N-Alkylprotoporphyrin Formation and Hepatic Porphyria in Dogs After Administration of a New Antiepileptic Drug Candidate: Mechanism and Species Specificity

Abstract

A new antiepileptic synaptic vesicle 2a (SV2a) ligand drug candidate was tested in 4-week oral toxicity studies in rat and dog. Brown pigment inclusions were found in the liver of high-dose dogs. The morphology of the deposits and the accompanying liver changes (increased plasma liver enzymes, increased total hepatic porphyrin level, decreased liver ferrochelatase activity, combined induction, and inactivation of cytochrome P-450 CYP2B11) suggested disruption of the heme biosynthetic cascade. None of these changes was seen in rat although this species was exposed to higher parent drug levels. Toxicokinetic analysis and in vitro metabolism assays in hepatocytes showed that dog is more prone to oxidize the drug candidate than rat. Mass spectrometry analysis of liver samples from treated dogs revealed an N-alkylprotoporphyrin adduct. The elucidation of its chemical structure suggested that the drug transforms into a reactive metabolite which is structurally related to a known reference porphyrogenic agent allylisopropylacetamide. That particular metabolite, primarily produced in dog but neither in rat nor in human, has the potential to alkylate the prosthetic heme of CYP. Overall, the data suggested that the drug candidate should not be porphyrogenic in human. This case study further exemplifies the species variability in the susceptibility to drug-induced porphyria.

Porphyria is a group of heritable metabolic disorders where the enzymes of the heme biosynthetic pathway are partially or totally deficient (Balwani and Desnick, 2012; Hift et al., 2011; Nordmann and Puy, 2002; Sassa, 2006; Sassa and Kappas, 2000; Thadani et al., 2000). As a consequence, porphyrins and heme precursors may accumulate to toxic levels. Their increased concentration in excreta confirms the diagnosis of porphyria and helps in distinguishing among the different forms of the disease. Depending on the location and the nature of the enzyme deficiency, porphyrias are classified as hepatic or erythropoietic (Poh-Fitzpatrick, 1998). Hepatic porphyrias are characterized by neurovisceral complications whereas erythropoietic porphyrias manifest as photocutaneous problems. In addition, hepatic porphyria may lead to hepatobiliary impairment as porphyrin precursors precipitate in the canaliculi (Bloomer, 1998) and/or impair canalicular secretory function (Meerman, 2000).

Most of the gene carriers of inherited porphyrias remain clinically asymptomatic. The disease mostly develops when the genetic defect is combined with precipitating factors such as infection, surgery, smoking, hormonal status, diet, and exposure to some drugs with cytochrome P-450 (CYP) inducing properties (Downey, 1999; Smith and De, 1980; Thadani et al., 2000). So far, very few xenobiotics were found to be porphyrogenic in nongene carrier individuals (Smith and De, 1980). This contrasts with laboratory animals where the disease can be induced by a number of chemicals. They include the herbicide 2-[1-(ethoxyimino)propyl]-3-hydroxy-5-(2,4,6-trimethylphenylcyclohex-2-enone (Brady et al., 1993), the allyl-substituted barbiturates secobarbital, sedormid, and the chemically related allylisopropylacetamide (AIA) (Ortiz de Montellano et al., 1984), griseofulvin (De Matteis et al., 1991; Holley et al., 1991), hexachlorobenzene (Wainstok de et al., 1989), dihydropyridine calcium antagonists (Schoenfeld et al., 1985), the antiarthritic sydnone 3-[(arylthio)ethyl]sydnone (TTMS) (McNamee and Marks, 1996; Sutherland et al., 1986), and the garlic derivative diallyl sulfone (Black et al., 2006). The mechanism of induced protoporphyria in laboratory animals involves mechanism-based inactivation of CYPs (De Matteis and Marks, 1996; Gamble et al., 2000; Marks et al., 1988; Marks et al.,1989). Reactive metabolites bind to the prosthetic heme of CYP through N-alkylation of one of the pyrrole moieties. The alkylated heme then dissociates from the apoprotein with the iron atom being liberated, yielding N-alkylprotoporphyrin IX (N-alkylPP). Once released, some N-alkylPPs have the potential to inhibit ferrochelatase, the terminal enzyme of the heme biosynthetic pathway. After loss of alkylated heme, CYP apoprotein reconstitutes with fresh heme. This phenomenon and the impaired heme biosynthesis are responsible for free heme pool depletion, which in turn stimulates an up-regulation of δ-aminolevulinic acid synthase (ALAS). As a result of all these changes, protoporphyrin IX accumulates in tissues, as characteristic dark pigments.

