THE J O U R N A L OF BIOLOGICAL CHEMISTRY Vol. 263, No. 7, Issue of March 5, pp. 3395-3401, 1988 Printed in U.S.A.

The Role of Mitochondrial Matrix Enzymes in the Metabolism and Toxicity of Cysteine Conjugates*

(Received for publication, October 23, 1987)

James L. StevensSQ, Nasib Ayoubill, and Joan D. Robbinsll From the $ W. Alton Jones Cell Science Center, Inc., Lake Placid, New York 12946 and the TCenter for Drugs and Biologics, Food and Drug Administration, Bethesda, Maryland 20892

The submitochondrial localization and identity of enzymes which metabolize cysteine conjugates were investigated. Glutamine transaminase K was purified from rat kidney mitochondrial soluble fraction and was shown to be a cysteine conjugate @-lyase. The purified mitochondrial enzyme is similar to the cyto- solic glutamine transaminase K whose @-lyase activity with S-( l,Z-dichlorovinyl)-~-cysteine (DCVC) is regu- lated by concurrent transamination (Stevens, J. L., Robbins, J. D., and Byrd, R. A. (1986) J. Biol. Chern. 261, 15529-15537). However, @-lyase activity in whole mitochondria is largely independent of regula- tion by cosubstrates for transamination suggesting that factors present in mitochondria are able to support the &lyase activity in the absence of added a-keto acid. Fractionation of mitochondria results in a loss of the independent @-lyase activity. However, the majority of the @-lyase activity can be recovered in the matrix if it is stimulated by the addition of a-keto-y-methiolbutyr- ate.

The data suggest that the regulation of @-elimination by the competing transamination pathway is different for each substrate and that multiple @-lyases may exist in rat kidney. S-(2-Benzothiazolyl)-~-cysteine (BTC) is a poor substrate for purified glutamine transaminase K from mitochondria and cytosol, but BTC is as active as DCVC in crude mitochondrial matrix suggesting that other enzymes may be present. In contrast to DCVC, with BTC as substrate, the @-lyase activity of the purified enzyme and enzyme(s) in the mitochon- drial matrix is largely a-keto acid-independent. The existence of multiple enzymes is also supported by the observation that a-keto acids which are not substrates for purified glutamine transaminase K from mitochon- dria and cytosol do stimulate @-lyase activity in the mitochondrial matrix fraction.

Mitoplasts were found to be sensitive to DCVC tox- icity consistent with the matrix localization of @-lyase activity. The possible role in cysteine conjugate toxic- ity of matrix enzyme regulation by a-keto acids is discussed.

Classically, mercapturate biosynthesis has been considered primarily a detoxication system (1,2). However, recent studies have shown that the mercapturic acid pathway produces toxic species from a variety of drugs and xenobiotics (3-5). The toxicity of some mercapturic acid pathway intermediates de-

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

I To whom correspondence should be addressed.

pends on their degradation to cysteine conjugates which are subsequently activated to toxic species by cysteine conjugate P-lyase, a pyridoxal phosphate (PLP)’-dependent enzyme (Equation 1 (3-19)). This mechanism was first proposed by Schultze and co-workers (6, 7 ) who suggested that S-(1,2- dichloroviny1)-L-cysteine (DCVC) is metabolized by P-elimi- nation (Equation 1) to a reactive thiol which binds covalently to cellular macromolecules. In recent studies this mechanism has been proposed to account for DCVC toxicity in several models (4, 5 , 8, 9, 14, E ) , and a similar mechanism has been suggested for the toxicity of other cysteine conjugates (4, 5, 12, 14, 18). NH,CH(COOH)CH,SR+ CH,(C=O)COOH + :NH, + RSH (1)

Stonard and Parker (8-10) first demonstrated in uiuo and in oitro that mitochondria might be the cellular target for DCVC. Recent studies of Jones et al. (12,131 with S-(1,2,3,4,4- pentachlorobutadieny1)glutathione and Lash et al. (14, 15) with S-(1,2-dichlorovinyl)glutathione (DCVG) and DCVC showed that mitochondria are a target in isolated rat kidney cells. With both S-(1,2,3,4,4-pentachlorobutadienyl)gluta- thione and DCVG, toxicity is blocked by inhibition of y- glutamyltransferase with AT125 or anthglutin or inhibition of cysteine conjugate @-lyase by aminooxyacetic acid (AOA) (12-15). Their data also indicate that perturbation of mito- chondrial calcium sequestration may be an early event related to toxicity.

