Magnefic Resonance hoging,Vol. 10. pp. 44-455. Printed in the USA. AU rights reserved.

1992

0730-725x/92 $5.00 + .Ou Copyright 0 1992 Pergamon Press Ltd.

0 Technical Note EVALUATION OF NONIONIC NITROXYL LIPIDS AS POTENTIAL ORGAN-SPECIFIC CONTRAST AGENTS FOR MAGNETIC RESONANCE IMAGING BERNARD

GALLEZ,* ROGER DEMEURE,~ RENE DEBUYST,~ DOMINIQUE LEONARD,* FERNAND DEJEHET,$ AND PIERRE DUMONT* *Department of Pharmaceutical Sciences, Laboratory of Medicinal Chemistry, Catholic University of Louvain, B-1200 Brussels, YDepartment of Radiology, Cliniques Universitaires Saint-Luc, B-1200 Brussels, and $Laboratory of Inorganic, Analytical, and Nuclear Chemistry, Catholic University of Louvain, B-1348 Louvain-la-Neuve, Belgium

Considering their intrinsic properties of accumulstion in the hepatic tissue, we have synthesized nitroxyl-containing lipids as potential organ-specific contrast agents for magnetic resonance imaging (MRI). Their resistance to reduction by ascorbate and in liver homogenates, and their relaxivity in different media were investigated and compared to these of free carboxy-Proxyl (3-carboxy-2,2,5,5-tetramethylpyrrolidine-l-oxyl) and Tempamine (4-amino2,2,6,6-tetramethylpiperidine-1-oxyl). With respect to the reduction rates by ascorbate, the lipid derivatives show the same well-known order of reactivity as carboxy-Proxyl and Tempamine, the five-membered nitroxyls being more stable than the six-membered compounds. However the binding of the piperidinoxyl compounds to the fatty acids confers to these lipid dertvatives a markedly increased stability. Similarly, in liver homogenates, the nitroxyl lipids remained unchanged more than 20 min, contrarily to carboxy-Proxyl and Tempamine. The measurements of spin-lattice relaxation time (T,) in biological media have demonstrated a higher relaxivity of nitroxyl lipids, which can be related to their interaction witb proteins. Tested in vivo, one of the synthesized compounds (0.75 mmol/kg) produced an enhancement of 44 f 12% of the hepatic signal 5 min after intraportal injection in Z’,weighted images. The potential applicability of the other nitroxyl lipids as contrast agents for MRI was limited in the in vivo studies by an unexpected toxicity. Work is currently in progress to improve the therapeutic index of the present class of nitroxyl lipids. Keywords: Nitroxyl; Lipid; Magnetic resonance imaging; Contrast agent; Stability; Relaxivity.

the approach of organ-specific contrast agents5 The development of vectorized nitroxyl molecules is also of great interest with regard to the method of imaging free radicals in aqueous solutions by proton-electron double-resonance (PEDRI).6,7 Finally, nitroxyl compounds have received considerable attention in consideration of their oxygen-dependent metabolization rate; this last property should confer them a choice place as diagnostic agent of hypoxic or anoxic tissues.‘-” The rational design of potential organ-specific nitroxyl compounds is based on the following issues: high organ specificity, multi-anchoring, l2 high nitroxyl content versus vector size ratio, absente of pharmacological activity, nonionic character, 13,14and possibility of binding with biological proteins. l5 Considering the well-known metabolic fate of lipids

Paramagnetic compounds produce enhancement of contrast in magnetic resonance imaging (MRI) mainly by their effects upon spin-lattice relaxation time (Ti). Gadolinium complexes (Gd-DTPA, Gd-DOTA) are routinely used as contrast agents of the extracellular fluid space thus indicating disorders of the blood brain barrier,’ infarctions, tumors,2 inflammatory diseases, and so on. Research in progress aims at the development of new products more specific of a given organ, tissue or pathology. In this perspective, recent developments led to potential hepato-biliary contrast agents. These metallic complexes are quickly eliminated via the biliary3,4 pathways, this elimination ensuring a relatively good tolerante. By binding nitroxyl spin labels to appropriate vector molecules, it is also possible to achieve

RECENED

9/19/91;

ACCEPTED

Medicinal Chemistry, School of Pharmacy UCL, Avenue

12/23/91.

Mounier 73.40, B-1200 Brussels, Belgium.

Address requests for reprints to B. Gallez, Laboratory of 445

446

Magnetic Resonance Imaging 0 Volume 10, Number 3, 1992

and their rapid accumulation in hepaticr6*” tissues, we have synthesized four nitroxyl lipid derivatives as potential organ-specific contrast agents for MRI and PEDRI. We report here the synthesis of these compounds. Stability and good relaxivity are two prerequisites in this class of products. Therefore the reactivity towards ascorbic acid, a reducing agent usually used in similar studies’8-2o as wel1 as their stability in liver homogenate were investigated; their contrast ability was studied by evaluating their influence on the Ti of various media. In both cases, the behavior of these nitroxyl lipids was compared to that of carboxyProxy1 (PCA) and Tempamine. These potential contrast agents were evaluated in vivo by imaging the livers of smal1 animals before and after parenteral administration. MATERIALS AND METHODS Chemistry Analyticalprocedures. IR spectra were taken on a Perkin-Elmer spectrophotometer, model 457. Mass spectra were recorded on a Kratos MS-80 mass spectrometer (FAB procedure). In most cases, [M+ (+l) +