The present work describes a new synaptic vesicle protein 2a (SV2a) antiepileptic drug candidate that was tested in 4-week oral toxicity studies in rat and dog. Brown pigments were found in the liver of dogs treated with 200 mg/kg/day of the compound. The pigment's characteristic red birefringence under polarized light, with occasional “Maltese cross” configurations, suggested accumulated hepatic porphyrin. These findings were accompanied by increased plasma liver enzymes. Although exposed to much higher drug levels, rats did not show any liver toxicity. Assays were conducted to explore the mechanisms underlying the dog liver findings and evaluate their clinical relevance.

MATERIALS AND METHODS

Chemicals

The SV2a antiepileptic drug candidate evaluated in the present study is a pyrrolidone derivative 2-[4-(2-chloro-2,2-difluoroethyl)-2-oxopyrrolidin-1-yl]butanamide (compound 1, MW: 268.69, Fig. 1) chemically related to levetiracetam (Carreno, 2007). The [¹⁴C]-labeled analog (53.8 mCi/mmol, 99% radiochemical purity) was synthesized by Amersham (Buckinghamshire, UK). Two metabolites were synthesized as reference standards, the carboxylic acid metabolite (compound 2) obtained from the hydrolysis of the acetamide side chain and a β-hydroxylated metabolite (compound 3). All the other chemicals were of analytical grade and were purchased from different commercial sources.

FIG. 1.

Structure of the SV2a antiepileptic drug candidate investigated in the present work (referred as compound 1) and its two major metabolites (compounds 2 and 3). Also shown are the chemically related compounds discussed in the article.

WEC culture medium refers to Williams’ E medium with Glutamax-I containing 100 IU/ml penicillin, 100 μg/ml streptomycin and 10% (vol/vol) fetal bovine serum. WBL medium refers to Williams’ E medium with Glutamax-I containing 100 IU/ml penicillin, 100 μg/ml streptomycin, 4 μg/ml bovine insulin, and 50μM hydrocortisone hemissuccinate.

Recombinant dog CYPs coexpressed with dog NADPH-cytochrome P450 reductase in Escherichia coli (Bactosomes) were purchased from Cypex (Dundee, UK). Dog liver microsomes (pool of 11 male beagle dogs) were obtained from Xenotech (Kansas City).

Animal treatment

The study design has been approved by the ethical committee and complied with the animal health and welfare guidelines. Beagle dogs (7–8-month-old, Harlan SARL, Gannat, France) were housed under a 12-h light/dark cycle in individual pens with free access to water. A fixed portion of pelleted diet (300 g/dog/day; Diet 125 C3, Safe) was given at least 1 h before the daily administration of the compound 1. Four males and four females per group received a daily oral administration of compound 1 at the dose levels of 0, 20, 60, and 200 mg/kg/day for 4 weeks. Four additional animals (two males and two females) were included in the control and high dose groups and were kept for a 2-week treatment-free period to examine the reversibility of any observed findings. The test compound was delivered by oral gavage as a suspension in 1% (wt/vol) methylcellulose. Wistar rats Crl:WI (Glx/Brl/Han, 6–7 weeks old) were supplied by Charles River Laboratories (l'Arbresle, France). At initiation of treatment, rats were 6–7 weeks old with free access to water and SDS RM1 (E) SQC controlled food. Rats (12 males and 12 females per group) received a daily oral administration of compound 1 at the dose levels of 0, 100, 300, and 1000 mg/kg/day for 4 weeks. Additional groups of animals were included to monitor recovery after 2 weeks and to measure plasma levels (referred to as satellite animals for toxicokinetics). Compound 1 was delivered by oral gavage as a suspension in 1% (wt/vol) methylcellulose.

Clinical and laboratory investigations

Animals were examined for clinical signs, body weight, food consumption, electrocardiogram (dogs only), and ophthalmoscopy. Blood, plasma, and urine samples for routine hematology and clinical chemistry parameters were obtained during the predosing period, at week 4 and at the end of the recovery period. Selected organs were collected at necropsy, weighed, and macroscopically examined. A portion of the liver was taken, washed with ice-cold saline, flash-frozen in liquid nitrogen and stored at −80°C until further analysis (i.e., CYP activities, ferrochelatase, porphyrins, and N-alkylPP). The remaining tissues were fixed and processed for histopathological examination. Plasma samples for toxicokinetic analysis were collected from main group (dog) or satellite animals (rat) after single dose on day 1 and on day 28.

