Rapid catabolism of 5-fluorouracil in freshly isolated rat hepatocytes as analyzed by high performance liquid chromatography.

The catabolism of 5-fluorouracil (FUra) has been investigated in freshly isolated rat hepatocytes in suspension by a new and highly specific high performance liquid chromatographic methodology. This technique permits rapid and simultaneous quantitation of FUra catabolites-dihydrofluorouracil (FIJI&), a-fluoro-/3-ureidopropionic acid (FUPA), and a-fluoro-b-alanine (FBAL) with FUra nucleosides and nucleotides. Analysis of intracellular and extracellular 3H was evaluated from 1 min to 2 h after exposure of the cells to 30 PM I3HlFUra. FUra is rapidly cleared from the incubation medium with only one-half of the unmetabolized drug remaining after 8 min when the cytocrit is as low as 3.5%. Transport of FUra into the hepatocytes is much slower than its catabolism to FUH, so that unmetabolized FUra is not detected within the cell as early as 1 min after exposure to this agent. Utilizing an initial c3WFUra concentration of 30 CM, intracellular FUH, (the major intracellular catabolite) reaches its peak level of 637.4 85.8 p~ within 11.5 min and subsequently decIines as FUHz leaves the cells and/or is further catabolized to FUPA and FBAL. The peak transmembrane concentration gradient of FUHz (642 -+ 93/1) is attained within 1 min. Low levels of FWA appear in the intracellular and extracellular fluids. A 50-fold transmembrane gradient for FBAL is generated within 3 min, which decreases as FBAL approaches equilibrium across the cell membrane over the next hour. FBAL is the main extracellular product of FUra catabolism after 120 min. Incorporation of 3H from r3H] FUra into RNA and protein is minimal (1% of the total intracellular 3H) and no drug was found to be bound to cellular macromolecules by a minicolumn chromatographic exclusion technique. Total 3H in the cell suspension between 1 min and 2 h is accounted for by the total of intracellular and extracellular FUra and catabolites, indicating that these catabolites are the major products of FUra metabolism by the hepatocyte. These studies indicate that FUra is rapidly transported into and catabolized by hepatocytes and confirm the critical role that the liver plays in the clearance of this fluoropyrimidine.

The catabolism of 5-fluorouracil (FUra) has been investigated in freshly isolated rat hepatocytes in suspension by a new and highly specific high performance liquid chromatographic methodology. This technique permits rapid and simultaneous quantitation of FUra catabolites-dihydrofluorouracil (FIJI&), a-fluoro-/3-ureidopropionic acid (FUPA), and a-fluoro-b-alanine (FBAL) with FUra nucleosides and nucleotides. Analysis of intracellular and extracellular 3H was evaluated from 1 min to 2 h after exposure of the cells to 30 PM I3HlFUra. FUra is rapidly cleared from the incubation medium with only one-half of the unmetabolized drug remaining after 8 min when the cytocrit is as low as 3.5%. Transport of FUra into the hepatocytes is much slower than its catabolism to FUH, so that unmetabolized FUra is not detected within the cell as early as 1 min after exposure to this agent. Utilizing an initial c3WFUra concentration of 30 CM, intracellular FUH, (the major intracellular catabolite) reaches its peak level of 637. 4 85.8 p~ within 11.5 min and subsequently decIines as FUHz leaves the cells and/or is further catabolized to FUPA and FBAL. The peak transmembrane concentration gradient of FUHz (642 -+ 93/1) is attained within 1 min. Low levels of F W A appear in the intracellular and extracellular fluids. A 50-fold transmembrane gradient for FBAL is generated within 3 min, which decreases as FBAL approaches equilibrium across the cell membrane over the next hour. FBAL is the main extracellular product of FUra catabolism after 120 min. Incorporation of 3H from r3H] FUra into RNA and protein is minimal ( ( 1 % of the total intracellular 3H) and no drug was found to be bound to cellular macromolecules by a minicolumn chromatographic exclusion technique. Total 3H in the cell suspension between 1 min and 2 h is accounted for by the total of intracellular and extracellular FUra and catabolites, indicating that these catabolites are the major products of FUra metabolism by the hepatocyte. These studies indicate that FUra is rapidly transported into and catabolized by hepatocytes and confirm the critical role that the liver plays in the clearance of this fluoropyrimidine.