number of nitroxyls] values are reported; other fragments, identified according to the literature,2’-23 are presented. Microanalyses were performed by Analytische Laboratorien H. Malissa und G. Reuter, GmbH, Gummersbach (Germany). EPR spectra (solution in chloroform) were recorded on a Bruker ER 200 tt EPR spectrometer. TLC analyses were performed on Silicagel60 F254(Merck). HPLC analyses were carried out with a Perkin-Elmer Series 10 Liquid Chromatograph; UV detection was accomplished by means of a Perkin-Elmer LC-85 B variable-wavelength spectrophotometer at 230 nm. A Nucleosil@ 5 CN column (Macherey Nagel) was used. The eluent was propan2-01 (Lab-Scan)/n-hexane (UCB) (1:9, v/v). A total of 20 ~1 of nitroxyl compound dissolved in the mobile phase (-2 mg/ml) was injected. Melting points were determined on a Thomas-Hoover capillary melting point apparatus and were uncorrected. The compounds have IR spectra, MS spectra and micro-analysis coherent with their structures. Syntheses.

Fig. 1.

CH,QWCY),,-NO, I cHQco(CY),,-w CH*OC(I(CY),,-m

l

Hz.NiR=y

CH,COOH

l

Hz . Ni Rmcy

lHF.kOH

b. ECC t

CH,-C-CO-(CY),,-NH-CO

Fig. 1. Synthesis schemes.

The synthesis schemes are shown in

Nonionic nitroxyl lipids as contrast agents 0 B. GALLEZ ET AL.

Synthesis of 1,2,3_propanetriol tris-[12-(2,2,5,5-tetramethyl-l-pyrrolidinoxyl-3-carbonylamino)dodecanoate] (compound A). A. 12-Nitrododecanoic acid (2.94 g, 12 mmol, Aldrich), 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (2.29 g, 12 mmol, = EDC, Aldrich) and 4-dimethylaminopyridine (147 mg, 1.2 mmol, Aldrich) were dissolved in 250 ml of dichloromethane (UCB) at 0°C. A solution of anhydrous glycerol (276 mg, 3 mmol, Fluka) dissolved in 200 ml of freshly distilled tetrahydrofuran was slowly added during one hour, and the reaction mixture was stirred 3 hr in an ice bath. The solvent was removed on a rotating evaporator at 25°C. The residue was redissolved in dichloromethane and washed twice by hydrochloric acid (15 ml, 1 M), sodium carbonate (15 ml, 5%), and a saturated solution of sodium chloride (10 ml). The solution was dried on magnesium sulfate, filtered, evaporated under reduced pressure, and the residue was purified by column chromatography on Silicagel (eluent : n-hexane/ethyl-acetate, 8:2, v/v). The purity control of the product (colorless liquid) was confirmed by TLC (n-hexane/ethyl-acetate 8:2, Rf = 0.4). The yield was 21%. IR (KBr) : v = 2930, 2860 cm-’ (alkyl), 1740 cm-’ (carbonyl ester), 1550 cm-’ (C-N02). B. 1,2,3_Propanetriol tris-( 1Znitrododecanoate) (500 mg) was dissolved in 50 ml acetic acid (Aldrich) containing 1 g Raney-Nickel catalyst (Aldrich). Hydrogenation was performed in a Parr-flask under hydrogen pressure (50 psi) during 6 hr. The solution was filtered on celite to eliminate the catalyst, and the solvent was removed on a rotating evaporator at 30°C. The oily green residue was dissolved in chloroform, washed with sodium carbonate, dried on magnesium sulfate, and evaporated under reduced pressure. The product, a white powder, was pure in TLC (ethanol/ammonia 28%, 8:2, Rf = 0.7, revelation by ninhydrin). The melting point (mp) was 64-66°C. The yield was 60%. IR (KBr) : v = 3380 cm-’ (N-H amine), 2900,283O cm-’ (alkyl), 1740 cm-’ (carbonyl ester), 1620 cm-’ (amine). C. A solution of 1,2,3_propanetriol tris-(lZaminododecanoate) (273 mg, 0.4 mmol) in dichloromethane was slowly added to a solution of carboxy-Proxyl (372 mg, 2.0 mmol, = compound E, Aldrich), l(3-dimethylaminopropyl)-3-ethylcarbodiimide (383 mg, 2.0 mmol, Aldrich) and 4-dimethylaminopyridine (50 mg, 0.4 mmol, Aldrich) in dichloromethane (200 ml) at 0°C. The mixture was stirred 3 hr at O”C, and 24 hr at room temperature. The solution was washed twice with hydrochloric

