxygen tension.

2.8.1. Anoxia vs. normoxia

Two 10 mL aliquots were taken from a homogenous cell suspension of EMT6 mammary carcinoma cells (50 mL, 5 × 10⁶ cells/mL) and each was equilibrated in a three-necked, glass round-bottomed flask at 37 °C under either anoxic (95% N₂, 5% CO₂) or normoxic (75% N₂, 20% O₂, 5% CO₂) conditions by passing a continuous flow of warmed humidified gas over the cells. The flasks were shaken throughout the experiment. After 30 min, when the oxygen concentration in the vessels had reached equilibrium, 50 μCi of the ⁶⁴Cu-labelled complex was added. At 1, 5, 15, 30 and 60 min, triplicate 200 μL samples of cell suspension were removed, and centrifuged to pellet the cells. The percentage cellular uptake of the radiolabelled compound was calculated as described previously [23,29]. As a control, the compounds were put through the identical methods without cells present to determine the extent to which the compounds adhere to vials due to the lipophilicity of the compounds. No appreciable adherence of any of the compounds was noted.

2.8.2. Oxygen concentration study

Aliquots (10 mL) of a homogenous cell suspension of EMT6 mammary carcinoma cells (5 × 10⁶ cells/mL) were equilibrated as above in three-necked, glass round-bottomed flasks at 37 °C at various oxygen concentrations (0%, 0.1%, 0.5%, 5%, or 20%). The flasks were shaken throughout the experiment. The gases used were 5% CO₂, the desired percentage of O₂, and the balance N₂. After 30 min, when the vessels had reached equilibrium, 50 μCi of the ⁶⁴Cu-labelled complex was added. At 15 min, triplicate samples of 200 μL of cell suspension were removed and centrifuged to pellet the cells. The percentage accumulation of the compound into the cells was then calculated. As a control, this experiment was performed with ⁶⁴Cu-ATSM and ⁶⁴Cu(acetate)₂.

2.9. Ex vivo biodistribution

All animal experiments were conducted in compliance with the Guidelines for the Care and Use of Research Animals established by Washington University's Animal Studies Committee. The biodistribution studies were conducted in female BALB/c mice (20 g, aged 5–6 weeks) (Charles River Laboratories, Wilmington, MA) that had been implanted with 6 × 10⁵ EMT6 cells in 100 μl medium into the right flank. The tumors were allowed to grow for 12-days post implantation, at which time the animals received 10 μCi of the desired complex in 100 μL of a saline/ethanol mix (ethanol < 5%) via lateral tail-vein injection. The animals were euthanized at desired time points (5, 20, 60 min, n = 4 per group) and the organs harvested (blood, lung, liver, spleen, kidney, heart, brain, bone, tumor, and intestines). Once the tissues and organs of interest were removed they were weighed and the radioactivity was measured in a γ-counter. The percent dose per gram (%ID/g) and percent dose per organ (%ID/organ) were then calculated by comparison to known standards.

2.10. Statistical methods

To compare the differences in cell uptake and biodistribution, Student's t test was performed. Differences at the 95% confidence level (P < 0.05) were considered significant.

