Comparison of metabolism and toxicity to the structure of the

Glutathione-Dependent Metabolism of the Antitumor Agent Sulofenur. Evidence for the Formation of p-Chlorophenyl Isocyanate as a Reactive Intermediate...
0 downloads 0 Views 977KB Size
Download PDF
Chem. Res. Toxicol. 1992,5, 667-673

667

Comparison of Metabolism and Toxicity to the Structure of the Anticancer Agent Sulofenur and Related Sulfonylureas William J. Ehlhardt,’ Joseph M. Woodland, John F. Worzalla, Jesse R. Bewley, Gerald B. Grindey, Glen C. Todd, John E. Toth, and J. Jeffry Howbert Lilly Research Laboratories, Indianapolis, Indiana 46285 Received May 4, 1992

The metabolic formation of p-chloroaniline from the oncolytic agent sulofenur [N-(5-in-

danesulfonyl)-N’-(4-chlorophenyl)urea, LY 186641, I] and from similar diaryl-substituted sulfonylureas, and its possible relevance to the compound’s toxicity, was studied. In previous studies it was found that significant amounts of metabolites such as 2-amino-5-chlorophenyl sulfate (11),which is also a metabolite of p-chloroaniline, are formed from sulofenur in mice, rats, monkeys, and humans. The metabolism of N-(4-tolyl)-N’-(2-hydroxy-4-chlorophenyl)urea (V) was studied, and V was not found to be an intermediate in the metabolic formation of I1 from the sulfonylurea N-(4-tolyl)-N’-(4-chlorophenyl)urea (LY181984,111). The amounts of this p-chloroaniline metabolite (11) formed in C3H mice from a series of diarylsulfonylureas were found to correlate with the compound’s propensities to form methemoglobin, one notable toxicity of p-chloroaniline. This metabolism was also found to correlate with the structure of the arylsulfonyl moiety of the sulfonylurea. Other evidence supports the hypothesis that p chloroaniline is directly formed by metabolism of sulfofenur and similar diarylsulfonylureas as well. Metabolic formation of p-chloroaniline thus appears to be a plausible explanation for the methemoglobinemia and anemia found to be dose-limiting toxicities of sulofenur in Phase I trials.

Introduction Sulofenur [I,N-(&indanesulfonyl)-N’-(4-chlorophenyl)urea, LY1866411 is one of a novel series of diarylsulfonylureas with potent antitumor activity in mouse in vivo

&?

‘un

resulting in a much longer half-life for sulfonylureas such as tolbutamide in this species (11-15). In addition to this major route of metabolism, it was also noted that trace amounts of p-chloroaniline, along with more significant amounts of its major metabolites, were formed after oral doses of [chZ~rophenyl-~~Clsulofenur were given to mice, rats, monkeys, and humans (10). The rate of sulofenur elimination appears to be determined primarily by the rate of its metabolism to benzylic hydroxylation products. However, the formation of p-chloroaniline-related metabolites appears to be a competing route of metabolism, which is more prominent in species where the elimination of sulofenur is slower, such as in monkey (IO). In other cases where alternative metabolic routes are slow or nonexistent, apparent cleavage of the sulfonylurea linkage has also been observed. Tolbutamide and similar sulfonylureas are metabolized in dogs, for example, to give toluenesulfonamide as the major urinary excretion product (11,14). However, it has not yet been shown that a direct cleavage of these substituted sulfonylureas is involved (see ref 10). For example, in the case of chlorpropamide (16), the propyl group of this aryl alkylsubstituted sulfonylurna is oxidatively degraded to eventually give the mono-substituted [(p-chloropheny1)sulfonyl]urea, which might itself be the precursor for the p-chlorobenzenesulfonamideobserved as a metabolite. Another way by which the sulfonylurea linkage might be cleaved is by chemical hydrolysis. For most of the sulfonylureas considered here, the hydrogen at the sulfonamide nitrogen is weakly acidic (see also Table 11), which may explain why such compounds decompose more slowly at neutral or slightly basic pH. For example, in 2% acetonitrile/% mM sodium phosphate buffer there is about 25% decomposition of sulofenur (pK, = 5.5) over 24 h at 37 “Cat pH 5 versus no detectable decomposition over 24

a=’

un

Sulofenur I

tumor models (1-3) and is currently being evaluated in Phase I and I1 clinical trials ( 4 , 5 ) . Other sulfonylureas such as tolbutamide, glyburide, and acetohexamide are used as hypoglycemic agents (6,7), and compounds such as sulfometuron or chlorsulfuron are herbicides known to inhibit acetolactate synthase (8). Sulofenur was found to be inactive in those models, however, and neither do typical examples of these two drug classes show activity in the in vivo antitumor screen (1). The basis of sulofenur’s potent antitumor activity in animals is not known but is apparently different from those of existing oncolytic drugs ( I 3, 9). The major route of sulofenur metabolism is through benzylic hydroxylation of the indanesulfonyl moiety, and the major metabolites in all species studied are LY264912 and LY227938 (XI1 and XIII, respectively, in Table 11) (10). In many respects its metabolism is thus similar to that of tolbutamide (11’12)and other sulfonylureas having a toluenesulfonyl group (13,14). The dog apparently lacks the ability to carry out this hydroxylation, however, Address correspondence and reprint requests to this author at 0825 Department of Drug Metabolism, Lilly Research Laboratories, Lilly Corporate Center, Indianapolis, IN 46285.