447

acid (15 ml, 1 M), sodium carbonate (15 ml, 5%), dried on magnesium sulfate, and evaporated under reduced pressure. The product recrystallized from a mixture of diethyletherlhexane (l:l, v/v) was pure in HPLC. The yield was 59%; yellow-orangey powder, mp = 43-45’C. IR (KBr): v = 3340 cm-’ (N-H amide), 2900, 2830 cm-’ (alkyl), 1740 cm-’ (carbonyl ester), 1650 cm-’ (carbonyl amide). MS: [M++4], [M+-3951, [M+-3961, [M+-4071, [M+-4081, [M+-4221, [M+-4231, [M+-7891, [M+--7901, [M+-7911, [M+-8031, [M+-8041, [M+-8051. EPR: g = 2.0059; AN = 14.8 gauss (3 lines). Microanalysis: calculated -C : 66.685, H: 10.091, N:7.071; found-C:66.42, H: 10.00, N: 6.92. Synthesis of methyl 12-(2,2,5,5-tetramethyl-lpyrrolidinoxyl-3-carbonylamino)dodecanoate (compound B). A. A solution of diazomethane in diethyl ether, prepared by means of a Diazald@ kit (Aldrich) by the reaction of potassium hydroxyde with p-tolylsulfonylmethylnitrosamide,24 was slowly added at 0°C to a solution of 12-nitrododecanoic acid (20.14 g, 82 mmol) in ethanol (1.6 L) until the acidic solution turned yellow, thus indicating stoichiometric amounts of the reagents. The reaction mixture was then stirred overnight at room temperature. After removal of the solvent on a rotating evaporator, the residue was dissolved in a mixture of hexane/ethyl-acetate (8:2, v/v), and filtered on Silicagel 60 (Merck); the solvent was evaporated under reduced pressure to give 20.38 g of methyl 12-nitrododecanoate, a colorless liquid, pure in TLC (n-hexane/ethyl-acetate, 8:2, Rf = 0.5-0.6). The yield was 96%. IR (KBr) : v = 2940,286O cm-’ (alkyl), 1740 cm-’ (carbonyl ester), 1550 cm-’ (C-N02). B. A mixture of methyl 1Znitrododecanoate (20.3 g, 72 mmol) with 10 g of Raney-Nickel catalyst in freshly distilled tetrahydrofuran and methanol (9: 1) was stirred 24 hr in a Parr-flask under hydrogen (50 psi). The mixture was filtered on celite to eliminate the catalyst, and the solvent was removed on a rotating evaporator to give 16.8 g of methyl 12-aminododecanoate pure in TLC (methanol/ammonia 28%, 8:2). The yield was 93%; white powder, mp = 84-86°C. IR (KBr): v = 3390 cm-’ (N-H), 2930, 2860 cm-’ (alkyl), 1740 cm-’ (carbonyl ester), 1620 cm-’ (amine). C. In 500 ml dichloromethane at 0°C were dissolved 12-aminododecanoate methyl ester (2.63 g, 11.5 mmol), carboxy-Proxyl (3 g, 16.3 mmol), 1-(3-

448

Magnetic Resonance Imaging 0 Volume 10, Number 3, 1992

dimethylaminopropyl)-3-ethylcarbodiimide (3.14 g, 16.3 mmol) and 4-dimethylaminopyridine (1 g, 8 mmol). The reaction mixture was stirred overnight, washed with 3 x 30 ml sodium carbonate 5% and 3 x 30 ml hydrochloric acid 0.5 M. The solvent was dried on magnesium sulfate, before removing the solvent on a rotating evaporator. The yellow residue was purified twice by column chromatography on Silicagel60 (eluents: dichloromethane/ acetone 1: 1 and dichloromethane/hexane 8:2). The yellow powder obtained was pure in HPLC. The yield was 65%; mp = 46°C. IR (KBr): u = 3360 cm-’ (N-H), 2980,2940,2860 cm-’ (alkyl), 1740 cm-’ (carbonyl ester), 1650 cm-’ (carbonyl amide). EPR: g = 2.0059; AN = 14.3 gauss (3 lines). MS: [M++2], [M++l], [M+], [M+-11, [M+-141, [M+-151, [M+-291, [M+-301, [M+-3 11, [M+-2851, [M+-3041. Microanalysis: calculated - C : 66.46, H : 10.40, N : 7.05; found C:66.31, H: 10.29, N:7.07.

mixture was stirred 2 hr at 25°C. After removing the tetrahydrofuran under reduced pressure, the residue was dissolved in dichloromethane (100 ml) and washed three times with hydrochloric acid (30 ml 0.5 M). The solution was dried on magnesium sulfate, and evaporated on a rotating evaporator. The compound was finally purified twice by column chromatography on silicagel (eluents: dichloromethane/acetone 1: 1, n-hexane/ethyl-acetate 8:2). The product (red sticky paste) was pure in HPLC. The yield was 67%; mp = 114°C. IR (KBr) : v = 3400-3300 cm-’ (N-H), 2980,2930,2860 cm-’ (alkyl), 1640 cm-’ (carbonyl amide). EPR: g = 2.0063; AN = 15.8 gauss (3 lines). MS: [M++3], [M++2], [M++l], VU+], [M+-14], [M+-15], [M+-161, [1M+-291, [M+-301, [M+-311, [M+-561, [M+-57].

Synthesis of N-(2,2,6,6-tetramethyl-1-oxyl-4-piperidiny dodecaneamide (compound C). To a solution of 4-aminotempo (1420 mg, 8.3 mmol, = compound F, Aldrich) and triethylamine (650 mg, 6.4 mmol, freshly distilled on potassium hydroxide) in tetrahydrofuran (200 ml, freshly distilled on LiAlHJ was slowly added a solution of lauroyl chloride (1.39 g, 6.4 mmol, Aldrich) in dry tetrahydrofuran (100 ml). The reaction mixture was stirred 3 hours in an ice bath, and the solvent was evaporated under reduced pressure at 25°C. The residue was redissolved in 100 ml chloroform and extracted four times with 30 ml hydrochloric acid 0.5 M; the solution was dried on magnesium sulfate, and then evaporated on a rotating evaporator. The residue was purified by column chromatography on Silicagel 60 (eluent: dichloromethane/acetone, 1: 1) and recrystallized from hexane. The red powder obtained was pure in HPLC. The yield was 79%; mp = 60°C. IR (KBr): u = 3300 cm-’ (N-H), 2930, 2860 cm-’ (alkyl), 1640 cm-’ (carbonyl amide). EPR:g = 2.0063; AN = 15.5 gauss (3 lines). MS:[M++2], [M++l], [M+], [M+-11, [M+-141, [M+-15], [M’-291, [M+-301, [M+-311. Microanalysis: calculated-C:71.33, H: 11.69, N:7.92; found-C:71.28, H: 11.52, N:8.07.