However, the available data do not explain a number of questions regarding the mechanism of toxicity and the organ specificity. For example, both DCVG and DCVC are nephro- toxins (11, 16) suggesting that events subsequent to degrada- tion of DCVG by y-glutamyltransferase and cysteinylglycine dipeptidase could be involved in the organ specificity. Fur- thermore, within the kidney the proximal tubule is the target and within the proximal tubule itself the pars recta (S3) segment is most sensitive (11, 18). Nash et al. (18) showed that the binding of radiolabel from hexa~hloro[’~C]butadiene was localized to S3 indicating that more conjugate might be activated there. A more complete understanding of conjugate delivery to the target tissue and the biochemical regulation of their activation in target cells and organelles may explain the factors which regulate the mechanism and specificity for toxicity. In this regard, the study of the segment-specific toxicity of cysteine conjugates may be a useful tool in inves-

The abbreviations used are: PLP, pyridoxal phosphate; MTB, 01-

keto-y-methiolbutyrate; DCVC, S-(1,2-dichlorovinyl)-L-cysteine; DCVG, S-(1,2-dichloroviny1)glutathione; DMOP, S-(l,Z-dichlorovi- nyl)-3-mercapto-2-oxopropionic acid; BTC, S-(Z-benzothiazolyl)-L- cysteine; TFA, trifluoroacetic acid; PMP, pyridoxamine phosphate; mGTK, mitochondrial glutamine transaminase K; cGTK, cytosolic glutamine transaminase K; AOA, aminooxyacetic acid.

3395

3396 Cysteine Conjugate Toxicity

tigating the biochemistry underlying heterogeneity in renal function.

We have been characterizing the cysteine conjugate @- lyase(s) from rat kidney (17, 19) and have made two obser- vations which may be important in understanding the bio- chemical basis for cysteine conjugate toxicity. First, a tissue- specific cytosolic transaminase, glutamine transaminase K, has been purified from rat kidney and shown to be a @-lyase. Second, cytosolic glutamine transaminase K catalyzes trans- amination as well as p-elimination, and DCVC partitions equally between the two pathways. The transamination re- action regulates the rate of @-elimination when an inadequate supply of cosubstrate, i.e. an a-keto acid, is present because the enzyme will accumulate in the pyridoxamine (PMP) form after half-transamination. DCVC cannot form a Schiff base with PMP enzyme, and @-elimination is inhibited. L-Amino acid oxidase (L-a-hydroxyacid oxidase (22)) cooperates with cytosolic glutamine transaminase K to support @-elimination by directly oxidizing DCVC to DMOP which serves as a source of a-keto acid (17).

Rat kidney contains a mitochondrial counterpart of cyto- solic glutamine transaminase K (20) which is located in the matrix (14) but whose properties as a @-lyase have not been studied. In this study, we characterize the submitochondrial localization of enzymes which metabolize DCVC and charac- terized the regulation of mitochondrial enzymes by transam- ination. The role in toxicity of @-lyases from different mito- chondrial compartments is investigated with particular atten- tion to the participation of mitochondrial glutamine trans- aminase K. Previous fractionation experiments by Lash et al. (14) suggested that mitochondrial p-lyase is localized predom- inantly in the outer membrane and that mitochondrial glu- tamine transaminase K is not the mitochondrial @-lyase. Our results differ from those of Lash et al. (14), and the differences are discussed. The possibility that fractionation experiments could be complicated by compartmentalization of factors which support the @-elimination reaction is considered.

EXPERIMENTAL PROCEDURES‘

RESULTS

Metabolism of DCVC and the Covalent Binding of Reactiue Species in Mitochondria and Mitoplasts-Initially, we inves- tigated the metabolism of [“C]DCVC in mitochondria and mitoplasts in the presence and absence of added a-keto-y- methiolbutyrate (MTB) (Table 1). Without MTB added to the incubations, activity decreased upon removal of the mi- tochondrial outer membrane (mitoplasts). However, when MTB was added to the mitoplasts, the specific activity was comparable to that found in mitochondria. @-Lyase activity in mitochondria was stimulated only 60% by MTB while in mitoplasts the activity increased 4-fold (Table 1). The differ- ence in stimulation was due to a loss of activity in the mitoplasts when no MTB was added to the incubations. It appeared that factors which support the @-lyase activity are lost when the outer membrane is removed with digitonin, and those factors can be replaced by MTB. \

The [“C]DCVC assay measures both @-elimination and transamination but does not distinguish between the two pathways (17). However, p-elimination from [”S]DCVC yields a reactive ””S-labeled cleavage fragment which cova-

* Portions of this paper (including “Experimental Procedures,” Tables 5 and 6, and Figs. 1-3) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press.

lently binds to macromolecules (5-7). Experiments with rat kidney amino acid oxidase, which converts DCVC exclusively to DMOP (17), the transamination product of DCVC, showed that DMOP is not reactive and does not bind to macromole- cules (see “Experimental Procedures”). Therefore, the binding of 35S label is a measure of the @-elimination pathway and not transamination. When we measured the binding of ”S label from [”S]DCVC to trichloroacetic acid-insoluble material in mitochondria and mitoplasts, there was little difference in the MTB stimulation of binding, -30% in both cases, or the specific activity (Table 1). Therefore, though total 14C-metab- olites are increased in mitoplasts when MTB is added, the difference is due primarily to an accumulation of [14C]DMOP with little change in the amount of DCVC metabolized by @- elimination. The mechanism of the differential stimulation of metabolism and binding will be addressed further under “Dis- cussion.”

