y-Glutamyl Hydrolase (Conjugase) PURIFICATION AND PROPERTIES OF THE BOVINE HEPATIC ENZYME*

Bovine hepatic y-glutamyl hydrolase (conjugase) has been purified to homogeneity. A feature of the purification procedure was the use of high affinity macromolecular polyanion enzyme inhibitors which formed tight complexes with the enzyme altering its solubility, gel filtration, and ion exchange properties. The enzyme, which cleaves the y-glutamyl bonds of pteroylpolyglutamates, has a molecular weight of 108,000. It is a glycoprotein with an acid pH optimum, properties consistent with its lysosomal localization. Zinc is essential for enzyme stability. The presence of highly reactive sulfhydryl groups was evident from the extreme sensitivity to oxidizing agents and organomercurials. Very little thermal denaturation occurs below 65”, but the enzyme is extremely sensitive to buffer anions, in keeping with the polyanionic nature of the substrate. In to study the mechanism of action of the a wide of pteroylpolyglutamates, N-t-Boc polyglutamates

Bovine hepatic y-glutamyl hydrolase (conjugase) has been purified to homogeneity. A feature of the purification procedure was the use of high affinity macromolecular polyanion enzyme inhibitors which formed tight complexes with the enzyme altering its solubility, gel filtration, and ion exchange properties. The enzyme, which cleaves the y-glutamyl bonds of pteroylpolyglutamates, has a molecular weight of 108,000. It is a glycoprotein with an acid pH optimum, properties consistent with its lysosomal localization.
Zinc is essential for enzyme stability. The presence of highly reactive sulfhydryl groups was evident from the extreme sensitivity to oxidizing agents and organomercurials. Very little thermal denaturation occurs below 65", but the enzyme is extremely sensitive to buffer anions, in keeping with the polyanionic nature of the substrate. In order to study the mechanism of action of the enzyme, a wide range of pteroylpolyglutamates, N-t-Boc polyglutamates and free polyglutamates were synthesized containing L-[U-"Clglutamic acid residues in different positions. Two pteroyltriglutamate derivatives were also synthesized in which an 01 bond replaced one of the two available y bonds. Time course studies of the products of the action of conjugase on these various substrates enabled us to draw the following conclusions about the enzyme: (a) peptide bond cleavage occurred only at y-glutamyl bonds and the presence of a COOH-terminal y bond was essential for enzyme action; (b) bond cleavage occurred with equal facility at internal points of the peptide chain and the enzyme should therefore be more appropriately classified as an acid hydrolase; (c) longer chain y-glutamyl peptides were preferentially attacked by the enzyme, the cleavage of diglutamyl peptides being extremely slow; and (d) cleavage of y bonds was independent of the NH,-terminal pteroyl moiety. Studies with polyanions such as the glycosaminoglycans and dextran sulfate supported the concept that the polyanion structure of the substrate was a major factor in substrate-active site interaction.
The group of enzymes which hydrolyze the y-glutamyl bonds of pteroylpolyglutamates, termed loosely "the conjugases", have an almost universal distribution throughout the phylogenetic spectrum. Initially identified some 30 years ago (I), relatively little was known about their properties until Baugh and his colleagues developed a sensitive efficient assay employing selectively radioactively labeled substrates synthesized by solid phase synthetic techniques (2). Although the enzymes were generally considered to be acid carboxypeptidases, certain tissues and lower organisms contain enzymes which were active at a neutral or alkaline pH. The recent assignation of the name glutamate carboxypeptidase (EC 3.4.12.10) is somewhat misleading in that it gives no indication either of the specificity of many of these enzymes for y-glutamyl bonds or, as in the case of the bovine hepatic enzyme at least, of the endopeptidase activity. The over-all biological role of conjugases is still unclear. They are certainly required during the process of digestion and absorption of dietary pteroylpolyglutamates, during which these molecules are reduced to their mono-or diglutamyl derivatives. Another species of the enzyme is required to effect the reduction to critical peptide chain length of pteroylpolyglutamates necessary for phage tail plate assembly (3). Central to the question of a major role for the enzyme is the clarification of the relative utilization of the oligo-and polyglutamate derivatives of the various reduced-and one-carbon substituted pteroylglutamates. It is evident that the polyglutamates constitute the bulk of the tissue stores and, in prokaryotes at least, it would appear that the enzymes involved in the interconversion of these compounds function concentration over an Amicon XM 100 filter provided an ideal method for enzyme stabilization and storage. The enzyme could not be concentrated other than as the enzyme-dextran complex because it bound to the membrane surface resulting in considerable activity losses.