Toxicokinetic analysis

Unchanged compound 1 and two metabolites (compounds 2 and 3) were quantified in dog and rat plasma using an LC-MS/MS method validated in a concentration range of 1.0–1000 ng/ml (Shah et al., 2000). Plasma samples were mixed with internal standard and loaded on solid phase extraction cartridges (Waters Oasis HLB, 30 mg, 1 ml, 96-well plate format). After loading, wells were washed and eluted with 500 μl of acetonitrile (ACN):water 80:20 (vol/vol) followed by 500 μl of ACN. The extracts were evaporated and reconstituted in 100 μl of mobile phase A. Analytes were quantified by selected reaction monitoring using a Micromass Quattro Premier mass spectrometer coupled to a Waters ACQUITY Ultra Performance Liquid Chromatography (UPLC) system with a Varian Polaris C18-column (3 μm, 50 × 2 mm) protected by a guard column Waters XBridge C18 (3.5 μm, 10 × 2.1 mm). Mobile phase A consisted of 95:5 water:ACN and mobile phase B consisted of 95:5 ACN:water, both containing 0.1% formic acid at pH 2.8. The flow was 0.4 ml/min at 35°C with a starting condition of 0% B, followed by a linear gradient of 15% B at 1.3 min, 25% B at 3.45 min, held for 1.05 min, and increased to 100% B, held for 1.35 min, and re-equlibration at the starting conditions from 5.25 to 6.5 min.

CYP marker activities

Liver samples were thawed, homogenized, and subjected to differential centrifugation to isolate the microsomal fraction (Levin et al., 1972). CYP marker activities were measured according to standard methods (Dutton and Parkinson, 1989; Pearce et al., 1996). 7-Ethoxyresorufin O-dealkylation (EROD), 7-benzyloxyresorufin O-dealkylation (BROD), and testosterone 6β-hydroxylation (6β-OHT) were measured in dog liver microsomes as markers of CYP1A1/2, CYP2B11, and CYP3A12, respectively. 7-Pentoxyresurufin O-dealkylation (PROD), testosterone 2α-hydroxylation (2α-OHT), and 6β-OHT were measured in rat liver microsomes as markers for CYP2B1/2, CYP2C11, and CYP3A1/2, respectively. Activities were expressed per mg microsomal protein. Levels in treated animals were compared with those in corresponding vehicle controls with data expressed as fold-induction over controls. In addition, pooled liver microsomes from treated dogs were analyzed for CYP2B content by Western blotting.

Hepatic porphyrin levels and ferrochelatase activity

Dog and rat liver samples (control and high-dose groups) were thawed, sonicated in 0.9% NaCl for 3 × 12 s and analyzed for total porphyrin levels using a spectrofluorimetric method (Poh-Fitzpatrick and Lamola, 1976). Data are reported in nmol protoporphyrin/g protein. In addition, the different porphyrins were separated and analyzed using HPLC with spectrofluorimetric detection (Rossi and Curnow, 1986). Ferrochelatase activity was quantified in dog liver samples on the basis of mesoporphyrin-zinc formation using a spectrofluorimetric method (Gouya et al., 2006). Data are reported in nmol of formed mesoporphyrin-Zn/mg protein h.

Hepatic N-alkylPP detection

Liver samples (500 mg) were homogenized and extracted three times with 500 μl of 0.01M HCl and 500 μl of ACN/dimethylsulfoxide (DMSO) (4/1, vol/vol). The insoluble material was removed by centrifugation at 10,000 rpm for 10 min. The supernatant fractions were collected, pooled, evaporated to half the original volume under nitrogen at 30°C, and then centrifuged at 3000 rpm for 30 min. Supernatants (10 μl) were injected into a Waters ACQUITY UPLC system coupled to a Waters Synapt G1 High Definition MS Q-TOF system. UPLC separation was achieved on a Waters HSS T3 C18 column (100 × 2.1 mm, 1.8 μm) with the column temperature set at 30°C and a flow rate of 300 μl/min. The mobile phase consisted of (A) water and (B) acetonitrile, both containing 0.1% formic acid. Elution started with 10% B for 1 min then the proportion of B was increased linearly to 90% in 14 min and held for 2 min. Total run time, including conditioning of the column to the initial conditions, was 20 min. For mass spectrometer, the capillary voltage was set to 3.0 kV, sampling cone 25 V, extraction cone 3.0 V, source temperature 120°C, and desolvation temperature 350°C. Nitrogen was used as desolvation and cone gas with the flow rate of 750 and 50 l/h. Compounds were analyzed in positive mode, data were collected in the V mode, with a scan accumulation time of 1 s and between m/z 100 and 1000. Collision induced dissociation (CID) was induced by applying the trap collision energy of 25 V and a transfer collision energy of 15 V. For accurate mass measurement, data were centroided during acquisition using an external reference (LockSpray) consisting of a 0.4 μg/ml of leucine enkephalin (Tyr-Gly-Gly-Phe-Leu) in 50:50 water:acetonitrile with 0.1% formic acid infused at a flow rate of 5 μl/min, generating a reference ion for positive ion mode ([M+H]⁺ m/z 556.2771) sampled at 10 s intervals.