0 1992 American Chemical Society

668 Chem. Res. Toxicol., Vol. 5, No. 5, 1992

h at pH 7.' Thus, blocking this ionization with, for example, a methyl substituent on the sulfonamide nitrogen, should result in a sulfonylurea much more susceptible to hydrolysis, and this has been shown in the case of a sulfonylurea herbicide (17). However, there appears to be no evidence for such a chemical hydrolysis in the metabolism of sulfonylureas in animals or man. The importance of these considerations in the metabolism of sulofenur is that anemia and methemoglobinemia were found to be notable toxicities in animals (18) and man ( 4 , 5 ) . While these toxicities are certainly mild in comparison with toxicities associated with conventional cancer chemotherapy, they were determined to be dose limiting for sulofenur in Phase I clinical trials (4, 5 ) . Various toxicities including hemolytic anemia have been firmly associated with anilines (19-24), and oxidation of hemoglobin to methemoglobin via N-hydroxy or nitroso metabolites is a well-studied characteristic toxicity of anilines (25). Since major metabolites of p-chloroaniline (26) have been observed as metabolites of sulofenur in animals and man as well (IO), it is possible that p-chloroaniline itself (through its usual metabolic activation) is responsible for much of the toxicity of sulofenur. The metabolism of N - (4-tolyl)-N'-(2-hydroxy-4-chloropheny1)urea (V) was studied, and the results demonstrate that ortho hydroxylation of the p-chloroaniline moiety does not account for the formation of I1 from the sulfonylurea N-(4-tolyl)-N'-(4-chlorophenyl)urea (LY181984,111).Evidence is presented to show that p-chloroaniline itself is formed from sulofenur and is associated with the toxicity of this and related sulfonylureas.

Materials and Methods Mass spectral data were obtained with a VG ZAB-3F instrument, and FT NMR2 spectra were recorded on a Nicolet 300MHz instrument. Liquid chromatography utilized a Waters Instrument Co. (Milford, MA) 600-990 system and WISP autosampler with a 5- X 250-mm Zorbax RX column (DuPont Chemicals, Wilmington, DE). A gradient solvent program was employed utilizing 5 % acetonitrile (Burdick and Jackson, HPLC grade)/0.025 M pH 7.0 NaP04 buffer (prepared from deionized water, Milli-Q system, Millipore Corp.) at 1.5 mL/min flow rate for 5 min followed by a linear change to 50% acetonitrile over 30 min. With the diode-array detector, UV spectra were recorded for each putative metabolite within the LC2run; after correction for background, the spectra were considered to match standards evaluated under the same LC conditions if the spectra were not greater than 5% different in absorption at any wavelength between 220 and 380 nm. Radiocarbon scintillation counting was carried out with PCS scintillant (Amersham, Arlington Heights, IL), except where noted, using a Tri-Carb System (Packard Instrument Co., Downers Grove, IL). In each sample, dpm were calculated from the external counting efficiencyand [I4C]scintillation spectrum; only samples containing dpm greater than 25% above background (30-40 dpm) were used for quantitation. For radiochemical LC analysis, either 0.5-min (0.75 mL) or 1.0-min (1.5 mL) fractions were collected (Foxy fraction collector, ISCO, Lincoln, NE), to which 0.5 mL of water and 10 mL of PCS solution were added for scintillation counting. Aliquots of 100-3OOpL urine or plasma were counted directly in 10 mL of PCS, whereas 200 pL of 30% hydrogen peroxide followed by 100 p L of glacial acetic acid and 1

William J. Ehlhardt, unpublished results.

* Abbreviations: FT NMR, Fourier-transform nuclear magnetic res-

onance; LC, liquid chromatography;FAB, fast atom bombardment; TLC, thin-layer chromatography; QSAR, quantitative structure-activity relationship; VDW,van der Waals radius.