9 To whom reprint requests should be addressed. 5-Fluorouracil, synthesized in 1957 (1) and studied extensively by Heidelberger and co-workers (2)(3)(4), is still today a major agent utilized in the treatment of several malignancies including carcinomas of the breast, gastrointestinal tract, and ovary. The cytotoxic effects of this antimetabolite are based upon two distinct biochemical mechanisms: 1) the conversion of FUra' to 5-ffuoro-2'-deoxyuridine 5'-monophosphate which binds to thymidylate synthetase and thereby inhibits de novo synthesis of dTMP and, hence, DNA synthesis (2,5-7); 2) the inhibition or alteration of RNA maturation and function as a consequence of 5-fluorouridine 5'-triphosphate incorporation into RNA (8,9). Studies over the past 2 decades have clarified the importance of these anabolic pathways; in contrast, knowledge of the catabolic pathway is limited (10). In previous studies on the mechanism of action of fluoropyrimidines, rapid degradation of FUra was indicated both in vivo and in vitro in normal tissues of mice and humans (11)(12)(13) and FUH, has been observed in plasma of patients given FUra (14, 15). The liver appears to be the major site of FUra catabolism (4) with the initial step postulated to be reductive degradation to FUH, followed by conversion to FUPA, a compound readily excreted in the urine (12,13). These two initial reactions utilize the same pathways as uracil and thymine (16-18). In the next proposed catabolic step, FUPA is converted to FBAL, with the release of CO, and presumably NH3 (13). In addition, previous studies indicate the presence of large amounts of radiolabeled urea in urine of patients exposed to [2-I4C]FUra (12,19). This latter finding led to the proposal that a-fluoro-@-guanidinopropionic acid is formed as well as FUPA with subsequent cleavage to urea and COZ (4,12,19).
Uncertainties regarding the pathways and extent of catabolism of FUra have been based, in part, upon the inadequacies of the analytical techniques that have been available. Previous methods for separating the various catabolites of FUra have included thin layer chromatography (20) and low pressure "ion exchange chromatography" (12,13). Both methods are limited by poor resolution; furthermore, these techniques would fail to measure FUHe because of the instability of this compound with the drastic alterations in pH and the length of time required to resolve FUra and its catabolites. The present studies describe a new and highly specific high performance liquid chromatographic method which, by its greater specificity and speed, permits the evaluation of the parent drug and its catabolites, including very unstable compounds such as FUH2, in 25 min.
This new HPLC technique has been utilized to assess liver ' The abbrevations used are: FUra, 5-fluorouracil; FUHz, dihydrofluorouracil; FUPA, a-fluoro-P-ureidopropionic acid; FBAL, a-fluorop-alanine; HPLC, high performance liquid chromatography; GC-MS, gas chromatography-mass spectrometry, 8171 catabolism and anabolism of FUra for the first time in the isolated rat hepatocyte system, a model for the study, at the cellular level, of biosynthetic, catabolic, and transport phenomena in the liver. This system permits analysis of drug-cell interactions within seconds, which is particularly important when there are rapid transport and catabolic events. In addition, the isolated hepatocyte system eliminates complexities of studies with the intact liver or liver slices such as alterations in blood flow, uncertainties about drug concentration at the cell membrane site because of large unstirred extraceklar spaces, and contributions to drug transport and metabolism by other cell types in the liver (ie. hepatic reticuloendothelial cells).
This report analyzes aspects of the kinetics of [6-"H]FUra catabolism in hepatocytes within I min after exposure of the cells to radiolabeled compound. This paper demonstrates (i) rapid catabolism of FUra by the freshly isolated hepatocyte, (ii) rapid release of FUra catabolites into the extracellular compartment, and (iii) insignificant anabolism of FUra by liver cells over an interval of u p to 2 h. These data indicate the crucial role of the liver in the rapid elimination of FUra and, hence, the important contribution of this tissue in determining the interval over which tumor cells are exposed to this drug in vivo.

