Immobilized derivatives of leucine aminopeptidase and aminopeptidase M. Applications in protein chemistry.

Abstract Leucine aminopeptidase and aminopeptidase M have been covalently bound to an arylamine derivative of porous glass. The bound forms of both enzymes retain 100% of their activities at saturating levels of substrate (leucine p-nitroanilide). For the hydrolysis at pH 7.3 and 25° κcat values for bound and free leucine aminopeptidase are 46 ± 5 s-1 and 46 ± 2 s-1, respectively; for aminopeptidase M (pH 7.5 and 25°) κcat values are 23 ± 2 s-1 and 21 ± 0.4 s-1, respectively. Although the Michaelis constants for both enzymes increase on binding, the pH and temperature dependencies of the bound enzymes remain unchanged. These data suggest that the environments and conformations of the enzymes are not significantly changed after coupling to the solid support. The apparent decrease in the binding of substrate could be explained by a decrease in the effective diffusion coefficient of the substrate. Both insoluble enzymes are active against polypeptide substrates. After treatment for removal of contaminating endopeptidases, the immobilized derivatives of leucine aminopeptidase and aminopeptidase M were used successfully in NH2-terminal sequence determination. The bound aminopeptidase M appears to be the better of the two for this purpose. Both bound enzymes will catalyze the hydrolysis of the aminoethylated A and B chains of insulin nearly to completion (≥87% recovery of free amino acids in all cases). These digests are carried out at pH values near neutrality in a volatile buffer with no activating metal. Immobilized pronase (Royer, G. P., and Green, G. M. (1971) Biochem. Biophys. Res. Commun. 44, 426) was used in concert with bound leucine aminopeptidase and bound aminopeptidase M for the hydrolysis of β-lactoglobulin. In both cases the recovery of free amino acids was 93%. These bound enzymes should be quite useful in amino acid composition determinations when acid-labile residues such as tryptophan, glutamine, asparagine, or certain "affinity labeled" side chains are present.

was used in concert with bound leucine aminopeptidase and bound arninopeptidase M for the hydrolysis of /3-lactoglobulin.
In both cases the recovery of free amino acids was 93 %. These bound enzymes should be quite useful in amino acid composition determinations when acid-labile residues such as tryptophan, glutamine, asparagine, or certain "affinity labeled" side chains are present.
It is also apparent that immobilized enzymes may serve as research tools. One area for such use would be proteiu structure studies.
Specifically, we hope to prepare and employ a series of immobilized proteolytic enzymes for total hydrolysis, sequence det.ermination, and proof of optical purity. Enzyme immobilization would permit the use of mixed proteases in high concentration for total hydrolysis. In all of the applications, the convenient separation of enzyme from digest is an advantage.
Also, the insoluble enzymes are stable, reusable, and may be used in high concentration without the risk of contaminating the digest.
hminopeptidases are estcnsivcly used for amino acid sequence determination (7)(8)(9) and have also been used in attempts to hydrolyze proteins and pcptidcs to free amino acids (10-12). Alt.hough the importance of NHz terminus sequencing by aminopeptidases has been diminished by the development of automatic sequencing systems, immobilized aminopeptidase derivatives could be quite useful in placement of amides, tryptophan determination, proof of optical purity, identification of acid-labile catalytic intermediates, and identification of affinity labeled side chains.

Materials
Lcucine aminopcptidasc was type IV from Sigma Chemical Company.
Aminolxptidase XI was the product of the HcnleJ Company.
The nrylnmino derivative of porous glass (13) was obtained from the Corning Glass Works.
Aminoel;hylation was carried out as described by Cole (14) ; the chains were separated by chromatography on  according to Humble et al. (15). Pronase bonded to glass was prepared as previously described (16). All other chemi cals were reagent grade.