In vitro metabolism in hepatocytes

Hepatocytes were prepared from 200 to 260 g male Wistar Hannover [Crl:WI (Han)] rats using a modification of Seglen's two-step collagenase perfusion technique (Seglen, 1976). Cells were seeded in WEC medium in multiwell plates precoated with a single film of collagen and then allowed to adhere for 3–4 h at 37°C in 5% CO₂:95% air humidified atmosphere. Plated fresh dog and human hepatocyte monolayers were obtained from Biopredic International. After the adherence period (rat) or at reception (dog and human), hepatocytes were transferred into WBL medium for an overnight incubation. Hepatocytes were incubated with 50μM [¹⁴C]-labeled compound 1 (ca. 3 μCi/ml) for 13 h at 37°C. The reaction was stopped by adding ice-cold ACN and the supernatant collected, diluted 1:1 with water and analyzed by radio-HPLC. The radio-HPLC system was constituted of an Agilent 1100 series coupled to a Packard Radiomatic 515 TR radiochemical detector. The column was a Waters Atlantis T3 (250 × 4.6 mm—5 μm) protected by a guard column Atlantis T3 (20 × 4.6 mm—5 μm). The flow was adjusted to 1 ml/min at 30°C. The mobile phase consisted of (A) water and (B) acetonitrile, both containing 0.1% formic acid. An AB gradient was run from 10 to 40% B in 40 min, followed by a gradient from 40 to 90% B in 10 min finished by an isocratic period of 10 min at 90% B. The radiochemical detector was fitted with a homogeneous scintillation counting cell of 500 μl. The scintillation cocktail (Ultima Flo M) was fed at 3 ml/min. The update time was 6 s. The counting window was set from 0 to 156 keV. The radioactive components were detected and identified by parallel radio-HPLC-MS (Agilent HP100 LC) system coupled to Micromass QTof-2 tandem mass spectrometer using reference standards when available.

In vitro incubation with recombinant dog CYPs

Compound 1 (1 and 10mM) was incubated for 60 min with 1000 pmol/ml dog CYP bactosomes (i.e., CYP coexpressed with NADPH-CYP reductase in E. coli) expressing CYP3A26, CYP3A12, CYP2D15, CYP2C41, CYP2C21, CYP2B11, or CYP1A1. The incubations were conducted according to the supplier's recommendations. Incubates were examined for N-alkylPP formation and compound 1 metabolism.

In vitro CYP induction in hepatocytes

Dog and human hepatocytes were prepared as described above and incubated in WBL medium for 72 h with compound 1 (1–1000μM). The culture medium was replaced every ca. 24 h. At the end of the incubation, cells were rinsed, harvested, and the microsomal fractions were prepared by differential centrifugation. CYP marker activities were measured according to the standard methods (Dutton and Parkinson, 1989; Faucette et al., 2000; Kanazawa et al., 2004; Pearce et al., 1996). EROD was measured as marker activity of dog CYP1A1/2 and human CYP1A1/2. BROD and bupropion hydroxylation were used as marker activities for dog CYP2B11 and human CYP2B6, respectively. 6β-OHT and midazolam 1′-hydroxylation were measured as marker activities of dog CYP3A12 and human CYP3A4, respectively. Activities were compared with those in corresponding vehicle treated cells with data expressed as fold-induction over controls.

In vitro CYP inhibition

Dog liver microsomes (0.1 mg protein /ml) and recombinant dog CYP2B11 (20 pmol/ml) were incubated with compound 1 at 1–1000μM in 50mM phosphate buffer pH 7.4 with a NADPH regenerating system and 7-benzoxyresorufin (1μM in dog liver microsomes and 0.5μM in recombinant dog CYP2B11) as probe substrate for measurement of CYP2B11 activity. After 5 min (recombinant dog CYP2B11) or 30 min (dog liver microsomes) incubation at 37°C, reaction was stopped by addition of an equivalent volume of ice-cold acetone. The incubation mixture was then centrifuged (5 min, 10,000 × g, 4°C) and supernatant used for determination of resorufin concentration using a fluorimeter (ex. 530 nm, em. 590 nm) and appropriate calibration curve and quality controls. Blank incubations were performed without NADPH regenerating system during the incubation step (NADPH added in the mixture after addition of acetone). The inhibition of dog CYP2B11 was determined by comparing the CYP2B11 activities (expressed as nmol resorufin formed/min/mg protein) in controls incubation and compound 1-treated conditions.

RESULTS

Liver Parameters and Histopathology

When orally delivered to dogs for 4 weeks at 20, 60, or 200 mg/kg/day, compound 1 produced liver histopathological changes consisting of brown pigment deposits and increased incidence of hepatocyte apoptosis (200 mg/kg/day; Fig. 2). The pigments were mostly localized in sinusoids and in Kupffer cell cytoplasm, and to a lower extent in hepatocyte cytoplasm. The deposits were granular to globular, showing red birefringency under polarized light with an irregular pattern, occasionally with “Maltese cross” configurations. These changes were paralleled with increases in plasma alanine aminotransferase (ALT), alkaline phosphatase (ALP), aspartate aminotransferase (AST), gamma glutamyl transferase (GGT) compared with control levels (up to 3.2-, 2.6-, 14-, and 1.7-fold, respectively) and slight elevation in bilirubin (up to +33%). These findings were found in both sexes and were only partially reversible after a 2-week treatment-free period.