Ehlhardt et al. 10 mL of PCS were added to aliquots of 100-200 pL of whole blood, red blood cell suspension, and cell precipitates. Materials. Sulofenur (I) was obtained and assayed (Eli Lilly and Co., lot E57-UQ6-127) and was found to be >98% pure. Other sulfonylurea compounds were prepared and characterized as reported previously (I). Uniformly labeled [chlorophenyl14C]s~lofenur (lot 497-1-213) and [~hlorophenyl-~~C]LY181984 (lot 497-1-202) were also prepared (D.Kau, Lilly Research Laboratories), and their radiochemical purities of >98% were confirmed by LC analysis (above method). p-Chloroaniline (caution: a possible carcinogen) of >98% stated purity was purchased from Aldrich Chemical Co. (Milwaukee, WI). Uniformly labeled [14Cl-p-chloroaniline,133.3pCi/mg, was obtained from Amersham (lot 02F9243), and its reported radiochemical purity of >98% was confirmed by LC analysis (above method). (A) Potassium 2-Amino-5-chlorophenylSulfate (Potassium Salt of 11). The procedure used was patterned after that of Boyland, Manson and Sims (27). To a mixture of p-chloroaniline (12.7 g, 100 mmol) and 100 mL of pyridine in 100 mL of 2 N KOH was added over 1.5 h a solution of potassium persulfate (27.0 g, 100 mmol) in 700 mL of water, after which the mixture was stirred 20 h a t room temperature. The supernatant was decanted from the insoluble tar formed and reduced to a volume of about 400 mL under vacuum a t 45 "C. The resulting brown supernatant was again decanted from insoluble material and washed two times with 100mL of ether, and water was evaporated under vacuum to give a solid, which was then triturated three times with 200 mL of methanol. Evaporation of methanol yielded 7.0 g of brown solid which was mainly the desired product. The material was recrystallized in 200 mL of 95% ethanol after treatment with 0.75 g of decolorizing carbon, from which the crystals were collected after refrigeration and washed with cold ethanol, yielding 3.7 g (14%). The material could be further purified by several recrystallizations with ethanol to give the ( 6 ) 205 (27 OM), 243 following: mp 213 OC dec; UV (EtOH) A, (11OOO),298 (2500)nm; MS [(+)-FAB] 262 (M + H)+;IR (KBr) 3407, 3319, 1630, 1588, 1492, 1418, 1289, 1237, 1200, 1057,928, 757 cm-I; 'H NMR (300 MHz, DMSO-ds) 6 4.83 (br s, 2 H, exch D20, NH2), 6.63 (d, 1 H, J = 8.5 Hz, aryl H), 6.81 (d of d, 1 H, J = 2.4, 8.5 Hz, aryl H),and 7.03 (d, 1 H, J = 2.4 Hz, aryl H). Anal. Calcd for CsH&lKN04S: C, 27.53; H, 1.92; N, 5.35; S, 12.25. Found: C, 27.57; H, 1.91; N, 5.10; S, 11.90. (B) N-(I-Toiyl)-W-(2-hydroxy-4-chlorophenyl) urea (V). The compound was prepared by modification of a procedure previously established (I). Alloperations wereconducted atroom temperature under dry nitrogen atmosphere. To a solution of in 7 mL of dry 0.810 g (5.64 mmol) of 2-hydroxy-4-chloroaniline tetrahydrofuran was added a solution of 1.113 g (5.64 mmol) of 4-toluenesulfonyl isocyanate in 5 mL of dry tetrahydrofuran over 2 min with stirring. After 2 h the solvent was evaporated under vacuum, and the resulting dark foam was treated with a small quantity of toluene. After standing 4 days, a crystalline solid had appeared; this was collected by filtration, washed twice with toluene, and dried under vacuum at 65 OC to give 1.882 g (95%) of a gray powder, which appeared homogeneous by TLC (EtOAc): UV (EtOH),,A ( 6 ) 292 (4570), 249 (11900), 210 (25 200) nm; MS (FAB) 341 (M + HI+;IR (KBr) 3330,3303,1678,1619,1599, 1563, 1553, 1508, 1415, 1156, 1090 cm-l; 'H NMR (300 NMR MHz, acetone-&) 6 2.41 (s,3 H, tolyl CH3), 6.81 (d of d, J = 2.2, 8.7 Hz, 1 H, aryl H),6.91 (d, J = 2.2 Hz, 1 H, aryl H), 7.42 (d, J = 8.1 Hz, 2 H, tolyl aryl H),7.95 (d, J = 8.2 Hz, 2 H, tolyl aryl H),7.96 (d, J = 8.5 Hz,1 H, aryl H).Anal. Calcd for CllH13C1N2O4S: C, 49.34; H, 3.85; N, 8.22; S, 9.41. Found: C, 49.60; H, 3.80; N, 8.06; S, 9.51. Methods. (A) Animal Studies. Female C3H mice (20 & 1 g, Charles River, Greenfield, IN) were placed in appropriate steel metabolism cages and given free access to water and food (Purina rodent chow 5001 or 5002),except that animals were fasted overnight prior to oral dosing. Suspensions of sulfonylureas were prepared in 10% acacia/water by sonication of the preparation for 5 min at room temperature (Branson Model 2500) followed by at least 5 min of continuous (magnetic) stirring. Dose

Chem. Res. Toxicol., Vol. 5, No. 5, 1992 669

Metabolism and Toxicity of Sulofenur

a ' '>

S02NHCONH

LY181984

Ill

Jy

i

K

S02NHCONH

[

--...--T

_._....---_..-.-M-oNHnc],

V

/ -03SO

VI Figure 1. Proposed scheme for metabolism of LY181984 (111) and V.

suspensions were prepared immediately before use, and stirring of the suspension was continued throughout the dosing period. Mice were dosed intragastrically by syringe with suspensions to 10 mg/mL of the sulfonylurea to achieve a dose of 100 mg/kg. sulofenur or LY181984 was When [chl~rophenyl-~~Cl-labeled used, aliquots of the dose suspensions were dissolved in 50% methanol/water and assayed for radiocarbon content by scintillation counting after dosing was complete. Three groups of 2-4 mice per group were used in each experiment. Control urine was collected for 24 h prior to dosing, and urine was collected for 24 h after dosing over dry ice and then stored a t -70 OC. Radioactivity was assayed as described above. (B) Urinary Concentration of 2-Amino-5-chlorophenyl Sulfate (11). For each compound, groups of 2-4 mice were dosed orally with 100 mg/kg of the sulfonylurea (molecular weights of these compounds varied from 310 to 420), and urine was collected form 0 to 24 h. The urine was weighed, and 200-mL filtered (0.2 pm) aliquots were analyzed directly by LC (conditions above). It was found in previous experiments (26) that recovery of radiocarbon from urine of animals treated with p-chloroaniline (in which the majority of radiocarbon is in the form of 11) was quantitative after filtration and LC analysis, as determined by counting LC effluent fractions. The peak corresponding to I1 (retention between 14 and 16 min), which was also confirmed by simultaneous UV analysis (Ams 240, 290 nm), was integrated (250 nm) and normalized to the total amount of urine collected 0-24 h. Two standard curves were used to determine the absolute amounts of I1 in urine samples; known amounts of I1 added to urine gave a linear response and were accurate within *lo% between 80 and 4000 pM, and amounts of I1 as determined by counting radiocarbon in the LC effluent fraction associated with its LC peak (2.5-25 nmol on column) matched these values within experimental error. Estimates of the amount of I1 in urine after treatment with the radiolabeled compounds (sulofenur and 111) using the above standard curve gave results within 20 % of those estimated by radiocarbon scintillation counting of LC fractions corresponding to 11,and repeated estimates of I1 after treatment with XI were also within 20% of the mean. At least two measurements of I1 were carried out with each sulfonylurea, and where the measurements differed by more than &20% of their average (usually due to partially interfering peaks in LC analysis) only one significant figure is given. (C)Methemoglobin Determination. Groups of 5-10 mice were given oral doses of 100 or 300 mg/kg of the sulfonylurea every day for 7 days and were then sacrificed by decapitation and >lo0 pL samples of blood collected. A standard method (28)was used to determine the amount of methemoglobin relative to total hemoglobin in blood (% MHb). The control value of 3.3 % (i0.3 % ) was substracted to give the increase in % MHb.