Properties
of Isolated Enzyme-Polyacrylamide gel electrophoresis of the purified enzyme revealed only a single protein band which also gave a periodic acid-Schiff reaction. Gel chromatography on either acrylamide (Bio-Gel P-300) or Sephadex G-150 indicated a molecular weight of 108,000 ( Fig.  1). This value was not altered by p-chloromercuribenzoate, Triton X-100, or sodium dodecyl sulfate, although a subunit structure might be anticipated.
The only published molecular weight data available were for the pancreatic conjugases of the rat and chicken for which the respective values of 60,000 and 52,000 were reported (12). The endopeptidase of chicken intestine conjugase has a reported molecular weight of 80,000 (13).
As might be predicted from the effects of anions on enzyme activity (uide infra) the pH activity profile varied with the nature of the buffer used. In 33 mM sodium acetate there was pH optimum at 3.9 with a rapid decrease below this level and above pH 4.5. In 33 mM sodium citrate, however, not only was the optimum at pH 4.5 but the shape of the profile was quite different. This altered profile could also perhaps have been related to a metal-chelation effect of the citrate ion, a potent enzyme inhibitor.
The K, value was determined for pteroyltriglutamate with the radioactive label on the terminal glutamic acid residue. The K, for this particular substrate was 1.7 FM, a value somewhat above the true figure as 20% of the radioactive diglutamate released would not be measured in the assay system. The turnover number was 732 y-glutamyl bonds 1. Molecular weight determination by acrylamide chromatography on Bio-Gel P-300 equilibrated in 5.0 mM fl-mercaptoethanol, 1.0 mM zinc acetate, 20 rnM sodium chloride, pH 6.5. The column dimensions were 86 x 1.5 cm and the flow rate was 1.7 ml per hour. Horse spleen apoferritin (Schwarz/Mann) (MW 480,000) was used for void volume determination.
Cation and Anion Effects on Enzyme Actiuity-The monovalent cations, sodium, lithium, ammonium, and potassium, had no effect on enzyme activity at concentrations up to 50 mM. The cations were tested as their chloride salts and beyond this concentration the anion itself was an inhibitor. Certain divalent cations (calcium, barium, magnesium, cobalt, nickel, and strontium) similarly had no effect on activity in concentrations up to 25 mM (tested as their chloride salts). Other cations, however, were enzyme inhibitors.
In increasing order of potency these were manganese, cadmium, iron (Fe3+), mercury (Hg'+) and copper (Cu"'). At a concentration of 10 mM, manganese chloride and cupric chloride produced 45 and 98% inhibition, respectively. Anion effects were assessed with the respective sodium salts. It was found that there was a small but significant decrease in activity when the concentration of the sodium acetate buffer used in routine assays was increased from 33 to 100 mM. The addition of 33 mM iodide, sulfate, and nitrate produced 40 to 50% inhibition.
This degree of inhibition was seen with 100 mM formate, fluoride, chloride, phosphate, and dimethyl glutarate. Sodium citrate was an extremely potent inhibitor, a concentration of 0.1 mM being sufficient to produce 50% inhibition; sodium cyanide was as potent in a concentration of 5.0 mM. The effect of these latter anions was almost certainly related to their zinc bonding property. Borate, in concentrations up to 300 mM, predictably produced no effect on enzyme activity as it was minimally ionized at pH 4.5 (14). Enzyme Activity as a Function of Temperature-These studies attempted to correct for thermal denaturation by using large amounts of enzyme coupled with short reaction times (1 min) in order to achieve a minimum of 25% substrate cleavage. Simultaneous "denaturation" profiles were carried out in which the enzyme was heated for 1 min at each temperature, rapidly cooled to 37", and assayed. Substrate was omitted during heat treatment as it was heat sensitive although additional experiments suggested that it protected the enzyme from heat denaturation.
It was shown that the apparent temperature optimum was 70" and in the light of the 40% loss of activity due to thermal denaturation at this temperature, the "true" optimum was certainly higher. Denaturation, in fact, only became significant over 65" and it was clear that this enzyme was highly heat-resistant in that only 30% loss of activity occurred at 30 min at 65".
Conjugase, a Zinc Metalloprotein-The addition of chelators such as EDTA and l,lO-o-phenanthroline to the assay did not inhibit the reaction although, as was discussed previously, cyanide and citrate were potent inhibitors.
Stability studies demonstrated that storage of the enzyme in 5.0 mM P-mercaptoethanol at 4" resulted in the loss of 40% of activity over 24 hours. This loss was prevented by the addition of 1.0 mM zinc acetate. The loss of 60% of activity when the enzyme was stored in water alone at 4' for 24 hours was reduced to 20% by the addition of the zinc ion.