AOA blocks the toxicity of DCVC in mitochondria (14) and other model systems (5, 12-15). AOA inhibited both the binding of 35S label and the metabolism of DCVC. The com- ponent of AOA-insensitive binding (Table 1) was somewhat higher than the value for metabolism. This is probably due to a degree of nonenzymatic breakdown of [”S]DCVC to reactive species.

Submitochondrial Localization of Glutamine Transaminase K and @-Lyase-Since mitoplasts did appear to retain @-lyase activity we investigated the submitochondrial distribution of the activity and compared it to the distribution of mitochon- drial glutamine transaminase K (Table 2). The majority of the P-lyase activity with DCVC and the mitochondrial gluta- mine transaminase K activity is located in the matrix. How- ever, in the absence of MTB, only 42% of the DCVC activity was recovered in all fractions, and only 21% was in the matrix. When MTB was included in the assays, recovery increased to 120% with 71% recovered in the matrix. The matrix DCVC activity was much more dependent on added MTB than the activity in whole mitochondria. Some of the @-lyase (31%) and mitochondrial glutamine transaminase K (24%) activities were recovered in the intermembrane space fraction suggest- ing that some of the mitoplasts ruptured during removal of outer membrane with digitonin. The results indicated that both cysteine conjugate @-lyase activity and mitochondrial glutamine transaminase K activity were located in the matrix fraction and that when mitochondria are fractionated, factors are lost that support the @-lyase activity. These factors can be replaced by the addition of MTB. Repeated attempts to stimulate the activity by recombining the various fractions were unsuccessful (data not shown).

Purification of Glutamine Transaminase K and Character- ization of Its Role as a @-Lyase-Since cytosolic glutamine transaminase K and mitochondrial glutamine transaminase K have similar substrate specificity and physical properties (20, 21), it seemed that the matrix activity might be due to mitochondrial glutamine transaminase. If this were true, then the stimulation of matrix @-lyase activity by a-keto acids should have a structure-activity relationship similar to that seen with cytosolic glutamine transaminase K (17,20,21), i.e. a predilection for hydrophobic 0-keto acids. When a variety of a-keto acids was added to the matrix fraction, phenylpy- ruvate and MTB were most stimulatory, but two of the hydrophilic a-keto acids (a-ketoglutarate and oxaloacetate), which are inactive with cytosolic glutamine transaminase K (see Table V in Ref. 17), also stimulated the matrix activity substantially (Table 3). When MTB was increased from 0.5 to 1.0 mM there was no increase in the stimulation suggesting that 0.5 mM MTB was a saturating concentration. However,

Cysteine Conjugate Toxicity TABLE 1

3397

Effect of a-keto-y-methiolbutyrate on @-lyase activity in mitochondria and mitoplasts Metabolism was determined with [‘‘Cc]DCVC and binding with [“S]DCVC as described under “Experimental

Procedures.” Since the error in the metabolism and binding data was greater between experiments than within an experiment, the specific activities shown are from triplicate determinations on one preparation of mitoplasts or mitochondria ( X & S.D.) and are representative of experiments on three separate preparations. However, the ratio (+MTB/-MTB) is the mean f S.D. of the ratios calculated from three separate preparations ( n = 3). AOA (0.2 mM) controls were not subtracted from the data prior to calculating the ratio (+MTB/-MTB; see text). This concentration of AOA is sufficient to inhibit purified enzyme (17) and mitochondrial metabolism (14) >98%; therefore, the difference between the activity +AOA and -AOA represents the PLP enzyme component of binding and metabolism.