(D) Substituent Parameter Analysis. Aromatic substituent parameters for the substituents on the arylsulfonamide ring were largely available from the parameter set developed previously for this series of compounds ( I ) . Additional required parameters were taken from the literature, or developed by procedures analogous to those used in ref I , either by summation of parameters for single substituents, extrapolation from the values of related substituents, or estimation via computer algorithms. The logarithm of the amount of I1 produced from the various analogs (urinary excretion in 0-24 h) was correlated against each of the parameters singly, and against certain pairs of parameters, using the linear regression analysis routine in the statistical package JMP (SAS Institute Inc., Cary, NC; Version 1.0.4, running on a Macintosh 11). The amount of I1 produced was analyzed as its logarithm because it is presumably proportional to the rate of a chemical or enzymatic process, and this rate would be expected to bear an exponential, rather than a linear, relationship to the free energy of interaction of each analog with a hypothetical reactive species. The degree of correlation was expressed as the square of the correlation coefficient, R2, which corresponds to the proportion of the total variance in the data set explained by that combination of parameters.

Results Metabolism of V. Compound V was synthesized and its metabolism studied to test whether it is a metabolite of LY181984 (111)and thus perhaps a metabolic precursor of I1 (Figure 1). First, it was apparent that V is not significantly more susceptible to hydrolysis than LY 18194. LY181984 is similar to sulofenur in its breakdown in aqueous solution (55% and 43 7% decomposition, respectively, at pH 5 over 48 h at 37 "C, none detectable for either at pH 7.41,and compound V decomposed to only a slightly greater extent (62% at pH 5,7% at pH 7 over 48 h at 37 "C). More significant is that the metabolism of V is not consistent with it being an intermediate in the metabolism of LY 181984. Table I shows that, when a 100 mg/kg oral dose of V is given to C3H mice, less of the metabolite I1 is excreted in 24 h than after a 100 mg/kg oral dose of LY181984. In addition, the major urinary excretion product of V is a compound (VI)3which is not observed at all as a product of the metabolism of LY 181984. Tentatively identified as VI, the sulfate derivative of V. Isolated VI (preparative LC; retention 25 min in system described under Methods, compare retention 32-33 min for V) gave UV, ,A a t 255 (sh 290) nm (compare UV, ,A at 255,290nm for V; both in LC buffer) and *HNMR (acetone-de): 6 2.40 (s, 3 H,tolyl CH3), 6.95 (d of d, J = 1.4, 8.5 Hz,1 H,

670 Chem. Res. Toxicol., Vol. 5, No. 5, 1992

Ehlhardt et al.

Table I. Urinary Excretion Products after Treatment of Mice with LY181984 (111) or Ve

5% of dose from compound

% of dose from compound metabolite I1 I11 IV V VI

I11 9

NDb ND ND

I11

V

40

V

'-'

>40

Table 11. Excretion of I1 and Increase in Methemoglobin after Treatment of Mice with Sulfonylurease

metabolite

0 Excretion of urinary metabolites in four C3H female mice after an oral dose of 100 mg/kg V or 100 mgikg, 4 pCi/ZO g mouse, [chlor0phenyl-1~C]-IIIin suspension. The % dose corresponding to each metabolite in urine collected 0-24 h postdose was calculated by quanitating radiocarbon for metabolites of 111,and by LC/UV area analysis for peaks corresponding to metabolites of V. See Figure 1 for structures. ND, none detected.

Measurement of 11. A number of sulfonylureas were available (1)which retain the p-chloroaniline moiety but differ in the arylsulfonyl substituent, and the structures of those used here are indicated in Table 11. To eventually test whether formation of metabolite I1 correlated with the amount of p-chloroaniline formed from these sulfonylureas, the amount of I1 excreted into the urine 24 h after 100 mg/kg single oral doses of each sulfonylurea was measured (see Methods). It was found that there is a wide variation in the propensity of these sulfonylureas to form metabolite 11. This ranged from 0.1 pmol of I1 excreted into the urine in 24 h after a dose of XXIII (accounting for about 1%of the administered dose) to 2.67 pmol (about 22 % of the administered dose) from XI, and the excretion of I1 from sulofenur (0.37 pmol) was found to be toward the low end of this range. Since these compounds differ in their molecular weights (from about 310 to 420))a somewhat different dose of each compound (always 100 mg/kg) was given. The levels of I1 listed in Table I1 are not corrected for the molecular weight difference (f15% of the midpoint). Note that the range of molecular weight is within f20 '?6 of the molecular weight of sulofenur (350), the compound used to calibrate the measurement of 11, and within the accuracy of these estimates (f20%). Correlation with Methemoglobin Formation. If the observation of I1is indicative of p-chloroaniline formation, then the formation of I1 after treatment of animals with various sulfonylureas should correlate with other properties of that aniline. Methemoglobin formation is one characteristic toxicity of anilines (25)which was also found to be related to the dose-limiting toxicity of sulofenur in Phase I trials (4, 5). Despite the greater amount of compound needed to cause a measurable response in mice after sulofenur treatment (about 250 mg), it was possible to measure the methemoglobin-inducing potential of six sulfonylureas which spanned most of the range of mearyl H),7.23 (d, J = 1.6 Hz,1H, aryl H),7.37 (d, J = 8.1 Hz, 2 H, tolyl arylH), 7.93 (d, J = 8.2 Hz,2 H, tolyl arylH), 7.99 (d, J = 9.1Hz, 1H, aryl H).The NMR data are similar to those for V (see Materials); the proton next to the phenolic group of V (6 6.91) is relatively the most shifted in VI (6 7.23), suggesting the phenol might be substituted with sulfate. A similar shift is seen for the corresponding proton in the NMR ofII(67.21inacetone-d6,ref26). TheMSdata [(-)-FAEl/glycerollrevealed signals (m/z 537 and 539 in a 3:l ratio) correspondingto VI plus NaOS03, the explanation for which is not clear. Incubation of a small amount of VI for 1 h at 37 "C in 0.2 M pH 5 NaOAc with 10 units of 0-sulfatase (Sigma S-9626), plus 100 mM saccharic acid l,.l-lactone to inhibit glucuronidase, resulted in conversion of about 25% of the sample to V, consistent with the assigned structure.