Prolonged dialysis (24 hours) of the enzyme against a solution containing 1.0 mM o-phenanthroline and 5.0 mM P-mercaptoethanol resulted in the loss of 70% of activity compared with a control dialysis in which 1.0 mM zinc acetate replaced the o-phenanthroline (Fig. 2). Further dialysis of the chelator-treated enzyme against the control dialysis fluid restored the activity to 70% of control levels within 24 hours and this activit~y remained stable for a further 10 davs. Continued dialysis against the chelator reduced the activity to 4% of control levels by 144 hours and no activity was detectable after 288 hours. The replacement of the chelating agent by different cations after 144 hours restored activity to a variable degree. Zinc restored activity to 32% of control levels and this activity remained stable for the succeeding 120 hours. Nickel also restored activity to approximately 20% of control values but the activity was not stable. Magnesium and calcium, but not cobalt, were also partly effective as reactivators.
Quantitation of the zinc content of two samples of the purified enzyme, carried out by atomic absorption spectroscopy, gave values of 4.47 and 4.15 zinc atoms per molecule of enzyme. These determinations were made following 8 days dialysis against distilled deionized water.
Reactive Sulfhydryl Groups-P-Mercaptoethanol was required for enzyme stabilization and p-chloromercuribenzoate (1.0 x lo-* M) produced instantaneous inhibition which could be as rapidly reversed by the addition of thiols even after several months exposure to the sulfhydryl reagent. The rapid inhibition and reversal implied that the sulfhydryl groups were both readily accessible and active site-related.
The long term stability of the inhibited enzyme suggested that bonding of these groups did not result in gross conformational changes. N-Ethylmaleimide was a far less efficient inhibitor, a concentration of 12.5 mM producing only 60% inhibition after several hours exposure. Thiol addition produced only partial recovery of activity. A (MW 55,000) inhibited the enzyme although a vast molar excess was required to produce a significant effect. Only 25% inhibition was achieved at a concentration of 0.8 ELM which in the assay system represented a 40-fold excess over the enzyme concentration.
The degree of inhibition was not increased by the addition to the assay of calcium or manganese ions (1.0 mM) which are critical for effective carbohydrate binding (15). The inhibition was not due to any substrate binding by concanavalin A as mixtures of the two could be readily resolved by gel chromatography.
Column chromatography of a mixture of concanavalin A and highly purified conjugase on Sephadex G-150 resulted in the elution of the enzyme at a position corresponding to a molecular weight of 160,000 (V,:V, = 1.15). wavelength 517 nm). Enzyme activity again corresponded to that of a protein of molecular weight 160,000, and to a major fluorescence peak. Chromatography of the fluorescent ligand alone confirmed that it coincided with a second major peak. The fluorescence emission peak for the free ligand was at 520 nm indicating a small blue shift on complexing with the enzyme. Glucoside was not required for removal of the free concanavalin A-fluorescein, but there was a significant degree of interreaction with the Sephadex in that the V,:V, was 4.5. This would suggest that the linking of the fluorophore to the concanavalin A had in some way modified the carbohydrate binding sites. Quantitation of the relative fluorescence in the two peaks from the column revealed a 2:l ratio between the free and conjugase-bound concanavalin A-fluorescein, supporting the 1: 1 stoichiometry of complex formation.
The 96% recovery of fluorescence from the column suggested that complex formation did not result in any significant quenching.
Substrate concentration was 10 FM and the specific radioactivity was equivalent for all compounds (0.48 to 0.50 Ci/mol). Samples were withdrawn from the assay system at appropriate time intervals and the release of free glutamic acid and diglutamate estimated in the usual assay system and by high voltage electrophoresis.
Quantitative ninhydrin estimations were performed in estimating the products of hydrolysis of the a-triglutamate.
It was evident that the rate of cleavage of the pteroyltriglutamate with the internal (Y bond (Pte-a-Glu-Glu-*Glu) was greatly reduced when compared with the standard substrate (Pte-Glu-Glu-*Glu), whereas barely any radioactivity was released from the substrate with the terminal a! bond (Pte-Glu-a-Glu-*Glu) (Fig. 3). This was confirmed by the radiochromatograms of the products of the reaction. Both glutamic acid and diglutamate were released from the standard substrate, whereas only glutamic acid was released from Pte-cu-Glu-Glu-*Glu and no detectable cleavage of Pte-Glu-a-Glu-*Glu occurred. This indicated not only that the internal o( bond could not be hydrolyzed but that the internal y bond was not split in the presence of a terminal LY bond. The very low affinity of the enzyme for Pte-Glu-a-Glu-Glu was emphasized by the observation that, even in a lo-fold molar excess, it failed to inhibit cleavage of the standard substrate. As might have been anticipated from this data conjugase could not hydrolyze the cY-triglutamate.