3. Results and discussion

3.1. Chemistry

The target bis(thiosemicarbazonato)copper(II)–nitroimidazole conjugates 1–3 and their ⁶⁴Cu labelled analogues were prepared directly from their respective proligands H₂L¹–H₂L³. Complex 1 incorporates 4-nitroimidazole, whereas 2 and 3 use 2-nitroimidazole, which, of the two isomers, has been shown to possess superior hypoxia selectivity (Scheme 1). The amide linkage was used (rather than the previously studied imine) to tether the nitroimidazole group to the bifunctional bis(thiosemicarbazone) in order to lessen the likelihood of hydrolysis under biological conditions. A simple propyl conjugate 4 (Scheme 1), was produced (via H₂L⁴) in order to model any structural and electronic effects due to the amide linkage and allow proper assessment of the effect of the nitroimidazole group on the properties of 1 and 2. H₂L¹ and H₂L⁴ were prepared by reaction of H₂ATSM/A with the N-hydroxysuccinimide esters of 3-(4-nitroimidazol-1-yl)propionic acid and propionic acid respectively (Scheme 2). However this method proved to be unsuitable for coupling with the 3-(2-nitroimidazol-1-yl)propionic acid, and gave a number of intractable products. Synthesis of H₂L² was achieved by direct coupling of H₂ATSM/A with 3-(2-nitro-imidazol-1-yl)propionic acid using the BOP reagent. In order to probe the effect of the hydrazine linker on the properties of complexes based on H₂ATSM/A, we devised a new bifunctional bis(thiosemicarbazone) – H₂ATSM/en (Scheme 3) – with an aminoethyl tethering group. Boc-1,2-diaminoethane was converted to the corresponding thiosemicarbazide by conversion to the methyl dithiocarbamate and subsequent hydrazinolysis. The 4-(boc-2-ethylamino)-3-thiosemicarbazide was condensed with diacetyl-2-(4-N-methyl-3-thiosemicarbazone) to produce the H₂ATSM/en-boc. Treatment with neat TFA gave H₂ATSM/en. Reaction of H₂ATSM/en with Zn(OAc)₂ gave the complex Zn-ATSM/en. Single crystals of this complex were grown by diffusion of atmospheric water into a DMSO solution and X-ray structure determination carried out (see supporting information). Synthesis of H₂L³ was achieved by direct coupling of H₂ATSM/en with 3-(2-nitroimidazol-1-yl)propionic acid using BOP.

Scheme 1.

Bis(thiosemicarbazonato)copper(II)–nitroimidazole conjugates in this study.

Scheme 2.

Synthesis of H₂ATSM/A derivatives. (i) N-hydroxysuccinimidyl-3-(4-nitro-1H-imidzol-1-yl)propanoate, Et₃N, DMSO; (ii) 3-(2-nitro-1H-imidzol-1-yl)propionic acid, BOP, DIPEA, DMF and (iii) N-hydroxysuccinimidyl propanoate, Et₃N, DMSO.

Scheme 3.

Synthesis of H₂ATSM/en and H₂L⁴. (i) CS₂, Et₃N, EtOH; (ii) MeI; (iii) H₂N₂, EtOH; (iv) HCl, EtOH; and (v) 3-(2-nitro-1H-imidzol-1-yl)propionic acid, BOP, DIPEA, DMF.

The cold copper compounds were prepared by reaction of the appropriate free ligand with 1.1 equivalents of CuCl₂·2H₂O in DMF at room temperature. The ⁶⁴Cu complexes were prepared by the direct reaction of ⁶⁴Cu(OAc)₂ with the corresponding proligand (H₂L¹–H₂L⁴), and their identity confirmed by HPLC co-injection of the radiolabelled complex with the authentic non-radioactive analogue and comparison of the HLPC retention times (R_t values were within ±5%). A summary of the HPLC data is shown in Table 2. In a number of previous studies ⁶⁴Cu-radiolabelling via transmetallation of the bis(thiosemicarbazonato)Zn(II) complex was used, since it resulted in complexes of higher radiochemical purity than direct labelling. It has recently been demonstrated that this method in combination with novel a solid phase purification technique which selectively extracts the Zn(II) precursor by axial coordination could be used to produce tracers of higher specific activity [43]. Such purification methods were not required in this study since direct labelling of the bis(thiosemicarbazone) ligands resulted in sufficient radiochemical purity (>95% as determined by radio-TLC) and specific activity (specific activities of 200–500 mCi/mg ligand and 40–50 mCi/mg ligand were used for biodistribution studies and cell culture studies respectively).

Table 2.

Partition coefficients (Log P), HPLC R_t values (min) and TLC R_f values.

Complex	Log Pa	Rt	Rt (64Cu)	Rfb
1	1.13 ± 0.02	12.96	13.24	0.84
2	1.11 ± 0.03	13.16	13.25	0.84
3	1.45 ± 0.01	13.25	13.36	0.82
4c	1.33 ± 0.05	13.32	13.46	0.86

^aThe Log P value of ⁶⁴Cu-ATSM under these conditions has previously been determined to be 1.85 ± 0.05.