% methemoglobin a t pmol compound substituent(s) R pKab of 11'

100 mg/kg

controld sulofenur 3,4-(CH2)3 (1) LY181984 4-Me (111) VI11 3,4-(CH= CHCH=CH) IX 3,4-(OCH20) X 4-OMe XI H LY264912 3,4- [CHzCHz(XII) CH(0H)I LY227938 3,4-[CHzCHz(XIII) C(0)I XIV 3,4-(CHzCHzO) xv 4-F XVI 4-C1 4-[C(O)CH31 XVII 3-CH3 XVIII 3,4-(C&)z XIX xx 3-C1,4-F 3,5-(Ch)z XXI XXII 3,5-C12 3,4,5-(CH3)3 XXIII 3,4,5-C13 XXIV

300 mg/kg

6.2

NDe 0.371

3.3 f 0.2 3.6 f 0.2 4.2 f 0.4 6.8 f 2.0

6.1

1.0y

6.0 f 0.3 20.3 f 2.0

0.55

4.7 f 0.6

5.8 6.3 6.4

0.73 0.93 2.67 0.33

4.5 f 0 . 5 11.8f 2.0 7.1 f 0.6 10.3 f 1.0 toxic

0.55

4.8 6.1 4.3 6.3 4.3

0.4 1.55 1.33 0.5 1.57 0.4 0.77 0.15 0.5 0.1 0.2

Groups of 2-4 mice were given 100 mg/kg oral doses of each sulfonylurea and combined urine was collected 0-24 h, and the amount of metabolite I1 excreted into the urine in 24 h per mouse was determined. In separate experiments, groups of 5-10 mice were given daily oral doses of either 100 or 300 mgikg for 7 days, and the amount of methemoglobin relative to total hemoglobin was determined. See Materials and Methods for details. Estimated in solutions of 2:l dimethylformamideiwater. pmol of I1 excreted in urine, 0-24 h. 200 pL of 10% acacia. e None detected. f Result confirmed by radiocarbon analysis.

tabolite I1 excretion. Oral doses of 100 or 300 mg/kg of each sulfonylurea were administered daily to C3H mice, and the percent methemoglobin found in blood was measured after 7 days (see Methods). The correlation of these results, listed in Table 11, with p-chloroaniline metabolite I1 formation is illustrated in Figure 2. The observation of this correlation (R2= 0.91) provides some evidence for both hypotheses involved, that the amount of I1 observed is indicative of p-chloroaniline formation in vivo and that this formation of p-chloroaniline may be responsible for the anemia with characteristic methemoglobinemia observed as a toxicity of sulofenur. QSAR2 Analysis. Since, as discussed above, the formation of I1 from the sulfonylureas tested varies over a fairly large range, and these compounds differed only in their arylsulfonyl substituent,the possible relationship of these substituents' properties to the formation of I1 was investigated. This was done using the classic linear free energy approach of Hansch (30),wherein observed biological activities are correlated against parameter scales for various properties of aromatic substituents. The urinary excretion of I1from the set of 19 compounds listed in Table I1 was measured, and the possible correlations were tested using simple linear regression analysis. All of these compounds showed in vivo antitumor activity ( I ) ,

Chem. Res. Toxicol., Vol. 5, No. 5, 1992 671

Metabolism and Toxicity of Sulofenur

0.5

,

i

MHb increase

Yo

-'1 I

I

-10

Table 111. Correlation (S)of Formation of 11 with Various Aromatic Substituent Parameters. correlation correlation with lo with lo parameter(s) ("01 of f 1 ~ parametab) ("01 of ~ I P ir

ir

+ *2

a a+ a-

field resonance

0.213 0.420 0.450

o,ooo

VDW vol 4-VDW VOI 3,5-VDW V O ~ 4-VDW VOI 3,5-VDW V O ~

0.733 0.065 0.657 0.819

0.OOO

ir2

0.788 0.686 0.833

0.004 0.168C 0.184'

I

0

10

I

20

I

30

= II

I

I

40

50

60

VDW volume

pmol II

Figure 2. Correlation between methemoglobin formation ( % MHb increase) and urinary excretion of I1 (pmol of 11). Mice were treated orally with 100mg/kg/day of various sulfonylureas for 7 days, and the percent methemoglobin formed was then measured, or the excretion of I1 was measured in urine collected 0-24 h after a single 100 mg/kg dose (data from Table 11). A correlation (R2)of 0.914 was found with these two parameters. Dashed lines indicate 95% confidence limits on linear fit (n = 6 compounds).

7r2

I

+ + VDW vol ~2 + 3,5-VDW V O ~ ir2-4-VDW V O ~+ 3,5-VDW V O ~

"All 19 compounds in Table I1 were fully parametrized and included in each correlation, except as noted. Excreted in urine, 0-24 h after dosing. First 15 compounds in Table I1 only.