Mefhods
Enzyme Coupling-Three hundred milligrams of the arylamino glass were suspended in 50 ml of 1 N HCl at 0" with mechanical stirring.
Drops of sodium nitrite (0.5 N) were added until an excess was indicated with starch-iodide paper. After 15 min, the glass was collected on a filter and washed with 200 ml of 3% sulfamic acid and 400 ml of distilled water.
The coupling was carried out by adding the glass to a test tube containing a solut'ion of aminopeptidase (3 mg of enzyme in 2 ml of 0.1 M Tris (pf1 7.3) which was 1 mM in MnC&) . The tube, connected to a constant torque stirrer, was rotated in an ice-water bath.
The stirrer was stopped periodically, and aliquots of supernatant were removed for assay. When reaction was complete the glass-enzyme derivative was filtered and washed with 200 ml of buffer. The preparation is stored at 4" in a moist cake. Determination 01 Amount of Enzyme Bound-The depletion of enzyme activity in the supernatant is used to calculate the amount of enzyme bound to the glass. Consideration should be made of any activity in the washings which represents loosely bound enzyme. For this particular system all of the activity was accounted for either in the form of tightly bound enzyme or free enzyme.
The amount of protein bound was also determined by amino acid analysis (17). A control and enzyme-glass derivative were hydrolyzed with 6 N HCl in an evacuated, sealed tube at 110" for 30 hours.

Enzyme
Assays-The hydrolysis of Leu-p-NOz-anilide was followed by monitoring the appearance of p-nitroaniline at 405 nm with a Cary rnodel 15 spectrophotometer fitted with a thcrmostat,ed cell holder.
The extinction coefficient employed was 9.9 X lo3 M-l cm-l.
The temperature was controlled to ~tO.2". The insoluble enzyme was assayed by filtering the reaction mixture at given time intervals and measuring the optical density of the filtrate.
The samples were stirred at uniform rates in a water bath with a "Tri-R" Ms.7 immersible stirrer. The experimental errors in the rate determinations were within =!=5%. The kinetic constants, kcat and Km(spp), were calculated by the method of Wilkinson (18 In the latter case, 280,000 represents an average of reported values. Peptide Hydrolysis-Digestions were performed with 1 rnnf solutions of peptide in 0.2 N N-ethylmorpholine acetate (pH 7.3 or 7.5). A screw cap tube (15 ml) with baffled sides was used as a reaction vessel. The tube containing 5 ml of protein solution a.nd 150 mg of insoluble enzvme was rotated in a constant temperature bath (35") by a constant torque motor. Samples (0.2 ml) withdrawn at various time intervals were lyophilized, dissolved in cit.ratc buffer (pH 2.2), and subjected directly to amino acid analysis (16). No interference from peptides was evident.
The enzymes employed in sequencing experiments wcrc treated with DFP and iodoacetnte by standard procedures (8) and stored under Tris buffer containing 1 mM MnCL For the hydrolysis of ,&lactoglobulin, 5 ml of a solution which was 1 rnnf in protein and 0.2 N in N-ethylmorpholine-acetnt'e buffer (pII 7.5) were placed in a screw cap tube with baffled sides. Immobilized Pronase (75 mg) was added, and the tube was rotated by a constant torque motor for 6 hours at room temperature.
The bound Pronasc was removed by filtration, and bound aminopeptidase (150 mg) was added. The tube was rotated for 24 hours, and the contents wcrc filtered, lyophilizcd, and subjected to amino acid analysis.