FIG. 2.

Liver section of a 200 mg/kg/day treated dog stained with hematoxylin and eosin showing brown pigment deposits and apoptotic hepatocytes. Original magnification objective: ×40.

When orally delivered to rat for 4 weeks, compound 1 did not produce any adverse effects up to the maximum tested dose of 1000 mg/kg/day. The observed changes in the liver were restricted to increased liver weight (maximum 50% increase in relative weight) and centrilobular hepatocellular hypertrophy, both considered as adaptive changes resulting from the induction of drug metabolizing enzymes. Plasma liver enzymes were modestly affected at the top dose with increases of ca. +30% in ALT (both sexes) and +32% in ALP (males only).

Toxicokinetic Determination

Plasma samples from the above described toxicology studies were monitored for parent drug and for the two metabolites, compounds 2 and 3 (Fig. 3). Animals were exposed throughout the day to the three analytes irrespective of the dose. Exposure increased with the dose and showed no gender effect. At the top doses, rats (1000 mg/kg/day) were more exposed than dogs (200 mg/kg/day) to the parent drug (mean AUC_{0–24 h} at week 4 of 2722 μg h/ml vs. 483 μg h/ml) and to the carboxylic acid metabolite compound 2 (55 μg h/ml vs. 28 μg h/ml). On the other hand, the exposure to the β-hydroxylated metabolite compound 3 was found to be lower in rat than in dog (mean AUC_{0–24 h} of 109 μg h/ml vs. 460 μg h/ml). The species difference was even larger considering the metabolite-to-parent drug AUC ratio (0.04 and 0.95, respectively). Exposure to compound 3 increased with treatment duration which was paralleled with a decrease in parent drug and suggested auto-induction of the β-hydroxylation metabolic route. Together, all these findings encouraged further investigations on the role of compound 3 in the liver changes observed in dog.

FIG. 3.

Plasma exposure after 4-week oral dosing with compound 1 to rat and dog. AUC was determined after a single dose (D1, AUC_inf) and after at the end of the treatment (D28, AUC_{0–24 h}) for compound 1 (A), carboxylic metabolite (compound 2, B), and β-hydroxylated metabolite (compound 3, C). Exposure values are means ± SD of three males and three females for rats, and 4–6 males and 4–6 females for dog, with genders combined.

Hepatic Drug Metabolizing Enzymes

Treatment of male and female rats with compound 1 produced a large dose-dependent increase in CYP2B1/2 and CYP3A1/2 activities (up to 39- and 6-fold, respectively) (Fig. 4). Treatment of dogs caused an increase in CYP2B11 marker activities in both sexes (up to 5-fold over controls) and to a much lower extent in CYP3A12 (up to 1.4-fold). Although the CYP3A12 increase appeared to be dose-dependent, the effect on CYP2B11 activity was more complex, being maximal at the mid-dose and returning to mean basal level at the top dose. Treatment had little or no effect on total CYP content or CYP1A1/2 marker activity. Overall, it was concluded that compound 1 was acting as a phenobarbital-like inducer in both species. The decreased CYP2B11 activity in the dogs from the high dose group (when compared with the mid-dose group) was regarded as resulting from metabolism-dependent inactivation of that particular isoform. Indeed, this finding was not accompanied by a parallel decrease in the CYP2B protein. Also, when tested in vitro in dog recombinant CYP2B11 and in dog liver microsomes, compound 1 did not inhibit CYP2B11 marker activity up to 1mM, ruling out direct competitive inhibitory effect.

FIG. 4.

Liver microsomal CYP marker activities after 4-week oral dosing with compound 1. Marker activities in rat (A) include PROD (CYP2B1/2, solid bars), 2α-OHT (CYP2C11, open bars), and 6β-OHT (CYP3A1/2, hatched bars). Marker activities in dog (B) include EROD (CYP1A1/2, open bars), BROD (CYP2B11, solid bars), and 6β-OHT (CYP3A12, hatched bars). Activities were compared with those in control animals. Results are means (+SD) of six males and six females for rat, and four males and four females for dog, with genders combined. Pooled dog liver microsomes were also analyzed for CYP2B content by Western blotting (insert; c: controls).

Hepatic Porphyrin Level and Ferrochelatase Activity

Biochemical investigations were performed to assess potential disruption of the heme biosynthetic pathway and confirm porphyrin accumulation. Total porphyrin level in the liver of high dose treated male and female dogs (200 mg/kg/day) was found to be dramatically increased when compared with control animals (ca. 6000-fold increase) (Fig. 5A). The increase was minor in the high dose rats livers (1000 mg/kg/day; ca. 2-fold increase). Examination of the HPLC chromatograms revealed that the vast majority of the increased porphyrin levels in dog liver was due to the increase in protoporphyrin IX. A subsequent assay showed that hepatic ferrochelatase activity in high dose treated dogs was decreased by ca. 75% (Fig. 5B).