T, Measuremen ts Nitroxyl compounds were dissolved in dichloromethane with Tween 80@. The solutions were evaporated under reduced pressure and the residues were then homogenized in aqueous medium to three concentrations 1.O, 2.0, and 4.0 mM (concentration of the surface-active agent : -20 mg/ml). The aqueous media considered were water (RQ, Millipore), solution of bovine albumin in water (40 g/liter, Sigma A-4503), human plasma, and liver homogenate (preparation described below). The TI relaxation time determinations were performed at 26 + 1°C by using an inversionrecovery pulse sequence (180”-k-90”) on a CXP-NMR spectrometer Bruker operating at 90 MHz. 90” and 180” RF pulses were optimized before the measurements by fine adjustment of pulse width. Relaxation curves were determined from 15 experiments, the pulse separation (recovery time) k varying in a range from 10 msec to 4.8 sec, and the repetition time TR being 5 sec in such a way that it is always greater then 5 x TI. For each experiment or k value, the height of the water peak was measured after Fourier transform of the free induction decay. From these data, T, relaxation time was obtained by fitting equation S = S,, [ 1-a x exp(-WTi)] to the experimental data with S, a, and TI as parameters. The fit was performed by employing a nonlinear chi-square minimization algorithm named Levenberg-Marquardt method.25 The relaxivity was defined according to M.A. Brown.26

Synthesis of N,N’-bis(2,2,6,6-tetramethyl-1-oxyl-4piperidinyl)dodecanediamide (compound 0). To a solution of 4-aminotempo (0.85 g, 5 mmol, Aldrich) and triethylamine (250 mg, 2.5 mmol, freshly distilled on potassium hydroxide) in tetrahydrofuran (200 ml, freshly distilled on LiA1H4) was slowly added a solution of dodecanedioyl dichloride (0.53 g, 2 mmol, Aldrich) in dry tetrahydrofuran (100 ml). The reaction

Reduction by Ascorbate 2 mM solutions of nitroxyl compounds were prepared in a potassium phosphate buffer (KH2P04 13.6 g/liter, pH adjusted at 7.4, glycerol 20% v/v) by the solubilization method described for relaxivity measurements. 500 ~1 of this solution were added to 500 ~1 of a freshly prepared ascorbate solution (10 mM in the same buffer). The reaction mixture was vortexed for

Nonionic nitroxyl lipids as contrast agents ??B.

five sec, and then placed in an EPR flat cell. The EPR scan was immediately initiated. Nitroxyl reduction rates were monitored with a Bruker ER 200 tt EPR spectrometer at room temperature. The decay of the high field peak height was followed as a function of time because the radical ascorbate anion interferes with measurements near the middle nitroxyl peak.19 The percentage of remaining nitroxyl was calculated by the ratio hl/ho x 100 where h, is the height at time t and ho is the height extrapolated at time 0. l9 Each experiment was repeated at least three times for each compound. Stability in Liver Homogenates Male Wistar rats, weighing -250 g, were used to prepare the liver homogenates. A freshly excised liver was homogenized in 3 ml of phosphate buffer (prepared as described above) per g of tissue with a Potter-Elvehjem tissue grinder. A solution of the nitroxyl compound (1.2 mmol/ml) equilibrated at 30°C was prepared as described for the reduction by ascorbate, but the surfaceactive agent used was Lubrol PX@ (Sigma, - 6 mg/ml). This nonionic detergent is used in solubilization of the cytochrome P-450 without (or with weak) denaturation at the present concentration.27 Just before the measurement, the liver homogenate (300 ~1) was mixed with a solution (at 30°C) containing Tris 5 x 10p2 M, sucrose 2 x 10-’ M, D-glucoseó-phosphate 9 x 10e3 M, NADH 5 x lom4 M, NADP 4 x 10w4M, and with 6 ~1 of glucosed-phosphate dehydrogenase (Boerhinger, grad 1): This system was used to regenerate the NADPH.28 The nitroxyl solution (500 ~1) was added, the reaction mixture vortexed, and immediately placed in an EPR flat cell. The evolution of the nitroxyl concentration was monitored by EPR spectroscopy observing the variation in the signal intensity of the low field peak. The intensity of the signal was estimated using the parameter Sio = i& x a where wie is the width of the line across one lobe of the derivative curve at height of l/lO of the amplitude a measured between the baseline and the summit of the peak.29,30 Each experiment was repeated at least three times for each compound. In Vivo Studies Vehicle of injection. Two vehicles of injection were used: solution 0.5 M in dimethylsulfoxide and IV emulsion (2%) prepared as follows : 40 mg of the nitroxyl compound were dissolved in 360 ~1 of trioleine (Sigma) containing 30 mg/ml of phosphatidylcholine (Sigma). This solution was heated at 90°C. 1.6 ml of an aqueous solution of glycerol (5%) and oleate (5 mM) equilibrated at pH = 7.4 was slowly added. The mix-

449

GALLEZ ET AL.