Preparation
and Characterization OJ Immobilized Aminopeptidases-A rcpresentativc time course of coupling appears in Fig. 1. The amount of enzyme coupled was determined both by depletion of activity in the supernatant and by amino acid analysis.
Good agreement was found for the two met'hods. The average values wcrc 1.2% (w/w) and 0.8 $$ (w/w) for leucine aminopeptidase and nminopeptidase M, respectively. We have used these enzyme preparations over many months wit'hout apparent loss of activity.
Roth soluble and insoluble forms of the enzymes follow Uchaelis-Mcnten kinetics ( Fig. 2 and 3). Values of &capp) and JZcst appear in Table I. These data point up the importance of establishing the substrate dependence of the enzyme-catalyzed ma&ion for purposes of comparing the activities of soluble and insoluble enzyme forms. For instance, in the case of leucinc aminopeptidasc the 11Zicll:Lclis;-lvlenten equation describes both reactions, but oi,,/Uao1 varies as a function of substrate concentration until saturation is reached.
This behavior reflects the differences in R,, for the two enzyme forms; i.e. when Vm,, is unchanged, Equation 1 holds.
(1) A plot of Oins/usol against So is iu fact hyperbolic and approaches unity as expected.
A change in pH dependence often accompanies insolubilization of enzymes (24). It is therefore important also to compare the activities of soluble and insoluble forms of an enzyme as a func- Activity was measured spectrophotometrically with leucine p-nitroanilide as substrate. In each case, the pI1 profile of the bound enzyme is virtually identical to that of the free enzyme.
The same observation was made previously in the study of a l'ronase-glass complex (16j. Arrhenius plots (In v versus l/T, So >> K,) arc linear and indicate that the bound and free forms of lcucinc aminopeptidase and aminopeptidase M exhibit similar temperature depcndences. For both enzymes the experimental activation energies of bound and free forms calculated from the slopes of these plots are identical within experimental error (Table I). It is interesting to note that for the hydrolysis of Leu-p-NOI-anilidc the experimental activation energy for the leucinc aminopeptidase-cntalyzed reaction is considerably lower than that of the reaction catalyzed by aminopeptidase 11. Also, values of kcst/Kim (Table  I) indicate that Leu-p-NOz-anilide is a considerably better When enzymes are stored in moist cake in the presence of 1 mM MnC&, no activating metals need be added to the digests. Note that the amino acids occur in the predicted order and that the separation of isoleucine and valine is less well defined in the digest with immobilized leucine aminopeptidase . The percentage of hydrolysis is judged by the recovery of amino acids listed.
Serine was not determined since asparagine and glutamine fall under this peak. dminoethylcysteine appeared on the short column as a lone peak between the expected points of appearance of lysine and histidinc; the color value for aminoethylcysteine was found to be 5% below that of Iysine. The appearance of aspartic acid is the result of a desamido contaminant in this particular preparation of insulin (Asn-21 -+ &p-21). The percentage of contamination found by our enzymatic method is in agreement with the manufacturers estimate based on electroyhoretic studies. The COOH terminus of the B chain of insulin has the sequence Thr-Pro-Lys-,41a-COOI-I.
Since the aminopcptidases used here release proline slowly or not at all, the B chain of insulin seemed a likely choice of substrates to determine the difficulty posed b3 the presence of prolinc; the results appear in Table III Notice that the residues on the carboxyl side of proline, lysine, and alaninc, are released to a greater cxt,cnt than proline.
In the case of aminopeptidase M hydrolysis, one would expect the complete release of threonine; this is not observed. In the case of the digestion by immobilized leucine aminopeptidase, one would expect the release of threonine to be at the same level as proline recovery. This is not. the case. Notice that alanine, the COOH-terminal residue of the insulin B chain, is released completely by bound leucine aminopeptidase but not by bound aminopeptidasc M. For the hydrolysis of P-lactoglobulin, we chose to pretreat with immobilized Pronase (16). Pronase is a mixture of proteolytic enzymes secreted by Streptomyces grisezrs; the mixture contains enzymes which encompass a very broad range of specificity. After several hours of incubation of &lxctoglobulin with bound Pronase, we see approximately 50 spots on a peptide map. The number of spots then decreases slowly as more peptides arc reduced to free amino acids. The degree of hydrolysis is about 65% after 6 hours. Subsequent treat.ment with bound leucine aminopeptidase or aminopeptidase IL1 increases the extent of hydrolysis to 937, in both cases; recoveries of proline are 25 to II'%, respectively (Table IV).