MTB AOA Metabolism +MTB/-MTB Binding +MTB/-MTB nmol. mg” .IO min”

Mitochondria - - 14.0 f 0.8 1.6 & 0.3 7.6 f 0.3 + 1.3 f 0.1

- 28.0 f 2.3 10.0 & 0.8 + 2.7 & 0.5 1.4 f 1.0

+ + 2.4 f 0.6 2.0 f 2.0 -

Mitoplasts - - 8.0 f 1.7 4.0 + 1.2 9.5 f 0.5 1.3 + 0.2 + - 29.0 f 3.1 13.0 * 1.7

+ 0 f 0.9 3.1 f 0.5 + + 2 & 2 1.4 f 0.7 -

TABLE 2 Localization of @-lyase and glutamine transaminase K activity i n mitochondrial subfractions

Mitochondria were fractionated by the method of Greenawalt (26). Glutamine transaminase K was assayed by the method of Cooper and Meister (20), and DCVC metabolism (nmol/lO min/mg) was assayed with the extraction procedure and 1 mM DCVC in the presence and absence of 0.5 mM MTB for 10 min at 37°C (see “Experimental Procedures”). The specific activities shown are in the presence of 0.5 mM MTB. The percent recovery of DCVC activity is expressed with regard to the appropriate mitochondrial value with or without MTB. Therefore, although the mitochondria are shown as 100% in both cases, the value in the absence of MTB is 2-fold less than in the presence of MTB (Table 1). Typically 250-400 mg of protein were used for one fractionation experiment, and the data are the mean f S.D. for four independent fractionations from pooled kidneys of four to eight rats.

Activity with DCVC Glutamine transaminase K

Fraction nmol/ Ratio, Recovery nmol/ Protein recovery

10 min/mg i/- MTB +MTB -MTB 10 min/mg Recovery

% % % Mitochondria 28 f 5 2.2 100 100 122 f 13 100 100 Outer membrane 9 f 1 3.0 5 3 70 f 16 8 Inner membrane

11 f 2 7 f 2 11.0 15 3 24 f 4 9

Matrix 98 f 30 14.0 71 21 474 f 70 65 18 f 6 49 f 8

Intermembrane space 54 f 9 3.4 31 15 186 f 34 24 15 f 6

TABLE 3 Stimulation of DCVC metabolism in mitochondria matrix

Mitochondrial matrix was prepared from 12 rat kidneys as de- scribed by Greenawalt (26) and stored a t -70 “C until used. Metab- olism was determined with DCVC (1 mM) using the extraction assay and an incubation time of 10 min at 37°C (see “Experimental Pro- cedures”). The data are the X & S.D. from triplicate determinations on one preparation of matrix and are representative of experiments from two preparations. Error within an experiment was much less than error between experiments; therefore, one experiment is shown. Experiment 2: no addition = 24.6 nmol/lO min.mg; 0.5 mM MTB = 133 nmol/lO min. mg; 0.5 mM a-ketoglutarate = 88 nmol/lO min. mg; 0.5 mM MTB + 0.5 mM a-ketoglutarate = 200 nmol/lO min. mg.

a-Keto acid mM nrnol/lO min . mg

None 13 f 2 a-Ketoglutarate 0.5 Oxaloacetate

50 f 2 0.5 46 f 3

Pyruvate 0.5 Phenylpyruvate 0.5

26 & 1

MTB 65 f 1

MTB 0.5 1.0

71 f 1 72 f 2

MTB + a-ketoglutarate 0.5 f 0.5 8 4 + 2 MTB + oxaloacetate 0.5 * 0.5 82 f 2

when 0.5 mM MTB was combined with 0.5 mM a-ketoglutar- ate or 0.5 mM oxaloacetate, the activity increased an addi- tional 15% over that seen with MTB alone indicating that the mixtures of a-keto acids had an additive effect. Therefore, although MTB stimulated activity, the structure-activity re- lationship for a-keto acid stimulation was not the same as for the purified cytosolic glutamine transaminase K. In order to characterize the role of mitochondrial glutamine transami- nase K as a P-lyase, we purified the enzyme responsible for MTB-stimulated DCVC activity and assayed the purification of both @-lyase and mitochondrial glutamine transaminase K activity at each step (Table 4). The two activities copurified after the pH 5 precipitation step.

Comparison of Mitochondrial and Cytosolic Glutamine Transaminase K-When the purified rat kidney mitochon- drial enzyme was compared to cytosolic glutamine transami- nase K (17) it was found to be similar in several ways. First, like cytosolic glutamine transaminase K, mitochondrial glu- tamine transaminase K was completely inhibited by 0.1 mM hydroxylamine or aminooxyacetic acid and 50% by potassium cyanide (data not shown), three inhibitors of PLP-dependent enzymes. Second, the P-lyase activity was markedly stimu- lated by the addition of hydrophobic a-keto acids (i.e. MTB and phenylpyruvate) but not by hydrophilic a-keto acids (i.e.

Cysteine Conjugate Toxicity

TABLE 4 Copurification of cysteine conjugate 8-lyase and glutamine transaminase K from rat kidney mitochondria

Purification was done as previous described (17). Glutamine transaminase K activity (transaminase, TRA) was assayed by the method of Cooper and Meister (20), and the cysteine conjugate /%lyase (CBL) activity with [“C] DCVC was assayed in the presence of 0.5 mM MTB using the extraction assay as described under “Experimental Procedures.” The ratio of the glutamine transaminase to cysteine conjugate &lyase activity is shown (TRA/CBL). Protein values were rounded off to the nearest mg, but the recoveries and specific activities were calculated from the protein data expressed to the nearest 0.1 mg.