MATERIALS AND METHODS
Preparation of Hepatocyte Suspension-Studies were performed utilizing rat hepatocytes in suspension isolated from male Sprague-Dawley rats by a modification of the collagenase perfusion technique of Berry and Friend (21) as previously described (22). Cell viability, as determined by trypan blue exclusion, was 90% or greater in these experiments. Hepatocytes were suspended to a final cytocrit of 3.5 to 5% and were incubated at 37 "C in Krebs-Henseleit buffer containing 0.25% gelatin and 10 m~ glucose. pH was maintained at 7.4 b.v passing warmed and humidified 95% 0 2 and 5% COS over the cell suspension.

Incubation Conditions and Extraction of Intracellular '3H-
Throughout the incubation, the hepatocyte suspension was stirred by a Teflon paddle in specifically designed flasks as described previously (23). The experiment was initiated with the addition of sufficient ["HI FUra (15 mCi/mM) to achieve a final concentration of 30 ~L M and portions of the cell suspension (0.5 ml) were layered on 400 pl of inert silicone oil of density 1.2 (24) in 1.5-ml plastic microcentrifuge tubes. The tubes were centrifuged at 15,000 X g in an Eppendorf model 5412 microcentrifuge for 15 s and the cell pellet was immediately frozen in dry ice/acetone. Times of incubation in the text represent the interval between introduction of ['HIFUra to the hepatocyte suspension and initiation of centrifugation. Portions of the extracellular medium (50 PI) were analyzed without further processing using the liquid chromatographic methodology described below. The frozen cell pellet was transferred to a plastic tube immersed in ice and subjected to sonic oscillation in 1 ml of 0.002 M potassium phosphate (pH = 7.4) with a 300 probe sonicator (Artek, Farmingdale, NY) for 30 s to release intracellular "H. The sonicate was centrifuged at 25,000 X g a t 0 "C in a Beckman 521 centrifuge for 15 min to pellet cellular debris. Fifty pl or 100 pi of the supernatant were analyzed by liquid chromatography (described below). Usually, 50-pl portions of the extracellular medium and the cell sonicate (after centrifugation) were analyzed to determine total ,'H in each compartment.
Extracellular and Intracellular Space Determinations-A portion of the hepatocyte suspension was incubated under the same conditions described above and exposed to 0.4 pCi (approximately 104,000 cpm/ml) of [carho~yl-'~C]inulin for 5 to 10 min. Portions of the cell suspension were layered onto oil and centrifuged at 15,000 X g to provide a measure of the extracellular space that accompanies cells in the pellet. This value was obtained from the ratio of the ["C] inulin content in the dry pellet to the inulin concentration in the supernatant. The determination of extracellular space permitted corrections to be made for FUra as well as the individual catabolites present in the extracellular space that accompany the intracellular radiolabeled substances in the cell pellet. Samples of the cell suspension exposed to [14C]inulin and spun in empty preweighted microfuge tubes were used to determine intracellular water volumes. Intracellular water was the difference between the wet and dry weights of the cell pellet less the ['4C]inulin space. This technique has been described in detail previously (23,25,26).
Analysis of Intracellular and Extracellular ['H]FUra Metabolites by High Performance Lqquid Chromatography-A high performance liquid chromatograph (Hewlett-Packard 1084B) was equipped with automatic injector, variable wavelength spectrophotometer, and chromatographic terminal (Hewlett-Packard 79850ALC). All analyses were performed on two reverse phase columns (25 X 0.46 cm) connected in series and packed with 5 pm of Hypersil ODS and 5 pm of Spherisorb ODs, respectively (Brownlee Labs, Santa Clara, CA). Elution was carried out isocratically at 1 ml/ min with 0.005 M tetrabutylammonium hydrogen sulfate and 0.0015 M potassium phosphate buffer (pH = 8). Column temperature was maintained at 25 "C; absorbance was recorded at 200 nm. Eluent from the columns was directed via a low dead volume connection into a LKB 2112 Redirac fraction collector (LKB Instruments, Rockville, MD) and timed fractions of 0.2 or 0.5 ml were collected into miniscintillation vials over 25 min. Total aqueous volume was made up to 0.5 with deionized water. After addition of 5 ml of Triton-based scintillation fluor, radioactivity was measured using a Beckman LS-8000 liquid scintillation counter. A quench correction was made using a standard quench curve with utilization of the external standardization process of this instrument. Under the conditions defined above, retention times (mean of three experiments k S. D.) of the unlabeled markers, FBAL, FUH2, FUPA, and FUra, were 6.71 & 0.15 min, 7.77 f 0.04 min, 11.3 f 0.19 min, and 17.2 rt 0.27 min, respectively. This methodology also completely resolves the nucleosides 5-fluorouridine and 5-fluoro-2'-deoxyuridine with retention times of 31 and 39 min, respectively. The nucleotide pool was strongly retained using this ion pair technique. 5-Fluorouridine 5'-monophosphate, 5-fluorouridine 5'diphosphate, and 5-fluorouridine 5"triphosphate with their deoxy derivatives were eluted with retention times of 54, 57, and 64 min, respectively, using a 10-min linear gradient of methanol from 0 to 50% starting at 40 min. The total radioactivity applied to the columns was recovered for both extracellular and intracellular compartments in 25 rnin (96.6 k 1.55% recovery based upon 48 runs).