Endopeptidase and Exopeptidase Action-This property of the enzyme was referred to above where it was observed that both glutamic acid and diglutamate were produced from Pte-Glu-Glu-*Glu.
The bulk of mammalian folates are stored either as their pentaglutamates (16) or even longer chain derivatives (17) and accordingly, pteroylpentaglutamate with a terminal ["Clglutamic acid residue was used for assessing the sequential mechanism of hydrolysis. It was shown (Fig. 4) that, initially, radioactive tetraglutamate (Glu-Glu-Glu-*Glu), diglutamate (Glu-*Glu), and glutamic acid were produced together with smaller amounts of triglutamate (Glu-Glu-*Glu) suggesting some sparing of 1% . the second innermost y bond. This latter observation was notable in that pteroyldiglutamate, the other product of cleavage at this particular bond, was only very slowly hydrolyzed and, in fact, acts as an inhibitor of the cleavage of longer peptide chain derivatives (uide infra).
The Glu-Glu-Glu-*Glu level did not fall until the bulk of the original substrate had been hydrolyzed, but then the level of Glu-Glu-*Glu rose steadily. indicating cleavage of the NH,-terminal y bond. This reaction resulted in the release of a nonradioactively labeled glutamic acid residue not detected in this analytical system. Similarly, cleavage of Glu-Glu-Glu-* Glu in the central bond resulted in the production of 2 diglutamate residues, only one of which was radioactive and hydrolysis of the COOH-terminal bond released unlabeled triglutamate but radioactive glutamic acid. Accordingly, there were relatively higher concentrations of glutamic acid, di-, and triglutamate present than was demonstrated in this analytical system. Prolonged hydrolysis resulted in the relatively rapid disappearance of Glu-Glu-Glu-*Glu and Glu-Glu-*Glu and subsequently the very slow disappearance of Glu-*Glu, the end product being glutamic acid. These radioactivity data were qualitatively confirmed by the rate of appearance and disappearance of ninhydrin positive material from the electrophoretogram.
The sequence of hydrolysis, namely Pte-Glu-Glu-Glu-Glu-*Glu, followed by tetraglutamate, triglutamate, and finally diglutamate, was consistent with the enzyme preference for longer chain glutamyl peptides.
Quantitation of the other products of this pattern of cleavage, that is the pteroylpolyglutamates present at various stages of the reaction, could not be effectively carried out due to the low substrate concentration and the unavailability of a suitably sensitive detection system. Thin layer chromatography on MN-300 cellulose, however, conclusively showed that Pte-Glu-Glu-Glu-Glu-*Glu rapidly disappeared to be loo- replaced at the end of the reaction by pteroylglutamic acid. The end products of the reaction were pteroylglutamic acid and glutamic acid.
It would appear that bovine hepatic conjugase had a preference for internal y bonds and this was illustrated by experiments with pteroyltriglutamates containing the radioactive label in the terminal and middle glutamic acid residues, Pte-Glu-Glu-*Glu and Pte-Glu-*Glu-Glu. With the routine assay system there was virtually no difference in the rate of release of radioactivity from these two substrates. A different perspective was obtained by examining the products of the reactions by high voltage electrophoresis and by thin layer chromatography.
The hydrolysis of Pte-Glu-Glu-*Glu was associated with the release of larger amounts of Glu-*Glu than glutamic acid initially indicating a preferential attack on the internal y bond. In a much slower reaction there was a progressive fall in the level of Glu-*Glu with a parallel increase in free glutamic acid. Virtually no radioactive glutamic acid was detected from the hydrolysis of Pte-Glu-*Glu-Glu until after some 10 min of incubation although there was a rapid release of *Glu-Glu. The subsequent slow decline in the level of *Glu-Glu was associated again with the appearance of glutamic acid. Preferential Cleavage of Longer Chain Polyglutamyl Peptides-This property of the enzyme was demonstrated by using pteroyltriglutamates and pteroylpentaglutamates containing "C labels in selected glutamic acid residues. The routine assay system was used with varying time periods of incubation and the study was facilitated by the exploitation of one of the major defects of the charcoal adsorption technique used in this assay, the retention of oligoglutamates by charcoal (see "Methods").
The effect of this was seen during a time course study of the hydrolysis of pteroylpentaglutamate containing a [14C]glutamate in the second innermost position, i.e. Pte-Glu-*Glu-Glu-Glu-Glu.