Step Protein P-Lyase Transaminase TRA/CBL Ratio,

mg amW total

min pmol/lO pmoll

total pmol. 10

min mg. 10 min mg.10 min

Mitochondria

pH 5.0 Supernatant

Ammonium sulfate DEAE Hydroxylapatite Sephacryl S-300 Q-300

4490 1157 822 118 39

7 2 1

0.01 0.03 0.05 0.15 0.46 2.2 2.9 3.9

44.9 34.7 41.1 19.5 17.9 14.5 6.1 3.3

0.09 0.26 0.29 0.95 2.4

10.1 12.0 20.7

382 300 238 112 94 67 25 18

9 9 5.8 6.3 5.2 4.6 4.1 5.3

TABLE 7 Comparison of BTC and DCVC activity

Mitochondrial matrix was prepared by the method of Greenawalt (26) as described under “Experimental Procedures” and stored at -70 ‘C until use. Purified mitochondrial and cytosolic glutamine transaminase K (GTK) were prepared as described under “Experi- mental Procedures.” BTC (2 mM) and DCVC (2 mM) were incubated under the standard assay conditions in the presence (+) or absence (-) of 0.5 mM MTB. The data are the mean & the range from two separate experiments using one matrix preparation from 12 rat kid- neys and one purified mitochondrial or cytosolic glutamine trans- aminase K preparation.

Enzyme MTB BTC DCVC nmolll0 min. mg

Mitochondrial matrix + 70 * 20 71 * 24 - 71 f 24 10 f 4

Mitochondrial GTK + 322 f 57 3960 f 460 - 211 f 30 550 f 550

Cytosolic GTK + 7 5 7 f 160 9805 f 1888 - 322 f 57 163 * 163

oxaloacetate and a-ketoglutarate) (Table 5). The dependence on MTB showed a similar behavior to cytosolic glutamine transaminase K, i.e. an increasing stimulation of DCVC ac- tivity up to about 1000 p~ MTB after which the activity decreased (Fig. 1). Third, the enzyme was stimulated by the addition of purified rat kidney amino acid oxidase (Fig. 2). Finally, by Ouchterlony double diffusion analysis, using affin- ity-purified antibody raised against the cytosolic glutamine transaminase K, no differences between the two proteins were detected (Fig. 3). The two forms were not separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (data not shown). There was a difference in the V,,, values for DCVC with cytosolic (22 pmol/lO min/mg (17)) and mitochondrial glutamine transaminase K (4 pmol/lO min/mg), but the K , values were similar, 0.3 (17) and 0.48 mM, respectively. The only detectable difference between the cytosolic and mito- chondrial glutamine transaminase K was their separation by elution from hydroxylapatite at 3.4 and 4.3 mmho, respec- tively. These data show that the purified mitochondrial glu- tamine transaminase K is very similar to cytosolic glutamine transaminase K, as reported by Cooper and Meister (20, 21) and that the 8-lyase characteristics of the two enzymes are similar.

The observation that the purified enzyme and the matrix

activity are stimulated by the addition of a-keto acids indi- cated that the matrix mitochondrial glutamine transaminase K may catalyze both transamination and 8-elimination reac- tions as has been described for the cytosolic glutamine trans- aminase K (17). If this were true, then DMOP should be a product. When the 14C-metabolites from incubations contain- ing purified enzyme, MTB, and [ “CC]DCVC were analyzed by HPLC, both DMOP and pyruvate were detected in a ratio of 1:l (Table 6). Therefore, the lack of activity of mitochondrial matrix P-lyase may be due to the absence of a-keto acids to complete the transamination reaction as suggested for cyto- solic glutamine transaminase K (17). More importantly, this mechanism accounts for the loss of 8-lyase activity when mitochondria are fractionated.

Comparison of BTC and DCVC Activities-BTC is a com- monly used substrate for the measurement of 8-lyase activity, and the activity of @-lyase with this substrate is similar to that exhibited with DCVC in mitochondrial matrix (25). In three preparation, 107 f 9% ( X f S.D.) of the BTC activity was recovered in the matrix fraction without MTB added to the incubations (n = 3). Less than 7% of the activity was recovered in the other fractions. We compared the activity of DCVC to that of BTC in the mitochondrial matrix and with purified mitochondrial and cytosolic glutamine transaminase K (Table 7). The activity with BTC in the matrix fraction was not dependent on the presence of MTB. With the purified enzymes, BTC had about 10% of the activity seen with DCVC, and the activity was stimulated only 2-fold by MTB. In contrast, MTB stimulated DCVC activity 7- and 60-fold with purified mitochondrial and cytosolic glutamine transaminase K, respectively. In the absence of MTB, the @-lyase activity was very low, and accurate measurement of metabolism was difficult as indicated by the large range of the data.