DISCUSSION
There are at least two types of explanations which could account for complete retention of activity when an enzyme is immobilized.
The side chains most accessible for coupling to the insoluble carrier could be sufficiently distant from the active center to allow covalent attachment without reaction of the groups at the active center. One might easily envisage available tyrosine and lysine residues on one side of the protein molecule and the active center on the opposite side. Alt.ernatively, reaction of the carrier with the enzyme could occur relatively close to the active center, if the catalytic groups were recessed in a crevice or hole. One related picture would be a toroid or donut-shaped molecule with the active site(s) on the inner wall of a hollow center. Again, reaction could occur at the active site only if a sufficiently small chemical modification reagent were used. Although no mention was made of the active site on the basis of low angle x-ray diffraction, Kretschmer and Kollin (26) suggested such a toroid structure for bovine lens aminopeptidase.
A second possibility would involve the inactivation of 1 molecule or subunit with commensurate activation of others. This situation would require the fortuitous coincidence of V,,, values of soluble and insoluble forms. Similarit'ies in pH dependences and activation energies would also tend t,o rule out this explanation. Morcorer,Schwabe (27) found that a noncovalent complex of calcium phosphate gel and leucine aminopeptidase from dent'al pulp retained full activity.
Although no comparisons of kinetic parameters for the soluble and insoluble forms of this enzyme were made, Schwabe suggested that t,he interaction of enzyme with support results in a specific orient,ation of the active site away from the support to face the solvent.
Although our system involves a different support and an enzyme from another source, we feel a similar orientation prior to formation of a covalent complex could also explain the kinetic results presented here.
The differences in the Km(app) values of the soluble and insoluble enzyme forms could be explained by diffusion limitations on the substrate.
The approach of substrate to the enzyme might be impeded by the support itself or an "unstirred lager" on the support part,icle (28, 29).
The use of insoluble enzyme supports as chemical modification reagent's seems attractive.
It is our hope to look also at soluble chemical modification reagents of low molecular weight to further probe the three dimensional structure of leucine aminopeptidase and aminopeptidase M. Results of both approaches taken t'ogether could yield valuable information on the topography of these and other proteins.
The data presented here indicate that bound leucine aminopeptidase and bound aminopeptidase M could be quite useful in peptide sequencing and tot'al hydrolysis.
Figs. 6 and 7 illustrate the use of bound aminopeptidases in sequencing. Isoleutine and valine are somewhat better resolved when bound aminopeptidase M is employed.
These immobilized enzymes would be especially valuable when examination of the residual peptide was desired. In this case, the reaction, termination, and enzyme removal is accomplished simply by filtration.
Also, the feature of enzyme reuse is valuable in that lcucine aminopeptidase is relatively expensive.
We are hopeful that a multiple aminopeptidase sequencing system can be devised in which other bound aminopeptidases, such as aminopeptidase P and proline iminopeptidase, for example, could be used in concert with bound leucine aminopeptidase.
When release of amino acids is slow because of X-Pro or Pro-X linkages, the digest could be exposed to aminopept,idase P or prolinc iminopeptidase. Enzyme insolubility would considerably reduce the operational difficulties inherent in such a system. IO is evident that bound aminopeptidases could be quite useful in the total hydrolysis of peptides and proteins.
When proline, o+N-acyl residues, and D isomers are absent, the digestion should be complete.
All of the advantages of enzyme immobilization given above would apply in total hydrolysis studies as well. In addition, mixed immobilized enzymes could be used without danger of contamination or enzyme inactivation. In addition, for total hydrolysis impurities of the enzyme preparation to be immobilized could be an advantage. It is clear that our lcucine aminopeptidase has a carboxypeptidase contaminant, since lysine and alanine are released to the extent of 80 and 95oj,, respectively, from the aminoethylated B chain of insulin. These residues are on the carboxyl side of proline and should, therefore, be recovered to the same extent as proline (60%). Delange,et al. (30) have demonstrated the presence of a carboxypeptidase contaminant in aminopeptidase M. Our results (Table III) indicate closer agreement among recoveries of proline, lysine, and alanine: 50, 60, and 65%, respectively.
The soluble preparation could be of higher purity or perhaps a carboxypeptidase contaminant could have been inactivated during coupling or simply coupled to a limited degree.
Our present direction is, of course, to solve the problem presented by the resistance of imide and amide bonds involving proline.
Much the same approach applicable in sequencing applies to total hydrolysis, except that bound derivatives of proline-releasing enzymes could be used simultaneously.