ture was stirred for 5 min and sonicated for 3 min. The quality of the emulsion was assumed by visualizing the preparation on microscope and by measuring the diameter of the micelles with a Coulter@ Model N4MD. The mean diameter size was 650 nm (95% limits = 595-705 nm). Acute toxicity. The acute toxicity was estimated by the approximate lethal dose (ALD) as defined by Lorke.31 Male NMRI mice (five to eight animals for each compound) received by intravenous injection different amounts of the tested compound (solutions 0.5 M in dimethylsulfoxide) in the range of 0.05-1.5 mmol/kg to define the lethal dose. Evaluation of the contrast enhancement. A Biospet@ (Bruker) imager was utilized for this study. The magnet operates at 4.7 T. A multislice spin echo technique was performed on male Wistar rats with acquisition of five transaxial sections, each 2 mm in thickness, a repetition time (TR) of 338 msec, an echo delay time (TE) of 13 msec, and a field of view (FOV) of 10 cm (acquisition time = 4’20”). The rats (n = 4), weighing 150-200 g, were anesthetized by IP injection of 40 mg/kg of pentobarbital. Images were taken before injection of the contrast agent. The rat was pulled out of the magnet. The parenteral administration of the compound B (sole tested compound on account of the acute toxicity of other molecules) was performed in the portal vein after incision of the abdomen. After suture of the wound, the rat was replaced in the magnet. The enhancement of the signal of the liver was estimated by measuring the intensity of brightness in regions of interest. The percentage signal enhancement was expressed as:

(Tissue post/Reference

post) - (Tissue prelleference

pre)

(Tissue pre/ Reference pre)

RESULTS AND DISCUSSION Relaxivity Relaxivity data are summarized in Fig. 2. In biological fluids, or similar media (solution of albumin), nitroxyl lipids present a higher relaxivity than the hydrophilic compounds E (carboxy-Proxyl) and F (Tempamine). On the contrary, the relaxivity in pure water does not differ significantly except in the case of compound A. More generally, this trinitroxyl glyceride-like compound possesses a higher relaxivity, although not strictly additive. Interestingly enough, this additivity trend is not shared by the binitroxyl lipid D, as already observed by Ehman et al. in the case of another structurally related diradical molecule.i2 This high relaxivity in biological media is appar-

Magnetic Resonance Imaging 0 Volume 10, Number 3, 1992

450

1.2 =T

1.0

**

g7

0.6

8 4 jj

o’4 0.2 0.8 l 0.0

t

? ?D

l-

Cl E

1 Water

Albumin

1.2 ’

1

=ì

1.0 -

6

1

0.8

7

0.6

;

0.4

G 1

0.2

0.8 ;

0.0

Pìúsma

Liver homogenate

Fig. 2. Relaxivity data per mole of compound (sec-’ mmol ‘). The measurements were performed at 90 MHz (26 ut 1°C) using an inversion-recovery sequence. Each value is an average of at least three experiments; the error bars represent the standard deviation. Significant differences between the nitroxyl-lipid and the hydrophilic nitroxyl lipids are presented: **: p < 0.05, *: p < 0.1, with regard to carboxy-Proxyl (compound E) $: p < 0.05, t: p < 0.1, with regard to Tempamine (compound F)

ently in contradiction with the hydrophobicity of the synthesized compounds, but can be explained, in biologica1 media, by the binding of these compounds to the proteins, especially to the albumin. Bennett et al. have recently demonstrated the importante of this binding on the relaxivity in their study on “doxylsterate.“” The relatively weaker relaxivity of the nitroxyl compounds with respect to the metal ion complexes may be also enhanced by immobilizing the nitroxide moiety on a macromolecule: The rotational correlation time increases and the inner sphere process becomes significant. The differences here observed in the relaxivities between hydrophilic and nitroxyllipid/albumin complex are less important than those noted by Bennett, but can be related to the high magnetic field used for the present study of relaxation time. The comparison of the relaxivity of different compounds is indeed more convenient by using NMRD profile, but the measurement of Tr by using a

higher magnetic field is more easily transposable to the usual clinical magnets. The comparable relaxivity of the lipophilic compounds in regard to their hydrophilic analogs in the pure water is more surprising. This behavior could originate from the formation of micelles thanks to the use of the surface-active agent, this micellar dispersion immobilizing likewise the nitroxide moieties. The relaxivity could be also increased in water through self-association into larger aggregates. However that may be, the relaxivity of lipophilic nitroxyl compounds in water was not really predictable. Stability

The values of the reduction rates of nitroxyl compounds by ascorbate are summarized in Table 1. The time course of reduction of nitroxyl compounds is presented in Fig. 3, where each point is the average of at least three measurements. We found again the wellknown order of reactivity: The five-membered rings

Nonionic nitroxyl lipids as contrast agents ??B. GALLEZ ETAL.

Table 1. Initial reduction rates of nitroxyl compounds by ascorbate (in micromoles per minute per millimole ascorbate). Each rate is the average of at least three measurements + standard deviation Compound

Rate

A B C D E F

5.6 f 2 5.7 11 0.4 21.4 + 7 111 f 4 3.6 + 1.4 189 f 17

are more stable than the six-membered rings. 18-*0In the case of the pyrrolidinoxyl radicals, a differente at 95% confidence level, although not highly significant (t-test), between the percentage of nitroxyl remaining at 22 min was observed, except for the pair of products B and E. With the piperidinoxyl compounds, we found the following order of stability at 2 min: C > D > F (highly significant differente at 95% confidence level). These results seem to indicate that the fatty acid chains are partly protecting nitroxyl radicals from the reduction. In liver homogenates, no detectable decay of the EPR signal intensity of the compounds A, B, C, and D was observed after 30 min. In the same conditions, the percentages of unchanged carboxyLProxy1 (E) and of Tempamine (F) remaining were respectively 90.6 f 0.2 and 65.6 f 5.1% (average of three experiments f standard deviation). The time course of reduction is presented in Fig. 4. The intensity of the EPR signal had to be calculated in liver homogenates by the pa-