GC-MS Analysis-
The identity and purity of the FUHl liquid chromatography peak were confirmed by GC-MS.' This analysis was performed using a Newlett-Packard 5980A GC-MS interfaced to a Hewlett-Packard 5934A data system. This study was carried out in chemical ionization mode using methane as reactant gas (ion source pressure 1 torr, source temperature 200 "C). The ionization energy and emission current were 15 eV and 300 FA, respectively. Helium served as the GC carrier (flow = 30 ml/min) and GC separations were accomplished using a glass column (2.5 m X 3 mm inner diameter) packed with 3% OV-275 on Chromosorb WHP (80 to 100 mesh). The GC oven was programmed at 210 "C. The HPLC fraction with a retention time of 7.76 min was collected, concentrated to dryness at 50 "C under stream of nitrogen, and derivatized using iodopentane.
The residue was dissolved in 100 pl of methanol, and portions of 2 PI were then injected into the GC/MS system. Incorporation. of ' H into RNA and Protein-After disruption of the cell pellet by sonic oscillation, the sonicate was acidified with 10% trichloroacetic acid. The acid precipitate was then separated into RNA and protein fractions by the methodology described previously (27). Radioactivity in each fraction was determined, and the incorporation of radiolabeled drug was expressed in picomoles/pg of RNA and picomoles/mg of protein contained in the trichloroacetic acid precipitate using the orcinol reaction (28) and the technique of Lowry et al. (29), respectively. Binding lo Cellular Macromolecules-After sonic oscillation of the cell pellet and centrifugation of the cellular debris, 100 pl of the supernatant were layered into a minicolumn of Sephadex (2-25 equilibrated with 5 mM sodium phosphate buffer at pH 7 and the minicolumn centrifuged at 1000 X g for 3 min. This procedure, described in detail previously (30), permits separation of the protein-ligand complex, which passes through the minicolumn, from the free ligand which is retained completely within the column.
Stability of FUra Catabolites-The pH stability of nonlabeled FUH, was assessed to determine the optimum conditions for preservation of this compound during the analytical procedures. Nonlabeled standard FUH, was analyzed by GC/MS (see above) and no degradation was observed in FUHl stored at -20 "C for I week when the pH was maintained between 7 and 8. However, rapid and complete C. Aubert, J. P. Sommadossi, and J. P. Cano, manuscript in preparation.
breakdown of the FUH, appeared with higher or lower pH values (data not shown). The stability of ["HIFUra catabolites in both the extracellular medium and in the cell extract was assessed using the HPLC methodology described above in samples stored at pH 7.4 and -20 "C over an interval of 6 h to 1 week (Fig. 1). In the cell extract, there was less than 3% breakdown of the FUH2 in 12 h and a degradation of about 20% in 1 week, while the amount of FBAL increased proportionately. These data indicate the relative instability of FUH2 in the presence of cellular protein and the requirement that analysis be performed within 12 h after the cell extract is obtained. In the extracellular medium, ["HIFUra and its catabolites FUHz, FUPA, and FBAL were found to be stable over 1 week at -20 "C.