The Role of Mitochondrial Matrix Enzymes in the Toxicity of DCVC-Our data (Table 2) led us to conclude that mito- chondrial matrix enzymes have 8-lyase activity with DCVC, and these mitochondrial enzyme(s) are retained in mitoplasts. Therefore, we investigated the role of matrix metabolism in the toxicity of DCVC by comparing the effect of DCVC on succinate-stimulated respiration in mitochondria and mito- plasts (Fig. 4). When the time course of DCVC toxicity was determined, DCVC inhibited both state 3 and state 4 respi- ration in mitochondria and mitoplasts. Therefore, we con- cluded that mitochondrial matrix enzymes can metabolize

Cysteine Conjugate Toxicity 3399

DCVC to toxic species which inhibit mitochondrial respira- tion.

DISCUSSION

Cysteine conjugate @-lyase activity is a property of cellular PLP-dependent enzymes (17, 32). One of these enzymes is cytosolic glutamine transaminase K, a kidney enzyme which catalyzes both @-elimination and transamination with DCVC as substrate. Because both reaction are catalyzed by one enzyme, the pathways interact through a complex regulatory scheme in which the accumulation of the PLP cofactor as P M P inhibits the flow of DCVC through the @-elimination pathway (see Ref. 17 for discussion). An appreciation of this scheme is important in understanding cysteine conjugate tox- icity. If cellular cofactors for transamination are limiting, enzyme could accumulate as P M P enzyme causing @-lyase activity and toxicity to decrease. Elfarra et al. (33) have suggested that MTB increases DCVC toxicity and BTC me- tabolism in mitochondria. However, since metabolism of DCVC and 35S binding were not measured, it is not clear that @-elimination of DCVC increased in the presence of MTB.

The results presented here suggest that enzymes located in the matrix fraction are responsible for the metabolism of DCVC in whole mitochondria. When mitochondria are frac- tionated using the digitonin method (26), the matrix enzymes become less active unless a source of a-keto acid is added to the incubation. However, intact mitochondria retain factors which support the @-lyase activity with DCVC. Presumably, these unknown factors provide the necessary a-keto acid to complete the transamination reaction and return PMP en- zyme to PLP enzyme. The data point toward two possible identities for these unknown factors. The first is L-amino acid oxidase which, when added to incubations, stimulates cyto- solic (17) and mitochondrial glutamine transaminase K (Fig. 2) by providing an external source of DMOP (17). The excess DMOP serves as the amino acceptor for P M P enzyme. L- Amino acid oxidase is concentrated in peroxisomes (22) which do not separate well from the mitochondrial fraction (34, 35), particularly in kidney (34). Damage to the peroxisomal mem- brane during tissue homogenization probably accounts for the purification of this enzyme from the cytosolic compartment (17, 22, 34, 35). The second possible source of a-keto acids are those normally present in mitochondria as Kreb’s cycle intermediates. Both oxaloacetate and &-ketoglutarate, as well as their precursors aspartate and glutamate, are normal com- ponents of mitochondria (36), and both of these a-keto acids stimulate metabolism in mitochondrial matrix (Table 3). Dur- ing removal of the outer membrane with digitonin, it is possible that these small molecules are lost from the mito- plasts resulting in an apparent loss of enzyme.

At first glance the binding and metabolism data in Table 1 appear paradoxical. How can metabolism decrease in mito- plasts if binding of “S label does not change to the same extent? The explanation lies in the recycling of DMOP to DCVC by the PMP enzyme. With the cytosolic enzyme (see Table V in Ref. 17) we showed that at limiting MTB concen- trations the DMOPpyruvate ratio decreased to 1:4 while at higher MTB concentrations the ratio was 1.3:l. The ratio changes because the DMOP pool decreased at limiting MTB concentrations while the amount of pyruvate, formed via the @-elimination pathway, remained constant. This is similar to the relationship between the 14C extraction and the 35S bind- ing data in Table 1 since covalent binding reflects only the @- elimination pathway while 14C extraction represents total metabolism (see “Experimental Procedures” for further dis- cussion). Therefore, the pool of DMOP is consumed during

the PMP enzyme-catalyzed amination of DMOP because @- elimination can occur from the quinoneimine complex formed between DMOP and PMP enzyme (see Fig. 1 in Ref. 17). If MTB is present in sufficient concentration to competitively inhibit the binding of DMOP to the PMP enzyme, then the pool of DMOP will increase as metabolism continues. Alter- natively, if DMOP is the only source of a-keto acid, it should decrease to a steady state level a t which its concentration will be determined by the rate constants for its formation, via half-transamination, and the rate of metabolism via amina- tion to DCVC or @-elimination. The @-elimination of DMOP which occurs from the quinoneimine complex formed by PMP enzyme and DMOP would result in loss of DMOP from the pool and the maintenance of the @-elimination pathway.