The internal position of the radioactive label resulted in the relatively late release of radioactive glutamic acid and Glu-*Glu (Fig. 5) which contrasted with the early and rapid release of detectable radioactivity from Pte-Glu-Glu-*Glu.
On incubation of conjugase with equimolar concentrations of these two substrates it was clear that the longer chain substrate was preferentially attacked.
In the routine assay systems the rate of release of radioactivity from the former substrate was of the order of 10% of that from the latter, and this was confirmed by examination of the reaction products. Pte-Glu-Glu was, however, effectively bound by the enzyme and this was reflected by its competitive inhibition of the hydrolysis of Pte-Glu-Glu-Glu with a relatively low Ki of 4.5 ph4.
Boc-diglutamate was similarly relatively slowly hydrolyzed but diglutamate cleavage was even slower. The low affinity for this latter substrate was reflected in the observation that it had a minimal effect on the hydrolysis of Pte-Glu-Glu-Glu.
Effect of Substrate Pteroyl Group on Enzyme Activity-The observation that the hydrolytic action of the enzyme was virtually independent of the presence of the pteroyl group was evident from the studies previously described. In a study in which the rates of cleavage of Pte-Glu-*Glu-Glu, Boc-Glu-Glu, and Glu-*Glu-Glu were compared, it was found that the Boc-substituted derivative was hydrolyzed faster than the pteroyl substituted and free triglutamates (Fig. 6). The pattern of cleavage, as determined by high voltage electrophoresis, was identical for all three substrates.
As the major form of pteroylglutamates in mammalian liver would appear to be 5-methyltetrahydropteroyl pentaglutamate or its longer chain analogues, the action of the enzyme on this compound appeared to be particularly pertinent. Using experimental conditions identical with those employed for the study on the hydrolysis of pteroylpentaglutamate, it was shown that the rate of release of radioactivity was identical for these two substrates containing a ["Clglutamate residue in the COOH-terminal position.
The sequence of y bond cleavage, with an identical release of glutamic acid and glutamyl peptides as a function of time, was confirmed by high voltage electrophoresis of the products of the reaction. Modification of the pteroyl group by reduction and methylation, therefore, had no effect on the action of the enzyme.
Polyanion Inhibition-Polyanions such as DNA and RNA inhibited hog kidney conjugase (18) and at a concentration of 1.0 pg/ml these nucleic acids produced 50% inhibition of the bovine enzyme at a substrate (Pte-Glu-Glu-*Glu) concentration of 10 PM.
We have systematically examined the inhibition produced by three other classes of polyanions: (a) the glycosaminoglycans (chondroitin sulfates A, B, C, and D, heparin, and hyaluronic acid); (b) the sulfonated dextran, dextran sulfate 2000; and (c) the sulfonated anthroquinone dyes Cibacron blue 3GA (and its dextran derivative blue dextran 2000) and Cibacron brilliant blue BRP.
Heparin was the most potent conjugase inhibitor (Fig. 7), producing 50% inhibition at a concentration 26.7 rig/ml, approximately equivalent to an S0,2-ion concentration of 0.1 pM (19). There was little difference between the degrees of inhibition produced by the various chondroitin sulfates. Fifty per cent inhibition was achieved at twice the concentration but TIME (min) . -FIG. 6 (left). The release of "'C radioactivity from Boc-Glu-*Glu-Glu (O), Pte-Glu-"Glu-Glu (0) and Glu-*Glu-Glu (A) by conjugase as a function of time. Substrate concentrations were 10 pM in the standard assay mixture and activity is expressed as disintegrations per min (x10-*) released. The blank values for the assays with Glu-*Glu-Glu as substrate were much higher as only 70% of the unhydrolyzed substrate was adsorbed to the charcoal. FIG. 7 (right). Inhibition of conjugase activity by the glycosaminoat the same SO,!+ ion concentration as heparin. Dextran sulfate 2000 produced 50% inhibition at a concentration of 42.5 rig/ml, equivalent to a S0,2-concentration of 277 nM. The binding of dextran sulfate 2000 to conjugase was studied in some detail as it proved to be an effective method of enzyme purification.
This polyanion formed a stable soluble complex with conjugase, as was illustrated by the column chromatography on Sepharose 4B. The polymer was excluded from this gel, eluting at the void volume while the much lower molecular weight enzyme was considerably retarded. When mixed with the polymer, however, the enzyme eluted at the void volume. The complex could be resolved by the addition of DEAE-dextran which bound dextran sulfate even more effectively than did the enzyme.