RESULTS
HPLCAnalysis of Pura a n d Its Catabolites FUHL, FUPA, a n d FBAL- Fig. 2 represents an HPLC chromatogram demonstrating separation of nonradiolabeled FUra and its presumptive catabolites FBAL, FUH2, and FUPA. This chromatographic technique unambiguously resolves FUra from these metabolic products with R values (mean of three exper-  As the FUH, was demonstrated to be an unstable derivative of FUra (see above), it appeared possible that the peak identified as FUH2 may actually have been a breakdown product of FUH2 degradation. Therefore, the identity of the FUH, peak was evaluated by GC/MS analysis. The FUHz chromatographic peak was derivatized as described under "Materials and Methods" and the analysis of this sample produced the mass spectrum shown in Fig. 3. The dipentylated derivative of FUH2 was identified from its methane chemical ionization labeled catabolites was comparable to that obtained with the nonradiolabeled compounds (Fig. 2). After 4.5 min, the major Within 1 min after the initial exposure of the hepatocyte suspension to ['HIFUra, intracellular FUPA and FBAL can be detected. By 30 min, FBAL has reached a steady state level of 31.5 f 4.9 PM within the intracellular water which is maintained for the remaining 1% h of the experiment. Intracellular FUPA appears transiently, reaching a maximum concentration of 37 k 1.4 ~L M in 5 min; by 25 min, FUPA is no longer detectable within the cell.

Analysis of the Time Course of Disappearance of Extracellular FUra and Appearance
of Extracellular Catabolites-Analysis of extracellular FUra and its catabolites over 1 min to 2 h after exposure of hepatocytes to r3H]FUra is illustrated in Fig. 6. The extracellular level of FUra declines to one-half of the initial level within 8 min and no unchanged drug is detectable after 30 min. As the level of extracellular FUra declines, catabolites of FUra that have been synthesized within the cells appear in the extracellular compartment. Extracellular FUHa gradually increases, reaching a peak level of 10.6 -+ 0.61 p~ at approximately 22 min, and subsequently declines. FBAL also appears rapdily in the medium, reaches levels equivalent to those of FUH2 by 22 min but remains the major extracellular catabolite as the FUH:! level declines. Extracellular FBAL finally reaches a steady state level of 18.5 All data presented represent the mean of three expenments -C S. D. FUra. The inset represents FUPA levels in the same experiment. At the indicated time, portions of cell suspension were separated by centrifugation and the total extracellular 3H assayed by HPLC as described under "Materials and Methods." 2 1.4 p~. Extracellular FUPA is detected within 1 min and then declines very slowly over 2 h. Total 3H added as FUra can be accounted for the sum of the catabolites in the extracellular and intracellular water; this indicates that FUra is converted essentially quantitatively to these catabolites.
Assessment of Intracellular Binding of FUra and Its Catabolites-Binding of intracellular 'H to cell macromolecules was assessed by the minicolumn technique of Fry et al. (30) in which free ligand remains in the column while ligand macromolecular complexes pass through the column during centrifugation. This procedure was carried out using cell extract taken at times ranging from 11 min when FUH:! has reached its maximum concentration up to 2 h when FUH:! has declined and FBAL is at steady state. All radioactivity from the cell extract that was applied to the minicolumn was trapped within the column, indicating that intracellular 'H was not bound to cellular constituents. This represents less than 1% of the total intracellular radiolabel. This, along with the absence of nucleotide derivatives of FUra, indicates that the catabolites formed from FUra are the main components of intracellular 3H and that anabolism is an insignificant factor following exposure of liver cells to FUra under these conditions.

DISCUSSION
These studies provide the first detailed analysis of the rapid formation of FUra catabolites and their disposition in the freshly isolated rat hepatocyte system. This approach was made possible by the development of a new HPLC methodology that permits rapid and simultaneous identification of all FUra catabolites and anabolites. The study of hepatocytes in suspension further permits quantitation of rapid transport and catabolic processes without complexities encountered in studies with the intact liver or liver slices, i.e. the presence of hepatic reticuloendothelial cells, large unstirred extracellular spaces, and changes in blood flow to the whole tissue.