^bReverse phase TLC (C18) using 80:20 acetonitrile:water.

^cHPLC performed without 0.1% TFA in mobile phase.

3.2. Electrochemistry

Cyclic voltammograms were measured for complexes 1–4 and Cu-ATSM in DMF solution, scanning from 0.3 V to −1.6 V (all potentials measured vs. SCE). All the complexes displayed a quasi-reversible reduction wave centered at approximately −0.60 V (Table 1), which is attributed to the Cu(I/II) redox couple. The most negative redox potentials were observed for Cu-ATSM and 3 (−0.64 V and −0.63 V respectively). The reversibility of the copper(I/II) couple was also measurably higher in these two complexes, with the separation between the cathodic and anodic peaks being between 0.21 and 0.25 V for complexes 1, 2 and 4, whereas the peak separation was 0.12 V for 3 and 0.10 V for Cu-ATSM. In addition to the metal centered reduction, complexes 1–3 each displayed a further, quasi-reversible reduction process at a more negative potential. These occur at approximately −1.3 V (for the 4-nitroimidazole compound, 1) and −1.0 V(for the2-nitroimidazoles 2 and 3) and are attributed to reduction of the nitroimidazole group. The electrochemistry of nitroimidazoles has been described in detail elsewhere [44,45].

Table 1.

Copper (I/II) redox potentials as determined by cyclic voltammetry. Studies were performed in degassed DMF with 0.1 M [TBA][BF₄] at a scan rate of 100 mV/s. Ferrocene was used as an internal reference. All values are in volts relative to SCE.

Complex	E½	Ep	Ep	ΔE
Cu-ATSM	−0.64	−0.69	−0.59	0.10
1	−0.58	−0.68	−0.47	0.21
2	−0.57	−0.69	−0.45	0.24
3	−0.63	−0.69	−0.57	0.12
4	−0.58	−0.70	−0.45	0.25

3.3. Lipophilicity measurements

The partition coefficients (log P) for the compounds are displayed in Table 2. Complexes 1 and 2, which differ only by the position of nitro group of the nitroimidazole, have very similar log P values (1.13 ± 0.02 and 1.11 ± 0.03 respectively), which is unsurprising given the similarities of their structures. Model complex 4 is more lipophilic (log P = 1.33 ± 0.05), as is 3 (log P = 1.45 ± 0.01), which contains the ethylenediamine linker. ⁶⁴Cu-ATSM is significantly more lipophilic than any of the amide conjugates used in this study (log P = 1.85 ± 0.05) [33].

3.4. Serum protein binding and media stability

Compounds 1–4 were incubated for 60 min in fresh mouse serum with samples taken at various time points. All compounds displayed similar results, with approximately 20% of the activity bound to proteins after 1 min. This remained constant at all time-points over the remaining 60 min. These results are comparable to the previously observed for Cu-ATSE in mouse serum, and the previously studied H₂ATSM/A imine and glucose conjugates [33,46]. No appreciable decomposition was detected for any of the radiotracers in serum-free media over the same time period.

3.5. Cell uptake studies

The apparatus used in these in vitro experiments has been used in previous hypoxia selectivity studies [23,46], and is specifically designed to keep variables such as cell concentration, temperature and pH constant. The uptake profiles for compounds 1–4 in EMT6 cells under both anoxic and normoxic conditions over 60 min are displayed in Fig. 1. All the compounds showed hypoxia selectivity, with statistically significant differences in cellular uptake between normoxic and hypoxic conditions at all time points (P ≤ 0.05). Considerable variation in the cellular uptake and degree of hypoxia selectivity was observed between the complexes. The model complex 4 showed 69.0 ± 2.1% cellular uptake under anoxic conditions after 1 min which increased to 85.6 ± 0.6% at 30 min and remained at a similar level thereafter. Uptake under normoxic conditions was significantly lower, but followed a similar profile, with 38.5 ± 1.7% uptake after 1 min, increasing to 64.8 ± 0.8% at 30 min and level thereafter. At 60 min the hypoxia selective index (HSI) value [30] of 4 was 0.51. The level of uptake is slightly lower under both normoxic and anoxic conditions than observed with previously studied imine conjugates using an identical experimental set-up [33]. The imine conjugates are considerably more lipophilic than 4 and it is likely that their higher cellular uptake is due to superior membrane permeability.