The polyanion-enzyme complex was precipitated in the presence of both the alkali earth (Mg2+, Ca*+, Ba2+) and transition (Zn2+, Co*+, Ni2+) elements due to the formation of a cation-enzyme-polyanion complex (20). This experiment was performed with crude enzyme preparations, and by appropriate adjustments to both the dextran sulfate and cation concentration, maximal recovery and purification of the enzyme could be achieved. This is illustrated in Fig. 8, where a concentration of 0.125 g/liter of dextran sulfate and 10 mM magnesium chloride resulted in the precipitation of over 75% of the enzyme activity with a 2Yz-fold increase in specific activity. It was also observed that protamine sulfate displaced the enzyme from this insoluble complex and this was exploited in the method evolved for enzyme purification. dextran produced 50% inhibition of the enzyme at a concentration of 0.67 pg/ml (335 PM), equivalent to a sulfonyl group concentration of 261 nM. The free dye also formed a soluble stable complex with the enzyme, but equivalent inhibition was only achieved at twice the concentration of the dextran-coupled dye. The dye Cibacron brilliant blue BRP, a structural analogue of Cibacron blue 3GA, formed a similar complex but the inhibition was only one third as potent. This dye had an average of one less sulfonyl group per molecule and a different orientation of the sulfonyl and amino groups on the phenylenediamine ring (21). The complexes in all three situations were readily dissociated by either bovine serum albumin or DEAEcellulose, both of which had a higher dye affinity than did the enzyme.

DISCUSSION
The conjugase group of enzymes appears to fall into two classes, one of which has an acid pH optimum and another smaller group has a neutral or alkaline pH optimum. The first group of enzymes is represented by the mammalian conjugases and these are presumably lysosomal in origin (22)(23)(24)(25). They include the enzymes from hog kidney (26,27), leucocytes (28), guinea pig intestine (29) human liver (22), and human placenta (30). The pH optimum for all of these enzymes falls in the range of 4.0 to 5.0. The second group includes the chicken intestine enzyme complex with maximum activity at pH 7.5 (31), the chicken pancreas enzyme at pH 7.8 (32), and the enzyme from Flavobacterium polyglutamicum, maximal between pH 8.0 and 8.5 (33). The pH of 4.5 was chosen for most of our routine assays with bovine conjugase as the enzyme was a little more stable at this pH during studies which required prolonged incubation with various substrates. It has been noted in earlier work on the human liver enzyme that the activity varied greatly in different buffers (22) and a direct correlation could be observed between the pK, of the buffer and the degree of inhibition of the enzyme at a particular pH level. Apart from the cases of the citrate and cyanide anions, the same phenomenon was observed with the bovine enzyme, activity being greater on the acid side of the buffer pK,, that is, at lower anion concentrations.
The anion effect probably results from a number of factors. At pH levels above 2.19 the substrate, pteroylpolyglutamate, almost certainly exists as a polyanion with an extended rod structure maintained by the free cu-carboxyl groups (34), and high anion concentrations could interfere with enzyme binding. It would appear that charge effects are important in determining conjugase activity. In relation to this, the ionization constants of protein groups vary with ionic strength (35) and changes in salt concentration could well alter the pK, values of active site-related groups. The marked thermal stability of the enzyme activity was regarded as evidence of the primitive origin of the protein consistent with the wide phylogenetic distribution of the enzyme.
The high temperature optimum placed beef hepatic conjugase within the group of thermophilic bacterial proteases, enzymes which are usually neutral proteases. One of these enzymes, thermolysin from Bacillus thermolyticus, has been shown to be a complex metalloenzyme which requires zinc for the maintenance of activity and calcium for thermal stability (36). Calcium, however, did not improve the heat stability of hepatic conjugase.
The zinc stabilization of conjugase and the zinc-mediated partial restoration of activity after prolonged treatment with o-phenanthroline suggested that the enzyme was a zinc metalloprotein. This was confirmed by the direct estimation of zinc in the purified protein at 4 atoms per molecule of molecular weight 108,000. It would appear that zinc was relatively firmly bound as prolonged exposure to a chelator was required to significantly decrease the activity. Furthermore, the localization of the zinc atoms may have imposed restriction on access for the relatively bulky phenanthroline molecule. This latter possibility was supported by the rapid inhibition by simple zinc ligands such as citrate and cyanide ions.
The removal of zinc also appeared to labilize the enzyme in that the full activity could not be restored by zinc addition. This phenomenon has been observed with carboxypeptidase A (37), carboxypeptidase B (38), carbonic anhydrase (39), and alkaline phosphatase (40). Similarly, other metal ions were able to substitute for zinc to varying degrees with these enzymes. In contrast to this, the removal of zinc from yeast alcohol dehydrogenase (41) and malic dehydrogenase (42) was irreversible.