*

except for LY264912 (XII) and LY227938 (XIII), which are the major metabolites of sulofenur in all species studies (10). Parameters used in the present analysis quantitate , in several ways the hydrophobicity ( T , T ~ ) electronic character (u, u+, u-, field, resonance), or size (VDW2 volume) of the substituents in a given pattern. The size of the substituents was considered both as the aggregate of all the substituents in a pattern (VDW volume) and also by calculating separately the bulk at the 4-position and that at the 3- and 5-positions combined (4- or 3,5VDW volume). None of the hydrophobic parameters strongly correlated with the production of 11, although 1r2 gave a modest correlation which was statistically significant (p < 0.05; Table 111). Combination with the related parameter T did not improve the analysis. None of the five electronic parameters tested gave any meaningful correlation. The only parameters which were clearly correlated with the production of I1 were those which measure the volume of the substituents. This is very evident graphically, as shown in Figure 3, where the amount of TI excreted decreases in a linear fashion as the aggregate VDW volume of the substituents increases. When the volumes at the 4-position and the 3,Ei-positions were analyzed separately, the great preponderance of the effect seemed to be conferred by the bulk of the substituents at the 3- andlor 5-positions. Use of the 4- and 3,5-VDW volumes in combination did not

Figure 3. Correlation between formation of I1 and VDW volume of substituents on arylsulfonamide ring. Dashed lines indicate 95% confidence limits on linear fit (n = 19 compounds). significantly improve the analysis relative to using total VDW volume as a single parameter, nor did combination of r 2 with any of the volume parameters improve the correlation relative to the volume parameter alone. As mentioned above, these sulfonylureas are susceptible to hydrolysis when in neutral form, at pH near to or less than their pK,. Although the pK;s of this series have not been systematically measured, those measured for 10 of these sulfonylureas vary from 4.3 to 6.4 (Table 11). No significant correlation of these values with the measurement of I1 formation (R2 = 0.31) was found.

Discussion Previous results on the metabolism of sulofenur suggested that the compound is cleaved in some fashion to give p-chloroaniline as a transient metabolite, along with the complementary products indanesulfonamide (from sulofenur itself) or 1-hydroxyindane-5-sulfonamide[from its major hydroxyindanyl metabolite LY264912 (XII);see Table I11 (IO). Not only was a trace of p-chloroaniline observed in the urine of mice, rats, monkeys, and a human patient given sulofenur, but greater amounts of 2-amino5-chlorophenyl sulfate (111,which is known to be the major metabolite of p-chloroaniline (261, were also found. The detection of 11, rather than p-chloroaniline, would be expected, even if sulofenur is cleaved directly to give pchloroaniline, since it was previously shown that the metabolism and elimination of p-chloroaniline are more rapid than those of sulofenur (IO, 26). While it is reasonable to assume that p-chloroaniline itself is an intermediary metabolite accounting for the presence of its major metabolite 11, one alternative hypothesis might involve initial oxidation of sulofenur at the ortho position of the p-chlorophenyl moiety. These two possibilities are illustrated for the closely related compound LY181984 (111) in Figure 1 (the major metabolite of LY181984 is VII). If, for example, the presence of such an o-hydroxy substituent provided anchimeric assistance, a compound such as V might hydrolyze rapidly, giving the o-hydroxyaniline precursor of I1 rather than the expected metabolite VI. However, another metabolite (VI), not observed in the metabolism of LY181984, was found to be the major urinary metabolite of V, and in fact less I1 is formed from V than from LY191984, so it was concluded that V is itself not an intermediate in the metabolism of this sulfonylurea.

672 Chem. Res. Toxicol., Vol. 5 , No.5, 1992

One line of evidence which would support the intermediacy of p-chloroaniline in the production of I1 would be the observation of biological activities and/or toxicities attributable to anilines following dosing of sulofenur and related sulfonylureas. A correlation between the production of I1 and aniline-like toxicities, if established, would also support the involvement of aniline metabolites in the known toxicities of these sulfonylureas, This assumes that formation of I1 is a reliable index of the amount of p chloroaniline formed and thus represents the degree of exposure to p-chloroaniline. In fact, measurement of sulfate metabolites eliminated into the urine has been used to estimate exposure of workers to anilines in industrial settings (29). One notable toxicity of anilines is methemoglobinemia, and the formation of methemoglobin (which is the hemoglobin-iron complex oxidized to FelI1) in blood is easily measured (28). The production of I1and the formation of methemoglobin were therefore measured in mice following dosing with a series of analogs of sulofenur. While mice are not the most sensitive species toward methemoglobin formation after treatment with anilines, they are the species in which the sulfonylureas were evaluated for oncolytic potential (1-3),and, because of this, toxicology (18)and metabolism (10)work has been carried out in this species. The results in Figure 2 illustrate the correlation of these two properties. Because the sulfonylureas listed in Table I1 differ only in the arylsulfonyl portion of their structure, the variations in this domain are presumably responsible for the great differences in the amount of I1 excreted (and, by implication, in the degree of methemoglobin formation). I t was therefore of interest to see whether a QSAR analysis could identify particular physical properties of substituents on the arylsulfonamide ring which might control the production of 11. In this analysis (Table I11and Figure 3) it was evident that only the size of the arylsulfonamide substituent was clearly correlated with the level of I1 measured, whereas there was no or only a very weak correlation of I1 formation with a number of electronic and hydrophobic parameters. Such a correlation was also discovered in an independent preliminary study of the metabolism of these sulf~nylureas.~ In considering these possible correlations, it should be noted that the response measured here (metabolic formation of 11) does involve more variables than those involved in a more directly measured response such as receptor binding, which is typically used in QSAR work. In particular, extensive (or at least similar) oral absorption of these compounds and rates of metabolism or elimination which result in half-lives on the order of 24 h or more (so that the compound is available in vivo for metabolism leading to the formation of 11)are necessary for consistent results. While the evidence, where available [sulofenur has been extensively studied (lo), and the plasma levels 111, XI, and XXIII are similar to those of sulofenur in mice, with similarly long half-lives of >24 hl], supports the assumption that these variables are similar for these sulfonylureas, such evidence is available for only a few of these sulfonylureas and is not presented here. Instead, a positive correlation, observed in this study with the size of the arylsulfonamide substituent, itself suggests that such a parameter is a key factor in the formation of I1 from sulfonylureas in vivo, just as it is clear that lack of James Wikel, Lilly Research Laboratories, personal communication.