FIG. 5.

Hepatic levels of total porphyrin (A) and ferrochelatase activity (B) after 4-week oral dosing with compound 1. Results are means (+SD) of six male and six female rats (open bars) or four male and four female dogs (solid bars), with genders combined. Treated animals are from the top dose groups (1000 and 200 mg/kg/day for rats and dogs, respectively).

N-alkylPP Detection in Liver Samples

Liver samples from high dose and control animals were processed and analyzed by UPLC-MS/MS to detect any adduct formed. The extracted ion chromatograms of samples from treated dogs revealed a peak with a retention time of ca. 10.1 min corresponding to a protonated molecular ion MH⁺ at m/z 845. The peak was detected in the four treated dogs examined (two males and two females) but was absent in the control animals (Fig. 6 shows results obtained in liver samples of one female treated dog and one female control dog). Similarly, this peak was not detected in any samples of rat liver examined (data not shown). Accurate mass determination for the protonated molecule resulted in experimental values in close agreement with the calculated values for an elementary composition for a MH⁺ ion corresponding to C₄₄H₄₈N₆O₇F₂Cl⁺. In any cases, errors were <0.005 u. The detected compound corresponded to the protonated adduct of protoporphyrin IX, compound 1 in a hydroxylated form. The CID ion mass spectrum of the protonated ion m/z 845 displayed three characteristic fragment ions: m/z 563 corresponding to the protoporphyrin moiety, m/z 283 corresponding to a hydroxylated metabolite of compound 1 and m/z 605 corresponding to the cleavage in α of the amide on the butyramide side-chain (Figs. 7A and 7B). Finally, the isotopic pattern clearly confirmed the presence of a chlorine atom (Figs. 7C and 7D). Taken collectively, the mass spectrometry (MS) data are consistent with the alkylation of one of four pyrrole nitrogens of protoporphyrin IX with an oxidized form of the butyramide side chain of compound 1.

FIG. 6.

Sum of the extracted ion chromatograms (m/z 285, corresponding to the protonated β-hydroxylated metabolite; m/z 563, corresponding to the protonated protoporphyrin IX; m/z 616, corresponding to the protonated heme; m/z 845, corresponding to the protonated adduct between protoporphyrin IX and the β-hydroxylated metabolite) of liver samples after 4-week oral dosing with compound 1 at 200 mg/kg/day for one female treated dog (A) and one female control dog (B). Liver samples were collected at necropsy, extracted, and analyzed by UPLC-MS. The chromatograms were calculated with 50 mDa mass tolerance. The extracted ion chromatograms are displayed on the same intensity scale with the time window between 4 and 12 min as a ×6 zoom in.

FIG. 7.

CID mass spectrum of ion at m/z 845 with exact mass measurement (mass resolution of ∼10,000 FWHM) (A). Structure of one of the possible N-alkylprotoporphyrin IX isomers (depending on which pyrrole nitrogen is alkylated) (B). Experimental (C) and theoretical (D) isotopic distribution pattern for chlorinated N-alkylprotoporphyrin.

In Vitro Metabolism in Hepatocytes

The metabolism of compound 1 was investigated in vitro using rat, dog, and human (3 donors) primary hepatocyte cultures. After 13 h incubation at 50μM, the test compound was only modestly transformed, with dog showing somewhat higher transformation rate than the other species (0.9, 4.4, and 0.5% parent drug transformed in rat, dog, and human hepatocytes, respectively) (Fig. 8). The major metabolite in dog corresponded to the β-hydroxylated derivative compound 3, followed by the acid metabolite compound 2 and two minor peaks. Rat and human hepatocytes produced the same low levels of acid metabolite but the β-hydroxylated metabolite was not quantifiable.

FIG. 8.

Radiochromatograms of rat (A), dog (B), and human (C) hepatocytes after incubation for 13 h with 50μM [¹⁴C]-labeled compound 1. Peaks 1, 2, and 3 correspond to the parent drug (compound 1), the carboxylic acid (compound 2), and the β-hydroxylated metabolite (compound 3), respectively. The radiochromatograms are displayed on the same intensity scale with parent drug as saturated peak.

In Vitro Incubation with cDNA-Expressed Dog CYP Isoforms

Attempts have been made to measure N-alkyl-PP formation following incubation of compound 1 with cDNA-expressed CYPs. No N-alkyl-PP could be detected following 60 min incubation of compound 1 (1 and 10mM) with NADPH-fortified dog CYP1A1, 2B11, 2C21, 2C41, 2D15, 3A12, or 3A26 isoforms. Under these conditions, only CYP2B11 metabolized the test compound significantly, with the β-hydroxylated derivative (compound 3) as the primary metabolite formed (data not shown).