Two separate lines of evidence suggest that multiple @-lyase enzymes may be present in mitochondria. First, the low specific activity of the purified mitochondrial and cytosolic glutamine transaminase K with BTC suggests that more than one enzyme may be metabolizing cysteine conjugates in mi- tochondria since BTC and DCVC activities are comparable in the matrix; mitochondrial glutamine transaminase K may be only one of these enzymes. Second, the a-keto acid data show that a-keto acids which do not stimulate the purified mitochondrial glutamine transaminase K do stimulate the 4- lyase activity in the crude matrix. This suggests that other enzymes whose activities are stimulated by more hydrophilic a-keto acids may also metabolize DCVC. The existence of multiple @-lyases could account for the inability of DCVC to inhibit BTC activity in rat kidney mitochondria (16). If multiple enzymes exist, it may be inappropriate to extrapolate from metabolism data with one substrate to toxicity data with another.

The lack of stimulation of BTC activity by 1 mM MTB (Table 7) is interesting in that it may indicate that the leaving group character plays a role in regulating the ratio of @- elimination to transamination. One might expect that a qui- noneimine intermediate formed from PLP enzyme and an amino acid with a good leaving group at the @-carbon would tend to collapse and undergo @-elimination more readily than

30 60 90 30 60 @I T h e ( rnin )

FIG. 4. Time course of DCVC toxicity in rat kidney mito- chondria and mitoplast. Mitochondria and mitoplasts were pre- pared as described under “Experimental Procedures.” The data are presepted as the percent of control respiration calculated as nano- atoms of oxygen/mg.min; mitochondria, state 3 = 175 f 48, state 4 = 39 * 13; mitoplasts, state 3 = 130 & 46, state 4 = 47 & 13 (nanoatoms/min mg; X & S.D., n = 3). Panel A, mitochondrial state 3 respiration; Panel B, mitochondrial state 4 respiration; Panel C, mitoplast state 3 respiration; PaneE D, mitoplast state 4 respiration. 0, control; 0, 1 mM DCVC.

3400 Cysteine Conjugate Toxicity

a complex formed between PLP and a substrate which does not have a good leaving group on the @-carbon. Leaving group character could play a significant role in toxicity by determin- ing the ratio of @-elimination and transamination and the ability of cofactors to regulate the rate of metabolism. How- ever, this must remain speculative until leaving group char- acter is correlated with the metabolism and toxicity of the conjugates.

Our results differ from those of Lash et al. (14) who re- covered most of the /3-lyase activity with BTC and DCVC in the outer membranes and <2% of either activity in the matrix fraction. Several differences are worthy of consideration. a- Keto acids were not added in their assays, and other differ- ences include the strain of rat, the specific activities reported (see “Experimental Procedures” for specifics for the @-lyase assay), and the amount of digitonin used in the fractionation. In our studies, concentrations of digitonin higher than 0.12 mg/mg mitochondrial protein resulted in uncoupling of State 4 respiration, and State 4 respiration is less sensitive to DCVC (Fig. 4). The differences in the specific activity might relate to differences in the assay methods or the method of protein determination (31, 37). It is likely that different mitoplast preparations show differences in the retention of mitochon- drial factors which support the @-lyase activity; therefore, several differences in the methods and model might account for the differences in the data.

In conclusion, our results suggest that metabolism of DCVC in mitochondria is regulated by the availability of a-keto acid. It is conceivable that differences in enzyme concentration or activity in the different segments could account for differences in the toxicity within the proximal tubule. Though glutamine transaminase K activity is highest in the cortex (38), the localization of @-lyase enzymes within the proximal tubule has not been determined. Other factors such as transport and the sensitivity of the tissue must also be considered. Eluci- dation of the biochemical factors which control the segment specific toxicity of DCVC may lead to a better understanding of biochemical heterogeneity in the proximal tubule and its relationship to renal function as well as nephrotoxicity.

Acknowledgment-We would like to thank Dr. M. W. Anders for critical reading of the manuscript.