Ehlhardt et al.

a perfect correlation (here R2= 0.733) indicates that other variables, possibly including oral absorption or half-life, must also affect this metabolism. The size of a substituent on a rigid aromatic ring usually has effects only locally in the molecule, as opposed to electronic effects, which can be transmitted through the structure to some distance. The point of attachment of substituents in these sulfonylureas is remote from the presumed site of cleavage leading to I1 (viap-chloroaniline), yet there is a correlation between the size of the group and the rate of cleavage. This suggests that the whole sulfonylurea molecule interacts during this cleavage with another entity at least as large as itself, perhaps considerably larger, through which the effect is communicated. In addition, the effect of substituent size is much stronger at certain positions on the aromatic ring, indicating the interaction is fairly stereospecific. Whether this entity is an enzyme or other macromoleculeis unclear and unknown at present (see below), but seems plausible on the basis of this analysis. The negative results with the electronic parameters also argue that in vivo cleavage occurs through a mechanism other than simple, uncatalyzed chemical hydrolysis. As discussed above, the lability of these sulfonylureas in solution is pH dependent, with much greater decomposition occurring around or below their pK, (ca. 5-6), suggesting it is the neutral form of the molecule which is labile. At any given pH, the relative proportions of the neutral and ionized forms of each sulfonylurea analog, and hence its relative rate of decomposition, would be determined by its PKa. Among the substituent properties examined, only the electronic effect of a substituent would be expected to have a significant influence on the pK, of the sulfonylurea linkage. The lack of correlation between the formation of I1 and the electronic parameters is thus supported by a similar lack of such correlation with the pKa estimates listed in Table 11. If pKa is not a factor in the cleavage of these compounds in vivo, this in turn supports a mechanism different from that seen in solution. This is also suggested by the previous finding that sulofenur does not decomposesignificantly during its passage through the stomach (IO), where it would presumably experience the greatest acidity, and hence potentiallability, of any site in the body. When the substituent effects on this ring were analyzed relative to antitumor activity ( I ) , it was found that the hydrophobicity ( T I , rather than the size of the substituents, correlated with activity. The desirable medicinal properties of this series and the propensity toward an undesirable metabolic pathway therefore appear to be functions of distinct physical properties of the substituents on the arylsulfonamide ring. Since in previous work it was found that any change in the sulfonylurea bridge resulted in almost complete loss of antitumor activity, the presence of a p-chloroaniline moiety or similar aniline moiety in that position was deemed essential for antitumor activity (1). The segregation of antitumor and metabolism correlations found in this study may however provide the basis for future design of analogs in which the undesirable metabolism is minimized relative to the desired activity. While the metabolic formation of I1appears to correlate well with the size of the arylsulfonamide substituent in this series of sulfonylureas and might therefore infer an enzymatic process, no conditions for the reproducible

Metabolism and Toxicity of Sulofenur

formation of p-chloroaniline from sulofenur, 111, XI, and related compounds in vitro at physiologic pH have yet been identified. Conditions which failed to produce either p-chloroaniline or I1 in incubations of these sulfonylureas included incubations in various C3H mouse tissue homogenates (including liver, kidney, heart, brain, spleen, pancreas, lung, skeletal muscle, thymus, plasma, whole blood, and 6C3HED lymphosarcoma tumor issue), liver and kidney cytosolic fractions, anaerobically cultured fecal pellets, or hydrolytic enzymes (such as aryl acyl amidase, aryl amidase, papain, trypsin, chymotrypsin, alanine aminopeptidase, elastase, and urease).' Since there is no direct evidence for a process forming p-chloroaniline from these sulfonylureas, one possible alternative route of metabolism should be mentioned, in which oxidation of the sulfonamide or anilide nitrogen could perhaps give an unstable sulfonyl-N(or N')-hydroxyurea. While there was no evidence for such a process in the experiments described above, and sulofenur was also found to be stable at 37 "C as a 2 % DMSO solution in 30 9% hydrogen peroxide for at least 24 h (also with horseradish peroxidase in 5 m M peroxide), such oxidations of various acetanilides are known (31,321,so such an alternate route cannot be ruled out. The results of these experiments suggest that diarylsulfonylureas such as sulofenur are metabolically cleaved, resulting in the release of anilines such as p-chloroaniline. The correlations described above also support the interpretation that the p-chloroaniline so liberated is responsible for the methemoglobinemia and hemolytic anemia seen as the dose-limiting toxicities of sulofenur, as these are known toxicities of anilines. A more detailed characterization of the anemia found with sulofenur, including direct comparison with the toxicities of p-chloroaniline itself, might directly confirm that this aniline is the root cause of the observed sulofenur toxicity.

References Howbert, J. J., Grossman, C. S., Crowell,T. A., Rieder, B. J., Harper, R. W., Kramer, K. E., Tao, E. V., Aikens, J., Poore, G. A., Rinzel, S. M., Grindey, G. B., Shaw, W. N., and Todd, G. C. (1990)Novel agents effective against solid tumors: the diarylsulfonylureas. Synthesis,activities, and analysis of quantitative structure-activity relationships. J. Med. Chem. 33,2393-2407. Grindey, G. B., Boder, G. B., Harper, R. W., Howbert, J. J., Poore, G. A., Rieder, B. J., Shaw, W. N., Todd, G. C., and Worzalla, J. F. (1986) Identification of diarylsulfonylureas as novel anticancer agents. Proc. Am. Assoc. Cancer Res. 27, 277. Grindey, G. B., Boder, G. B., Grossman, C. S., Howbert, J. J., Poore, G. A., Shaw, W. N., Todd, G. C., and Worzalla, J. F. (1987)Further development of diarylsulfonylureas as novel anticancer drugs. Proc. Am. Assoc. Cancer Res. 28, 309. Taylor, C. W., Alberta, D. S., Ketcham, M. A., Satterlee, W. G., Holdsworth, M. T., Plezia, P. M., Peng, Y.-M., McCloskey, T. M., Roe, D. J., Hamilton, M., and Salmon, S. E. (1989) Clinical pharmacology of a novel diarylsulfonylurea anticancer agent. J. Clin. Oncol. 7, 1733-1740. Hainsworth, J. D.,Hande, K. R., Satterlee, W. G., Kuttesch, J., Johnson, D.H., Grindey, G. B., Jackson, L. E., and Greco, F. A. (1989)Phase I clinical study of N-[(4-chlorophenyl)amino]carbonyl2,3-dihydro-lH-indene-5-sulfonamide (LY186641). Cancer Res. 49, 5217-5220. Levine, R. (1984)Sulfonylureas: background and development of the field. Diabetes Care 7 (Suppl. I), 3-7.