In Vitro CYP Induction in Hepatocytes

When incubated for 72 h with primary cultures of dog hepatocytes, compound 1 dose-dependently induced the CYP2B11 marker activity BROD (Fig. 9). At 100μM, the activity was increased ca. 10-fold over controls. The test compound similarly induced the CYP3A12 probe 6β-OHT (ca. 4-fold at 100μM) but did not affect EROD activity (CYP1A1/2). When tested on human hepatocytes, compound 1 had no effect on EROD (CYP1A1) or midazolam hydroxylase (CYP3A4/5) up to the highest tested concentration of 100μM. Treatment increased bupropion hydroxylase (CYP2B6). However, the effect was of low amplitude with maximal 2.5-fold increase over controls at 100μM, which corresponds to 16% of the positive control response (500μM phenobarbital).

FIG. 9.

In vitro induction of CYP marker activities in hepatocytes incubated with compound 1. Marker activities in dog hepatocytes (A) include EROD (CYP1A1/2, open bars), BROD (CYP2B11, solid bars), and 6β-OHT (CYP3A12, hatched bars). Marker activities in human hepatocytes (B) include EROD (CYP1A1/2, open bars), bupropion hydrolase (CYP2B6, solid bars), and midazolam 1′-hydroxylase (CYP3A4, hatched bars). Activities were compared with those in vehicle-treated cells. Results are means (+SD) of three wells. The positive control omeprazole produced a 20- and a 10-fold induction of CYP1A1 in dog and human hepatocytes, respectively. Phenobarbital produced a 51- and a 16-fold induction of CYP2B in dog and human hepatocytes, respectively. Finally, rifampicin produced a 5- and a 2.4-fold induction of CYP3A in dog and human hepatocytes, respectively.

DISCUSSION

The present study showed that oral administration of a new antiepileptic drug candidate, compound 1, to beagle dogs for 4 weeks elicited the formation of dark deposits in the liver, accompanied by an increased incidence of hepatocyte apoptosis and elevation in plasma liver enzymes. The deposits were demonstrated to result from protoporphyrin IX accumulation. The concomitant decrease in hepatic ferrochelatase activity and evidence of CYP inactivation at the high dose indicated drug-induced hepatic protoporphyria through alkylation of CYP2B11. This hypothesis was further validated by the detection of N-alkylPP in dog liver samples. MS data were consistent with N-alkylation of protoporphyrin IX via an oxidized form of the butyramide side-chain of compound 1.

The structural similarity with AIA, a prototypical porphyrogenic agent (Fig. 1), suggests the involvement of a reactive alkene metabolite of compound 1. The π-bonds in alkenes are known to react with CYPs, alkylating the pyrrole nitrogens, or active site amino acids. Examples include AIA (Bornheim et al., 1987; Wong and Marks, 1999), secobarbital (He et al., 1996a,b), diallyl sulfone (Black et al., 2006), and the 4-ene metabolite of valproic acid (Sadeque et al., 1997). The mechanism underlying porphyrin N-alkylation by terminal olefins was reviewed by B. Meunier (Meunier et al., 2004). It could be hypothesized that the carbon atom C₃ of a butyramide alkene side-chain could react with the oxygen atom of the heme leading to a radical intermediate on C₄. This free radical might then bind to one of the heme nitrogens. Finally, the loss of the iron would lead to N-alkylPP with an alcohol function in C₃ (Fig. 10).

FIG. 10.

Proposed mechanism for compound 1 oxidation, terminal olefin formation, heme N-alkylation, and release of N-alkylprotoporphyrin.

Although not detected either in vitro or in vivo, a terminal alkene metabolite of compound 1 could be potentially formed by direct desaturation of the butyramide moiety (Ortiz de Montellano, 1989). Examples of CYP-catalyzed desaturation reactions include testosterone (Korzekwa et al., 1989), valproic acid (Fisher et al., 1998; Rettie et al., 1995), lauric acid (Guan et al., 1998), indoles including zafirlukast (Skiles and Yost, 1996; Sun et al., 2007), nevirapine (Wen et al., 2009), benidipine (Yoon et al., 2007), capsaicinoids (Reilly and Yost, 2005), and ezlopitant (Obach, 2001). Alternatively, an alkene metabolite could also be potentially formed from the dehydration of the β-hydroxylated metabolite. Dehydration of alcohol functions can be supported by CYP, hydrolase, hydrase, or general acid-catalyzed and has been reported for haloperidol (Fang and Gorrod, 1991), atropine, scopolamine (Wada et al., 1994), and ezlopitant (Obach, 2001). The alkyl side-chain of valproic acid can be transformed to both 4-ene- and 4-hydroxy-valproic acid metabolites (Ortiz de Montellano, 1989; Rettie et al., 1988). The unsaturated metabolite of valproic acid is chemically related to AIA (Baillie, 1988). Not surprisingly, valproic acid has been documented to be porphyrogenic in human (McGuire et al., 1988). Compound 1 might undergo the same two parallel reactions, β-hydroxylation to compound 3 and desaturation of the butyramide (Fig. 10). This latter alkene derivative may not be detectable because of its reactivity or the low levels formed.