It was clear that the maintenance of reduced sulfhydryl groups was important for enzyme activity and 5.0 mM P-mercaptoethanol was quite effective for this purpose. This thiol is, however, one of a group of sulfur-containing ligands which include BAL and cysteine which have a high affinity for the transition and Group IIB elements, particularly iron, copper, and zinc (43). Cysteine most effectively removed zinc from the carboxypeptidase A active site, indicating a low stability constant for this metalloenzyme (44). Although P-mercaptoethanol was required to stabilize conjugase it almost certainly removed the zinc from the protein as the addition of 1.0 mM zinc acetate to the thiol-stabilized enzyme resulted in the maintenance of full activity for 3 months at 4". This observation would indicate that, despite the formation of zinc mercaptide, the concentrations of both free sulfhydryl groups and Zn2+ were sufficient to maintain the reduced sulfhydryl groups and to maintain an equilibrium between the protein-bound zinc and the surrounding medium. Studies on conjugase from other sources, although not extensively pursued, supported the concept that these were sulfhydryl enzymes. The soybean (45) and hog kidney (18) enzymes were inhibited by sulfhydryl reagents, the latter being activated by cysteine (46). The human hepatic enzyme was sensitive to oxygen during tissue homogenization (47), whereas sulfhydryl reagents blocked the dipeptidase activity of the chicken intestinal group of conjugases (31). The bovine hepatic enzyme was similar to the soybean enzyme in that the heavy metals were inhibitory while the metals of the alkali and alkali earth group exerted little effect (48).
Some indication that the enzyme was a glycoprotein in keeping with its lysosomal localization (49)  suggest that if more than 1 saccharide residue was available for binding then it was bonded to the other available site on the concanavalin A molecule (50) and that steric factors related to the size of the interacting proteins prevented additional binding.
The specificity of conjugase for y-glutamyl bonds has been investigated for several of the enzymes from different species. The enzyme from Flavobacterium polyglutamicum was able to hydrolyze only y-glutamyl bonds (33), whereas the chicken pancreas enzyme, after initial reports of y bond specificity (51), was subsequently reported as being able to hydrolyze LY bonds at one-half the rate (52). Human liver enzyme failed to hydrolyze cu bonds (22). The same investigators also demonstrated that, while the human enzyme hydrolyzed pteroylglutamyl-glutamyl-leucine, it could not hydrolyze pteroylglutamyl-glutamyl-oc-leucyl-leucine.
This experiment was interpreted as demonstrating that the enzyme was an exopeptidase with terminal y bond specificity (47). This requirement for a terminal y bond, demonstrated also for the bovine enzyme, actually prohibited any conclusions being drawn about possible internal bond cleavage using substrates with terminal cy bonds.
The conjugase group of enzymes has generally been regarded as consisting of carboxypeptidases.
This generalization largely evolved from Pfiffner's observation that methyl esterification of pteroylhepataglutamate rendered it insusceptible to attack by hog kidney conjugase (53). This methyl ester, however, was a polymethylester with methyl groups potentially esterified to each of the seven free cu-carboxyl groups as well as to the one free y-carboxy group of the molecule. This would result in a major alteration in the polyanion nature of the substrate which we consider essential for enzyme action. The evidence for the human liver enzyme being exclusively an exopeptidase has been discussed in the preceding section and this also we do not feel is conclusive. Baugh and Krumdieck (22) also demonstrated the isolation of all of the possible pteroylpolyglutamate intermediate breakdown products of the human conjugasemediated hydrolysis of pteroylheptaglutamate. This data could just as readily support the concept of the enzyme randomly cleaving any of the six available substrate bonds, i.e. acting as a hydrolase, as the theory that there was sequential COOH-terminal cleavage of glutamic acid residues. The end point of conjugase hydrolysis of pteroylpolyglutamates has been determined for several species of the enzyme. Pte-Glu-Glu has been reported to be the pteroyl derivative end product of the chicken pancreas enzyme (52). This same end product has been demonstrated for conjugase from Flavobacterium polyglutamicum (33), although a peptidase has been isolated from another species of flavobacterium which could cleave Pte-Glu-Glu (54). It has not yet been demonstrated whether Pte-Glu-Glu was a substrate for the human hepatic enzyme although that inference could be drawn from the data provided (22). It is clear that pteroyl glutamic acid and glutamic acid are the end products of the action of the bovine hepatic enzyme.