Fig. 1.

Graphs showing uptake of radiolabelled complexes 1–4 in EMT6 cells under normoxic (dashed line) and anoxic (solid line) conditions. Errors (one stan. dev.), if not indicated, are within symbols.

The 4-nitroimidazole, 1 showed a markedly different uptake profile, and lower overall uptake compared to both 4 and the previously studied imine conjugates. Uptake for 1 under anoxic conditions was 22.5 ± 0.8% after 1 min, and increased to 45.6 ± 0.8% at 60 min. Under normoxic conditions uptake was 16.6 ± 0.7% after 1 min and remained at a similar level over the time course of the experiment. The HSI of 1 at 60 min was 0.64, somewhat higher than for 4. Thus the addition of the bio-reducible nitroimidazole group to the bis(thiosemicarbazonato)Cu(II) complex appears to lead to an increase in hypoxia selectivity at the expense of a reduction in cell permeability. One factor causing the latter effect could be the lower lipophilicity of 1 (log P = 1.13) compared to 4 (log P = 1.33).

The 2-nitroimidazole, 2 showed a similar uptake profile to its 4-nitroimidazole analogue 1. The observed uptakes of 64.5 ± 1.3% and 26.2 ± 1.7% at 60 min under anoxic and normoxic conditions respectively, correspond to an HSI of 0.71. The enhanced selectivity of 2 compared to 1 is due to greater accumulation in anoxic cells. It therefore appears that this small change in substitution generates increased hypoxia selectivity. Lipophilicity is unlikely to be a factor since the log P of 2 (1.11) is similar to that of 1. However previous work has shown that the reduction potential of the nitro group in 2-nitroimidazoles is less negative than that of 4-nitroimidazoles. This is thought to lead to the superior hypoxia selectivity observed with 2-nitroimidazole derivatives compared to their 4-nitroimidazole analogues [47]. The cyclic voltammetry results confirm that the reduction potentials of the 2-nitroimidazole groups in 2 and 3 are less negative (by 300 mV) than that of the 4-nitroimidazole in 1. Taking this into account the increased uptake of 2 in anoxic cells is a strong indication that the nitroimidazole is playing an active role in the redox trapping of these complexes.

Complex 3, which is based on the ligand H₂ATSM/en instead of H₂ATSM/A, showed the strongest hypoxia selectivity with an HSI value of 0.84 (at 60 min). After 1 min 15.8 ± 0.5% and 4.08 ± 0.5% had accumulated in cells under anoxic and normoxic conditions respectively, which increased to 63.6 ± 0.4% and 20.3 ± 1.9% respectively after 60 min. The increase in HSI compared to 2 results from reduced accumulation normoxic cells (anoxic uptake for both tracers was comparable). This behavior could be a result of the ethylenediamine linker group leading to a more negative Cu(I/II) redox potential in 3 compared to 1, 2 and 4. Previous uptake studies on bis(thiosemicarbazonato)Cu(II) complexes in EMT6 cells by Dearling et al. have shown a correlation between hypoxia selectivity, the Cu(I/II) redox potential and the reversibility of this reduction process, with more negative redox potentials showing the best hypoxia selectivity [30,42,48]. Close inspection of the reported data reveals that this trend is also due to the complexes with more negative redox potentials generally showing lower accumulation in normoxic cells (as opposed to greater accumulation in hypoxic cells).