451

rameter Sr0 (and not by the peak height as in the experiment of reduction by ascorbate) owing to the variations in the width of the peaks. For example, the height of the peak of compound D was systematically increasing, but the parameter Sr0 of the peak was not significatively changed; this behavior was more variable with the compounds B and C, and was not observed with compounds A, E and F. This variability in the ratio height/width of the peak is probably in relation with changes in the repartition (binding to biological proteins, presumably) of the nitroxyl compounds within the liver homogenate. Couet et al. have demonstrated that the relative rates of reduction in the presence of ascorbate were often parallel to those observed in tissue homogenates.*’ However, ascorbate only had a minor share in the reduction by the tissues. It must also be noted that the bioreduction site of the nitroxyl compounds is not yet clearly known. The two major reduction sites are apparently the microsomes and the mitochondria. By using an experiment on entire homogenate without subcellular fractionation, the image obtained is probably the most similar to the in vivo situation. The present results thus do not mean that the synthesized nitroxyl lipids are not affected by the metabolism (see below in vivo studies), but they certainly demonstrate that, in the present experimental conditions, the nitroxyl lipids are much more stable than the classica1 nitroxyl molecules E and F. This very high stability in biological fluids can be related to the localization of the chain-ending nitroxyl radical;33 this position seems to hinder the accessibility of the nitroxyl compounds to the bioreduction sites. These data should not be directly extrapolated to the living systems, nev-

, COMPOUND: A R C D E

E0

5

10

15

20

25

Time (min) Fig. 3. Time course of reduction of nitroxyl compounds

by ascorbate. Each experiment was repeated at least three times.

452

Magnetic Resonance Imaging 0 Volume 10, Number 3, 1992

Fig. 4. Time course of reduction of nitroxyl compounds in liver homogenates. Each experiment was repeated at least three times. Note that the four nitroxyl lipids remained unchanged more than 20 minutes in these conditions.

ertheless they represent classically a valid indication of

the resistance to bioreduction, and thus of the prolongation of their effect on the MRI contrast. This high stability should be interesting in fhe perspective of a prolonged examination with a contrast agent, but is less appreciable as a marker of perturbations in the metabolization rate, oxygenation for example. Zn Vivo

Studies An unexpected acute toxicity at relatively weak dose was observed with the synthesized compounds. The approximate lethal dose (ALD) for mice are presented in the Table 2. The symptoms before the death for al1 the compounds were seizures, opisthotonos, and tetany. This acute toxicity has limited US to evaluate in magnetic resonance imaging only the compound B, the least toxic nitroxyl lipid synthesized. Injected in the portal vein at 0.25 mmol/kg as IV emulsion or as solution in DMSO, no significant contrast enhancement was observed. Nevertheless the administration of 0.75 mmol/kg (solution in DMSO) to the rats produced an

enhancement of the signal intensity of the liver; the time course of this enhancement is summarized in the Fig. 5. Pre- and postcontrast illustrations are presented at the Fig. 6. The IV emulsion was not tested at the dose of 0.75 mmol/kg because the volume of injection became too important. We have also injected the compound B with the same dose in the tail vein of NMRI mice and evaluated its effect in Ti-weighted images; the contrast effect in these conditions was less important and the time course of the enhancement of the liver was more variable from one mouse to another. The use of a strong magnetic field is also an inconvenience to evaluate finely the T, contrast.

Table 2. Approximate lethal dose (ALD) in mmol/kg observed with the nitroxyl lipids administered in the tail vein of male NMRI mice (solution 0.5 M in DMSO) Compound

ALD Time (min)

A B C D

0.075 1 0.4 0.75

Fig. 5. Time course of enhancement of the liver after intraportal administration of the compound B (0.75 mmol/kg, solution OSM in DMSO). Each point is the average of the enhancement observed with four rats f standard deviation.

Nonionic nitroxyl lipids as contrast agents 0 B.

GALLEZ ET AL.

453

Fig. 6. Transversal slices showing the rat Ever image taken before contrast (on the left) and 5 minutes after intra-portal injection of 0.75 mmol/kg of the compound B (on the right).

CONCLUSION Up to now, nitroxyl products have not yet found clinical applications as MRI contrast agents. However some studies tend to demonstrate their potential applicability as contrast agents,33-35 although the hydrophilic nitroxyl compounds used had a lesser efficacy on the imaging contrast than the metallic complexes. Their usefulness could be increased if they were more

effective on relaxation processes, more stable and specific. Surprisingly, if a lot of in vitro studies concerning relaxivity and stability of the nitroxyls were undertaken, little work dealing with the organ specificity was published. We have chosen the lipids as vectors on the one hand because they accumulate predominantly in the liver after parenteral administration. On the other hand, the fundamental studies of Bennett on the relax-