Although the structure of N-alkylPP formed in dog was not fully elucidated, it undoubtedly involves the oxidation or activation of the butyramide side-chain, as indicated by MS. This finding is in line with the absence of alkylated protoporphyrin adduct or protoporphyria in rat, a species less prone to oxidize the test compound (e.g., lower levels of β-hydroxylated metabolite both in vitro and in vivo). The in vitro metabolism data also indicate that the oxidation of the butyramide side-chain is low in human. Thus, the porphyria observed in dogs should have limited clinical relevance, if any.

Species variability in drug-induced porphyria is not uncommon and has been reported for other xenobiotics. 1-[4-(3-Acetyl-2,4,6-trimethylphenyl)-2,6-cyclohexanedionyl]-O-ethyl propionaldehyde oxime and ETC produce hepatic protoporphyria in mice, not in rat or hamster (Brady et al., 1993; Frater et al., 1993). A new antipsychotic drug candidate was reported to induce porphyria in dog but not in rat (Greijdanus-van der Putten SW et al., 2005). The authors argued that the drug metabolism differed between the two species with some potential toxicological impact. In some circumstances, strains, and genders might also differ in their response to porphyrogenic agents. Wistar rats are more sensitive to the porphyrogenic activity of hexachlorobenzene than rats from the CHBBTHOM strain (Wainstok de et al., 1989). When compared to females, male rats were demonstrated to form much more N-alkylPP following administration of TTMS or 3,5-diethoxycarbonyl-1,4-dihydro-2,6-dimethyl-4-ethylpyridine (ethylDDC) (Wong et al., 1998). Similarly, male mice accumulated greater amounts of protoporphyrin than females after treatment with griseofulvin (Holley et al., 1990). This gender/strain/species variability in susceptibility is primarily due to differences in CYP, complicating the extrapolation of the animal toxicology data to the human situation. For this reason, some authors proposed in vitro studies with cDNA-expressed single CYP to identify the responsible isoform in animal and look in its human orthologs (Lavigne et al., 2002; McNamee et al., 1997). This approach showed that AIA is prone to alkylate rat CYP1A2 but not its human counterpart. Similarly, TTMS forms N-alkylPP when incubated with rat CYP1A2, 2C6, and 2C11, not with the human orthologs 1A2, 2C9, and 2C19 (Gamble et al., 2002). In the present work, CYP2B11 was the only dog CYP isoform able to oxidize the butyramide side-chain of compound 1. However, no alkylated protoporphyrin adduct could be detected, even by using MS detection and using recombinant dog CYP2B11 at relatively high concentrations. A similar sensitivity issue with N-alkylPP detection has been evoked elsewhere (Wong and Marks, 1999) and might be due to the overall low metabolic clearance of compound 1. Alkylation of CYP2B11 is however likely as highlighted by the inactivation observed in vivo. Of interest, rat CYP2B1 (which has 75% amino-acid identity with dog CYP2B11 (Graves et al., 1990)) is frequently involved in the alkylation and inhibition by alkene-containing drugs such as secobarbital (He et al., 1996a) and AIA (Wong and Marks, 1999). The reason why CYP2B activates compound 1 in dog but not in rat or human remains unknown.

Finally, compound 1 was also found to induce dog CYP2B11 both in vivo and in vitro. This finding is likely to have aggravated the hepatic protoporphyria in that species by further increasing the formation of the CYP2B11-mediated reactive metabolite and N-alkylPP. In addition, the increased CYP synthesis most likely depleted the hepatic heme reserve (as shown in Fig. 6), activating the heme biosynthetic cascade leading to the accumulation of the porphyrin precursors. Of notice, compound 1 did not substantially induce human CYPs in vitro. The only measurable effect was low amplitude (i.e., 2.5-fold over controls, 16% of phenobarbital response) and only observed at concentrations largely exceeding the anticipated therapeutic plasma levels.

In summary, compound 1 is suggested to produce protoporphyria in dog liver through N-alkylPP formation, ferrochelatase inhibition, and CYP induction. Data suggest CYP2B11 could support the formation of a terminal alkene metabolite, chemically related to the reference porphyrogen AIA. These findings were not observed in rat presumably because of difference in metabolism between the two species. Taken collectively, data also indicate compound 1 should not induce protoporphyria nor activate quiescent inherited porphyria in humans. This case study further illustrates how to extrapolate animal data to the human situation with porphyrogenic agents.