These results clarify a number of key steps in the catabolic pathway of FUra in the rat hepatocyte (Fig. 7). (i) The initial degradation reaction is thought to be the reduction of FUra to FUH2 by dihydrouracil dehydrogenase (4). Low levels (0.8% of the total radioactivity) of FUHz have been detected previously in high speed supernatant fractions of mouse liver exposed to [2-14C]FUra (12). In contrast, under the conditions of these experiments, FUH2 was shown to be the major constituent of intracellular '3H after incubation of the hepatocytes with 30 ~L M r3H]FUra. FUHe achieves a maximal transmembrane chemical gradient of 642 f 93/1 with respect to extracellular FUHz within 1 min, and a maximum intracellular level of 637.4 & 85.8 ~L M is achieved in 22 min. Since no catabolites of FUra appeared to be bound to cellular constituents, the data suggest that FUH2 exits from the cells very slowly relative to the rate at which it is synthesized, and that the rate of FUHn synthesis is much faster than its rate of degradation. Furthermore, because intracellular FUH, never declined to equilibrium levels across the cell membrane (a transmembrane gradient of 22.3 5.6/1 was still present after 2 h of incubation), active transport for this compound into the cell is possible; however, longer intervals of exposure would be required to determine whether, in fact, a steady state gradient for FUH, is actually sustained.
(ii) The second step of FUra catabolism is thought to proceed via degradation of FUH2 to FUPA (4). The FUPA levels in these experiments were low both in the intracellular (37 f 1.4 p M ) and extracellular (1.14 f 0.025 p~) compartments, suggesting that in the liver, FUPA is a transient intermediate in the transformation of FUHz to FBAL. Because FBAL accumulates to a much greater level than FUPA, the data indicate that the rate of FUPA conversion to FBAL is rapid compared to the rate of FUPA formation from FUH2.

FBAL
These findings differ from results published previously which indicated that FUPA was the major catabolite of FUra degradation in liver homogenates (12). A possible explanation for the differences in these studies is that the instability of FUHz did not permit the measurement of this compound by the techniques employed, and the measured levels of FUPA actually represented the sum of FUHz and FUPA.
(iii) Subsequently, FUPA is converted to FBAL (4). In the present studies, FBAL appears within 1 min in the cell water at a level of 8.1 & 1.8 p~, indicating the rapidity of FUra catabolism. A 50-fold transmembrane gradient for FBAL is generated within 3 min, which decreases during the 1 h to approach equilibrium across the cell membrane. The quantitative increase in the formation of FBAL as FUra and the other catabolites decline indicates that FBAL is probably the final catabolite of [B-'H]FUra formed by the hepatocytes. Previous studies (32) also suggested that FBAL is the main extracellular product of FUra catabolism.
Since FUra is labeled in the 6-hydrogen, it has not been possible, as yet, to study the fate of the remainder of the molecule such as the formation of COz or urea (12,13) or possible degradation of FBAL (13,32,33). However, the total radioactivity applied to the HPLC system was recovered for both intracellular and extracellular ' H as FUra and its catabolites FUH2, FUPA, and FBAL so that formation of other catabolites that might involve the 6(C-H) position of the FUra molecule by the liver must be negligible. Heidelberger et al. (12,13) suggested the possible formation of another catabolite, a-fluoro-/3-guanidinopropionic acid, simultaneously with FUPA. However, our studies indicate that this compound is not formed in hepatocytes under these conditions. Absolute verification of the absence of a-fluoro-/3-guanidinopropionic acid would require an a-fluoro-p-guanidinopropionic acid standard to exclude the unlikely possibility that a-fluoro-pguanidinopropionic acid co-elutes with FUra or the other catabolites described herein.
These findings indicate the rapid catabolic conversion of FUra to FUHz, FUPA, and FBAL and insignificant conversion of FUra to nucleosides and nucleotides in liver cells. The data support the critical role of the liver as a major factor in the clearance of FUra from the circulation. As some pharmacokinetic studies indicate that the plasma clearance of FUra exceeds the hepatic blood flow (34, 35), degradation of FUra in other tissues in vivo is possible as well. The isolated hepatocyte system provides an opportunity for a highly quantitative assessment of the transport and metabolic steps involved in FUra degradation by the liver and will be utilized subsequently as a model for exploring how these processes might be modulated to increase the chemotherapeutic effectiveness of this agent,