454

Magnetic Resonance Imaging 0 Volume 10, Number 3, 1992

ation processes affecting the nitroxyl compounds have demonstrated that the inclusion of the nitroxyl moiety in fatty acid chains is a favorable factor to enhance the relaxivity of these compounds.15,36 His work concerned essentially the “doxyl” groups, but these moieties are too quickly reduced to be used as in vivo contrast agents. We have therefore synthesized other nitroxyl Iipid analogs containing piperidinoxyl and pyrrolidinoxyl radicals. We have obtained the same order of efficiency in relaxivity with the nonionic nitroxyl lipids discussed in the present work; we have also observed that the binding of the nitroxyl moiety with a lipid structure confers to the radicals an increased stability. The acute toxicity observed at the doses usually administered did not allow USto verify their organ specificity. The origin of the toxicity being probably in the centra1 nervous system, the opportunity to employ, in this precise case, entirely nonionic products is perhaps to be reconsidered. If the nonionic contrast agents are often better tolerated because they produce a weaker osmotic charge, it is likely that this nonionic character promotes the penetration of the compounds across the blood-brain barrier eventually resulting in the toxicity observed. The binding of some products to glyceride or fatty acid structures is indeed sometimes used to facilitate their passage through the hemato-encephalic barrier. 37-40We are currently exploring nitroxyl lipids containing ionizable groups assuming that the polarity enhancement could decrease the acute centra1 toxicity, preserve the potential hepatic accumulation, and permit a formulation as stable intravenous emulsion with a greater concentration of the nitroxyl products. REFERENCES 1. Carr, D.H.; Gadian, D.G. Contrast agents in magnetic resonance imaging. Clin. Radiol. 36:561-568; 1985. 2. Carr, D.H. The use of proton relaxation enhancers in

magnetic resonance imaging. Magn. Reson. Imaging 3: 17-25; 1985. 3. Lauffer, R.B.; Greif, W.L.; Stark, D.D.; Vincent, A.C.; Soini, S.; Wedeen, V.J.; Brady, T.J. bon-EHPG as an hepatobiliary MR contrast agent: Initial imaging and biodistribution studies. Nucl. Med. Biol. 15(1):47-50; 1988. 4. Kawamura, J.; Endo, K.; Koizumi, M.; Watanabe, Y.; Saga, T.; Konishi, J.; Horiuchi, K.; Yokoyama, A. Gadolinium-phthalein complexone as a contrast agent for hepatobiliary MR imaging. J. Comput. Assist. Tomogr. 1:67-70; 1989. 5. Brasch, R.C. Work in progress: Methods of contrast enhancement for NMR imaging and potential applications. Radiology 147:781-788; 1983. 6. Lurie, D.J., Bussell, D.M.; Bell, L.H.; Mallard, J.R. Proton-electron double magnetic resonance imaging of

7.

8.

9.

10.

ll.

12.

13.

free radical solutions. J. Magn. Reson. 76:366-370; 1988. Lurie, D. J.; Hutchison, J.M.S.; Bell, L.H.; Nicholson, 1.; Bussell, D.M.; Mallard, J.R. Field-cycled protonelectron double-resonance imaging of free radicals in large aqueous samples. J. Magn. Reson. 84:431-437; 1989. Swartz, H.M.; Bennett, H.; Brown, R.D., 111; Morse, P.D., 11; Pals, M.; Koenig, S.H. Feasibility of measuring oxygen and redox metabolism in vivo NMR: Effect of paramagnetic materials and their cellular metabolism on relaxation times of protons of water and lipids. Period. Biol. 87(2):175-182; 1985. Iannone, A., Hu, H., Tomasi, A.; Vannini, V.; Swartz, H.M. Metabolism of aqueous soluble nitroxides in hepatocytes: Effects of cel1 integrity, oxygen, and structure of nitroxides. Biochim. Biophys. Acts 99190-96, 1989. Swartz, H.M.; Sentjurc, M.; Morse, P.D., 11. Cellular metabolism of water soluble nitroxides: Effect on rate of reduction of cell/nitroxide ratio, oxygen concentrations and permeability of nitroxides. Biochim. Biophys. Acts 888:82-90; 1986. Chen, K.; Glockner, J.F.; Morse, P.D., 11; Swartz, H.M. Effect of oxygen on the metabolism of nitroxide spin labels in cells. Biochemistry 28(6):2496-2501; 1989. Ehman, R.L.; Brasch, R.C., McNamara, M.T.; Eriksson, U.; Sosnovsky, G.; Lukszo, J.; Li, S.W. Diradical nitroxyl spin label contrast agents for magnetic resonance imaging. A comparison of relaxation effectiveness. Znvest. Radiol. 21:125-131; 1986. Almen, T. Development of nonionic contrast media. Znvest. Radio/.

2O(l):SI-S9; 1985.

14. Grodd, W.; Paajanen, H.; Eriksson, U.G.; Revel, D.;

Terrier, F.; Brasch, R.C. Comparison of ionic and nonionic nitroxide spin labels for urographic enhancement in magnetic resonance imaging. Acts Radiol. 28(5):593600; 1987. 15. Bennett, H.F.; Brown, R.D.; Keana, J.F.W.; Koenig, S.H.; Swartz, H.M. Interactions of nitroxides with plasma and blood: Effect on l/T, of water protons. Magn. Reson. Med. 14:40-55; 1990. 16. Weichert, J.P.; Longino, M.A.; Schwender, S.W.; Counsell, R.E. Potential tumor- or organ-imaging agents. Polyiodinated 2-substituted triacylglycerols as hepatographic agents. J. Med. Chem. 29:1674-1682; 1986. 17. Belfrage, P.; Edgren, B.; Olivecrona, T. The tissue distribution and metabolism in the rat of intravenously injected labeled fat emulsions. Acts Physiol. Stand. 62:344-351; 1964. 18. Couet, W.R.; Brasch, R.C.; Sosnovsky, G.; Lukszo, J.; Prakash, 1.; Gnewuch, C.T.; Tozer, T.N. Influence of chemical structure of nitroxyl spin labels on their reduction by ascorbic acid. Tetrahedron 41(7):1165-1172; 1985. 19. Keana, J.F.W.; Pou, S.; Rosen, G.M. Nitroxides as potential contrast enhancing agents for MRI application: Influence of structure on the rate of reduction by rat hepatocytes, whole liver homogenate, subcellular frac-