As complex 3 demonstrated the most promising initial results, its cellular uptake was tested at various oxygen concentrations (Table 3). At oxygen concentrations less than 0.5% no statistically significant variation in uptake (15 min incubation) between the three concentrations measured was observed (67.2 ± 1.0%, 66.1 ± 0.9%, 64.3 ± 1.5% uptake at 0%, 0.1% and 0.5% O₂ respectively). However, at an oxygen concentration of 5%, uptake was dramatically lower (43.3 ± 1.1%). In this experiment, under normoxic conditions (20% O₂) cellular uptake of 3 was decreased further (24.1 ± 2.1%). ⁶⁴Cu-ATSM showed higher uptake than 3 under anoxic and hypoxic conditions (92.5 ± 0.2% and 36.0 ± 0.8% respectively at 0% and 20% O₂ respectively) but displayed a lower percentage uptake than compound 3 at 5% O₂ (39.5 ± 0.9%). The results for ⁶⁴Cu-ATSM correspond to those obtained previously, as did the results for ⁶⁴Cu(acetate)₂ which showed minimal uptake, irrespective of oxygen concentration, and was not selective for hypoxic cells [33].

Table 3.

Percentage cellular uptake in EMT6 cells for 3 and ⁶⁴Cu-ATSM at various oxygen concentrations.

[O2] ppm	%Cellular uptake
	64Cu-ATSM	3
10	92.6 ± 0.2	67.2 ± 1.0
1000	81.9 ± 1.3	66.1 ± 0.9
5000	73.5 ± 0.4	64.3 ±1.5
50000	39.5 ± 0.9	43.3 ± 1.1
200000	36.0 ± 0.8	24.1 ±2.1

For comparison, uptake profiles of 3 and ⁶⁴Cu-ATSM were plotted using the ratio between the uptake of the compound at each respective oxygen concentration and the uptake of the compound under normoxic conditions against log₁₀ of the oxygen concentration (Fig. 2). This analysis is preferable to using percentage uptake or HSI values since image contrast in PET ultimately depends upon the ratio of uptake between cells with a particular condition (in this case hypoxia) and those without (in this case normoxia). The plot clearly demonstrates that the ratio of normoxic to hypoxic uptake was higher for 3 than for ⁶⁴Cu-ATSM at all oxygen concentrations (excluding normoxia). Moreover, the uptake ratio for 3 did not change significantly from 0% to 0.5% O₂, whereas the ratio for ⁶⁴Cu-ATSM decreased considerably. From 0.5% to 5% O₂ the uptake ratios of both tracers decrease at a similar rate, however the high selectivity of 3 at 5% O₂ is particularly interesting. A major limitation of ⁶⁴Cu-ATSM is its poor delineation of mild hypoxia [46], and this is illustrated by these results which show only a 10% increase in uptake at 5% O₂ in comparison to normoxia. In contrast 3, shows an 80% increase in uptake between the same two oxygen concentrations. These preliminary results suggest that 3 could have potential for imaging hypoxia in diseases such as cardiac ischemia which are generally associated with less severe hypoxia than the radiobiologic effect relevant to tumors [21,46].

Fig. 2.

Graph showing ratio of hypoxic cellular uptake to normoxic cellular uptake for compound 3 and ⁶⁴Cu-ATSM in EMT6 cells at varying oxygen concentrations. Errors (one stan. dev.), if not indicated, are within symbols.

3.6. Ex vivo biodistribution

Biodistribution studies were performed in BALB/c mice bearing EMT6 tumors using tracers 2 and 3 and the data obtained are shown in Table 4 (%injected dose/organ data is also provided in the supporting information). This in vivo model was chosen to allow qualitative comparison with the results of previous studies on ⁶⁴Cu-ATSM, its analogue ⁶⁴Cu-ATSE, the bis(selenosemicarbazonato)⁶⁴Cu complex ⁶⁴Cu-ASSM, and ⁶⁴Cu-ATSM/A imine conjugates (these studies vary only in the site and age of the implanted tumors) [23,33,46]. Both 2 and 3 showed rapid clearance from the blood (4.03 ± 0.48%ID/g and 3.00 ± 0.95%ID/g respectively at 5 min, decreasing to 2.74 ± 0.17%ID/g and 2.18 ± 0.68%ID/g respectively after 60 min). This is similar to all the previously reported tracers, except ⁶⁴Cu-ASSM which showed higher residual activity in the blood after 60 min. The general pattern of uptake in the other organs was also similar, with the liver, kidneys, intestines and lungs showing the highest levels of retention. Initial liver uptake of 2 was high (42.3 ± 7.2%ID/g at 5 min), and did not change significantly thereafter. Liver uptake of 3 was significantly lower than this (26.8 ± 3.7%ID/g at 5 min) and decreased over time (to 21.6 ± 0.5%ID/g at 60 min).