Chem. Res. Toxicol., Vol. 5, No. 5, 1992 673 Balant, L. (1981) Clinical pharmacokinetics of sulphonylurea hypoglycemic drugs. Clin. Parmacokinet. 6,215-241. LaRossa, R. A.,and Schloss, J. V. (1984)The sulfonylurea herbicide sulfometuron is an extremely potent and selective inhibitor of acetolactate synthase in Salmonella typhimurium. J.Biol. Chem. 259,8753-8757. Houghton, P. J., Houghton, J. A., Myers, L., Cheshire, P., Howbert, J. J., and Grindey, G. B. (1989)Evaluation of N-(5-indanylsulfonyl)N'-(4-chloropheny1)ureaagainst xenografts of pediatric rhabdomyosarcoma. Cancer Chemother. Pharmacol. 26,84-88. Ehlhardt, W. J. (1991)Metabolism and dispositionof theanticancer agent sulofenur in rat, mouse, monkey and human. Drug Metab. Dispos. 19,370-375. Wittenhagen, G., Mohnike, G., and Langenbeck, W. (1959)Beitrige zu Abbau und Ausscheidung von N-[4-methyl-benzosulfonyl]-N'butyl-harnstoff (D 860). Hoppe-Seyler's Z.Physiol. Chem. 316, 157-163. Thomas, R. C., and Ikeda, G. J. (1966)The metabolic of tolbutamide in man and in the rat. J. Med. Chem. 9,507-510. Lin, C., Foester, R., Lim, J., Hsu, T., Morton, J. B., and Symchowicz, S. (1978)Absorption, metabolism, and excretion of 14C-tosifen in the dog and rat. Drug Metab. Dispos. 6,50-58. Scholz, J., and HHussler, A. (1964) Metabolic fate of several sulphonylureas in different species. Proc. Eur. SOC. Study Drug Toxic. 4, 23-29. Back, D.J., and Orme, M. L. (1989)Genetic factors influencing the metabolism of tolbutamide. Pharmacol. Ther. 44, 147-155. Thomas, R. C.,and Judy, R. W. (1972)Metabolic fate of chlorpropamide in man and in the rat. J. Med. Chem. 15,964-968. Ferguson, D. T., Schehl, S. E., Hageman, L. H., Lipone, G. E., and Carraro, G. A. (1985)DPX-L5300 a new cereal herbicide. Proceedings-British Crop Protection Conference-Weeds, Vol. 1,pp 43-48, BCPC Publications, Surrey, U.K. Todd, G. C. LY186641 toxicology data from IND filing no. 29,309. Lilly Research Laboratories, Department of Pathology, Greenfield IN 46285. Bus, J. S.,and Popp, J. A. (1987)Perspectives on the mechanism of action of the splenic toxicity of aniline and structurally-related compounds. Food Chem. Toxicol. 25,619-626. Harrison, J. H., and Jollow, D.J. (1986)Role of aniline metabolites in aniline-induced hemolytic anemia. J. Pharmacol. Exp. Ther. 238, 1045-1054. Grossman, S. J., and Jollow, D. J. (1988)Role of dapsone hydroxylamine in dapsone-inducedhemolytic anemia. J. Pharmacol. Exp. Ther. 244, 118-125. Ponte, C. D.,Lewis, M. J., and Rogers, J. S. (1989)Heinz-body hemolytic anemia associated with phenazopyridine and sulfonamide. DICP, Ann. Pharmacother. 23, 140-142. Rieder, M. J., Uetrecht, J., Shear, N. H., and Spielberg, S. P. (1988) Synthesis and in uitro toxicity of hydroxylamine metabolites of sulfonamides. J. Pharmacol. Exp. Ther. 244, 724-728. Cribb, A. E., Miller, M., Tesoro, A., and Spielberg, S. P. (1990) Peroxidase-dependentoxidation of sulfonamides by monocytes and neutrophils from humans and dogs. Mol. Pharmacol. 38,744-751. Kiese, M. (1974)Methemoglobinemia: A comprehensive treatise. CRC Press, Cleveland, OH. Ehlhardt, W. J.,and Howbert, J. J. (1991)Metabolismanddisposition of p-chloroaniline in rat, mouse and monkey. Drug Metab. Dispos. 19,366-369. Boyland, E., Manson, D., and Sims, P. (1956)The preparation of o-aminophenylsulfates. J. Chem. SOC. 1956,3623-3628. Tietz, N. W. (1986)Textbook of Clinical Chemistry, pp 1536-1540, W. B. Saunders, Philadelphia, PA. Eadsforth, C. V., Conveney, P. C., Hutson, D. H., Logan, C. J., and Samuel, A. J. (1986)The metabolism of o-fluoroaniline by rata, rabbits and marmosets. Xenobiotica 16,555-566. Hansch, C., Leo, A., Unger, S. H., Kim, K. H., Nikaitani, D.,and Lien, E. J. (1973)'Aromatic" substituent constants for structureactivity correlations. J. Med. Chem. 16, 1207-1212. Fries, W., Kiese, M., and Lenk, W. (1973)Oxidation of polycyclic N-arylacetamides to glycolamides and hydroxamic acids in rabbits. Xenobiotica 3,525-540. Lim, L. O.,Corbett, B. R., and Corbett, M. D. (1987)Irreversible inhibition of the cytosolic metabolism of N-hydroxy-2-acetylaminofluorene by its glycolyl analog. Cancer Lett. 37,205-211.