Nonionic nitroxyl lipids as contrast agents ??B. GALLEZ

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31. 32.

tions, and ascorbate. Magn. Reson. Med. 5:525-536; 1987. Couet, W.R.; Brasch, R.C.; Sosnovsky, G.; Tozer, T.N. Factors affecting nitroxide reduction in ascorbate solution and tissue homogenates. Magn. Reson. Imaging 3: 83-88; 1985. Morrison, A.; Davies, A.P. Mass spectrometry of piperidine nitroxides, a class of stable free radicals. Org. Mass Spectrom. 3:353; 1970. Davies, A.P.; Morrison, A.; Barratt, M.D. Mass spectrometry of stable free radicals: Pyrrolidine nitroxides. Org. Mass Spectrom. 8~43-48; 1974. Sosnovsky, G.; Rao, N.U.M.; Li, S.W.; Swartz, H.M. Synthesis of nitroxyl (aminoxyl) labeled probes for studies of intracellular environment by EPR and MRI. J. Org. Chem. 54~3667-3674; 1989. De Boer, T.J.; Backer, H.J. Diazomethane. In: Organic Syntheses, Collective Volume 4. New York: John Wiley & Sans, Inc.; 1963: 250-253. Press, W.H.; Flannery, B.P.; Tenkolsky, S.A.; Vetterling, W.T. Numerical Recipes: The Art of Scientific Computing. Cambridge, England: Cambridge University Press; 1986: 523-528. Brown, M.A. Effects of the operating magnetic field on potential NMR contrast agents. Magn. Reson. Imaging 3:3-9; 1985. Isa, M.; Cumps, J.; Fossoul, C.; Atassi, G. Solubilisation des cytochromes P-450 humains par diverses classes de détergents. J. Pharm. Belg. 43(4):246-252; 1988. Cumps, J.; Razzouk, C.; Roberfroid, M. MichaelisMenten kinetic analysis of the hepatic microsomal benzpyrene hydroxylase from control, phenobarbital- and methyl-3-cholanthrene-treated rats. Chem.-Biol. Interactions 16:23-38; 1977. Kwan, C.L.; Yen, T.F. Electron spin resonance study of coal by line width and line shape analysis. Anal. Chem. 51(8):1225-1229; 1979. Goldberg, I.B. Chemical analysis by EPR. In: Electron Spin Resonance, vol. 6. London: The Royal Society of Chemistry , Burlington House; 198 1: 1-23. Lorke, D. A new approach to practica1 acute testing. Arch. Toxicol. 54:275-287; 1983. Baldassare, J.J.; Robertson, D.E.; McAfee, A.G.; HO, H.C. A spin-label study of energy-coupled active trans-

33.

34.

35.

36.

37.

38.

39.

40.

ET AL.

455

port in Escherichia coli membrane vesicles. Biochemistry 13(25):5210-5214; 1974. Ehman, R.L.; Wesbey, G.E.; Moon, K.L.; Williams, R.D.; McNamara, M.T.; Couet, W.R.; Tozer, T.N.; Brasch, R.C. Enhanced MRI tumors utilizing a new nitroxyl spin label contrast agent. Magn. Reson. Imaging 3:89-97; 1985. Brasch, R.C.; London, D.A.; Wesbey, G.E.; Tozer, T.N.; Nitecki, D.E.; Williams, R.D.; Doemeny, J.; Dallas Tuck, L.; Lallemand, D.P. Work in progress: Nuclear magnetic resonance study of a paramagnetic nitroxide contrast agent for enhancement of renal structures in experimental animals. Radiology 147:773-779; 1983. Herfkens, R.; Davis, P.; Crooks, L.; Kaufman, L.; Price, D.; Miller, T.; Margulis, R.; Watts, J.; Hoenninger, J.; Arakawa, M.; McRee, R. Nuclear magnetic resonance imaging of the abnormal live rat and correlations with tissue characteristics. Radiology 141:21 l218; 1981. Bennett, H.F.; Brown, R.D., 111; Koenig, S.H.; Swartz, H.M. Effects of nitroxides on the magnetic field and temperature dependence of l/T, of solvent water protons. Magn. Reson. Med. 4:93-111; 1987. Jacob, J.N.; Shashoua, V.E.; Campbell, A.; Baldessarini, R.J. y-Aminobutyric acid esters. 2. Synthesis, brain uptake, and pharmacological properties of lipid esters of y-aminobutyric acid. J. Med. Chem. 28:106-110; 1985. Jacob, J.N.; Hesse, G.W.; Shashoua, V.E. y-Aminobutyric esters. 3. Synthesis, brain uptake, and pharmacologica1 properties of C-18 glyceryl lipid esters of GABA with varying degree of unsaturation. J. Med. Chem. 30: 1573-1576, 1987. Jacob, J.N.; Hesse, G.W.; Shashoua, V.E. Synthesis, bram uptake, and pharmacological properties of a glyceryl lipid containing GABA and the GABA-T inhibitor y-vinyl-GABA. J. Med. Chem. 33:733-736; 1990. Mergen, F.; Lambert, D.M.; Saraiva Gonçalves, J.C.; Poupaert, J.H.; Dumont, P. Antiepileptic activity of 1,3dihexadecanoylamino-2-valproyl-propan-2-01, a prodrug of valproic acid endowed with a tropism for the CNS. J. Pharm. Pharmacol. 43:815-816; 1991.