Table 4.

Biodistribution (%ID/g ± SD, n = 4) at 5, 20 and 60 min of 2 (top) and 3 (bottom) in BALB/c Mice Bearing the EMT6 Mammary Tumor.

Tissue	5 min	20 min	60 min
Blood	4.03 ± 0.48	3.17 ± 0.20	2.74 ± 0.17
Lung	12.6 ±1.4	11.9 ± 1.4	11.7 ± 0.8
Liver	42.3 ± 7.2	37.9 ± 3.5	37.4 ± 5.1
Spleen	6.33 ± 1.06	5.91 ± 0.51	5.76 ± 0.88
Kidney	25.1 ± 3.9	21.1 ± 0.9	17.2 ± 1.5
Muscle	2.36 ± 0.50	1.64 ± 0.10	1.39 ± 0.20
Heart	5.55 ± 0.51	5.04 ± 0.26	4.86 ± 0.25
Brain	0.66 ± 0.09	0.71 ± 0.09	0.72 ± 0.03
Bone	1.56 ± 0.22	1.64 ± 0.18	1.94 ± 0.24
Tumor	2.17 ± 0.11	2.85 ± 0.69	2.97 ± 0.76
Intestines	8.82 ± 1.88	11.5 ± 1.1	15.3 ± 1.4
Blood	3.00 ± 0.95	2.19 ± 0.22	2.18 ± 0.68
Lung	6.10 ±0.77	6.07 ± 0.34	6.98 ± 0.61
Liver	26.8 ± 3.7	25.9 ± 3.4	21.6 ± 0.5
Spleen	2.91 ± 0.27	2.77 ± 0.35	3.33 ± 0.33
Kidney	21.7 ± 0.4	17.0 ± 1.4	13.7 ± 0.8
Muscle	1.51 ± 0.48	1.21 ± 0.16	1.37 ± 0.26
Heart	3.03 ± 0.25	2.80 ± 0.19	3.14 ± 0.12
Brain	0.28 ± 0.06	0.26 ± 0.03	0.29 ± 0.03
Bone	1.15 ± 0.10	1.12 ± 0.18	1.37 ± 0.21
Tumor	1.68 ± 0.70	1.74 ± 0.16	2.56 ± 0.57
Intestines	18.1 ± 4.3	23.9 ± 1.6	22.7 ± 1.6

Both 2 and 3 differ from ⁶⁴Cu-ATSM and ⁶⁴Cu-ATSE, which show significantly increased accumulation in the liver over time. Hepatic clearance was observed for the ⁶⁴Cu-ATSM/A imine conjugates, but these appeared to give higher liver uptake than all the other tracers. The clearance of 3 from the liver is reflected in its increasing uptake in the intestines (rising from 18.1 ± 4.3%ID/g at 5 min to 22.7 ± 1.6%ID/g at 60 min). Initial intestinal uptake of 2 was considerably lower (8.82 ± 1.88%ID/g at 5 min) but also increased with time (to 15.3 ± 1.4%ID/g at 60 min). It was proposed that the high liver uptake of ⁶⁴Cu-ATSM/A imine conjugates could be due to their high lipophilicity and/or hydrolysis of the imine linkage. In the present study both these factors can be ruled out since hydrolysis of amides is unlikely, and 3, which utilizes the alkylamino linker, shows lower liver uptake than 2 despite its higher log P value.

Accumulation of activity in the liver could occur by a number of possible mechanisms. The liver is the target organ for unchelated copper where it is trapped by metallothioneins [49]. Loss of copper from thermodynamically stable chelates via reduction and transchelation to liver enzymes has also been documented [48,50]. Parts of the liver are mildly hypoxic and high uptake has also been observed in the 2-nitroimidazole class of hypoxia markers [51,52]. The Cu(II/I) reduction process plays a direct role in the latter two mechanisms and it is therefore probable that the less negative potential and lower reversibility of this process in 2 compared to 3 results in its higher liver uptake.