A possible explanation for the preference for the longer chain substrates, also a property of the human enzyme (22), may lie in the relative charge of the substrates. The pK, values for the cr-carboxyl and y-carboxyl groups of glutamic acid are 2.19 and 4.25, respectively (55). At the assay pH of 4.5, the free cu-carboxyl groups would be strongly anionic and the total charge on the longer chain derivative would be greater. As mentioned earlier, there is good evidence that these relatively short chain polyglutamates exist as extended rodlike polyanions at these pH levels (34). The evidence for charge interaction being a factor in substrate binding and catalysis was supported by the studies with polyanion inhibitors of the enzyme.
The glycosaminoglycans basically consist of repeating units of a uranic acid and hexosamine. Hyaluronic acid contains no sulfate residues, the chondroitin sulfates contain an average of 1 sulfate residue per repeating unit and heparin has 2 sulfate residues per repeating unit (19). The minimal effect of hyaluronic acid on enzyme activity indicated that the SO,'-ions were the prime inhibitory determinants, and that other structural differences within this group of compounds were not significant.
The inhibition by hyaluronic acid would indicate the extent of any effect due to free carboxyl groups. It would also appear that the position of the S0,2m ions on the carbohydrate backbone was not critical. The potency of the sulfated glycosaminoglycans as enzyme inhibitors, producing virtually total inhibition of activity at concentrations of SO,'ions 2 orders of magnitude lower than the concentration of the substrate, indicated a high degree of specificity in their interreaction with the enzyme. Furthermore, it implied an irreversibility of the binding. This was emphasized by a consideration of the relative concentrations of enzyme and heparin in the assays. Assuming an average molecular weight of 12,000 for heparin, 26.7 ng ml was equivalent to a concentration of 2.25 nrd. The enzyme concentration in the assays was approximately 5.0 nM. These figures would indicate a virtual 1: 1 stoichiometry of binding. Extensive studies on the conformation of the sulfated polysaccharides have shown that they consist of unbranched extended polysaccharide chains fringed with charged side groups (56, 57). This is essentially the configuration assigned to short chain polyglutamates by Kovacs et al. (34) with the free oc-carboxyl groups extending out from the linear y peptide chain. The maintenance of the extended rodlike structure is dependent upon the environmental ionic strength, a more random coil configuration developing as the effects of charge repulsion between adjoining charged groups (either S0,2-ions or carboxyl groups) is minimized.
This alteration in substrate structure could well contribute to the decrease in enzyme activity observed with increasing ion strength of simple anions such as SOa2m and acetate.
Studies with the macromolecular polyanion, dextran sulfate 2000 (average MW 4.3 x 106) supported the concept that substrate configuration was important for enzyme action. This polymer contains an average of 2.3 SOd2-ions per glucosyl residue, a much higher sulfate content than the glycosaminoglycans. The higher concentration of anions required for an equivalent degree of inhibition by dextran sulfate when compared with the sulfated glycosaminoglycans could be related to a number of factors. One of these is the branched chain structure of the dextran molecule leading to a charge distribution which would interfere with the assumption of any extended rod configuration.
On the other hand, the high charge density and the size of the enzyme itself would produce steric problems limiting the potential number of SO,'-ions actually capable of enzyme binding. This latter point becomes particularly relevant when one considers that the actual polymer concentration producing 50% inhibition was approximately 10 PM.
As the enzyme concentration was 5 nM, the conclusion was that each dextran sulfate polymer molecule was capable of binding a minimum of 250 enzyme molecules. Blue dextran 2000 has been found to bind via its chromophore to other enzymes, including phosphofructokinase (21) and pyruvate kinase (58). Investigation of the reactions between phosphofructokinase and both Cibacron blue 3GA and Cibacron brilliant blue BRP have indicated that these were based on the structural homology between the former dye and the enzyme substrate ATP (21). These studies also confirmed that the coupling of the dye to the dextran facilitates enzyme binding.
The physical chemistry of the enhanced binding properties of polyanions compared with the free or "monomeric" anions has been discussed by Scott (59).
The failure of the enzyme to hydrolyze either of the bonds of Pte-y-Glu-a-Glu-Glu suggested an absolute requirement for a terminal y bond. It is proposed that the substitution of this terminal cy bond resulted in a perturbation of the linear extended rod structure of the substrate which could no longer align correctly on the active site. In addition, the free LY carboxyl group adjacent to the internal y bond of Pte-Glu-Glu-Glu was now involved in the formation of the terminal cy bond and this group would appear to be critical for y bond cleavage. The requirement for a terminal y bond has not been examined with longer chain polyglutamates, however, and it is possible that internal y bonds remote from a terminal LY bond may be