REFERENCES 1. Jakoby, W. B., and Habig, W. H. (1980) In Enzymatic Basis of

Detoxication (Jakoby, W. B., ed) p. 63, Academic Press, Or- lando, FL

2. Chasseaud, L. F. (1976) In Glutathione: Metabolism and Function (Jakoby, W. B., and Arias, I. M., eds) p. 77, Raven Press, New York

3. Igwe, 0. J. (1986) Biochem. Pharmacol. 35,2987-2994 4. Green, T., and Odum, J. (1985) Chem. Biol. Interact. 54 , 15-31 5. Stevens, J., Hayden, P., and Taylor, G. (1986) J. Biol. Chem.

261,3325-3332

6. Derr, R. F., and Schultze, M. 0. (1963) Biochem. Phurmacol. 12,

7. Bhattacharya, R. K., and Schultze, M. 0. (1972) Arch. Biochem.

8. Stonard, M. D., and Parker, V. H. (1971) Biochem. Pharmucol.

9. Stonard, M. D., and Parker, V. H. (1971) Biochem. Pharmucol.

465-474

Biophys. 153, 105-115

20 , 2417-2427

20,2429-2437 10. Parker, V. H. (1965) Food Cosmet. Toxicot. 3,75-84 11. Terracini, B., and Parker, V. H. (1965) Food Cosmet. Toxicol. 3 ,

12. Jones, T. W., Wallin, A., Thor, H., Gerdes, R. J., Ormstad, K., and Orrenius, S. (1986) Fed. Proc. 45, 1933 (abstr.)

13. Jones, T. W., Wallin, A., Hjordis, T., Gerdes, R. G., Ormstad, K., and Orrenius, S. (1986) Arch. Biochem. Biophys. 261,504-513

14. Lash, L. H., Elfarra, A. A., and Anders, M. W. (1986) J. Bid. Chem. 261,5930-5935

15. Lash, L. H., and Anders, M. W. (1986) J. Biol. Chem. 2 6 1 , 13076-13081

16. Elfarra, A. E., Jakobson, I., and Anders, M. W. (1986) Biochem. Pharmacol. 35 , 283-288

17. Stevens, J. L., Robbins, J. D., and Byrd, R. A. (1986) J. Biol. Chem. 261,15529-15537

18. Nash, J. A., King, L. J., Lock, E. A., and Green, T. (1984) Toxicol. Appl. Pharmucol. 73,124-137

19. Stevens, J. L. (1985) Biochem. Biophys. Res. Commun. 129,499- 504

20. Cooper, A. J. L., and Meister, A, (1985) Methods Enzymol. 113 , 344-349

21. Cooper, A. J. L., and Meister, A. (1981) Comp. Biochem. Physiol. 69B, 137-145

22. Hamilton, G. A. (1985) Adu. Enzymol. Relut. Areas Mol. Bid. 57 ,

23. McKinney, L. L., Picken, J. C., Weakely, F. B., Eldridge, A. C., Campbell, R. E., Cowan, J. C., and Biester, H. E. (1959) J. Am. Chem. SOC. 81,909-915

24. Hayden, P., Schaeffer, V. H., Larsen, G., and Stevens, J. L. (1987) Methods Enzymol. 143,228-234

25. Dohn, D. R., and Anders, M. W. (1982) A d . Biochem. 120 ,

26. Greenawalt, J. W. (1974) Methods Enzymol. 31 , 310-323 27. Schnaitman, C., and Greenawalt, J. W. (1968) J. Cell Biol. 38 ,

28. Maizel, J . V., Jr. (1971) Methods Virol. 5 , 179-246 29. Stevens, J. L., and Jakoby, W. B. (1985) Methods Enzymol. 113 ,

30. Larsen, G. L., and Stevens, J. L. (1986) Mol. Pharmucol. 29,97-

31. Bradford, M. M. (1976) Anal. Biochem. 7 2 , 248-254 32. Stevens, J. L. (1984) J. Biol. Chem. 260 , 7945-7950 33. Elfarra, A. E., Lash, L. H., and Anders, M. W. (1987) Mol.

34. Badhuin, P. (1974) Methods Enzymol. 31 , 356-368 35. Leighton, F., Poole, B., Beaufay, H., Baudhuin, P., Coffey, J . W.,

36. Greenbaum, A. L., Gumaa, A., and McClean, P. (1971) Arch.

37. Read, S . M., and Northcoate, D. H. (1981) Anal. Biochem. 116 ,

38. Kopyt, N., and Narins, R. G. (1986) Miner. Electrolyte Metab.

6 7- 74

85-177

379-386

158-175

510-515

103

Pharmucol. 31,208-212

Fowler, S., and de Duve, C. (1968) J. Cell Biol. 37,482-513

Biochem. Biophys. 143,617-663

53-64

12 , 347-351

Cysteine Conjugate Toxicity 3401

c

0.6 E /- L[, I I

a 5 0.2

0 1 2 3 4

c .. i