The initial uptake values of 2 and 3 in the kidneys were not significantly different to each other (25.1 ± 3.9%ID/g and 21.7 ± 0.4%ID/g respectively at 5 min), but renal clearance was marginally faster for 3 (13.7 ± 0.8%ID/g at 60 min compared to 17.2 ± 1.5%ID/g for 2). Renal clearance was also observed with ⁶⁴Cu-ATSM, ⁶⁴Cu-ATSE and ⁶⁴Cu-ASSM, whereas kidney uptake of the ⁶⁴Cu-ATSM/A imine conjugates, which at 5 min was apparently lower than all the other tracers, did not change significantly over time. The complexes also differ from each other in their lung uptake, with significantly lower%ID/g values observed for 3 in comparison to 2 (6.98 ± 0.61%ID/g and 11.7 ± 0.8%ID/g at 60 min respectively). Neither complex showed any appreciable change in lung uptake over the 60 min period. For a potential PET imaging agent it is obviously beneficial to reduce accumulation in non-target organs (i.e. the liver, intestines, kidneys and lungs), and from this point of view complex 3 is superior to complex 2. In this case the small structural change from the H₂ATSM/A to the H₂ATSM/en backbone results in improved biodistribution characteristics. Again it is possible that the higher non-target uptake of 2 compared to 3 is due to its lesser redox stability.

In the tumor a similar level of uptake was observed for both tracers (2.97 ± 0.76%ID/g and 2.56 ± 0.57%ID/g at 60 min for 2 and 3 respectively). This appears to be lower than for ⁶⁴Cu-ATSM,⁶⁴Cu-ATSE, and the ⁶⁴Cu-ATSM/A imine conjugates, and may be due to the lower lipophilicity of the nitroimidazole conjugates reducing their tissue penetration. However, it could also be a result of variations in tumor hypoxia between the different studies. In this study we were not equipped to measure the level of tumor hypoxia and therefore it is not possible to make any assessment of in vivo hypoxia selectivity.

4. Conclusions

Three bis(thiosemicarbazonato)copper(II)–nitroimidazole conjugates were successfully synthesized in their cold and ⁶⁴Cu radio-labelled forms. These potential combination agents for hypoxia detection showed improved hypoxia selectivity in EMT6 cellular uptake experiments when compared to a model propyl conjugate. In particular the observed difference in anoxic uptake between the 4- and 2-nitroimidazole isomers, 1 and 2, indicates that the nitroimidazole group actively contributes to hypoxia trapping. The detailed oxygen dependent cellular uptake study carried out on 3 demonstrates that this complex is more sensitive to mild hypoxia than Cu-ATSM.

In vivo the 2-nitroimidazole conjugates 2 and 3 showed significantly different biodistributions in comparison to each other and in comparison to previously studied Cu-ATSM derivatives. Moreover uptake of the Cu-ATSM/en conjugate 3 in non-target organs was considerably lower than for the derivatives based on Cu-ATSM/A.

This study provides further evidence that a degree of control over the hypoxia selectivity and biodistribution characteristics of bis(thiosemicarbazonato)copper(II) complexes can be achieved by functionalisation at the exocyclic nitrogen and that the nature of both the linker and the conjugate group used have a significant impact on these properties. Future work will focus in detail on the in vivo hypoxia selectivity of 3 and investigate new bioconjugates based on Cu-ATSM/en.

Abbreviations

Boc

tert-butoxycarbonyl

Cu-ATSE

diacetyl-bis(N⁴-ethylthiosemicarbazonato)copper(II)

DCC

N,N′-dicyclohexylcarbodiimide

DMF

dimethylformamide

EDC

N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide

PBS

phosphate buffered saline

SCE

saturated calomel electrode

SPECT

single photon emission computed tomography

TBA

tetrabutylammonium

TFA

trifluoroacetic acid

THF

tetrahydrofuran